Antecedents and Consequences of Engagement with Television Content in a Social Media Context

MISSING IMAGE

Material Information

Title:
Antecedents and Consequences of Engagement with Television Content in a Social Media Context A Study of Primetime Network Programming
Physical Description:
1 online resource (263 p.)
Language:
english
Creator:
Guo, Miao
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Mass Communication, Journalism and Communications
Committee Chair:
Chan-Olmsted, Sylvia M
Committee Members:
Elias, Troy Rawle Clive
Ostroff, David H
Weitz, Barton A

Subjects

Subjects / Keywords:
engagement -- social
Journalism and Communications -- Dissertations, Academic -- UF
Genre:
Mass Communication thesis, Ph.D.
Electronic Thesis or Dissertation
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )

Notes

Abstract:
Today television audiences are experiencing greater control over how they consume television in a multiple media environment. In particular, how they interact with television content through multiple social media platforms has emerged as a noteworthy phenomenon. This study investigated the social viewing experience of audiences by introducing the social engagement construct and validating its measurement scale. Two online consumer panels of 655 social media users were sampled to complete the three-stage research plan. Through conceptualization and operationalization of social engagement, this study identified four underlying dimensions in social engagement as vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influence. The four dimensions represent a continuum in which audience social engagement behaviors range from the lower levels (i.e., vertical involvement and diagonal interaction) to the higher levels (i.e., horizontal influence and horizontal influence), suggesting that the social viewing experience starts with the relationship between audiences and the branded television content and continues to extend that relationship to other audience members. The findings of this study illustrated that the social engagement process is a composite result, which is determined by multiple components jointly under the integrated framework of active audience behavior. Specifically, audience attributes were found to be most salient in predicting the four social engagement dimensions. The audiences who demonstrate higher innovative tendencies and more social activities in their real lives are more likely to engage in different levels of social engagement behaviors surrounding television programming. Further, audiences' affinity towards programming, genre preference, and program involvement were found to significantly predict the overall social engagement experience. Finally, the higher levels of social engagement dimensions exhibited more salient predictive effects on program loyalty, audience satisfaction, and product purchase likelihood compared to the lower levels of social engagement behaviors. It appears that the broadcast culture of the late twentieth century is quickly evolving into a multi-media community of communication on many levels. The digital multi-media culture of the 21st century seems to be growing at near-light speed. Therefore, the theoretical and practical implications of this study have immediate applications as well as challenges for future research.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
General Note:
Description based on online resource; title from PDF title page.
General Note:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2012.
General Note:
Adviser: Chan-Olmsted, Sylvia M.
General Note:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2014-05-31
Statement of Responsibility:
by Miao Guo.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0043607:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 ANTECEDENTS AND CONS EQUENCES OF ENGAGEME NT WITH TELEVISION CONTENT IN A SOCIAL MEDIA CONTEXT: A STU DY OF PRIMETIME NETW ORK PROGRAMMING By MIAO GUO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORI DA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 201 2

PAGE 2

2 201 2 Miao Guo

PAGE 3

3 To my parents, who made all of this possible, for their endless love, support and encouragement

PAGE 4

4 ACKNOWLEDGMENTS It is my great pleasure to thank everyone who helped me write my dissertation successfully. I am sincerely and heartily grateful to my advisor and chair of this dissertation, Dr. Sylvia Chan Olmsted, for her guidance and support throughout this process. As an excellent mentor for me to learn from in life, she continually and convincingly conveyed a spirit of adventure in regard to research and scholarship, and an excitement in regard to teaching. Without her encouragement, supervision, and grant funding sup port, this dissertation would not have been possible. I could not ask for a better chair and mentor! I am fortunate to have an excellent committee, and would like to express my special thanks to my committee members, Dr. David Ostroff, Dr. Barton Weitz, an d Dr. Troy Elias. Dr. Ostroff challenged me to think deeply about television theories and helped me gain a broad perspective of mass communication. Dr. Weitz expanded my knowledge to the marketing field and enriched my ideas of consumer behavior and audien ce activities. Dr. Elias introduced me to several theories from the field of advertising and advanced my knowledge of research methods. The encouragement from all of them helped me gain confidence as an academic researcher. This dissertation would not have been possible without assistance from industry. My research for this dissertation was made more efficient but also much more extensive through the use of online survey software and consumer panels. Thus, I would like to thank Jeff von Liebermann and Ian R iner from the uSamp TM company for their quick feedback and in kind support. I am also grateful for the technical assistance from the staff working at Qualtrics Labs, Inc (Provo, UT).

PAGE 5

5 My graduate studies have been benefited intellectually, emotionally, an d spiritually from the network of many supportive professors, staff s and friends at the University of Florida. I am especially grateful to Dr. James Algina from the College of Education for his unwavering patience and coaching on the data analysis techniq ues through the complex structural equation modeling I was also greatly inspired pedagogically through the lectures by Dr. Julie Dodd, as well as Professor Gary Corbitt, with whom I cooperated to teach an exciting course of Telecommunication Research I s end a special thank you to Kim Holloway and Jody Hedge at the Graduate Division for their kindness and support. I need to express my special gratitude and deep appreciation to Pastor Danny Austin and his wife, whose friendship, knowledge, and piano tutori ng have encouraged, enlightened, and entertained me throu ghout the dissertation journey. My special thanks also go to Jing Zhang, who has stood by me no matter where I was and what I was going through. I also wish to acknowledge Huaxia Zhang, Jun Shan, Hon gwei Yu, Zhengpeng Wang, Byul Hur, Li Fan, Liping Chen, Chunsik Lee, Soo Yeon Kim, and Sangwon Lee for their consistent support and encouragement. Finally, I cannot thank my parents and family enough for giving me endless love, support, and encouragement. financial support made this happen. I am deeply grateful for having such a supportive older brother, who has been one of my biggest supporters throughout my whole life. My uncle was such a tremendous b lessing by offering suggestions and advice to help me go through each step during the journey. Dear family, I could not have done without your love and care in difficult and happy times. We finally did it!

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 10 LIST OF FIGURES ................................ ................................ ................................ ........ 12 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 15 Emerging Multimedia Consumption Pattern ................................ ............................ 15 Purpose and Overview of Study ................................ ................................ ............. 18 Theoretical and Practical Contributions ................................ ................................ .. 21 2 THEORETICAL FRAMEWORK ................................ ................................ .............. 23 A Comprehensive Understanding of Audience Behavior ................................ ........ 23 The Audience Behaviorist Research Tradition ................................ ................. 23 The Uses and Gratifications Approach ................................ ............................. 25 Theory of Television Program Choice ................................ .............................. 28 Technology Acceptan ce Model and Innovation Diffusion Theory ..................... 30 Media Social Presence Theory ................................ ................................ ......... 32 An Integrated Model of Television Audience Behavior in a Social Media Context .. 34 3 LITERATURE REVIEW AND RESEARCH QUESTIONS ................................ ....... 38 Social Engagement ................................ ................................ ................................ 38 Viewer Engagement ................................ ................................ ......................... 39 Viewer Engagement, Attitude, and Involvement ................................ ............... 42 Engagement with Different M edia Platforms ................................ .................... 43 Perceptions of Television Program ................................ ................................ ......... 48 Program Genre Preference ................................ ................................ .............. 48 Program Affinity ................................ ................................ ................................ 51 Program Involvement ................................ ................................ ....................... 52 Perceived Characteristics of Social Media ................................ .............................. 54 Perceived Ease of Use ................................ ................................ ..................... 54 Compatibility ................................ ................................ ................................ ..... 55 Social Presence ................................ ................................ ............................... 56 Audience Attributes ................................ ................................ ................................ 58 Motives ................................ ................................ ................................ ............. 58 Innovativeness ................................ ................................ ................................ 60 Social Characteristics ................................ ................................ ....................... 61

PAGE 7

7 Consequences of Social Engagement ................................ ................................ .... 63 Program Loyalty ................................ ................................ ............................... 63 Audience Satisfaction ................................ ................................ ....................... 66 Product Purchase Likelihood ................................ ................................ ............ 67 4 METHODS ................................ ................................ ................................ .............. 71 Study Design Overview ................................ ................................ ........................... 71 Online Survey ................................ ................................ ................................ ......... 73 Stage One: Social Engagement Scale Development ................................ .............. 76 Item Generation ................................ ................................ ................................ 76 Pilot Test ................................ ................................ ................................ .......... 77 Descriptive Stat istics of Pilot Test ................................ ................................ .... 79 Exploratory Factor Analysis ................................ ................................ .............. 80 Scale Description ................................ ................................ ............................. 86 Scale Reliability ................................ ................................ ................................ 91 Stage Two: Social Engagement Scale Confirmation ................................ .............. 92 Confirmatory Factor Analysis ................................ ................................ ........... 92 Discriminant Validity ................................ ................................ ......................... 98 Stage Three: Antecedents and Consequences Tests ................................ ............. 99 Main Test ................................ ................................ ................................ .......... 99 Television Program Sample ................................ ................................ ........... 101 Measures ................................ ................................ ................................ ........ 103 Program genre preference ................................ ................................ ....... 103 Program affinity ................................ ................................ ........................ 104 Program involvement ................................ ................................ ............... 104 Perceived ease of use ................................ ................................ ............. 105 Compatibility ................................ ................................ ............................ 105 Social presence ................................ ................................ ....................... 105 Innovativeness ................................ ................................ ......................... 106 Motives ................................ ................................ ................................ .... 106 Social characteristics ................................ ................................ ............... 107 Prog ram loyalty ................................ ................................ ........................ 107 Audience satisfaction ................................ ................................ ............... 108 Product purchase likelihood ................................ ................................ ..... 108 Media use and demographic information ................................ ................. 109 Descriptive Statistics of Main Test ................................ ................................ .. 113 Participants ................................ ................................ ................................ ..... 115 Data Analysis Strategy ................................ ................................ ................... 118 Social engagement with different television genres ................................ 118 M otivations behind social engagement behavior ................................ ..... 120 Antecedents and consequences tests ................................ ..................... 120 5 RESULTS ................................ ................................ ................................ ............. 123 Social Engagement Scale Explanation ................................ ................................ 123 Social Engagement with Different Television Program Genres ............................ 128

PAGE 8

8 Motivations behind Social Engagement ................................ ................................ 131 Antecedents and Consequences of Four Social Engagement Dimensions .......... 137 Meas urement Model Assessment ................................ ................................ .. 138 Structural Model Testing ................................ ................................ ................. 145 Antecedents to four social engagement dimensions ................................ 149 Consequences of four social engagement dimensions ............................ 155 Antecedents and Consequences of the Overall Social Engagement .................... 161 Measurement Model Assessment ................................ ................................ .. 161 Structural Model Testing ................................ ................................ ................. 162 Antecedents to the overall social engagement ................................ ......... 163 Consequences of the overall social engagement ................................ ..... 166 6 DISCUSSION AND CONCLUSION ................................ ................................ ...... 174 Summary of Findings: Social Engagement ................................ ........................... 174 Summary of Findings: Social Engagement with Different Program Genres .......... 177 Summary of Findings: Antecedents to Social Engagement ................................ .. 178 Summary of Findings: Consequences of Social Engagement .............................. 180 Theoretical Implications ................................ ................................ ........................ 181 Benefits of the Integrated Framework for Active Audience Behavior .............. 181 Contribution s of Social Engageme nt and I ts Measurement Scale .................. 183 Different Tendencies in Social Engagement with Television Program Genres ................................ ................................ ................................ ........ 187 Audience Innovativeness as a Critical Determinant of Social Engagement ... 190 Importance of Audience Social Characteristics as Predictors of Social Engagement ................................ ................................ ................................ 191 Impacts of Program Affinity, Involvement, and Genre Preference on Social Engagement ................................ ................................ ................................ 195 Different Influences of the Perceived Social Media Characteristics as Predictors of Social Engagement ................................ ................................ 197 Instrumental Motivations behind Social Engagement ................................ ..... 202 Salient Effects of Social Engagement o n Program Loyalty, Audience Satisfaction, and Product Purchase Likelihood ................................ ........... 205 Practical Implications ................................ ................................ ............................ 207 Implications for the Au dience Research Industry ................................ ........... 207 Implications for the Television Industry ................................ .......................... 209 Implications for the Social Media Industry ................................ ...................... 214 Limitations ................................ ................................ ................................ ............. 215 Future Research ................................ ................................ ................................ ... 219 APPENDIX A LITERATURES ON ENGAGEMEN T AND MEASUREMENT SCALES ................ 225 B ASSENT SCRIPT ................................ ................................ ................................ 232 C QUESTIONNAIRE ................................ ................................ ................................ 233

PAGE 9

9 LIST OF REFERENCES ................................ ................................ ............................. 246 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 263

PAGE 10

10 LIST OF TABLES Table page 3 1 Defi nition of the attributes of engagement ................................ .......................... 46 4 1 List of original social engagement scale ................................ ............................. 77 4 2 Models and goodness of fit indices by exploratory factor analysis ..................... 84 4 3 Means, standard deviations, and correlations for exploratory factor analysis ..... 85 4 4 Facto r structure matrix by exploratory factor analysis ................................ ........ 90 4 5 Models and goodness of fit indices by confirmatory factor analysis ................... 94 4 6 Means, standard deviations, and correlations for confirmatory factor analysis ... 95 4 7 Factor structure matrix by confirmatory factor analysis ................................ ...... 96 4 8 The list of primetime network programs in the main test ................................ .. 102 4 9 Constructs and operational definition ................................ ............................... 109 4 1 0 The order of television programs by social media users ................................ ... 114 4 11 The comparison of demographics and media usage in the two tests ............... 117 5 1 Social engagement dimensions, definitions and scale items ............................ 126 5 2 Descriptive statistics and correlations of four social engagement dimensions 128 5 3 Descriptive statistics of different genres by social engagement dimensions .... 131 5 4 Exploratory factor analysis for motives behind social engagement behavio r .... 134 5 5 Descriptive statistics, internal consistency values, and intercorrelations for motive factors ................................ ................................ ................................ ... 136 5 6 Confirmatory fac tor analysis for measurement model with four dimensions ..... 140 5 7 Descriptive statistics and internal consistency values for constructs ................ 145 5 8 Correlations matrix for antecedent variables ................................ .................... 147 5 9 Antecedents to the four social engagement dimensions ................................ .. 154 5 10 Consequences of the four social engagement dimensions .............................. 156

PAGE 11

11 5 11 Antecedents to the overall social engagement ................................ ................. 165 5 12 Consequenc es of the overall social engagement ................................ ............. 167

PAGE 12

12 LIST OF FIGURES Figure page 2 1 Antecedents and consequences of social engagement with television content .. 37 4 1 ................................ ......... 72 4 2 A model of social engagement with television content ................................ ....... 91 4 3 Four factor path model and four factor with one higher order factor path model ................................ ................................ ................................ .................. 97 5 1 Antecedents and consequences of the vertical in volvement dimension ........... 157 5 2 Antecedents and consequences of the diagonal interaction dimension ........... 158 5 3 Antecedents and consequen ces of the horizontal intimacy dimension ............. 159 5 4 Antecedents and consequences of the horizontal influence dimension ............ 160 5 5 Ante cedents and consequences of the overall social engagement .................. 168 5 6 Visual depiction of the salient results of the vertical involvement dimension .... 169 5 7 Visual depiction of the salient results of the diagonal interaction dimension .... 170 5 8 Visual depiction of the salient results of the horizontal intimacy dimension ...... 171 5 9 Visual depiction of the salient results of the horizontal influence dimension ..... 172 5 10 Visual depiction of the salient result s of the overall social engagement ........... 173

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ANTECEDENTS AND CONS EQUENCES OF ENGAGEME NT WITH TELEVISION CONTENT IN A SOCIAL MEDIA CONTEXT: A STU DY OF PRIMETIME NETW ORK PROGRAMMING By Miao Guo May 201 2 Chair: Sylvia M. Chan Olmsted Major: Mass Communication Today televisi on audiences are experi encing greater control over how they consume television in a multiple media environment In particular, how they interact with television content through multiple social media platforms ha s emerged as a noteworthy phenomenon. This study investigated the so cial viewing experience of audiences by introducing the social engagement construct and va lidating its measurement scale. Two online consumer panels of 655 social media users were sampled to complete the three stage research plan. T hrough conceptualization and operationalization of social engagement this study identified four underlying dimensions in social engagement as vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influence. The four dimensions represent a continuum in w hich audience social engagement behaviors range from the lower levels (i.e., vertical involvement and diagonal interaction) to the higher levels (i.e., horizontal influence and horizontal influence), suggesting that the social viewing experience starts wit h the relationship between audiences and the branded television content and continues to extend that relationship to other audience members.

PAGE 14

14 The findings of th is study illustrated that the social engagement process is a composite result, which is determine d by multiple components jointly under the integrated framework of active audience behavior. Specifically, audience attributes were f ound to be most salient in predicting the four social engagement dimensions. T he audiences who demonstrate higher innovativ e tendencies and more social activities in their real lives are more likely to engage in different levels of social engagement behaviors surrounding television programming. Further, a programming, genre preference, and program inv olvement were found to significantly predict the overall social engagement experience Finally, t he higher levels of social engagement dimensions exhibited more salient predictive effects on program loyalty, audience satisfaction, and product purchase like lihood compared to th e lower levels of social engagement behaviors. It appears that the broadcast culture of the late twentieth century is quickly evolving into a multi media community of communication on many levels. The digital multi media culture of the 2 1 st century seems to be growing at near light speed. Therefore, the theoretical and practical implications of this study have immediate applications as well as challenges for future research.

PAGE 15

15 CHAPTER 1 INTRODUCTION Today televisi on audiences are experiencing greater control over how they consume television in the platforms that best suit their needs. Particularly, as online social media like blogs and social networks gradually enter the mainstream and reach a broad demographic spectrum (Stephen & Galak, 2009 ), consumers interacting with television content through an expanding array of social media has emerged as a noteworthy phenomenon. The marriage between traditional television and the emerging social media platforms can be attributed to the gro wing adoption of social media tools by the consumers and their increasing cross platform multitasking media consumption patterns (Nielsen, 2010a, 2010b ; Toy, 2010). The trend of simultaneous television/Internet usage is especially meaningful for broadcast ers, program producers, and advertisers in their justification of investment in content, retaining and acquiring customers, enhancing brand affinity and program loyalty, as well as identifying and marketing the most valuable audiences (Epps, 2009; Harris I nteractive, 2008). Emerging Multimedia Consumption Pattern In recent years, research behavior, especially watching television while surfing the Internet, is becoming the norm ( Nielsen, 2010a, 2010b ) In 2 009, around 59% of Americans used the Internet while watching television once a month, with both media simultaneously for three and a half hours per day (Nielsen, 2010 a ). Meanwhile, driven by the increasing adoption of social media platforms among diverse demographic segments, Nielsen (2010 b ) reported that Americans spend 43% more time on social media than a year ago, making social networking and blogs more popular than personal email among the top online activities.

PAGE 16

16 In addition, industry survey revealed th at nearly 43% of the Americans utilize social media and other websites to opine about television programs and a third of the respondents vent their feelings about shows after the program airs (Harris Interactive, 2011). Particularity, Twitter TM and Faceboo k TM ranked among the top social media sites that people tend to visit when they are media multitasking. The degree of multitasking Academy Awards (Toy, 2010). In this sense, soci al media offer a rare platform with attractive potentials in developing relationship s with audiences and/or marketing television content. In reviewing the emerging multiplatform consumption pattern, industrial practitioners and academic scholars have sugg ested various benefits of cross media engagement fo r advertisers and broadcasters. engaged viewers in the multi media environment are more likely to remember advertisement, internalize the message, and be motivated by it than those who are less engaged (Epps, 2009). In addition, advanced Web based technologies and their online applications, especially various social media platforms, are useful in nurturing customer relationships with brands among those who are most tig htly engaged with the program. Because more involved viewers tend to engage in cross media activities, advertising campaigns utilizing cross media platforms might increase the likelihood of targeting the most involved viewers (Harris Interactive, 2008; Net worked Insights, 2010). Hence, the value of advertising grows as viewers connect television program ming and marketing message in multi media platforms.

PAGE 17

17 Furthermore, viewer engagement with television programs through various social media platforms can offe r never before opportunities for broadcasters to connect with their increasingly fragmented audiences. Major broadcasting and cable networks have adopted real time platforms like Twitter TM to drive tune in and take advantage of the growing trend of simulta neous television/Internet usage by dispensing online information in tandem with the airing of the programs (Leavy, 2010). More recently, several new entertainment focused social platforms, such as GetGlue, which allow fans to connect through mobile and onl ine platforms and share their opinions about certain shows across their social profiles, are gaining popularity. In a sense, the use of social media to enhance audience engagement has tremendous marketing potentials as the socially engaged viewers are more likely to stick with a show, talk about the show, and spread word of through real time interaction can humanize broadcasters, which enables them to listen to, affirm, and amplify the opinio ns of their fans (Leavy, 2010). In addition to deepening relationships with television viewers, social engagement enhance viewer loyalty to the program. Particularity, e stablishing a presence for certain programs on social networks, such as Facebook TM and Myspace TM could build affinity for the program brand by providing a platform for discussion among devoted fans (Leavy, 2010). When viewers share an active connection wi social network, conversation is facilitated around the show, providing incentive for viewers to tune in frequently. For example, major cable networks such as ABC Family, Bravo, and USA Networks are experimenting with audience re wards programs via

PAGE 18

18 social media platforms, hoping that the viral nature of Facebook TM in syndicating and reposting content would be an effective way of driving audience loyalty (McBride, 2010). Purpose and Overview of Study Given the supposed significance of viewer engagement with television content and increasing multitasking media consumption trend, the purpose of this investigation is to build an active audience behavior model that facilitate s the understating of the interaction between television and so cial media consumption by validating the proposed social engagement scale and testing its antecedences and conseque nces in a social media context. More specifically, this investigation explores why certain consumers increasingly choose social media platfor ms in relation to television content, how they utilize the different social media platforms to interact with specific shows, and what the actual effects of such an engagement are. scholars and industry practitioners have struggled to understand what engagement is, how it works, and what its practical outcomes might be. While there is no universa l definition on engagement, it is usually described in terms of mental, emotional, attitudinal, and/or behavioral connection between a viewer, a media vehicle, and a brand message, examining viewer recollection of and reaction to programs, product placemen t, promotions, and commercials in multiple media environments (Askwith, 2007). Prior research has pointed out that there are diverse attributes of viewer engagement under different media environments (Calder, Malthouse & Schaedel, 2009); therefore, audien ce engagement with specific social media platforms could

PAGE 19

19 exhibit similarities and/or differences with viewer engagement with traditional mass media. It is suggested that engagement with the Internet or various social media platforms possesses a certain sim ilar nature with that of traditional media in terms of information utility, inspiration, connectivity, and interaction (Kilger & Romer, 2007; Russell, Norman, & Heckler, 2004a). However, the engagement process is enriched by other special experiences such as participating, socializing, community building, and influencing (Takahashi, 2010, Yanga & Kangb, 2009). Thus, the social engagement with television content in the mul ti media contexts. It has been challenging for industry practitioners and academic scholars to operationalize and develop valid, complete, and systematic scales to measure the engagement construct. Diverse measurement scales mainly facilitated by the survey method have been employed but varied greatly in specific scenarios. For example, Haven (2007) identified four dimensions of audience engagement with media content, each incorporating quantitative and qualitative data sources, such as involvement, in engagement with television content should be measured as a function of viewer attitudes, viewer behavior, and viewer attentiveness. Although previous measurement metrics offer some general terms to study social engagement with television content, there is still a void of comprehensive indicators to systematically examine this important concept. Furthermore, it is logical to identify the factors that might drive television audience s to utilize social media platforms to engage with television content, as well as the

PAGE 20

20 possible consequences of that behavior. Prior studies proposed that audience media exposure should be investigated under an integrated framework, and the combined effects from the diversity of media platforms, media content, and media users may complicate the exposure process (McQuail, 200 6 Webster, Phalen, & Lichty, 2006). The current study subscribes to this framework and investigates the soc ial engagement process from three dimensions: 1) social media characteristics (e.g., social presence, compatibility, and perceived ease of use), 2) television program characteristics (e.g., genre preference, program affinity, and program involvement), and 3) audience characteristics. Because media use is a sociable activity, motivations, inter personal interaction, and social activity, are likely to play a role in the social enga ging process (Haridakis & Hanson, 2009). Broadcasting and cable networks that regard engagement as an important by lining up an expanding a rray of social media platforms. Questions remain, however, if the engagement with television content in a social media context would yield the expected effects, and if so, to what extent. S everal industry surveys have revealed that cross media engagement could increase the value of tele vision program ming and advertising as well as provide additional opportunities for branded content and product placement (Harris Interactive, 2008; Networked Insights, 2010) Nevertheless, it is still unclear what might be the exact effects of social engag ement on the following viewing activities, including product purchase likelihood, audience satisfaction, repeat viewing behavior, and program loyalty

PAGE 21

21 In summary, this investigation employs a multiple research strategy, including three focus groups and two online consumer surveys, to 1) develop and validate the construct of social engagement and its measurement scales, 2) explore the factors that among program characteris tics, social media characteristics, and audience characteristics, 3) examine the impacts of social engagement with television content on viewing activities, including viewer loyalty, audience satisfaction, and product purchase likelihood, and 4) investigat e the possible differences and/or similarities in the aforementioned antecedents and consequences based on the different types or levels of social engagement, if any. Theoretical and Practical Contributions This research sheds light on the importance of a n emerging television consumption behavior engagement with television content in a social media context from both theoretical and empirical perspectives. Starting with the potential for practical contributions, as the television industries increasingly compete against alternative distribution platforms while facing a fragmented audience with decreasing loyalty, it is critical for media organizations to develop a more long term relationship with their viewers. Th is study, therefore, empirically addresses the issue of whether television broadcasters/advertisers should devote resources to develop a social engagement strategy and how they should go about it. This is an important consideration as the careful deployment of resources is most essential in a comp etitive environment. Theoretically, the conceptualization and operationalization of social engagement with television content synthesize several bodies of literature to help validate an emerging active audience behavior theory and establish a more compreh ensive picture

PAGE 22

22 for viewer engagement in a social media context. This exploration is unique in that it simultaneously integrates different segments of literature from mass communication, marketing, and information systems, thereby providing an opportunity f or a comprehensive examination and tighter integration of the components in the media choice and use process. In addition, most research in the audience behaviorist tradition has been limited by the separate examination of engagement in different media pla tform contexts and the lack of valid measurement indicators in an integrated fashion. T his investigation, therefore, enrich es active audience behavior perspectives, by introducing a multi dimensional engagement concept in relation to television audience engrossed experience to catch emerging multitasking media consumption pattern in new media environments.

PAGE 23

23 CHAPTER 2 THEORETICAL FRAMEWOR K A Comprehensive Understanding of Audience Behavior The Audience Behaviorist Research Tradition There are several no table traditions in audience research, such as the structural tradition, the behaviorist tradition, and the cultural tradition and reception analysis. Each of these traditions suggests a different explanation for media use behavior and involves diverse the oretical foundations, research strategy, and methods (McQuail, 2006; Webster, Phalen, & Lichty, 2006). The structural tradition proposes that the media systems and the social systems shape media user behavior, while the social cultural approach emphasizes the particular context in which an audience is located and the process of giving a meaning to cultural products and experiences. The behavioral direct media effects, has tra nsformed from the source dominated to active audience perspectives, with an emphasis on individual needs, motives, and circumstance as the starting point (McQuail, 2006). The active audience perspectives are now achieving premier status in the behavioral audience research domain due to the continuously evolving media environment and audience media consumption patterns. Active audience, a key concept in this field, is viewed as a more or less active and motivated set of media users/consumers, who are in cha rge of their media experiences and choose their preferred media in given situations or preferred content within a given medium to meet their specific consumption needs (Katz, Blumler, & Gurevitch, 1974). Under the basic active oriented assumption in audien ce behavior, the term audience activity has been

PAGE 24

24 conceptualized as a variable construct with varying kinds and degree of activity, including selectivity, utility, intentionality, resistance to influence, and involvement (Biocca, 1988; Blumler, 1979). More specifically, selectivity describes audiences who plan their media use or choice actively to reflect their existing interests and preferences. Intentionality signifies audiences who actively engage in the cognitive processing of information and experience directed by their prior motivations. Utility defines audience as the self interested consumer to satisfy various needs. Resistance to influence emphasizes that audience members are obstinate and actively avoid certain types of media influence. Involvement There are several critics of the behavioral research approach from the structural tradition, who argued that the functionalist a pproach fails to provide much successful prediction or causal explanation of media choice and use, due to the fact that much media usage is actually very circumstantial and weakly motivated (McQuail, 2006). In addition, Webster (1998) suggested that defini behaviorist tradition focuses too much on the micro level questions of how individuals interact with media texts, thus undervaluing the role of habit in audience behavior and minimizing the concept of audience as mass. A ccordingly, an integrated model of audience behavior incorporating both individual audience and structural factors is proposed as a better way to comprehensively understand the continuing process of media choice (McQuail, 2006; Webster & Wakshlag, 1983; We bster, Phalen, & Lichty, 2006).

PAGE 25

25 television content in a new social media environment, the characteristics emphasized vary differently from that of the traditional television v iewing habits perspectives. Therefore, the researcher here holds that an individual behavioral approach on the social engagement process is more appropriate for this investigation. Thus, building upon the following theoretical foundations, a holistic frame work is proposed to The Uses and Gratifications Approach U ses and gratifications is often seen as a conceptual approach that provides functional typologies for various media usage in both traditional and the newer, continually evolving, interactive digital environment (Ruggiero, 2000). It is one of the underlying theoretical foundations of this research. The principal elements in the uses 1) the social and psychological origins of (2) needs, which generate (3) expectations of (4) the mass media or other sources, which lead to (5) differential patterns of media exposure (or engagement in other activities), resulting in (6) need gratification s and (7) other consequences, perhaps mostly 510). A contemporary view of uses and gratifications is grounded in several assumptions that highlight the role of audience initiative and activity. Specif ically, communication behavior is generally seen as goal directed, purposeful, and motivated. Audience members are variably active participants who initiate the selection of media and content from an array of communication alternatives in response to their expectations and desires. Social and psychological factors guide, filter, or mediate behavior, while audience members are conscious of the media related needs, which are

PAGE 26

26 constrained by personal traits, social contexts, and interaction. People are typicall y more influential than media in this process, and individual initiative mediates the patterns and consequences of media use (Palmgreen, 1984; Palmgreen, Wenner, & Rosengren, 1985; Rubin, 200 9 ). Uses and gratifications researchers propose that different l evels of audience activity are associated with various media orientation. Levy and Windahl (1984) categorized active television audience viewing behavior into three types intentionality, selectivity, and involvement. Rubin and Perse (1987a, 1987b) furthe r found that various audience members could exhibit different levels of activity before, during, and after television exposure. Diverse viewing motives such as instrumental and ritualized use are related positively or negatively to audience intentionality, selectivity, and i nvolvement with local news or soap opera s Ritualized media use focuses more on using media habitually to consume time and for diversion, which entails greater exposure to and affinity with the medium, while instrumental media use center s on seeking specific media content for informational reasons. Researchers on uses and gratifications also pointed out that mass media compete with other forms of communication or functional alternatives for a finite amount of time among limited audience s (Kaye & Johnson, 2003; Rubin, 200 9 ). The relationship psychological circumstance, including lifestyle, life position, and personality. Prior studies, in particular, examined the ro le of individual life position attributes, such as personality, mobility, lifestyle, inter personal interaction, social activity, economic security, and need for cognition, in shaping the choice of communication alternatives,

PAGE 27

27 motives to communicate, strateg ies for seeking information and diversion (DiMaggio, Hargittai, Neuman, & Robinson, 2001; Hamilton & Rubin, 1992; Perse & Rubin, 1990). Furthermore, differences in personality, cognition, social affiliation, and motivations are found to affect exposure, cu ltivation, and satisfaction of media use (Haridakis, 2006; Harwood, 1999). As the Internet and online applications such as email and various social media mediated communication offers a va st continuum of communication behaviors for the researchers of uses and gratifications to examine. Ruggiero (2000) pointed out three distinctive characteristics of online communication complementary to traditional mass media communication interactivity, demassification, and asynchroneity. As a core notion of audience activity, interactivity describes the degree to which audience members in the com munication process have control over ; demassification illustrates that the ability of the Internet in providin g selectivity attributes to allow individuals to tailor messages to their needs, and asynchroneity signifies the convenience of communicating among senders and receivers at different time s A meta analysis of new technology use found that the uses and gra tifications approach is helpful in studying a wide range of new media, including the Internet (Papacharissi & Rubin, 2000; Stafford, Stafford, & Schkade, 2004), email (Boneva, Kraut, & Frohlich, 2001; Dimmick, Sikand, & Patterson, 1994), and the online con tent sharing community (e.g., YouTube ) (Haridakis & Hanson, 2009). In subscribing to the uses and gratifications approach, the current study attempts to synthesize various motivations of traditional television, the Internet, and new media technologies to assess

PAGE 28

28 emerging social media platforms to engage with traditional television content. In e social activity, interpersonal interaction, and innovativeness are examined to better understand the media engagement process. Theory of Television Program Choice There are two mainstreams of theories on individual television viewing choice including th e functionalist perspective and media economics and marketing. The first working functionalist theories were originated from the uses and gratifications approach and enriched by media professional practices (Webster, Phalen, & Lichty, 2006). Researchers on the working theories of program choice contended that people have consistent preferences for certain content type in determining their choice of media are their likes. In other words, audiences recognize their dislikes more easily than what they may like. In addition, program preferences are closely identified with demographic attributes of the audience (Webster, Phalen, & Lichty, 2006). However, several researchers argu ed that the decision of whether to watch is largely passive, while the decision of what to watch appears more active. This two step decision process highlights the limitation of using program preference as the only predictor of viewing choice and stresses the important role of habit in explaining au dience behavior (Heeter, 1985). The second stream of theories on program choice is developed by economists and marketing researchers, which represents a more formidable explanation in program choice viewing behav ior. Applying a conventional consumer product choice model into

PAGE 29

29 the television program consumption scenario, Steiner (1952) developed an economics model of television program choice and posited the existence of a thoroughly active audience in which audienc the extent to which hypothetical industry structures maximize viewer satisfaction or predict audience viewing pa tterns (Owen & Wildman, 1992; Rust, Kamakura, & Alpert, 1992). These theories from the economics and marketing perspectives are premised upon two important assumptions of audience program choice. First, it is assumed that there are certain content charact supported television operating within the bounds of available program content p. 431). With these two assumptions in place, it is logical to predict the distribution of the audience across channels. When there are only a few competitors, similar programs are likely to be offered across channels. On the other hand, when the number of c ompetitors increases, program content is likely to become more differentiated, which leads to audience fragmentation (Webster, Phalen, & Lichty, 2006). Webster and Wakshlag (1983) integrated the two disparate theoretical approaches into a single comprehensive model of television program choice based on the fundamental assumption that specific program preference is a cause of program choice. The model captured the likely interaction among programming s tructures, program preference, viewer availability, and viewer needs and awareness. The authors proposed

PAGE 30

30 that the program schedule and viewer availability are predictive of viewing choice. Particularity, the structure of program options, such as channel lo yalty, inheritance effects, and repeat viewing, predict program choice, while program genre preferences seem to predict specific program preferences. Nevertheless, the authors admitted that the advent of new technologies that allow viewers to schedule thei r own programs which will eventually force the revision of the model structure in the new digital communication era. The theories in television program choice model clearly establish the significant role of program type in audience choice and underline the need to incorporate program related factors in the proposed study. Technology Acceptance Model and Innovation Diffusion Theory rather than traditional television as a way to be involved with television content, the innovations as one of the important explanatory and predictive variables for the use behavior. Specifically, two streams of t heories from the technology acceptance model (TAM) and the innovation diffusion approach are employed due to their robust and behavior. The TAM approach has been extensivel technology systems (Davis, 1989; Davis, Bagozzi, & Warshaw, 1989; Hendrickson & Collins, 1996). Davis (1989) proposed that people tend to use a system to the extent that they believe it will help them perform their job better (perceived usefulness ). The author also suggested that the beliefs of persons concerning the efforts required to use a system can directly affect system usage behavior (perceived ease of use). Prior

PAGE 31

31 research further revealed that both perceived ease of use and perceived usefulness have a direct influence on the behavior intention to use a technology system, which directly leads to actual system use. Moreover, perceived usefulness mediates the direct effect between perceived ease of use and behav ioral intention (Venkatesh & Davis, 1996). Thus, perceived ease of use and perceived usefulness are two core predictors of technology acceptance intention and actual use behavior in both organizational and individual settings (Schepers & Wetzls, 2007). In process by which an innovation is communicated through certain channels over time 5). Like perceived characteristics of a te chnology in TAM, Rogers (2003) conceptualized the perceived characteristics of an innovation as relative advantage, compatibility, complexity, trialability, and observability. Prior studies indicated that complexity and relative advantage are salient in pr edicting adoption of communication technologies (Lin, 1998, 2001) and supported the critical role of relative advantage in Internet adoption and uses (Atkin, Jeffries, & Neuendorf, 1998; Busselle, Reagan, Pinkleton, & Jackson, 1999). It should be noted tha t complexity and relative advantage are defined similarly as the two constructs of perceived ease of use and perceived usefulness in TAM (Moore & analysis of innovation adoption, the authors discovered that an innovation is more likely to be adopted when it is study includes perceived ease of use in TAM along with perceived compatibility in the innovation diffusion theory as two major independent variables, but excludes job

PAGE 32

32 performance related construct of perceived usefulness as well as trialability and observability in the hypothetical model due to their inconsistent impacts on innovation adoption. Media Social P resence Theory To better understand social behavior in a mediated environment, researchers have advocated the use of social presence theory due to its robust predictive power (Biocca, Harms, & Burgoon, 2003). Social presence is defined as the degree of sal ience (i.e., medium (Short, Williams, & Christie, 1976, p. 65). Biocca and Harms (2002) further to moment awareness of co presence of a mediated body and the sense (p. 14). In other words, social presence is a sense that others are psychologica lly present and that communication exchanges are warm, personal, sensitive, and active. Social presence is postulated to have three levels: 1) the perceptual level describing the detection and awareness of the co he subjective level depicting the extent of accessibility to the others attentional engagement, emotional state, comprehension, and behavioral interaction, and 3) the inter subjective level illustrating the perceived symmetry of social presence (Biocca & H arms, 2002). Based on the premise that media have different capacities to carry interpersonal communicative cues, different media platforms vary in their social presence (Short, Williams, & Christie, 1976) or media richness (Daft & Lengel, 1984). Prior res earch on telecommunication systems use in organizational and interpersonal communication

PAGE 33

33 settings has provided insights into the nature of computer mediated communication, and addressed how the limited number of nonverbal cues in shaping the model of commu nication in a computer mediated context (Hazemi & Hailes, 2001; Steeples & Jones, 2002). Regarding television consumption, the availability of media technologies, especially in the forms of social media, has enabled this traditional medium to significantly opportunities to interact with others and groups beyond the immediate physical surroundings. Thus, audience television viewing behavior is evolving with the integration of these newer and r icher social platforms. In a sense, there is an increase in the mediated social interaction in the television consumption context. ion systems and interfaces are progressively designed to improve human communication for collaborative work (Weiming, Kremer, Ulieru, & Norrie, 200 3 ), online education (Hazemi & Hailes, 2001; Steeples & Jones, 2001), and social services or ecommerce (Save, Guazzelli & Poucet 2001). Examples of evolving social presence technologies include mobile and wireless telecommunication, high bandwidth teleconferencing interfaces, agent based ecommerce and help interfaces, and 3D social virtual environments (Biocca, Harms, & Burgoon, 2003). However, few researchers have expanded the social presence theory into the emerging online communication systems social media. The examination of social presence of various online social media platforms in the context of televis ion consumption provides a notable approach to assess antecedents of the behavioral interaction between traditional television and emerging social media platforms.

PAGE 34

34 An Integrated Model of Television Audience Behavior in a Social Media Context By drawing upo n the aforementioned theories of the uses and gratifications approach, television program choice model, TAM, innovation diffusion theory, and social presence perspectives, this investigation adopts the active audience behaviorist approach and aims to build a holistic, comprehensive model of active audience television engagement behavior in a social media context. From a managerial point of view, the integrated model attempts to serve as a guide for television broadcasters and advertisers, deciphering patter ns of social engagement with television content and exploring potential benefits and risks involved in the deployment of social media platforms. Theoretically, the goals of the integrated framework are to establish a set of valid social engagement measures that reflect conceptually the interaction between television and social media platforms, identify specific factors that affect this engagement process, and assess possible outcomes of the active television viewing experience. Figure 1 is the schematic rep resentation of those factors in the proposed active television audience social engagement model. The core component of this model is the au dience media exposure behavior, that is, using a social media platform to engage with television content. As suggest ed in the audience behaviorist research tradition, audiences are variably active across several qualitative dimensions and along the temporal dimension before, during, and after media exposure. In contrast to the traditional passive television viewing beha vior, active television viewers utilize various social media platforms to engage with television content, indicating that the media use is purposive and planned; the media content is selected; and the viewing experience is involved. Conceptualizing this ne w, active television consumption behavior and validating accordingly the measurement scales

PAGE 35

35 that capture the interaction between the two platforms presents a new perspective on television viewing in a parasocial television environment. Building upon the a forementioned theories and a pproaches, this investigation i dentifies three categories of explanatory factors to predict the social viewing experience. They are perceptions o f television programs, perceived characteristics of social media, and audience attr ibutes. First, the investigation postulates that individual f specific television programs may influence his/her television viewing experience. Specifically, audience preferences for particular television program genre, program affi nity, and personal involvement can play a role in the selection and engagement of programming. Second, the perceived characteristics of social media, such as perceived ease of use, compatibility, and social presence, may predict dia adoption, acceptance, and usage behavior. In addition, media exposure is assumed to be originated from several social and viewing motives, innovativeness, and interpers onal and social activities compose the third factor of audience characteristics. While these antecedents might be useful in identifying the effective means of improving the social engagement act, industrial practitioners and executives alike are more li kely to value the possible outcomes of the emerging consumption behavior. This investigation proposes several consequences from the marketing and advertising perspectives. The first consideration involves program loyalty defined from both the behavioral di mension and attitudinal dimension. Prior industry research found that establishing a presence for certain programs on social media could build affinity for the

PAGE 36

36 program by providing a platform for discussion among devoted fans and eventually enhance viewer loyalty to the program (Leavy, 2010). The second possible outcome centers on audience satisfaction or gratifications, which is assumed to be a significant outcome in consumer behavior from the marketing and communication perspectives. The last set of possi ble consequence is product purchase likelihood. Advertisers and marketers have posited that the value of advertising grows as viewers connect television program and marketing message across media platforms (Harris Interactive, 2008; Networked Insights, 201 0)

PAGE 37

37 Figure 2 1. Antecedents and consequences of social engagement with television content

PAGE 38

38 CHAPTER 3 LITERATURE REVIEW AN D RESEARCH QUESTIONS cons platform engagement. The chapter then reviews the relevant factors that predict the social engagement process, including perceptions o f television programs (i.e., program genre pre ferences, program involvement, and program affinity), perceived social media characteristics (i.e., perceived ease of use, compatibility, and social presence), and audience attributes (i.e., user motives, innovativeness, and personal social characteristics ). The chapter then ends with a discussion of the possible impacts of social engagement on viewing activities, including program attitudinal and behavioral loyalty, audience satisfaction, and product purchase likelihood. Social Engagement Social engagemen t in this study refers to the degree of interactions and connections that a viewer develops with television content through social media platforms. The co re component of the construct, social engagement, is engagement It was suggested that engagement is p rimarily driven by program content in the television consumption context, and the deepest engagement experience happens at the content level ( Epps, 2009; study and includes the program content itself, characters or celebrities in the show, and related staff such as writers, directors, or producers, etc.. The word, social indicates the very nature of the social media platforms which possess distinct charac teristics from the traditional television medium, injecting a

PAGE 39

39 with television content. To provide a comprehensive background of investigation, the current study examines the social engagement construct from both the perspectives of Viewer Engagement The engagement construct has been defined in many ways by a range of disciplines. In marketing, engagement refers to the d egree of attitude or efforts accorded an advertising message (Zaichkowsky, 1985). Goffman (1974) posited that involvement is similar to engagement and occurs when a person pays attention to and describe the degree of absorption or immersion attached to involvement in the social experience. In turning on a prospect to a brand idea enhanced by the surrounding contex technology that is characterized by challenges, aesthetic and sensory appeal, feedback, novelty, interactivity, perceived control and time, awareness, motivation interest, and af 949). It appears that engagement has varied attention, emotion, or experience with a specific object. In the television consumption context, viewer engagement develops within a variety of audience research traditions, such as selectivity of programs, involvement with the characters, use of television personally and socially, and interpretation of content empiric ally or critically (Takahashi, 2010). Askwith (2007) proposed that a behavior, attitude, and desire in relation to a given media, content, or advertising brand.

PAGE 40

40 Moreover, McCl a cross as many different platforms as possible. From an industry perspective, Nielsen (2006) defines viewer engagement as the focu sed mental and emotional connection between a consumer, a media vehicle, and a brand message, examining viewer recollection of and reaction to programs, product placement, promotions, and commercials in multiple media environments. Russell, Norman, and He ckler (2004a) proposed the connectedness construct akin to engagement to capture the parasocial relationship between television viewers with intensity of the relationshi p(s) that a viewer develops with the characters and contextual 152). In addition, the authors (2004b) emphasized the social nature of television viewing and deciphered the connectedness co nstruct into three dimensions: vertical connections (viewer program described as the commitment that individual viewers feel toward their favorite programs ) ; horizontal connections (viewer viewer focused on the interpersonal relationship that viewers fo rm with others around the show ) ; and vertizontal c onnections (viewer character, defined as the imagined and parasocial interactions that viewers develop with characters in their favorite programs ) Considering the current multiplatform television consumpt ion pattern, Askwith the horizontal aspect (viewer viewer) of connections could be audience community building, which is facilitated and enabled primarily through the creation of onl ine social groups and

PAGE 41

41 activities. The vertical interaction (viewer celebrity) describes the increasing opportunities for individual audiences to interact with television celebrities, which is often facilitated by various social media platforms like Twitter TM and Facebook TM The third aspect of social interaction is diagonal interaction (viewer character), which represents two increasingly popular types of engagement touchpoints diegetic extensions and experiential activities. The author concluded that the present opportunities for diegetic interaction are alternate reality games and blogs. The above explication of the connectedness construct points out the social interaction nature of viewer engagement and its multiple platform applications driven by a ran ge of existing digital extensions. To empirically validate the connectedness construct, Russell et al. (2004a) further subdivided connectedness into six second order factors, including escape, modeling, fashion, imitation, aspiration, and paraphernalia. T he escape dimension defined the cathartic element that connects a viewer to a television program. The modeling dimension measures a social learning process by capturing the degree to which individuals relate their lives to the lives of characters. The fash ion dimension represents patterns. The aspiration dimension identifies how people aspir e to actually be on the show or meet with the characters. The last dimension paraphernalia measures the degree to which people collect items to bring the show into their real world (Russell, Norman, & Heckler, 2004a).

PAGE 42

42 Viewer Engagement, Attitude, and Involvement To conceptually and operationally define the engagement construct, it is necessary to differentiate viewer engagement with the other two similar constructs attitude and involvement. The attitude construct has been extensively used to meas ure the degree of favor or disfavor toward an attitudinal objective (Cacioppo, Gardner, & Berntson, 1997). In regard to television consumption, prior studies used programming liking, a summary evaluation of the experience of viewing a television program, t o measure the attitude toward a program and further examine the program effect on s or subsequent repeat viewing behavior (Barwise & Ehrenberg, 1987; Murry, Lastovicka, & Singh, 1992). However, as Russell et al. ( 2004a) argued, the construct of viewer connectedness or engagement goes beyond an overall evaluative response to the program F urthermore, the attitude construct cannot capture the fact that such a parasocial relationship would emerge, although a positive attitude toward a program may mediate the development of connectedness. While both viewer engagement and audience involvement emphasize the personal relevance of a television program, it is important to differentiate engagement from the construct of invol vement. Rubin and Perse (1987 b ) proposed that viewer involvement consists of three dimensions affective involvement, cognitive involvement, and behavioral involvement. Specifically, involved television viewers may feel empathy toward those in need on th e show (i.e., affective involvement), consider the messages of the show (i.e., cognitive involvement), and talk about the show with others (i.e., behavioral involvement) during and after exposure. Park and McLung (1986) revealed that program involvement is related to personal relevance or importance with response to particular television content measured by cognitive, affective, and functional

PAGE 43

43 dimensions. Nevertheless, prior research concluded that television program connectedness may start by fostering end urin g involvement with the program over the course of repeat viewing, but end up absorbing their audiences in parasocial relationships with the characters in the program, i.e., viewer engagement or connectedness (Russell et al., 2004a). Engagement with Di fferent Media Platforms Considering the growingly interwoven video platform environments, it is essential for this study to examine viewer engagement with different media platforms to better understand the social engagement construct. Kilger and Romer (20 07) proposed multiple media engagement dimensions when investigating the relationship of media engagement wi th product purchase likelihood. Those dimensions includ e inspiration, trustworth iness life enhanc ement social involvement, personal timeout, and a dvertising attention receptivity. The study specifically suggested two extra attributes for the television vehicle: 1) personal connection measuring personal association with ment of the degree a program i s the context. The same study also developed two more dimensions associated with the Internet platform such as interactivity/community and enjoyment/attraction. Whe n exploring the relationship between online engagement and advertising effectiveness, Calder, Malthouse, and Schaedel (2009) provided a systematic approach to measure engagement with the Internet through eight different online experiences Those experience s include stimulation and inspiration, social facilitation, temporality self esteem and civic mindedness, intrinsic enjoyment, utilitarian ism, participation and socialization and community. The authors further grouped the eight online experiences

PAGE 44

44 into tw o types of engagement with the Internet personal and social interactive engagement. P ersonal engagement is manifested in experiences that are derived from stimulation and inspiration, social interaction, self worth, intrinsic enjoyment, and utilitarian f acilitated by the site, while social interactive engagement is motivated by participation, socializing, as well as community building. The two general dimensions of personal and social interactive engagement with online media in fact provide a useful lens to examine viewer engagement with television content through various social media platforms. Several recent industry studies have investigated how media companies adopt social media, video formats, rich Internet applications, and other initiatives to driv e involvement, interaction, intimacy, and influence that an individual has with a brand over metrics such as visitors to a site or application, page views or page view equivalents per visitor. The second dimension interaction may include metrics such as video s played, community contribution s ratings, reviews, vote s submitted, photo s or video s uploaded, and/or text message s sent Int imacy is the third dimension in measur ing the sentiment in blog post s blog comments, and discussions in online forums. The last dimension of influence metrics consist s of tracking forward content, tagged content, widget and video embeds, and friends and fans in social networks (Epps, 2009). With respect to audience engagement with specific social media platforms, Takashi (2010) developed a model of social networking sites engagement (e.g., Myspace TM and Facebook TM ) within a framework of audience engag ement. The first

PAGE 45

45 dimension is the information seeking activity and selectivity, which identif ies such activities as seeking, collecting, and sharing information relevant to daily life of close friends, school life, or general issues, news, or events. The s econd dimension is related to connectivi ty, measuring connection formation among people and groups, transnational and trans age connectivity. The third dimension of bricolage emphasizes the creation of bricolage of friends and images from different communi ties or managing impression with profiles. The last dimension is focused on participation and is characterized by the involvement in various online communities. Regarding blog ging, one of the most popular social media platforms in addition to social netwo rk sites, Yanga and Kangb (2009) developed and validated a measurement scale of blog engagement. The authors explicated the concept of blog engagement as the likelihood and outcome of interactive blog communication, and suggested four attributes of blog en gagement, i.e., contingency interactivity, self company connection (the cognitive dimension), company attitude (the attitudinal dimension), and WOM intensions (the behavioral dimension). The study further concluded that interactive blogs can enhance self c ompany connection s positive attitudes toward the company, and supportive WOM intention s In summary, drawing together the threads of various studies on engagement with media content in both the offline and online media contexts, the engaging viewing exp erience is manifested in multiple dimensions measur ed by diverse scales. Th ose dimensions are temporal, utilitarian, enjoyment, inspiration, participation, involvement, connectivity, interaction, socializing, community, intimacy, and influence The synthes es of the attributes of engagement serve as the foundation contributing to the

PAGE 46

46 conceptualization and operationalization of the social engagement construct in this study. The definition for each attribute of engagement and its measure scales are presented i n Table 3 1 and Appendix A. Table 3 1. Definition of the attributes of engagement Attributes Definition Temporal ity Media consumption is part of a routine or regular schedule; escaping or diversion from problems Utilitarian ism Finding out about relevan t events and conditions in surroundings, society and the world; seeking advice on practical matters or opinions and decision choices; satisfying curiosity an d general interest; learning, self education; gaining a sense of security through knowledge Enj oyment Relaxing, getting intrinsic cultural or aesthetic enjoyment, filling time, emotional release Inspiration Stimulating, inspiring, fascinating, and made me curious or enthusiastic Participation/Involvement Actively involved or participating in i nterested community or various media touchpoints Interaction/ Interactivity Interacting with the characters and contex tual settings; getting feedback from people and groups Connectivity Having a personal a ssociation with the characters/ situations in th is vehicle; forming a connection with people or groups Socialization Finding a basis for con versation and social interaction; having a substitute for real life companionship; identifying with others and gaining a sense of belonging Community Community participation, building, and contribution; enabling one to connect with family, friends, or the society Intimacy Em otional investment and affective response Influence The likelihood of reco mmending advocating, word of mouth Conceptually, this inves tigation defines social engagement as the degree of intensity or types of connections that audiences develop with television content through social media platforms over time. This social engagement experience extends beyond the traditional, passive televis ion viewing pattern, representing both active behavioral

PAGE 47

47 engagement and emotional connection that viewers develop with television characters or program contextual settings through social media platforms. The s ocial media platforms consist of an expanding a rray of online media applications which facilitate information sharing, knowledge distribution, and opinion exchanges. Typical examples are socia l networks (e.g., Facebook TM and Myspace TM ), blogs (e.g., WordPress ), microblogs (e.g., Twitter TM ), content sh aring communities (e.g., YouTube and Flickr ), online discussion forums, podcasts, Really Simple Syndication (RSS) feeds, online social tags and bookmarks (e.g., Digg TM and Delicious), and mobile texting, etc. The term television content covers a broader scope, including television program ming and its relevant information, the characters or celebrities related to the program, and professional working staff such as producers or directors of the show. To operationalize the social engagement construct, the present study adopts the inductive reasoning approach, which consists of making specific observations and measure ment s, detect ing patterns and regularities, formulat ing some tentative hypotheses, and finally developing some general conclusions or theories (Holland, Holyyoak, Nisbbett, & Thagard, 1989). Guided by the reasoning and logic, Churchill (1979) outlined a paradigm suggesting a comprehensive procedure for developing better multi item measures of marketing constructs through eight steps. The first s tep is to specify the domain of construct through a literature search, intensity or types o f connections that audiences develop with television content through social media platforms along behavioral and functional dimensions. The second step in

PAGE 48

48 the procedure is to generate a sample of items which capture the domain as specified in step one by u sing multiple research methods. Specifically, this study employs focus groups interviews and literature reviews to generate a pool of items identifying social media use experience in relation to television content. The third and fourth stage is to collect data and purify measures by conducting factor analysis and testing coefficient alpha. The first four steps of the scale development are completed to address the content validity, dimensionality, and the internal consistency of the set of scales developed. The next four steps of collecting data, assessing reliability and validity, and developing norms are focused on assessment of reliability with new data and issues concerning criterion and construct validity. Specifically, this investigation validates the factor structure of the scales and tests the predictive ability of social engagement with the other important variables proposed in the aforementioned integrated model. Under the scale development and validation framework, the first research question in th is study attempts to investigate if there are different dimension s or levels in the social engagement construct as addressed below: RQ1: Are there different dimensions or levels of social engagement with television content in a social media context? Perce ptions of Television Program Program Genre Preference program type or genre among a set of available program types or genres (e.g., soap opera, sports, drama, news, etc.) (Y oun, 1994). Scholars and industry practitioners have previously concluded that television genre is an important predictor in viewing

PAGE 49

49 choice because the industry relied heavily on imitation (Bielby & Bielby, 1994; Gitlin, 1983). Program genres are regarded production of television series and provide heuristics for estimating the potential success of proposed programs based on the success of previous programs in the same ge 205). The common kn owledge in program choice behavior is that conventional program types, such as drama, situation comedies, and so on, bear systematic relationship s to program preferences (Geerts, Cesar, & Bulterman, 2008; Webster & Wakshlag, 1983). In most models of televi sion program choices, individual audiences are assumed to have relatively consistent preferences for program types, and these general dispositions tend to determine preferences for specific programs (Webster & Wakshlag, 1983; Webster, Phalen, & Lichty, 200 6). Prior (2005) demonstrated that audiences choose a program that best satisfies their preferences, and people who prefer a certain program genre tend to watch that program type over other program types across various media platforms. In addition, Wober a nd Gunter (1986) pointed out that people are likely to stick with favorite genres although their preferences for specific shows and particular episodes may vary. The preferences of different types of content could stimulate diverse social viewing experien ces and communication patterns surrounding certain programs. Specifically, genre preferences can impact the way viewers talk, chat, or interact with each other while watching television or afterwards (Geerts, Cesar, & Bulterman, 2008). Sports programming i s often cited as one of those genres that is best suited for stim ulating social interaction. O ther types of content such as cooking programs and movies also

PAGE 50

50 present this type of sociability (Harrison & Amento, 2007). A prior study also revealed that there are several program genres that television audiences talk about most after watching (e.g., including news, sports, soap opera, docusoap, reality show, talk show, comedy series, and quizzes) S ome genres have been shown to be motivat ors of viewer engagement experience s such as sharing these genre program videos or viewing experiences with others (Geerts, Cesar, & Bulterman, 2008). Simmons multi media engagement study (2008), a specific study focused on the relationship between program genres and television viewer engagement, surveyed six broadcast programs and found that reality shows score higher than dramas in most of the measured engagement dimensions, especially in terms of the dimension of social interaction, trustworth iness, and life enhancement The results suggest that viewers tend to be more actively involved with reality programs over scripted dramas through social interaction with their colleagues, friends, and family. Based on the above discussion on the sociability and engagement nature of vario us program genres and the relationship between genre preference and viewer engagement, the following hypothesis and research questions are proposed: H1: Program genre preference will be positively related to the overall social engagement with the program. RQ2: Does the overall social engagement or the different dimensions of social engagement with the program vary among different genres (i.e., drama, reality show, sitcom, game/talk show, and animated comed y )? Note that the above research question and hypo level of social engagement, that is, all engagement activities via various social media

PAGE 51

51 platforms. The different dimensions or levels of social engagement and their relationships with the various sets of antecedents will be addressed separately in RQ6. Program Affinity Affinity, defined as the level of importance one attached to the medium or media content, is one attitude that has received considerable research attention in broadcasting and electronic media areas (Rubin, 1 983, 200 9 ). Rubin (1983) proposed that television viewing patterns should be examined from the perspectives of viewing behaviors (i.e., viewing levels and program preferences) and television attitudes (i.e., affinity and realism). Along with the television realism construct, television affinity is the second dimension of television attitudes, defined as the perceived importance of television in the lives of television viewers (Rubin, 1983). With respect to specific television programming, Rubin and Perse (1 987 a ) measured program affinity as the Several researches revealed that affinity has been associated with diverse media use behavior and viewing motives (Haridaki s & Hanson, 2009; Rubin, 200 9 ). For example, Rubin (200 9 ) found that more habitual and less active media users tend to attach an affinity with the medium of their choice, whereas instrumental and active media user s are more likely to show an affinity with content. In addition, affinity with particular media content such as television programs is found to link to post viewing discussion activity (Rubin & Perse, 1987 b ). Furthermore, prior studies demonstrated that affinity with the television medium or progra m content such as soap opera s is positively related to viewing motives such as arousal, information seeking, pass ing time, escap ing and entertainment (Rubin, 1983; Rubin & Perse, 1987 b ).

PAGE 52

52 Regarding new online media platforms, Papachrissi and Rubin (2000) showed that interpersonal utility motivation is positively predictive of Internet affinity when investigating potential predictors of Internet use. Haridakis and Hanson (2009) examined one of the most popular social med ia platforms content sharing commun ity, and regarded affinity as one of a range of independent variables predicting such viewing behavior as co viewing videos on YouTube and sharing video s with others. Thus, the current study anticipates that the more affinity of a particular television pr ogram an audience possesses, the more likely it is for the audience to exhibit greater engagement with the program through a range of social media vehicles. H2: Program affinity will be positively related to the overall social engagement with the program. Program Involvement The involvement concept has been examined in a variety of research domains. In communication fields, involvement, or a sense that certain communication content has personal relevance, has been conceptualized as an active psycholo gica l processing of content The involvement concept has been categorized into two levels the degree to which audiences perceive a connection between themselves and the mass media content, and the degree to which they interact psychologically with the medium and its message (Levy & Windahl, 1984). In advertising research, program involvement refers & Smit, 2007, p. 127). In marketing areas, telev ision program involvement focuses on program, and is usually defined as personal relevance or importance of the program (Feltham & Arnold, 1994).

PAGE 53

53 Prior research suggested that viewer involvement is a multi dimension al construct representing different manifestations of how viewers connect and develop relationships with television content. The manifestation cognitive versus emotional involvement (Perse, 1990), demanding versus relaxing programs involvement (B arwise & Ehrenberg, 1987), cognitive, affective, and functional involvement (Park & McClung, 1986), and cognitive, func tional, and demanding involvement process (Feltham & Amold, 1994). Cognitive involvement is usually driven by utilitarian functions, whil e affective involvement tends to be expressive, emotional, and experiential in nature. In summary, all of these dimensions in involvement suggest that an involved individual would be more likely to find a television program personally engaging, fascinating intriguing, engrossing, and riveting. In the television program consumption context, Rubin and Perse (1987a, 1987b) categorized viewer involvement with soap opera s or news as affective involvement, cognitive involvement, and behavioral involvement. Spec ifically, an involved television viewer may feel affective toward those in need on the show (i.e., affective involvement), consider the messages of the show (i.e., cognitive involvement), and talk about the show with others (i.e., behavioral involvement) d uring and after the exposure. Additionally, in the reality program con text, Hall (2009) proposed audience involvement as a three dimension al construct to capture the current reality program consumption in a cross media environment. includ e social involvement, cognitive involvement, and online involvement (2009) The author further suggested that each form of involvement is associated with enjoyment, an important element of viewer engagement. Furthermore, Green and colleagues posited that, within the field of

PAGE 54

54 narrative theory, the highly involved audiences tend to have strong emotions about a They suggested th at the involved state in a story could contribute to audience enjoyment by distracting viewers from per sonal concerns or stress, allowing them to learn new things and fostering a sense of connection with the characters (Green & Brock, 2000; Green, Brock, & Kaufman, 2004). As discussed in the social engagement construct section, engagement and involvement are two different concepts in nature. Program involvement in this study is focused on personal relevance or importance with response to particular television content measured by cognitive, affective, and functional dimensions (Feltham & Arnold, 1994; Park & McClung, 1986). In addition, television program connectedness may start by fostering simple involvement with the program. Over the course of repeat viewing, it may end up absorbing their audiences in parasocial relationships with the characters in the pr ogram (Russell, Norman, & Heckler, 2004a). Thus, the current study expects that the highly involved audiences will be more likely to possess greater engagement with the program content through a range of social media vehicles. H3: Program involvement will be positively related to the overall social engagement with the program. Perceived Characteristics of Social Media Perceived Ease of Use Many scholars have attempted to explain and predict user acceptance and use of new communication technologies. A co mmon theme underlying these various research streams is the inclusion of the perceived characteristics of a technology as key independent variables. The theory of TAM has been extensively applied to predict tems, and two core constructs

PAGE 55

55 perceived usefulness and perceived ease of use are suggested in the model to jointly affect intention to use the technology (Davis, 1989; Davis, Bagozzi, & Warshaw, 1989). Specifically, perceived usefulness is defin person believes that using a particular system would enhance his or her job is, 1989, p. 320). The present study employs perceived ease of use as one of the constructs to predict the intension of using social media to engage with te levision content. P rior enjoyment on cell phone usage (Kwon & Chidambaram, 2000), online learning systems adoption (Saade & Bahli, 200 5 ), and mobile Internet applications acceptance (Che o ng & Park, ively new online communication technologies such as social media systems would be related to the adoption of them to interact with television content. The study thus posits the following hypothesis: H4: The perceived ease of use of social media will be pos itively related to the overall social engagement with the program. Compatibility Given the innovative nature of social media systems, the theory of innovation diffusion offers a heuristic framework to investigate how individual audiences use social media platforms to engage with television content. Rogers (2003) conceptualized the perceived characteristics of an innovation as relative advantage, compatibility, complexity, trialability, and obersvability. C which the adoption of a technology is compatible with existing values, past experiences,

PAGE 56

56 and n 15). Therefore, compatibility may refer to compatibility with the values or norms of the potential adopters or it may imply congruence with the existi ng practices of the adopters (T o r nat z ky & Klein, 1982). The first description of compatibility suggests a type of normative or cognitive compatibility (compatibility with what people feel or think about a technology), while the second descriptio n implies a more practical or operational compatibility (compatibility with what people do). In either case, the compatibility of innovation to the potential adopters is, theoretically, positively related to adoption and impl ementation of the innovation (T o r nat z ky & Klein, 1982). Innovation diffusion research has found that compatibility is salient in predicting the adoption of new communication technologies (Chen, Gillenson, & Sherrell, 2002; analysis of innovation adoption, the authors discovered that an innovation is more likely to be adopted when it is compatible of compatibility on alternative video platform vi ewing pattern s, such as online video streaming and mobile television, and found that viewers who see online video streaming as compatible with their lifestyles tend to engage more frequently in online video viewing. Thus, the following hypothesis is posite d: H5: The perceived compatibility of social media will be positively related to the overall socia l engagement with the program. Social Presence The theory of social presence is primarily used to measure how users sense the existence of other people in the mediated environment. The concept of social presence

PAGE 57

57 experience others as bein 1987, p. 531). Lombard and Ditton (1997) described extent to which a mediu m is perceived as sociable, warm, sensitive, personal, or intimate when the medium is used to interact with other people. Accordingly, a communication medium that involves more human senses will generate stronger feelings of social presence. For example, S hort, Williams and Christie (1976) found that experiment participants tend to possess a stronger impression of social presence after an audio visual task based interaction than after one based on audio alone The conceptualization of social presence was f irst proposed as an unidimensional construct by measuring the self Williams, & Christie, 1976, p. 65). The authors employed a set of semantic differential scales, which attempt ed to evaluate the social and emotional capabilities of the medium. It should be noted that the measurement s indicate d impression on the medium itself, not their judgment on the experience with others. On the other hand, some scholars asserted that social presenc e should be conceptualized as multidimensional construct rather than in terms of the direct attributions about medium per se (Biocca, Harms, & Burgoon, 2003). For example, a previous study found that media form and media content could influence various dim ensions of presence (Dillon, Keogh, & Freeman, 2002). Shen and Khalifa (2008) further proposed affective and cognitive social presence in the online community context. Affective social e online

PAGE 58

58 community, while cognitive social presence refers to the perceived mutual understanding existing among communicators. The performance of social presence varies along a range of communication technologies, but is suggested that compute r mediated c ommunication is lower in social presence compar ed to face to face communication due to a lack of nonverbal cues (Perse & Courtright, 1993; Rice, 1993). In addition, Garramone, Harris and Anderson (1986) found that social presence is positively associated w ith personal identity satisfaction, such as expressing, commenting, and interacting opinions with others, when exploring political user behavior of online bulletin board sy stems. As social media platforms, such as social networks and microblogs exhibit th e capacity of interpersonal communication, the present study expects that the perceived social presence of online social media will stimulate audience members to actively engage in these platforms with other viewers in the context of television viewing. Th us, the following hypothesis is proposed: H6: The perceived social presence of social media will be positively related to the overall social engagement with the program. Audience Attributes Motives The uses and gratifications approach has historically pr ovided functional typologies for many media use situation s including traditional mass media and new communication technologies (Ruggiero, 2000). Regarding traditional television viewing motives, prior studies have identified habit, relaxation, companionsh ip, passing time, information/learning, arousal, social interaction, escape, and entertainment as major drivers for tele vision viewing ( Palmgreen & Ra y burn, 1979; Rubin, 1983).

PAGE 59

59 In terms of the usage of new alternative platforms to television content viewi ng such as mobile television and webcasting (or Internet television), prior research confirmed several motives shared between traditional television and alternative platforms. At the same time, these studies also identified a number of additional, new moti to the usage of mobile video devices in various situations beyond the simplistic notion of ing ile TV own space to manage the relationship with others around in the public setting is a new reason for mobile television usage, along with traditional needs such as e ntertainment, information, relaxation, and killing time. With regard to watching television content on the Internet (i.e., Internet television or webcasting), Lin (2001) found that entertainment appears to be less potent than the other two motives, inform ation learning and escape/interaction when exa mining online services adoption. H owever, w ith further exploration of webcasting adoption at a later time the author concluded that entertainment plays a more critical role than news and information learning (Lin, 2004, 2006). Yang and Kang (2006) conducted an exploratory for using Internet television in Taiwan, and concluded that the more respondents used the Internet for entertainment and social reasons the more likely they would watch television online. Furthermore, audience motives are found to predict various viewing activities (Rubin & Perse, 1987 a, 1987b ). Specifically, the more strongly viewers are motivated, the more actively they engage in var ious audience activities before viewing (e.g.,

PAGE 60

60 viewing intention), during viewing (e.g., attention and involvement), and post viewing (e.g., discussion) (Lin, 1993). In addition, more salient viewing motivations, especially exciting entertainment and socia l utility, are found to be related to parasocial interaction, post viewing cognition, and post viewing discussion in the soap opera consumption context (Rubin & Perse, 1987 b ). Prior studies have indicated that alternative media platforms and traditional television viewing share a majority of motives such as entertainment, information, diversion, personal communications, and passing time. On the other hand, due to other in nate media characteristics associated with the Internet and its online applications there are additional motives involved with these online platforms, such as convenience, immediate access, and social interactions. T he current study therefore synthesizes various motives of traditional television, the Internet, and new media technologies t o assess the social and physiological origins of the socially engaging experience, and poses the following research questions : RQ3: What motives do audiences have for using social media to engage with the program? RQ4: What specific motives will be posit ively related to the overall social engagement with the program? Innovativeness media could also help determine how a television viewer might use social media to engage wit h television content. Since a majority of social media are still recognized as newly emerging communication platforms and are in the early stage of diffusions, the innovativeness attributes that a television viewer possesses could contribute to his or

PAGE 61

61 her social media adoption decision making. According to the innovation diffusion theory, early adopters are characterized as having a higher degree of personal innovativeness (Rogers, 1995, 2003). Prior research show ed that both innate innovativeness (the soci al cognitive foundation) and actualized innovativeness (the social situational basis) of an ind associated with the adoption of an innovation (Midgley & Dowling, 1978). attribute to his or her adoption tendency of the Internet and its online applications. In particular, Lin revealed that greater need for innovativeness is a significant predictor for personal computer adoption (1998) and webcasting (2004). Likewise, Bussel le and colleagues (1999) s use. Furthermore, Sun, Youn, Wu, and Kuntaraporn (2009) concluded that innovativeness is an important predictor of online social activities su ch as forwarding content and chatting with others. More importantly, one relative study that focused on the social media platform, YouTube showed that personal innovativeness predicts viewing and sharing of video in the content sharing community website (Haridakis & would be relevant to the use of social media to engage video content. Thus, this study develops the following hypothesis: H7: Audience innovativeness will be p ositively related to the overall socia l engagement with the program. Social Characteristics As suggested in the uses and gratifications approach media compete with other forms of communication or functional alternatives for a finite amount of time among

PAGE 62

62 limited audiences (Kaye & Johnson, 2003; Rubin, 200 9 ). The relationship between s offl ine activities like interpersonal interaction and social activities are suggested to have an impact on their online media use behavior (DiMaggio, Hargittai, Neuman, & Robinson, 2001). Papacharissi and Rubin (2002) found that the greater satisfaction with p ersonal interaction such as face to face communication people have, the more likely they are to use the Internet for information purpose s ; whereas those who are not satisfied with face to face interaction tend to use the Internet for interpersonal intera ction in the virtual world. In the social media context, Haridakis and Hanson (2009) empirically concluded that socially active audiences, particularly those watching for purposes of social interaction and co viewing, use YouTube as a way of sharing onlin e activities with family/friends and with persons with whom they have existing social ties. s are hypothesized to be salient when using social media to engage with television c ontent. however, are not consistent according to different research findings. For example, some studies revealed that the heavy Internet users tend to have small social cir cles, less interpersonal communication, and loneliness (Kraut, Patterson, Lundmark, Kiesler, Mukopadhyay, & Scherlis, 1998). On the other hand, several scholars argued that Internet users have wider social networks than nonusers, since the Internet enlarge s existing social circles (Hampton & Wellman, 2003). Such opposite research findings

PAGE 63

63 interpersonal interaction or social activity in relation to their online media use behavio r, regardless of the position taken. Thus, this investigation proposes the following research question: social activities) relate to the overall social engagement with the program? By extensively reviewing engagement experience with media content and media platforms, the present study speculates that there are different dimensions or levels in the social engag ement experience with television content through a range of social media platforms. If the social engagement construct in this study does present multi dimensional levels, will the aforementioned sets of antecedents all be related to the multiple dimensions or vary greatly along these levels? With this question in mind, this investigation next delves into the more complicated relationships between the aforementioned sets of antecedents and different dimensions indicated in social engagement behavior ( if there are any), and formulates the following research question: RQ6: How do audience perceptions of television programs (i.e., program genre presence, program affinity, and program involvement), perceived characteristics of social media (i.e., perceived ease of use, compatibility, and social presence), and audience attribut es (i.e., motives, innovativeness, and social characteristics) relate to the different dimensions or levels of social engagement with the program? Consequences of Social Engagement Program Loyalty Loyalty is viewed as one of the most important concept s i n marketing and consumer research, categorized into brand loyalty (Jacoby, 1971; Jacoby & Chestnut, 1978), product loyalty (Olsen, 2007), customer/service loyalty (Dick & Basu, 1994, and

PAGE 64

64 chain/store loyalty (Macintosh & Lockshin, 1997). The loyalty constru ct is defined and measured in either a behavioral or attitudinal dimension, and is therefore suggested as a multi dimensional concept (Dick & Basu, 1994; Jacoby & Chestnut, 1978). Dick and hip between an 99). Oliver (1997) referred to 392). The behavioral dimension of loyalty is usually measured as a proportion of purchase (Cunningham, 1966), purchase sequence (Kahn, Kalwani, & Morrison, 1986), or a probability of purchase (Massey, Montgomery, & Morrison, 1970). The attitudinal aspect of loyalty is defined as attitude toward th e loyalty/disloyalty act ( Rundle Thiele & Mackyay 2001) brand preference (Guest, 1994, 1995), and commitment (Hawkes, 1994). In addition, three types of antecedents to br and loyalty involves, c ognitive antecedents (i.e., accessibility and confidence), af fective antecedents (i.e., emotion, feeling states, satisfaction, and involvement), and conative antecedents (i.e., switching cost, sunk cost, and expectation) (Dick & Basu, 1994; Macintosh & Lockshin, 1997; Oliver, 1999). In the context of television con sumption, b u ilding upon pr ior marketing and consumer resea rch, television program loyalty, in this study is centered on both television programs. Brosius, Wober and Weimann (1992) defined television viewer loyalty (a) general loyalty to watching television, (b) channel (or

PAGE 65

65 324). The authors further operati onalized the viewing loyalty construct as the ratio of the individual numbers and the numbers of watched programs, and programs within program genres or channel s using British consumer panels. Although repeat purchasing behavior is distinguished from bran d loyalty in marketing research (Jacob & Kyner, 1973), the concept of repeat viewing in television and advertising research indicates the critical dimension of viewer loyalty to specific television programs (Cooper, 1996; Sabavala & Morrison, 1977). In fac t, the phenomenon of repeat viewing has been extensively examined by communication researchers in the past three decades. Repeat viewing refers to the number of members of an audience who watch an episode of a program and then choose to watch a subsequent episode of the same program when given an opportunity to do so actual program choice and viewing patterns, while attitudinal program loyalty identifies the extent to which a viewer intends to engage in certain television programs. Furthermore, research from increasing cross platform, multitasking media consumption patterns would help promote program loyalty. When investigati ng the relation ship of cross media usage with television viewer loyalty, Ha and Chan Olmsted (2004) assessed the usage of enhanced features on television websites such as online video streaming and message boards, and found that the increase in the number of website features usage positively predicts viewer loyalty (i.e., attitudinal loyalty). Lu and Lo (2007) further reported that television audience satisfaction one element of viewer engagement strongly predict s repeat viewing intention (i.e., behavio ral loyalty). Thus, it is logical to consider that

PAGE 66

66 highly engaged viewers using multiple social media platforms to interact with television content are more likely to be loyal to their chosen programs. H8: The overall social engagement with the program wil l be positively related to the behavioral loyalty to the program. H9: The overall social engagement with the program will be positively related to the attitudinal loyalty to the program. Audience Satisfaction Satisfaction in marketing and consumer res earch is often seen as a multi dimensional construct defined in many ways, including components of pleasure, need fulfillment, evaluation of products/services, and benefits (Oliver, 1997; 1999). Olsen (2007) suggested that individual satisfaction toward a product category is a cumulative consumption outcome, as well as the overall experience and evaluation of satisfaction summary of (direct) consumption experience, based on the discrepancy between prior 146). In mass media research, satisfaction has been studied from both the interpersonal and mediated communication contexts. From the interpersonal communica tion point of view, satisfaction is derived from fulfilling expectations through interaction (Hecht, natural outcome of media use viewed as overall satisfaction with the medi um, with a specific genre or communication activity, or with specific content during (Patwardhan, 2004, p. 419). More formally, according to the uses and gratifications approach, Palmgreen and Rayburn (1985) defined audience satisfaction as a gen eral feeling of contentment result ing from repeated exposures to a particular content genre

PAGE 67

67 (e.g., television news). In addition, Perse and Ferguson (1993) suggested that audience satisfaction should emphasize outcomes and benefits of media use, rather tha n expectations which are predicted to be less important in understanding television satisfaction. Based on prior satisfaction propositions from the marketing and mass media domains, audience satisfaction in this study is defined as the overall evaluative s ummary of the (direct) viewing experience with particular television content. Prior media research posited that viewing engagement or involvement has much impact on media use and effects, specifically influencing the satisfaction that people receive from media use (Levey & Windahl, 1984), and further subsequent planned media e xposure (Rubin & Perse, 1987 a ). Following previous research on audience satisfaction (Lin, 1993; Palmgree & Rayburn, 1985; Perse & Rubin, 1998), the current study proposes that audien ce viewing behavior is a temporal gratification seeking process, and expects that the more strongly motivated viewers would more actively engage in various audience activities to connect with the characters and contextual setting of a program throughout th e viewing process and thus receive greater viewing satisfaction afterwards. Accordingly, the pres ent study hypothesizes that highly engaged audiences in social media contexts are more likely to possess greater satisfaction with the television program. H10 : The overall social engagement with the program will be positively related to audience satisfaction with the program. Product Purchase Likelihood The eventual effects of viewer engagement with a television program on consumer purchase behavior are the chief considerations of advertisers and marketers. Kilger and Romer (2007) proposed that media engagement, advertising engagement, and brand

PAGE 68

68 posited that there are three mech anisms operating to link engagement to consumer and subsequent purchase, i.e., cognitive, emotional, and behavioral attention to program and advertising content. When further investigating a set of dimensions of engagement with three media channels (i.e., television, magazines, and the I nternet ) the authors revealed that there is evidence of a strong relationship between engagement in the media vehicle and the likelihood of purchasing a product advertised within that media vehicle (Kilger & Romer, 2007). Ha and Chan Olmsted (2001) suggested that there are two types of merchandise available on television networking sites: fan based items and non fan based items. The fan based items are items relevant to the network or its shows and stars. Several examples a re products used in the television shows. Non fan based items are essential products of the advertisers of the broadcasting/cable networks. The authors noted that the more television website visitors are exposed to enhanced television features; the more li kely they are to show an interest in buying products that have been advertised in the network shows or on the websites. Thus, it is reasonable to postulate that viewer engagement with television content through various social media platforms is associated with the purchase intention of re levant products advertised on the station/network sites. The corresponding hypothesis is: H11: The overall social engagement with the program will be positively related to product purchase likelihood. This study postula tes that the social engagement construct possesses multiple dimensions, and speculates there are relation ship s between a set of antecedents and

PAGE 69

69 the different dimensions of social engagement. Likewise, this investigation conjectures that the different dimen sions/levels of social engagement will play a role in t he above proposed consequences. T hus, the following research question is addressed: RQ7: How do the different dimensions or levels of social engagement predict program behavioral and altitudinal loyalt y, audience satisfaction, and product purchase likelihood? Based on the above discussion, the following hypotheses and research questions are submitted: Social engagement RQ1: Are there different dimensions or levels of social engagement with television content in a social media context? Antecedents to Social Engagement Television Program Genre, Affinity, and Involvement H1: Program genre preference will be positively related to the overall social engagement with the program. RQ2: Does the overall social engagement or the different dimensions of social engagement with the program vary among different genres (i.e., drama, reality shows, sitcoms, game/talk shows, and animated comedies)? H2: Program affinity will be positively related to the overall s ocial engagement with the program. H3: Program involvement will be positively related to the overall social engagement with the program. Perceived Characteristics of Social Media H4: The perceived ease of use of social media will be positively rel ated to the overall social engagement with the program. H5: The perceived compatibility of social media will be positively related to the overall social engagement with the program.

PAGE 70

70 H6: The perceived social presence of social media will be positively related to the overall social engagement with the program. Audience Attributes RQ3: What motives do audiences have for using social media to engage with the program? RQ4: What specific motives will be positively related to the overall social engagem ent with the program? H7: Audience innovativeness will be positively related to the overall social engagement with the program. social activities) relate to the overal l social engagement with the program? Antecedents to the Different Dimensions or Levels of Social Engagement RQ6: How do audience perceptions of television programs (i.e., program genre presence, program affinity, and program involvement), perceived char acteristics of social media (i.e., perceived ease of use, compatibility, and social presence), and audience attributes (i.e., motives, innovativeness, and social characteristics) relate to the different dimensions or levels of social engagement with the pr ogram? Consequences of Social Engagement H8: The overall social engagement with the program will be positively related to behavioral loyalty to the program. H9: The overall social engagement with the program will be positively related to attitudinal loyalty to the program. H10: The overall social engagement with the program will be positively related to audience satisfaction with the program. H11: The overall social engagement with the program will be positively related to product purchase like lihood. Consequences of the Different Dimensions or Levels of Social Engagement RQ7: How do the different dimensions or levels of social engagement predict program behavioral and altitudinal loyalty, audience satisfaction, and product purchase likeliho od?

PAGE 71

71 C H A PTER 4 METHODS Study Design Overview The purpose of this study is two fold: 1) build an active audience behavior model that aims to decipher the emerging multimedia television consumption pattern and 2) examine the social media practices curren tly implemented by broadcasters and advertisers by validating a set of social engagement scale s and testing the engagement antecedences and consequences. More specifically, this investigation explores why certain consumers increasingly choose so cial media platforms in relation to television content, how they utilize the different social media platforms to interact with specific shows, and what the actual effects of such an engagement are. With this in mind, this study develops and validates the social engagement scale based on measures of marketing constructs Specifically, Churchill argued for the development of multi item measures as more robust ways to measure marketin g phenomena, and put forward an eight step procedure to achieve it. The eight steps are: 1) specify the domain of the construct, 2) generate a sample of items to capture the domain as specified 3) collect data for pretest, 4) purify measure, 5) collect ne w data, 6) assess reliability with new data 7) assess construct validity, and 8) develop norms. Churchill (1979) further recommended various coefficient s and operational techniques that are appropriate for each stage to realize the conceptualization and o perationalization process For the first two steps, several methods were suggested, including literature search, experience survey s critical incidents, focus group s and interview s For the step of purifying measure, two statistical techniques coefficie nt alpha and factor analysis were

PAGE 72

72 recommended to remove redundant or non reflective items. For the fifth and sixth steps, split half reliability and criterion validity were involved (Churchill 1979). The schematic representation of the procedure is indi cated by Figure 4 1. Figure 4

PAGE 73

73 Follow procedure this investigation employs a three staged research plan to develop and validate the measurement in struments of social engagement and further test the proposed antecedents and consequences in the integrated, active audience behavior model. The first stage is to develop the measurement scales for the social engagement construct This process includ es ite m generation, exploratory factor analysis, and scale description. Specifically, three focus groups of twenty seven undergraduate students (nine students per group) who participate d on a voluntar ily basis were conducted Drawing upon the qualitative analysi s from the focus groups plus relevant scales adapted from the previous engagement research, this study generates a pool of items that c haracterize the connections that individuals form with television content through various social media platforms. The fin al set of statements is transformed into a questionnaire of five point Likert scales, and adminis tered to an online consumer panel of 161 participants. During the next stage, this study surveys another online consumer panel of 49 4 qualified respondents who have certain social media experiences to confirm the factor structure generated from the exploratory factor analysis results. The final stage includes scale applications and tests of the relationships between the social engagement behavior and its anteced ents and consequences in the proposed model. Online Survey This investigation employs online survey as the main research method because the Internet is a rich domain for conducting survey research related to technology oriented and computer mediate d commu nication (Wright, 2005). Topics as diverse as mobile Internet adoption and its applications (Cheong & Park, 2005), online gaming (Hsu & Lu, 2004), and W eb based learning (Jiang & Ting, 2000) have been studied

PAGE 74

74 using online survey s In addition, Internet sur veys have been widely applied to the full communication spectrum from interpersonal (Wright, 200 5 ), group (Hobman, Bordia, Irmmer, & Change, 2002), organizational (Ahuja & Carley, 1998), health (Rice & Katz, 2001), and mass communication (Flanaqin & Metzge r, 2001). Considering the topical relevance of online social media engagement with television content from an individual seems appropriate. Online survey s are generally advan tage ou s over traditional paper based, mail in survey s by providing access to a unique population, by saving time and expenses and by removing geographical restriction s More specifically, online survey s can take advantage of the ability of the Internet an d virtual communities to provide access to specific groups and individuals who share particular interests, attitudes, beliefs, and values regarding an issue, event, or activity (Wright, 2005). A second advantage of the online survey is that this survey met hod provides time and cost benefits for researchers. As mentioned by Wimmer and Dominick (2006), an online survey can quickly gain access to a large number of individuals by posting invitations to participate in newsgroups, chat rooms, and message board co mmunities with economic cost. Moreover, online survey s are particularly attractive when the population under study is distributed across large geographic regions. In addition, Selm and Jankowski (2006) suggested that the object of study and particular char acteristics of the population may request the use of an online survey. Given the goal of this study is to measure online social media experience in relation to television content consumption among connected

PAGE 75

75 consumers, the cost, ease, and speed of delivery and response weight in favor of the Internet as a delivery method for this survey research. Online survey s however, should be implemented cautious ly due to several disadvantages associated with the survey research method. The greatest concern involv ing o nline survey is sampling issues, which implies that Internet surveys are lacking a central registration of users on the Web. Unlike the telephone numbers and home addresses, it seems impossible to construct a random sample from email addresses due to the v ariation s in address construction (Selm & Jankowski, 2006). The same problems also apply to the samples generated from virtual groups and organizations as well as online communities (Wright, 2005). To overcome the problem, the current study constructs two screened, purposive samples by using online consumer panels which are administrated by a research company (uSamp TM ). The company requests that all respondents must have a valid email address to join their panels, and employs an administrative tool that use s advanced technology and unique identification algorithms to aggressively remove duplication and fraudulent respondents to online surveys. By using a branching function provided by online professional survey software ( Qualtrics Labs, Inc. Version 2009 of the Qualtrics Research Suite ) to filter out potential respondents, this investigation constructs screened samples by collecting relevant screening data in the survey response so that only responses from the required sample are analyzed. Self selection bi as is another concern when conducting online survey s which is not only shared with other traditional survey research but more salient in online research environments (Stanton, 1998; Wright, 2005). In any given online

PAGE 76

76 circumstances, it is no doubt there is a tendency that some individuals are more likely to respond to an invitation than others to participate in an online survey, leading to a systematic bias and desensitizing respondents to worthwhile survey request posts on the websites. To avoid this self selection issue, the present study particularly excludes participants who have completed any surveys in the past week to reduce the bias. Stage One : Social Engagement Scale Development Item Generation Through the qualitative process of focus groups resea rch and extensive review of the engagement literature, the goal of this exploratory stage is to generate a pool of items to characterize the connections that individual audiences form with television content via diverse social media platforms. The study fi rst conducted three one hour session focus groups at a southeastern university during the Fall of 2010 E ach focus group consisted of nine undergraduate students. The discussion questions included media us e experience, and the experience of using social media to interact with television content. Discussion results were transcribed and analyzed in a systematic fashion. Furthermore, this study adapted some relevant scales measuring the engagement expe rience of the Internet (Calder, Malthouse, & Schaedel, 2009; Epps, 2009; Haven, 2007), social networks (Takashi, 2010), blogs (Yanga & Kangb, 2009), and television (Russell, Norman, & Heckler, 2004b) to complement the results from the focus groups. Guided by the proposed active audience behavioral model, this qualitative analysis yielded a final set of nineteen items focusing on behavioral statements while eliminated such statements as feelings or physiologi cal descriptions ( Table 4 1).

PAGE 77

77 Table 4 1. List of original social engagement scale 1. I have watched the program(s) in video sharing community sites (e.g., YouTube ). 2. I am a follower of the program(s) (including actors, writers, and producers, etc.) in Twitter TM 3. I have used my m obile phone to watch video clips, check photos and texts alters, or p lay games relevant to the program(s). 5. I have used check in apps for the program(s) in Foursquare TM Miso Starling, or GetGlue, etc.. 7. I have submitted ratings, reviews, or votes related to the program(s). 8. I have uploaded or forwarded videos or photos relevant to the program(s). 9. I have sent mobile messages about the program(s) with my friends or family. 10. I have read posts relevant to the program(s) i n blogs. 11. I have written or commented on blog posts relevant to the program(s). in online discussion forums. TM ). blogs (e.g., Twitter TM ). 16. I have used social bookmarks (e.g. D igg TM and Delicious) to tag the program(s). 18. I am a fan of the program(s) and share it with my frien ds in social networks (e.g., Facebook TM and Myspace TM ). Facebook TM and Myspace TM ). Pilot Test The pilot test responses were gathered from an online consume r panel managed by the research company uSamp TM using an online survey instr ument facilitated by the Qualtrics Labs, Inc. software (Version 2009 of the Qualtrics Research Suite) during July 2011. The researcher specified a general sample frame as active online consumers over eighteen years old with a range of ages and demographics. The data collection was terminated when the numb er of completed surveys met the quota of 150 respondents. Upon entering the survey website, the respondents completed two initi al set s of of experiences with the social media

PAGE 78

78 used among the listed twenty nine social media tools, which were categorized in to social networks (e.g., Facebook TM Myspace TM Bebo Friendster Hi5 StumbleUpon, Foursquare TM Gowalla Miso, Philo, Starling, GetGlue, and Ning), blogs (e.g., WordPress and Xanga ), microblogs (e.g., Twitter TM and Tumblr ), online discussion boar d s/ forums, social bookmarks (e.g., Digg TM Delicious, Reddit, and Tagged ), content sharing communities (e.g., YouTube FunnyOrDie, Vimeo and Flickr ), podcasts, RSS feeds, mobile texting and applications, and widgets. These social media sites were chos en based on the data of online traffic, registration numb ers, or popularity 10 The l media experience measured how often the respondents used their chosen social media platforms through a five point Likert scale (1 = very rarely, 5 = very frequently). If the respondents did not have any experience with the listed twenty nine social media they were disqualified to participate in this study. However, if the participants had used at least one social media platform in the list, regardless of their use frequen cies they were qualified to continue the survey. The next screening question measur ed whether the respondents ever utilized their chosen social media to comment, post, watch, or read anything about television shows or programs. If the respondents had no such social media use experience they were disqualified and automatically filtered o ut from this study. Following these two screening questions, the qualified respondents were further asked to identify the specific program titles that they used social media to interact with,

PAGE 79

79 as well as indicate their level of agreement with the nineteen proposed social engagement statements on a five point Likert scale ( 1 = strongly disagree, 5 = strongly concerning media ownership, television subscription type, and the Internet service and concluded with demographic questions, such as age, gender, ethnicity, education, household income, employment situation, and current marital status. Descriptive Statistics of Pilot Test A total of 435 individuals responded to the online pretest survey during the first week of July 2011 Among the 435 respondents, 161 were qualified to further complete the whole survey, providing an incident rate of 37.0%. It should be noted that this was a purposive sample with individuals chosen on the basis o f their social media experience of media experience questions, the study found that the social network, Facebook TM is the most popular social media platform with 91% p enetration rate, followed by content sharing community YouTube (52%), Myspace TM (37%), Twitter TM (27%), mobile texting and applications (15%), and online discussion boards/forums (14%), respectively. Some recently emerged, entertainment focused social net works (e.g., Miso, Philo, and Starling) and social bookmarks (e.g., Delicious) are rarely used by the participants In term of the frequency of social media usage, it turned out that mobile texting and applications are most frequently used ( M = 4.05, n = 6 4), followed by Facebook TM ( M = 3.92, n = 398), YouTube ( M = 3.29, n = 231), and online discussion forums ( M = 3.20, n = 60). The study also asked respondents to identify whether they utilized social media to comment, post, watch, or read anything about television content; and if they did, what

PAGE 80

80 specific programs or shows they interacted with. Among the total of 4 02 online social media users 59% never used social media in this way, while 30% used social media to interact with television content aft er watching t he program 17% before watching the program and 12% while they were watching the program With respect to specific television programs or shows, the final dataset consisted of 83 titles These programs were categorized into the following genr es ranked by their popularity: reality shows (e.g., The Bachelor / Bachelorette Survivor Big Brother ), drama (e.g., Lost Law & Order The Glades CSI Blue Bloods White Collar NCIS ), sitcoms (e.g., Rule of Engagement The Big Bang Theory ), animated com edies (e.g., The Simpsons Family Guy ), game/talk shows (e.g., The Next Talk Show Star Top Chef ), science fiction (e.g., Star Trek ), and soap opera (e.g., All My Children Bold and Beautiful ). Particularity, the current study subscribed to the five most p opular genres, including reality shows, drama, game/talk shows, sitcoms, and animated comedies, to construct a final list of twenty television programs for the main test. Exploratory Factor Analysis Factor analysis is a collection of approaches used to ex amine how underlying constructs influence the responses on a number of measured variables, including exploratory factor analysis (EFA) and confirmatory factor analysis (CFA), as well as hybrids invoking exploratory factor extraction followed by confirmator y rotation (Thompson, 1992) or confirmatory maximum likelihoo d factor analysis (Jreskog, 1969). In particular response s determines cts is influencing responses 1998, p. 1). The current study adopted the hybrid approach to develop and validate t he social engagement construct by using EFA

PAGE 81

81 in the scale development stage followed by confirmatory maximum lik elihood factor analysis in the scale confirmation stage. There are two general approaches to conduct exploratory factor analysis: principal components analysis (PCA) and common factor analysis. The primary difference between the two approaches is their theoretical assumption. In common factor analysis, the factors are viewed as the causes of the observed variab 1995, p. 287). On the contrary, in variances of the observed variables in as economical a fashion as possible, and no latent variables underlying the observed va 287). More important, estimate s based on com mon factor analysis may generalize better to those obtained using CFA than in components analysis. Therefore, Floyd and Widaman (1995) concluded that common factor analysis is preferred over component analysis if the research aim s to discover a domain of p henomena concerning a smaller number of underlying, latent variables. In the present study, common factor analysis, termed as EFA, was employed to understand the latent dimensions underl ying the social engagement sc ale. Several recommendations regarding sa mple size in factor analysis have been propose d stating i n terms of either the minimum necessary sample size, N, or the minimum ratio of N to the number of variables being examined p (MacCallum, Widaman, Zhang, & Hong, 1999). Comrey and Lee (1992) suggest ed a rough rating scale for adequate sample size in factor an alysis: 100 = poor, 200 = fair, 300 = good, 500 = very good, 1,000 or more = excellent. In terms of the guide for the N: p ratio, prior

PAGE 82

82 studies had diverse recommendations, such as being in the ra nge of 3 6 (Cattell, 1978), a minimum ratio of 5 (Gorsuch, 1983), or at least 10 (Everitt, 1975). On the other side, some empirical research indicated that the adequacy of factor analysis relies more on the data characteristics such as communaliti e s than on the sample size employed (MacCallum, Widaman, Zhang, & Hong, 1999). More importantly, i n the extensive review of scale development practices in the study of organizations, Hinkin (1995) conclude d able for sc ale 974). The statistics employed to assess model fit for the EFA and CFA included the robust Chi square, 2 Comparative Fit Index (CFI), Turker Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), a nd Standardized Root Mean Square Residual (SRMR). It is the weakness of Chi square statistics to be nearly always large and statistically significant for a complex model, especially when the sample size is large and variables are considerabl y skewed; thus less weight is given to the Chi square statistics compared to the other model fit indices (e.g., CFI, TLI, RMSEA, and SRMR). Regarding the cutoff criteria of goodness of fit indices, TLI or CFI statistics greater than .90 are considered as an del fit, and values greater than .95 are deemed as a 1999). It should be noted that the cutoff criteria o f CFI, TLI, RMSEA, and SRMR are only validated for CFA models, whereas CFI and TLI values closer to 1.00 and SRMR and RMSEA closer to zero are considered better for EFA models (Norberg, Wetterneck,

PAGE 83

83 Sass, & Kanter, 2011). In addition, these cutoff values ar e somewhat arbitrary and should serve as a rule of thumb rather than fixed criteria (Bollen, 198 9 ). There is no certain threshold of factor loadings which is usually an arbitrary choice according to different research domains with diverse subjects. Jresko g and Srbom (1993) suggested three criteria on factor loadings: 1) weak convergence requiring the elimination of indicators that did not have a significant factorial regression coefficient greater than 2.58 ( p = .01); 2) strong convergence forcing the eli mination of those indicators that are not substantial, for example, those whose standardized coefficient is less than .5 0 ; and 3) a selective elimination of indicators that least contribute to the explan ation of the model with the cut off value of R 2 less than .03. In addition, Floyd and Widaman (1995 ) recommended that factor loadings of all variables should be reported on all factors ; p articularly, i n EFA, factor loadings are generally considered to be meaningful when they exceed .30 or .40. Accordingly, t he researcher in the present study retained items with factor loadings above .50 for the exploratory factor analysis purpose thus taking a conservative approach To discover the dimensions underlying the proposed social engagement measures, this investiga tion employed the EFA procedure to analyze the 161 pretest responses and construct ed a scale on the basis of the resulting factor loadings using the data analysis program Mplus (Version 6. 0 ) There were two purposes involved in the EFA : 1) determining the number of common factors affecting a set of measures, and 2) assessing the strength of the relationship between each factor and each observed measure (DeCoster, 1998). Analyses were performed on a polyc horic correlation matrix using the maximum likelihood with mean and variance estimation procedure through an

PAGE 84

84 oblique Geomin rotation. The Geomin rotation was selected as it was designed to minimize cross loading, while reducing the interfactor correlation (Brown, 2001; Sass & Schmitt, 2010). By analyzing th e screen plots and goodness of fit indices, a series of models was estimated and compared, and a four factor model showed the best fit with a comparison of other four models, suggesting a scale with four underlying dimensions or levels of the s ocial engage ment construct T he other four models examined were nul l model or zero factor model, single factor model in which all nineteen i tems comprised one factor (Model A), two factor model (Model B), an d three factor model (Model C). A one factor model was first estimated resulting in a 2 = 706.17 with 152 df CFI = .650, TLI = .606, RMSEA = .150, and SRMR = .111. This lack of fit indicated that a single factor could not adequately explain the covariance among the indicators. The two factor and three factor stru cture s also did not result in an improved fit over the one factor model. The four f actor model, which resulted in a 2 = 220.50 with 101 df CFI = .924, TLI = .872, RMSEA = .086, and SRMR = .042, indicated a substantial improvement in the model fit. Howeve r, it should be noted that the fit indices showed that the four factor model fit the da ta adequately but not good The five factor model was also tested but none of the factor loadings was greater than .50 on the fifth factor. The model estimations and ite m intercorrelations are indicated in Table 4 2 and Table 4 3 respectively. Table 4 2 Models and goodness of fit indices by exploratory factor analysis Model 2 df CFI TLI RMSEA SRMR Null Model 1753.54 171 Model A One factor Model 706.17 152 .650 .606 .150 .111 Model B Two factor Model 404.83 134 .829 .782 .112 .064 Model C Three factor Model 296.37 117 .887 .834 .098 .048 Model D Four factor Model 220.50 101 .924 .872 .086 .042 Note: n = 161

PAGE 85

85 Table 4 3. Means, standard deviations, and correl ations for exploratory factor analysis M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1.RSS 2.00 1.18 1.00 2.MOB 1.86 1.24 .47** 1.00 3.UPL 2.23 1.31 .47** .50** 1.00 4.CHK 1.75 1.11 .57** .57** .45** 1.00 5.BKM 1.80 1.07 .52** .47** .52** .60** 1.00 6.WID 1.94 1.23 .59** .55** .52** .63** .77** 1.00 7.TWI_1 2.32 1.43 .49** .38** .33** .35** .31** .40** 1.00 8.TWI_2 2.17 1.31 .42** .38** .46** .45** .50** .57** .63** 1.00 9.TWI_3 2.09 1.27 .58** .58** .54** .58** .62** .69** .60** .76** 1.00 10.BLG_1 3.61 1.20 .17* .10 .17* .08 .09 .19* .30** .29** .21** 1.00 11.BLG_2 2.93 1.43 .42** .25** .33** .29** .30** .36** .36** .38** .42** .54** 1.00 12.FRM_1 3. 14 1.38 .29** .20* .25** .20* .23** .30** .29** .32** .34** .66** .59** 1.00 13.FRM_2 2.73 1.39 .37** .27** .34** .31** .32** .43** .33** .48** .48** .47** .78** .65** 1.00 14.SNT_1 3.53 1.28 .13 .14 .26** .24** .24** .23** .15 .27** .25** .14 .31** .22** .26** 1.00 15.SNT_2 3.20 1.40 .26** .21** .28** .29** .31** .33** .14 .27** .28** .23** .51** .30** .37** .65** 1.00 s RSS feeds or podcasts; MOB = using mobile phone to watch video clips, check photos an d text alerts, or play games relevant to the program(s); UPL = uploading or forwarding videos or photos relevant to the program(s); CHK = using check in apps for the program(s) in GetGlue, Foursquare TM Miso, Philo, or Starling, etc.; BKM = using social bo okmarks (e.g., Digg TM and Delicious) to tag (s) (including actors, writers, producers, etc.) in Twitter TM ; TWI_2 = reading the prog s tweets (including actors, writers, producers, etc.) in Twitter TM ; s tweets (including actors, writers, producers, etc.) in Twitter TM ; BLG_1 = reading blog posts relevant to the program(s) BLG_2 = wri online discussion forums s posts in online discussion forums; SNT_1 = a fan of the program(s) and sharing it (them) with friends in social networks (e.g., Facebook TM and Myspace TM s posts in social networks (e.g., Facebook TM and Myspace TM ). p < .05, p < .01 ( two tailed) n = 161

PAGE 86

86 Scale Description The four dimensions indicate the different manifestations of how tele vision audiences use an expanding array of social media platforms to engage with television activity, or str ategic offering that allows the media consumer to engage with a television time or time shifted (D 53). According to the previous explication of surrounding television content could be manifested into three types: 1) interacting with a core program content and/or ancillary content (vertical dimension), 2) interacting with other te levision viewers (horizontal dimension), and 3) interacting with the characters or celebrities related to the programs (diagonal dimension) ( Askwith, 2007; Russell, Norman, & Heckler, 2004b). Building upon the three social interaction pillars, the social e ngagement construct represents the nature of how television audiences take advantages of each social medium capability to develop a deep, perpetual engagement with television program content and related information, characters or celebrities, and other tel evision viewers over time The manifestations are identified as vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influence. The first dimension, vertical involvement, measures the degree in which television viewers actively use a range of social media platforms to be involve d with their favorite programs Their involvement may be manifest ed through such behaviors as: 1) using social bookmarks like Digg TM or Delicious to tag the program, 2) using widgets to embed s video clips or photos online, 3) using check in apps for the program in several entertainment focused social networks such as Foursquare TM Miso, Philo,

PAGE 87

87 Starling, or GetGlue, etc., 4) using a mobile phone to watch video clips, check photos and text alert s RSS feeds or podcasts, and 6) uploading or forwarding videos or photos relevant to the program. The vertical involvement dimension characterizes the participatory behavior in relatio n to the core content and/or ancillary content of a program. The involvement activities are more one way oriented but critical because they cover a range of touchpoints that an individual could have with the program content. The program involved could be t he core program content, and/or ancillary content which include new erall knowledge, such as critiques gossip about the stars, back scene interviews, and televis ion promos. T he second dimension, diagonal interaction, measures the degree of social interaction that viewers develop with characters or celebrities related to their favorite programs in a social media context. Such engagement behaviors are mainly facilitated by micro blog s such as Twitter TM writers, director, producers, or other professionals in Twitter TM s tweets relevant to the characters or celebrities in Twitter TM and 3) writing or commen Twitter TM Originally, television audiences were more likely to show interests in communicating with and being acknowledg ed trend is that tel evision viewers are increasingly taking a more active interest in television professionals who work behind the scenes (Askwith, 2007). Powered by the Internet relay chat function embedded in Twitter TM the participation of celebrities and characters

PAGE 88

88 of tel evision programs in the social media platform could prompt involvement from new audiences who might not be interested in show centered activities and further stimulate the two way social interaction. For example, over the past few years, broadcasters have been eager to embrace Twitter TM to provide new ways to drive viewer engagement O ne of the important new approaches is to recruit a long list of stars involved in their shows along with their overall social television campaigns. The third dimension, hori zontal intimacy, measures the extent to which individual viewers emotionally respond to a television program and the affection of the viewers toward the branded content with other audiences in social media environments. This dimension captures a deeper and more intimate connection between the viewers and the diegetic, narrative text depicted in a program through one or more of the following peer to peer social media activities: 1) reading blog posts relevant to the program, 2) writing or commenting on blo in online discussion forums. In recent years, a growing number of television programs have launched their own off icial message boards or related blogs in the major broadcast and cable network websites or local television station sites D evoted viewers have ed blog posti ng and commenting sponsore d by a third party. Through either approaches, the horizontal intimacy dimension characterizes a mode of engagement that satisfies the television program. Driven by the context ual and extratextual immersion, such peer to

PAGE 89

89 from other viewers s with the branded television program through his/her presence in the online communities The last dimension, horizontal influence, measures the degree of identification and belonging, as well as the extent of meaningful influence in the direction or outcome of television program ming in a peer related space like social networks (e.g., Facebook TM and Myspace TM ). Such typical activities include: 1) indicating to be a fan of the program by sharing it with friends in social networks (e.g., Facebook TM and Myspace TM ), and 2) sts in social networks (e.g., Facebook TM and Myspace TM ). The horizontal influence dimension involves peer to peer interaction between members of the program audiences and potential meaningful influence on non program audiences. Facilitated by the relations hip focus and the identity nature in social networks like Facebook TM to some degree indicates that the viewer draws upon the progr am as part of his/her self and social identity as well as adding meaning to his/her relationship with others. In addition, influential opportunities also exist in relation to the shows when the particular television ng, opinion input, and program sharing in social networks. This activity has a friends, regardless of whether they are members of the program audiences or not. In this sense, the television viewers be come an ambassador on behalf of the television brand to advocate, recommend, and finally personally promote certain television content. The exploratory factor loadings of the four social engagement dimensions are shown in Table 4 4 and its schematic repres entation is indicated by Figure 4 2.

PAGE 90

90 Table 4 4. Factor structure matrix by exploratory factor analysis Factor L oading Factor Item Vertical Involvement Diagonal Interaction Horizontal Intimacy Horizontal Influence Vertical Involvement I have used social bookmarks (e.g., Digg TM and Delicious) to tag the program(s). .877 .060 .042 .023 s video clips or photos online. .854 .003 .056 .002 I have used check in apps for the program(s) in Foursquare TM Miso, Philo, Starling, or GetGlue, etc.. .724 .032 .036 .042 I have used my mobile phone to watch video clips, check photos and text alerts, or play games relevant to the program(s). .614 .111 .027 .023 s RSS feeds or podcas ts. .582 .072 .151 .028 I have uploaded or forwarded videos or photos relevant to the program(s). .526 .105 .068 .043 Diagonal Interaction I am a follower of the program(s) (including actors, writers, producers, etc.) in microblogs (e.g., Twitte r TM ). .011 .722 .068 .044 s tweets (including actors, writers, producers, etc.,) in microblogs (e.g., Twitter TM ). .109 .770 .009 .071 I have written or commented on the pr s tweets (including actors, writers, producers, etc.) in microblogs (e.g., Twitter TM ). .416 .578 .009 .004 Horizontal Intimacy I have read blog posts relevant to the program(s). .167 .049 .716 .020 I have written or commented on blog posts relevant to the program(s). .042 .046 .790 .180 s posts in online discussion forums. .001 .037 .798 .023 I have written or commented on the progr s posts in online discussion forums. .085 .034 .793 .013 Horizontal Influence I am a fan of the program(s) and share them with my friends in social networks (e.g., Facebook TM and Myspace TM ). .016 .119 .039 .649 s posts in social networks (e.g., Facebook TM and Myspace TM ). .038 .051 .034 .973

PAGE 91

91 Figure 4 2. A model of social engagement with television content Scale Reliability Scale reliability focuses on the proportion of variance in a measure, which is attributable to the true score on the latent construct that is being measured (DeVellis, 1991). Churchill (1979) concluded that the reliabil ity of a measure is high when agree 65) O n the other hand a low coefficient alpha show items performs poorly in capturing the construct which 68). C oefficient alpha is normally used to estimate the reliability of a multi item reflective scale by Churchill (1979) also proposed that coefficient alpha should be cal culated for each dimension as well as the whole construct during the measure purification stage. Specifically, this study chose the

PAGE 92

92 most commonly accepted measure to estimate internal consistency reliability for the whole s cale and its four dimensions. The suggested was greater than .70 ( Kline, 20 11 ; Nunna lly, 1978 902 and the values for the four dim ensions i.e., vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influence were .8 76, 852 .8 65, and .78 2 respectively This suggest s that the fifteen sample items perform ed well in capturing the proposed social engagement c onstruct indicated by high intern al consistency reliability. In addition, according to the EFA results and reliability test four proposed items describing viewing the program in video sharing communities (e.g., YouTube ), joining ommunities, sending mobile messages about the program, and submitting ratings/reviews/votes about the program were removed from the social engagement measure scales. Stage Two : Social Engagement Scale Confirmation Confirmatory Factor Analysis Having esta blished the fifteen item social engagement scale through exploratory f actor analysis, this study proceeded to the second stage and collected another large set of consumer data (n = 494) reliability CF A requires a sample size of five to ten times the number of items or the minimum necessary sample size (good = 300), hence the current sample size of 494 satisfied both criteria. The CFA was performed on the 494 responses with the item correlation matrix a s input and maximum likelihood as the model estimation technique. T he intercorrelations of the fifteen indi cators were all significant rang ing from .24 to .81. Skewness for the scale item s ranged between .795 and .981 and Kurtosis between

PAGE 93

93 1.361 and .332 (both within the recommended 2 to +2 range), which suggested reasonably good distribution properties for the empirical data. By conducting the CFA procedure, a series of models was estimated and the four factor model was validated (Table 4 5) All model s were compared to a null model in which each manifest indicator was treated as an independent, orthogonal component. A single factor model, comprising fifteen items to tap a common construct, was estimated, resulting in a 2 = 1303.90 with 90 df CFI = .7 79, TLI = .742, RMSEA = .165, and SRMR = .090. The lack of fit indicated that a single factor could not adequately explain the covariance among the indicators. The four factor model (Model B) in which correlations between factors were allowed resulted in a substantial fit with a 2 = 377.79 with 84 df CFI = .946, TLI = .933, RMSEA = .084, and SRMR = .042, which suggested that the model fit the data adequately. As this study regarded social engagement as a higher order construct explained by a number of rel ated dimensions, this investigation further tested a higher order model (Model C), where a second order factor represented the overall construct of social engagement. As shown in Table 4 5 this model fit s the observed data also adequately but not better t han the four factor model. While both the first order and second order representation s of the social engagement construct fit well with the observed data, the first order model is preferred by virtue of its simplicity and better fit. Therefore, the present study retained both the four factor model and the four factor model with on e higher order factor as the social engagement scale for the subsequent antecedents and consequences testing Construct validity for each scale was assessed by examining the stand ardized CFA factor loadings for its hypot hesized items which were originally derived from the

PAGE 94

94 EFA procedure For acceptable construct validity, it is proposed that each item should have a minimum factor loading of .60 on its hypothesized latent factor (Nun nally, 1978) The norm was met for both the four factor Model and the four factor model with one higher order factor All factor loadings in these two models were greater than .60 with a significant level at p < .001 through two tailed test. For the vertic al involvement dimension, the program (UPL) had the lowest factor loading compar ed to other five items, suggesting that the item was acceptable but not perfectly indicative of the latent construct. Regarding the horizontal influence dimension, the variable a fan of the program and share it with my friends in social networks (SNT_1) had a lower factor loading in relation to the second indicator (SNT_2), which means that the a for the dimension of horizontal influence. In addition the factor loading compar isons on the four dimensions in the Model C suggest that the social engagement behavior was more characterized by the one way participatory involvement activities and social interaction with characters / celebrities than the influencing activities among audience members. T he correlation matrix with means and standard deviat ions for the CFA procedure are presented in Table 4 6 The confirmatory factor loadings are presented in Table 4 7 and the path models are indicated in Figure 4 3 Table 4 5 Models and goodness of fit indices by confirmatory factor analysis Model 2 df CFI TLI RMSEA SRMR Null Model 5589.62 105 Model A One factor model 1303.904 90 .779 .742 .165 .090 Model B Four factor model 377.785 84 .946 .933 .084 .042 Model C Four factor model with one higher order factor 486.052 86 .927 .911 .0 97 .063 Note: n = 494

PAGE 95

95 Table 4 6 Means, standard deviations, and correlations for confirmatory factor analysis M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1. RSS 2.22 1.27 1.00 2. MOB 2.32 1.44 .65** 1.00 3. UPL 2.51 1.40 .55* .59** 1.00 4. CHK 2.03 1.29 .62** .60** .51** 1.00 5. BKM 2.17 1.29 .67** .57** .54** .68** 1.00 6. WID 2.21 1.29 .66** .62** .57** .69** .73** 1.00 7. TWI_1 2.59 1.43 .56** .49** .43** .56** .58** .54** 1.00 8. TWI_2 2.47 1.39 .56** .54** .45** .62** .67** .67** .66** 1.00 9. TWI_3 2.27 1.34 .66** .62** .53** .70** .77** .72** .62** .81** 1.00 10. BLG_1 3.37 1.36 .33** .34** .34** .25** .37** .37** .38** .42** .39** 1.00 11. BLG_2 2.80 1 .40 .50** .52** .51** .47** .56** .58** .41** .52** .60** .67** 1.00 12. FRM_1 3.13 1.40 .46** .47** .42** .39** .45** .48** .38** .49** .49** .70** .6 9 ** 1.00 13. FRM_2 2.68 1.38 .52** .50** .52** .50** .56** .61** .44** .52** .59** 60 ** .80** .72 ** 1.00 14. SNT_1 3.56 1.31 .30** .31** .38** .24** .31** .34** .27** .34** .31** .41** .43** .45** .4 2 ** 1.00 15. SNT_2 3.14 1.44 .39** .42** .52** .35** .45** .49** .32** .43** .47** .4 9 ** .64** .52** .6 1 ** .63 ** 1.00 Note: p < .05, p < .01 ( two tailed) n = 494

PAGE 96

96 Table 4 7 Fact or structure matrix by confirmatory factor analysis Note: VINVOL = vertical involvement, DINTER = diagonal interaction, HINTIM = horizontal intimacy, HINFLU = horizontal I nfluence All factor loadings are significant at p < .001 level (two tailed) n = 4 94 Standardized Factor Loading Four factor m odel Four factor model with one higher order factor Manifest indicator Vertical Involvement Diagonal Interaction Horizontal Intimacy Horizontal Influence Vertical Involvement Diagonal Inte raction Horizontal Intimacy Horizontal Influence Second order RSS .787 .787 MOB .742 .744 UPL .668 .675 CHK .795 .791 BKM .849 .846 WID .853 .855 TWI_1 .702 .701 TWI_2 .866 .868 TWI_3 .930 .929 BLG_1 .727 .726 BLG_2 .899 .895 FRM_1 .804 .807 FRM_2 .886 .889 SNT_1 .672 .668 SNT_2 .931 .936 VINVOL .966 DINTER .926 HINTIM .768 HINFLU .632

PAGE 97

9 7 Figure 4 3 Four factor path model and four factor with one higher order factor path mode l

PAGE 98

98 Discriminant Validity Discriminant validity is provided by the lack of a significant relations hip with constructs that should not, nomologically, be related. Specifically, t o investigate nomological validity of the proposed social engagement scale, this study employed simple correlations between social engagement and the measures of television prog ram affinity and program involvement to validate the scale Program affinity, the attitudinal construct, has been used extensively to reflect the perceived importance of watching (Rubin, 1983; Rubin & Perse, 1987 a ) As Russell, Norman, and Heckler (2004 b ) suggested, television audiences could develop positive attitude or affinity toward a program in a short period of time, whereas program engagement or connectedness may take time to form and develop. Th erefore, the authors concluded that a positive attitude toward a program may mediate the development of connectedness or engagement but the attitude construct cannot capture the parasocial relationship formed between audiences and their favorite televisio n programs. Likewise, the current study proposed that program affinity and social engagement with television program s are separate and distinct constructs. Involvement is another often misused construct with engagement ( or connectedness ) In the tele visio n consumption environment, program involvement reflect s personal relevance of a television program to the audiences (Park & McLung, 1986). Russell, Norman, and Heckler (2004 b ) also posited that television programs may start by fostering lasting involvement with the program, over the course of repeat ed viewing, but end up absorbing their audiences in a deep, perpetual, and connected relationships with the core program content, ancillary information, characters and celebrities and/or other television audienc e members Accordingly, program

PAGE 99

99 involvement was also proposed in this study to test the nomological validity of the social engagement construct. The test of disciminant validity was performed by calculating correlation coefficients and 95% coefficient int ervals between social engagement, program affinity, and program involvement. b ) four item program affinity shows, resulting in semantic differential items were applied on a five point scale to measure program involvement with .913 coefficient alpha. The social engagement scale from s alpha. Given the large sample size, all correlations were statistically significant at the .01 level and examined for their practical significance. The correlation between social engagement and program affinity was .376, with a 95% confidence interval be tween .298 and .449. This small correlation between television program social engagement and program affinity provides initial evidence of discriminant validity. The correlation between social engagement and program involvement was .236, with a 95% confide nce interval rang ing from .151 to .318. All the correlation analys e s provide d strong evidence that neither program affinity nor program involvement should, by theory, be related to program social engagement and they are not. Stage Three : Antecedents and Consequences Tests Main Test The pilot test offered a theoretical rationale for the proposed social engagement construct, the main test was conducted to confirm the scale and test its antecedents and consequences by surveying the online consumer panel of 494 qualified

PAGE 100

100 respondents. The online survey instrument for the main test was also facilitated by the online survey program Qualtrics and administrated by the uSamp TM company Similar to the pilot test, the first screening question in the main test focuse d on the social media use experience. Different from the second screening question in the pretest, the main test asked respondents whether they ever used their chosen social media to comment, post, read, and read anything about a specific show from a twenty program list provided by the researcher. If the respondents did not have experience with any of the social media platforms nor ever used their chosen social media to interact with the listed television program s they were disqualified to con tinue the main survey. Following the two screening questions, as a means of assessing the social behavior taking place around television program s the survey asked the respondents when they typically used social media to comment, post, watch, or read an ything about these programs. The main survey then instructed the respondents to choose one program among their socially engaged shows as their favorite, and the subsequent set of questions w as based on the favorite television show. Specifically, using a fi ve Likert scale, the fifteen social engagement items, along with the program affinity scale and genre preference measures, were constructed. This set was followed by seven semantic differential items using a five point scale to measure program involvement. The next set of questions w as also based on the respondent s program using five point scales, including program behavioral loyalty and program attitudinal loyalty (1 = strongly disagree, 5 = strongly agree), program satisfaction (1 = n ot at satisfied, 5 = very satisfied), and product purchase likelihood (1 = definitely not, 5 = definitely).

PAGE 101

101 The main survey assessed the perceived characteristics of social media by asking ence about the general social media concept. Perceived ease of use and compatibility were measured using a five point Likert scale, while five semantic differential items using a five point scale ( unsociable/social, impersonal/personal, insensitive/sensiti ve, cold/warm, and passive/active ) to measure perceived social presence. The next sets of questions innovativeness, and their offline social characteristics. In particular a total of 49 motive items with a five point Likert scale (1 = strongly disagree, 5 = strongly agree) were constructed, covering diverse motives behind television viewing (Rubin, 1983), the Internet use ( Papacharissi & Rubin, 2000), and YouTube video viewing (Harid akis & Hanson, 2009). Along with the motivation measures, the respondents were further asked to assess their innovativeness toward the social media platforms, and to indicate their offline social characteristics such as inter personal interaction and social activities. Finally, media ownership and use along with basic demographic information was collected as described in the pilot test. Television Program Sample The main test selected twenty primetime television program s delivered through broadcast and ca ble networks B ased o n the five most popular genres indicated in the pilot test results, these program genres were reality shows, drama, game/talk shows, sitcoms, and animated comedies. The networks included all eight national English language broadcast ne tworks, ABC, CBS, NBC, FOX, CW, MyTV, ION, and PBS, as well as the top twenty five cable programming networks by subscriber counts. The specific program list was composed by referring to an online database, Social

PAGE 102

102 Television Charts (http://trendrr.tv/), wh ich is a comprehensive television index that incorporates multiple social and syncopated data sources tracking all major networks and shows. The index includes such social media activit ies as public Facebook TM posts, Twitter TM mentions, GetGlue check ins, and Miso check ins. The index is updated daily and weekly, providing insights and better understanding of the tightly coupled two screen synchronous social behavior taking place around television. By referring to the social television index from August 29 t h to September 4 th of 2011, the week before the main survey was implemented; th e present study constructed the final list of primetime network programs which is presented in Table 4 8 Table 4 8 The list of primetime network programs in the main test Pr ogram Title Broadcast/Cable Network Program Genre Big Brother CBS Reality show Keeping Up with the Kardashians E! Entertainment Reality show Jersey Shore MTV Reality show Teen Mom MTV Reality show Glee FOX Drama NCIS CBS Drama The Vampire Diaries The CW Drama Pretty Little Liars ABC Family Drama Gossip Girl The CW Drama True Blood HBO Drama How I Met You Mother CBS Sitcom The Office NBC Sitcom The Big Bang Theory CBS Sitcom NBC Game show Monday Night Raw USA Game sh ow Conan TBS Talk show Family Guy FOX Animated comedy South Park Comedy Central Animated comedy The Simpsons FOX Animated comedy

PAGE 103

103 Measures The current study employed multiple items validated from prior studies to measure the theoretical constructs exc ept for program behavioral loyalty and audience satisfaction. Although there was no empirical agreement on which is superior between multi item measures or single item instrument s most researchers preferred to employ multiple item measures when requiring respondent self reports of attitudes, beliefs, perceptions and the like (Gard n er, Cummings, Dunham, & Pierce, 1998). Program genre preference Prior research on program genres pointed out that program type is a reasonably valid an d reliable scheme to use for categorizing television content (Bielby & Bielby, 1994; Cohen, 2002; Gitlin, 1983). Through analyzing the television programs identified in the pretest sample, the current study chose the five popular genres that audiences used social media platforms most to comment, post, watch, or read about, including reality shows, drama, game/talk shows, sitcoms, and animated comedies. In addition, based on the specific program that the respondent selected in the main test, this study evalu ated the participant s the program genre that the specific show belongs to. As discussed in the theory of television program choice model, program genre preference is viewed as one of the basic premises in program choice; however, h ow to operationalize it through a valid and systematic measurement scale is still needed in the audience behaviorist research tradition. Some television program genres studies used the amount of attention paid in watching shows of particular genres as the basis for viewer genre preference (Hawkins, et. al., 2001; Moyer Gus, 2010), while others Moyer Gus, 2010) or watching

PAGE 104

104 likelihood ( Rubin, 1983) to approach the concept. Thus, the present study adapted two sta tements focused on viewing attention and enjoyment experience using a five point Likert scale scale (1 = not at all, 5 = extremely). Specifically, the respondents were asked to assess how much attention or how much enjoy ment they experienced when watching each of the following types of programs: reality shows, drama, game/talk shows, sitcoms, and animated comedies. Program affinity Two sets of measures of Televisi on Affinity Scale (Rubin, 1983) and program affinity (Rubin & Perse, 1987 a, 1987b ) were adapted about their favorite television shows with which they interacted using various social media platforms. The three item affinity scale was used to assess how important and how much affinity the respondents felt watching the ir favorite shows using statements such The r espondents were asked to indicate their level of agreement with each of the statements using a five point Likert scale (1 = strongly disagree, 5 = strongly agree). Program involvement To assess the personal cognitive, affective, and functional dimensions of involvement with a particular television program, seven semantic differential items were applied on a five point scale including irrelevant/relevant, means nothing to me/means er /matters to me, uninterested/interested, insignificant/significant, superfluous/vital, and nonessential/essential (Park & McClung, 1986).

PAGE 105

105 Perceived ease of use The cons individual believes that using a particular system would be free of physical and m ental 323). Three items were adapted from prior studies to assess perceived ease of use of a general social media system in terms of learning, skillfulness, and usage through a five point Likert scale (1 = strongly disagree, 5 = strongly agree) (Davis, Bagozzi, & Warshaw, 1989; Wu & Wang, 2005). Compatibility Perceived compatibility meas technology is compatible with existing values, past experiences, and needs of potential 15). This study used three items borrowing from Tronataky and Klein (1982), Chen, Gillenson, and Sherrell (2002), and Chan Olmsted and Chang (2006). A five with each of the statements assessing the variable of perceived compatibility with social media system s in general. Social presence Social presence is defined as co presence, co or psychological involvement ( Biocca, Harms, & Burgoon, 2003 ). The construct was measured by using a semantic differential technique on bipolar items such as unsoci able/sociable, impersonal/personal, insensitive/sensitive, cold/warm, and passive/active (Papacharissi & Rubin, 2000; Short, Williams, & Christie, 1976). Social media having a high degree of social presence were judged as being sociable, personal sensitive warm, and active. The present study constructed a social presence index by summing and averaging the five responses.

PAGE 106

106 Innovativeness media systems. The domain specific scale reflected the tendency to learn about and adopt innovations within a specific domain of interest, which was found to be a valid and reliable measure for different innovations in transnational settings ( Goldsmith & F lynn, 1992 ). The present study modified the six items to reflect the social media context and asked respondents to rate their level of agreement with each statement using a five point Likert scale, including ast in my circle of friends to know the 1, 3, and 5 were reversely cod ed. Motives The social and psychological needs of using social media to interact with television content were mainly driven by the television program itself, the Internet, and diverse online applications. Therefore, the current study compiled forty nine i tems of motives behind television viewing (Rubin, 1983), the Internet use (Papacharissi & Rubin, 2000), and YouTube video viewing (Haridakis & Hanson, 2009). The final set of the 49 item scale represented a range of motives identified by prior studies, in cluding relaxation, companionship, habit, pass ing time, entertainment, social interaction, information seeking, arousal, escape, convenience, and personal utility. Specifically, this study asked the respondents to indicate how much each of the forth nine m otive statements was like their own reasons behind using various social media platforms to engage with

PAGE 107

107 television content through a five point Likert scale (1 = strongly disagree, 5 = strongly agree). Social characteristics Adapted from the previous stud ies on contextual age scales (Rubin, 1986; Rubin & Rubin, 1982, 1989), the present study measured two dimensions of social characteristics of the respondents, including the level of interpersonal interaction and offline social activities. The respondents r ated their level of agreement (1 = strongly disagree, 5 = strongly agree) with four statement s assessing their interpersonal statements to measure their offline social meetings or activities of clubs, lodges, recreation centers, churches, or other Program loyalty The present study operationalized program loyalty as attitudinal program loyalty and behavioral p rogram loyalty. Regarding the attitudinal loyalty measure, adapted from the scales of brand loyalty (A a ker, 1991), service loyalty (Ganesh, Arnold, & Reynolds, 2000), and store loyalty (Campo, Gijsbrechts, & Nisol, 2000), this study used a five point Liker level of agreement with three statements. The original items were modified according to the different attributes involved in television content consumption The three item scale focused on stated recommendation s preference s or probabilit ies of viewing by the audiences, thus emphasizing the cognitive element of program loyalty. The behavioral dimension of loyalty is usually operationalized as the actual purchases observed over a certain time period, as this actual purchase measure is

PAGE 108

108 directly related to the performance and existence of the firm (Melles, Dekimpe, & Steenkamp, 1996). Applying the operational definition to television program behavioral loyalty, this study asked the regarding their favorite shows over a specific time period. Specifically, the study asked the any episodes of the program when they broa point Likert scale (1 = strongly disagree, 5 = strongly agree). Audience satisfaction This study defines audience satisfaction as a global evaluative summary of (direct) viewing experience with particular television content. The respondents marked how satisfied they were when they watch ed a specific television program, using one item point semantic differential scale, the two anchors being not at all satisfied (1) and very satisfied (5). Product purchase likelihood There are usually two types of merchan dise available on television station/networking sites: fan based items and non fan based items. The fan based items are items relevant to the networks or its shows and stars, while non fan based items are essential products of the advertisers of the broadc ast/cable networks (Ha & Chan Olmsted, 2001). he two types of products were dis covered by three items using a five point Likert scale (1 = definitely not, 5 = definitely) to assess whether t he respondents would be more likely to buy memorabilia

PAGE 109

109 /merchandise either related to the television station/network or the television show/stars, as well as products shown in the program. Media use and demographic information This study incorporated me dia use and demographic information as two types of control variables. Considering the technology driven nature of the research, several technology ownership and media use variables related to online social media and video consumption were measured. Specif ically, respondents were asked to indicate whether they own a digital video recorder (DVR) and other new communication technologies at home, using dichotomous yes or no response categories. In addition, respondents were ask to check their media use such as the types of Internet service (i.e., dial up or high speed Internet) and television subscription types (i.e., over the air broadcasting only, basic cable, premium cable, satellite television, and the Internet Protocol Television (IPTV)). Several demograph ic factors such as age, gender, income, education, ethnicity employment, and marital status were included at the end of the survey. Table 4 9 summarized the constructs included and their operational definitions. Table 4 9 Constructs and operational defi nition Construct Operational Definition Source Program genre preference How much at tention do you typically pay when you watch each of the following types of programs: reality shows, drama, game/talk shows, animated comedies, and sitcoms? How much do yo u enjoy watching each of the following types of programs: reality shows, drama, game/talk shows, animated comedies, and sitcoms? Moyer Gus (2010) Program affinity I would feel lost without the program to watch. m, I really miss it. Watching the program is one of the most important things I do each day or each week. Rubin (1983); Rubin & Perse (1987 a, 1987b )

PAGE 110

110 Table 4 9 Continued Construct Operational Definition Source Program involvement Irrelevant Releva nt Means nothing to me Means a lot to me r Matters to me Uninterested Interested Insignificant Significant Superfluous Vital Nonessential Essential Park & McClung (1986) Perceived ease of use Learning to use social media to commen t, post, watch, or read anything about the television program is easy for me. It is easy for me to become skilled at using social media to comment, post, watch, or read anything about the television program. It is easy to use social media to comment, post, watch, or read anything about the television program. Davis (1989); Davis, Bagozzi, & Warshaw, (1989); Wu & Wang (2005) Compatibility Using social media to comment, post, watch, or read anything about the television program is compatible with most aspe cts of my television viewing. Using social media to comment, post, watch, or read anything about the television program fits my lifestyle. Using social media to comment, post, watch, or read anything about the television program fits well with the way I li ke to engage in television viewing. Chan Olmsted & Chang (2006); Chen, Gillenson, & Sherrell (2002); Tronataky & Klein (1982) Social presence Unsociable Sociable Impersonal Personal Insensitive Sensitive Cold Warm Passive Active Papacharissi & Rubin (2000); Short, Williams, & Christie (1976) Innovativeness In general, I am among the last in my circle of friends to use a new social media platform when it appears. If I heard that a new social media platform was available online, I would be i nterested enough to try it. Compared to my friends I use few of the social media platforms. heard of it yet. In general, I am the last in my circle of friends to know the names of the latest so cial media platforms. I know more about new social media platforms before other people do. Goldsmith & Hofacker (1991); Goldsmith & Flynn (199 2)

PAGE 111

111 Table 4 9 Continued Construct Operational Definition Source Motives Because it relaxes me Because it allows me to unwind Because it makes me feel less lonely Because I just like to use it When I have nothing better to do bored Because it gives me something to do to occupy my time Because it entertains me enjoyable Because it amuses me ends come over So I can be with other members of the family or friends who are watching the program Because it helps me to learn things about myself and others done before So I could learn about what could happen to me Because it peps m e up So I can forget about school or other things So I can get away from the rest of the family or others So I can get away from what I Because I wonder what other people said Because I can meet people with my interest Because it easier to get information Because I can search for information Because I can get information for free Because it provides a new and interesting way to do research So I can keep up with current issues and events So I can see what is out there So I can learn about useful things So I can learn about unknown things Because they are convenient to use Because I can get what I want for less effort Haridaki s & Hanson (2009); Papacharissi & Rubin (2000); Rubin (1983)

PAGE 112

112 Table 4 9 Continued Construct Operational Definition Source Motives Because I can use it anytime, anywhere Because I want to show others encouragement Because I want to communicate with friends and family Because I want to belong to groups with the same interests as mine Because I want to let others know I care about their feelings Because I can express myself freely Because I can talk a s long or as short as I want Because I can participate in the discussion Because I can meet new people Haridakis & Hanson (2009); Papacharissi & Rubin (2000); Rubin (1983) Social characteristics: Interp ersonal interaction I g e t to see my friends as of ten as I would like I spend enough time communicating with my friends and family by telephone or mail I have ample opportunity for conversations with others I can always find someone to speak with when I need to talk. Rubi n (1986); Rubin & Rubin (1982 1989) Social characteristics: social activity I often travel, vacation, or take trips with others. I often visit with friends, relatives, or neighbors in their homes. I often participate in the meetings or activities of clubs, lodges, recreation centers churches, or other organizations. I often go places to socialize with others. I often participate in games, sports, or activities with others. Rubin (1986); Rubin & Rubin (1982, 1989) Program behavioral loyalty Over the past month, I have not misse d any episodes of the program when they broadcast on television. Constructed by the author Program attitudinal loyalty I would recommend the program to others. I think of myself as a loyal viewer of the program. I would be willing to watch the program r ather than other shows. Aaker (1991) ; Ganesh, Arnold, & Reynolds ( 2000) ; Campo, Gijsbrechts, & Nisol ( 2000 ) Audience satisfaction Overall, how satisfied were you with watching the program? Ferguson & Perse (2004)

PAGE 113

113 Table 4 9 Continued Construct Opera tional Definition Source Product purchase likelihood If any of the following items were available in the would you be more likely to buy them? (1) Memorabilia/merchandise of the television station/ne twork (2) Memorabilia/merchandise of the television show or stars (3) Products shown in that television show Ha & Chan Olmsted (200 1) Note: indicates the items that are reversely coded Descriptive Statistics of Main Test The online survey was active for two weeks (between September 9 and 23, 2011 ) A total of 1,427 individuals responded to the main test and 494 were qualified to complete th e whole survey, yielding a 34.6% incident rate. Regarding the first screening the social media platforms is the same with the pilot test results. Specifically, Faceboo k TM is the most commonly used social media tool with the highest penetration rate of 93% (n = 1,328), followed by YouTube (52%, n = 746), Myspace TM (39%, n = 559), Twitter TM (30%, n = 421), and mobile texting and applications (16%, n = 226), respectively. R egarding the social media usage frequency, the results were also identical with the pretest, showing that mobile texting and applications are most frequently used ( M = 4.02, n = 1,314), followed by Facebook TM ( M = 4.02, n = 1,314), YouTube ( M = 3.40, n = 735), and online discussion boards/forums ( M = 3.37, n = 176). For the second screening question, the study asked the respondents to identify whether they had used their chosen social media to comment, post, read, or read anything about the listed twe nty television programs. A mong the total of 1 314 online

PAGE 114

114 social media users, around 41 % of them had utilized social media to interact with these television programs at least once The most socially involved shows were NCIS ( n = 15 0) and nt ( n = 146), whereas Monday Night Raw and Pretty Little Liars invoked the least social activities around these shows. Among the se social ly involved programs, the social media users were further asked to select one as their favorite. Table 4 10 present s th e order of the socially involved television program s and their favorite r ank ing s based on the number of social media users Table 4 10. The order of television programs by social media users The Number of Social Media User The Favorite Number Genre Pr ogram Title Broadcast/ Cable Network 150 68 Drama NCIS CBS 146 63 Game show Got Talent NBC 145 33 Animated comedy Family Guy FOX 127 25 Animated comedy The Simpsons FOX 122 37 Drama Glee FOX 122 43 Drama True Blood HBO 119 22 Animated comedy South Park Comedy Central 111 43 Sitcom The Big Bang Theory CBS 85 19 Sitcom How I Met You Mother CBS 82 27 Reality show Big Brother CBS 81 23 Reality show Jersey Shore MTV 67 12 Reality show Teen Mom MTV 67 8 Sitcom The Office NBC 61 14 Drama The Vampire Diaries The CW 60 10 Reality show Keeping Up with the Kardashians E! Entertainment 55 11 Talk show Conan TBS 51 7 Drama Gossip Girl The CW 45 12 Game show Monday Nig ht Raw USA 40 10 Drama Pretty Little Liars ABC Family

PAGE 115

115 As a means to assess the social behavior pattern, the survey next asked the respondents when they typically used their selected social media t o interact with these programs. The pattern was simi lar but not totally the same with the results revealed in the pilot test, 57% respondents using social media to interact with the shows after the y watch ed the programs 23% respondents while the y watch ed the programs and 20% before they watched the progra ms However, the pretest results showed that the pattern was after (30%), before (17%), and during (12%). In addition, the findings from a national survey conducted between March 11 th and 15 th 2011 (Harris Interactive, 2011) revealed that, among these 80 some million people, 33% had the social media experience after watching a television show and that fewer had done so before watching (18%) or while watching (17%) a television program. Summarily, the present study along with the national survey indicated that the majority of online adults prefer to use various social media platforms to interact with television program s after they watch ed the shows Part icipants There were 484 respondents who completed the demographic information in the main test and 161 respondents in the pilot test. The profile of the participants in these two tests is described and compared below The average age in the main test was 38.62 ( SD = 15.28 ) a younger sample than those in th e pretest with mean age of 43.01 ( SD = 16.45 ) Male s (n = 151) account ed for 3 0 6 % while females were 6 7.4 % (n = 333) i n the main test. Although the gender structure was shown to be s imilar in the main test, there was a higher percentage of females (80.1%) than males (19.9%) in the pretest. When it comes t o ethnicity in the main test, while white Caucasians accounted for 74.5%, African Americans and Asians had the same weight (7.4%), followed by

PAGE 116

116 Latino/Latina/Hispanics (6.0%). The ethnical structure in the pretest showed a little different result with white Caucasians ma king up for 83.2%, followed by African Americans (6.2%), Hispanics (4.3%), and Asians (2.5%), respectively. yearly household income s were under $30,000, while 28.9% ranged between $30,000 and just under $50,000. The rest of the 31.2% respondents had an income above $50,000. Regarding the education level in the main test, out of 484 respondents, 3 5.8 % of the participants completed some college (n = 177). Anothe r 3 1.8 % and 27. 1 % h e ld college graduate degree s or more (n = 157) and high school diplomas (n = 134), re spectively. While 30.8% of the respondents in the main test were full time employ ee s outside the home, the second biggest section was from retired peopl e (13.2%). Both tests showed that the majority of the respondents were married. In addition, a similar sample structure in household income education level and employment status was also indicated in the pretest. When it comes to media ownership and usa ge in the two tests, the most commonly owned new media technology among the respondents was a DVD player with an approximate 88 .0 % penetration rate, followed by cell ul ar phone, computer, videogame systems, HDTV, and iPod or other portable MP3 players, resp ectively. The least owned media technologies were the newly emerg ed tablet (e.g., iPad) and portable video player (e.g., video iPod). With respect to media use in the main test 4 8 .0 % of the respondents said they subscribe d to basic and expanded basic cabl e television service, whereas 3 7.2 % of the respondents had satellite television service. The over the air only and IPTV services (e.g., U verse and FiOS) accounted for 13. 2 %

PAGE 117

117 and 7.9 %, respectively. The majority of the respondents (9 2. 3 %) had high speed Int ernet connection at home, whereas the dial up connection made up for 6. 5 % and no Internet connection was 1. 2 %. Table 4 11 presents t he demographics and media usage in the two tests Table 4 11. The comparison of demographics and media usage in the two tests Main Test Pilot Test % n % n Gender Men 30.6 151 19.9 32 Women 67.4 333 80.1 129 Age 18 29 35.5 172 28.6 46 30 49 37.2 180 32.3 52 50 64 21.3 103 28.6 46 65+ 6.0 29 10.6 17 Race/Ethnicity White 74.5 368 83.2 134 African American 7.3 36 6.2 10 Hispanic 5.9 29 4.3 7 Asian 7.3 36 2.5 4 Household Income Less than $30,000 37.9 187 39.8 64 $30,000 $49,999 28.9 143 25.5 41 $50,000 $74,999 16.0 79 17.4 28 $75,000+ 15.2 75 17.4 28 Education Level Less than high sch ool 3.2 16 2.5 4 High school graduation 27.1 134 24.2 39 Some college 35.8 177 37.3 60 College+ 31.8 157 36.0 58 Employment Status Employed outside the home full time (30 hours or more per week) 30.8 152 27.3 44 Employed outside the home part tim e (1 29 hours per week) 12.6 62 9.9 16 Retired 13.2 65 21.1 34 Full time homemaker 13.0 64 16.8 27 Temporarily unemployed 11.9 59 8.7 14 Full time student 10.3 51 9.3 15 Other 6.4 31 6.8 11 Marital Status Married 45.3 224 48.4 78 Single, nev er married 36.2 179 42.9 69 Other 16.4 81 8.7 14 Television Subscription Over the air only 13.2 65 8.7 14 Basic and expanded basic cable 48.0 237 54.0 87 Satellite 37.2 184 29.8 48 IPTV (e.g., U verse, FiOS) 7.9 39 6.8 11 Internet Connection Hi gh speed 92.3 456 97.5 157 Dial up 6.5 32 1.9 3 No Internet connection 1.2 6 1.2 2

PAGE 118

118 Data Analysis Strategy Besides the EFA and CFA procedures discussed in the stages of social engagement scale development and validation, several other statistics test s were used in this study to answer the research questions and evaluate the hypotheses Specifically, the following statistics were used: multiple planned comparisons for the five different television genres along each social engagement dimension, the EFA procedure for the motives behind social engagement behavior, and a structural equation modeling (SEM) for the antecedents and consequences tests. Social engagement with different television genres The RQ2 asked whether the overall social engagement or t he different dimensions of social engagement with the program vary among different genres (i.e., drama s reality shows, sitcoms, game/talk shows, and animated comedies). To construct the social engagement index of each television program genre, this study first asked the respondents to indicate one particular show with which they have utilized social media next recoded the program into different program genres. For exa favorite show was among the following s : Glee, Criminal Minds, NCIS, Gossip Girls, Pretty Little Liars, True Blood, and The Vampire Diaries this study recoded it as the drama category. If the respondent selected Big Brother, J erse Talent, Keeping Up with the Kardashians, or Teen Mom the present study recoded it as the reality s how genre. After classifying each favorite show into different types of programming this study calculated the means and standard dev iations of different program genres along each social engagement dimensions.

PAGE 119

119 A s discussed in the stages of scale development and validation, w hile both the first order and second order representation of the social engagement construct fit well with the ob served data, the first order model was preferred by virtue of its simplicity and better fit. In addition, the EFA and CFA results in the first two research stages indicate d that a single factor model, comprising fifteen items to tap a common construct, cou ld not adequately explain the covariance among the indicators. The present study therefore preferred not to use the sum and average s approach to t he fifteen items when construct ing the social engagement index for the following planned comparison s analys e s. Thus this study evaluated specifically how the five different television genres vary along each social engagement dimension including vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influence. The multiple comparisons for one factor designs were planned to answer the RQ2 The one factor was each dimension of the social engagement construct in the current study context. Planned contrasts between any two television genres rather than Post hoc comparisons were performed. The reason why planned (a priori) tests were chosen rather than Post hoc contrasts are : 1) planned comparisons are more powerful than Post hoc tests; 2) planned comparisons are more specific than the omnibus test which usually uses a one way analysis of varia nce ( ANOVA ) procedure to evaluate the omnibus hypothesis; and 3) planned comparisons are more targeted, which means only a few of all possible comparisons are needed (Myers & Well s 2003) Therefore, t he familywise error was selected for control. The Shaff er Holm procedure was used so that the pairwise comparisons would be reasonab ly powerful and incorporate the omnibus test.

PAGE 120

120 Motivations behind social engagement behavior The RQ 3 investigated what motives audiences have for using social media to engage w ith television content. Different from prior television audience behavior studies (e.g., Haridakis & Hanson, 2009; Papacharissi & Rubin, 2000; Rubin 1983), the present study employed the EFA procedure rather than PCA to analyze the fo rty nine motive state ments. As discussed earlier, the main purpose of PCA is to generate a minimum number of components that accounts for the maximum amount of variance in the original data when the researcher feel s confident there is limited error variance among the variables Therefore, no latent variables underlying the observed variables need to be invoked. On the other hand, EFA is used to i dentify latent dimensions represented among the original variables, and among th e observed variables, and the factors are viewed as the causes of the observed variab 287). T he current social engagement motive scales were adapted from the previous research, covering diverse motivations behind television c onsumption, the Internet use, and YouTube video viewing However, there seem ed to be no established scales for each motive and several motivation items were used in an inter changeable fashion. Thus, it was necessary for th is study to i dentify latent facto rs represented among the forty nine motive items by employing the EFA. Specifically, the factor analyses w ere performed on a polychoric correlation matrix using maximum likelihood with mean and variance estimation procedure through an oblique Geomin rotati on by Mplus (Version 6.0 ) program. Antecedents and consequences tests The rest of the research questions (RQ4 to RQ7) and hypotheses (H 1 through H 11 ) were all related to the antecedents and consequences of the social engagement

PAGE 121

121 behavior The SEM was perf orm ed to examine the proposed antecedents and consequences by using the data analysis program Mplus (Version 6.0 ). The proposed model contains nine exogenous variables and five endogenous variables. The endogenous variables are : 1) social engagement incl uding the second order overall social engagement structure and the first order four dimensions structure, 2) behavioral program loyalty, 3) attitudinal program loyalty, 4) audience satisfaction, and 5) product purchase likelihood. The exogenous variables a re : 1) program genre preference 2) program affinity, 3) program involvement, 4) perceived ease of use, 5) compatibility, 6) social presence, 7) innovativeness, 8) motives, and 9) social characteristics. Structural equation modeling in practice is perfor med through a two step approach (Anderson & Gerbing, 1998) The first step is to use CFA procedure to test measurement model, which examines relationships between latent and multiple observed indicators. Latent variables are the variables that are not dire ctly measured but indicated by one or more observed variables or indicators (Hair, Anderson, Tatham, & Black, 199 5 ). In building measurement models, multiple indicator measurement ning to single indicator of some construct is available, the single item construct is acceptable in the SEM. The purpose of conducting measurement model ing is to define t he indicators for each construct and assess the reliability of each construct for the causal relationships. After the measurement model is specified to be unidimensional and reliable the second step is to use the simultaneous equation model (also called s tructural equation model) to test the hypothesized relationships among latent variables.

PAGE 122

122 The structural model is examined with the significance of estimated coefficients. Specifically, by performing the SEM to test the proposed antecedents and consequences the present study first assessed the four dimensions of social engagement, followed by the overall social engagement that is treated as a second order factor When conducting the SEM, there should be an awareness of the missing data issue in study desig ns since the incomplete data could cause the analysis to be biased if covariance matrix input (Byrne, 2001). There are two approaches to deal with missing values: 1) using a listwise deletion technique, and 2) using full information analysis. The advantag e of the listwise deletion technique is eas e of use, which simply removes the observations with missing data and makes the data set complete. The disadvantage of th is approach is less power and accuracy due to the reduction in the sample size which also p roduce s inconsistent estimates even if the data are not missing completely at random. By contrast estimation s that use all available data (full information analysis) do not remove any observations, resulting in better power and accuracy than listwise dele tion. Thus, all missing data in the current study were treated with full information maximum likelihood (FIML) given that this procedure is robust when data are missing completely at random (MCAR) or missing at random (MAR), or the percentage of missingnes s is minimal ( Muthen, Kaplan, & Hollis, 1987). For this study, the percentage of missing data was negligible (0.1 % 0.2%), thus this study adopted the FIML estimation in models with missing data through using Mplus (Version 6.0 ) program

PAGE 123

123 CHAPTER 5 RE SULTS This chap ter presents the results of hypothesis testing and answers to the research questions. The statistical procedure for each test is also briefly described. More specifically, the results are grouped into the following topics: 1) a brief explan ation of the social engagement scale, 2) social engagement with different television program genres, 3) the motives behind social engagement behavior, 4) the antecedents and consequences of the four social engagement dimensions, and 5) the antecedents and consequences of the overall social engagement. Social Engagement Scale Explanation The RQ1 asked about whether there are several dimensions or levels in social engagement with television content. Through the scale development and validation approach, this study found the four underlying dimensions of social engagement behavior : vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influence. The four dimensions indicate the different manifestations of how television audiences use a n expand ing array of social media platforms to connect with television content over time. Thus, this investigation on the social engagement construct and its measure ment scale contribute s to the understanding of the consumption of television programming in the new social media environment s The first dimension, vertical involvement, measures the degree in which television viewers actively use a range of social media platforms to be involve d with the core program content and its relevant information. The ver tical involvement dimension characterizes the participatory behavior in relation to the core content and/ or ancillary content of a program. In particular, t he involvement activities are more one way oriented

PAGE 124

124 but critical because they cover a range of socia l media touchpoints that an individual could have with program content These social media touchpoints i nclud e R SS feeds, podcasts, mobile video and applications, check in applications in the entertainment focused social networks, social bookmarks, online widgets, and video and photos uploaded or forwarded in social media. The second dimension diagonal interaction, measures the degree of social interaction that viewers develop with characters celebrities and working staffs related to their favorite show s in microblogs such as Twitter TM The possibilities of developing interactive relationship s between viewers and characters/celebrities of the program are limited in the one way television consumption pattern H owever, as more celebrities performers and professional working staffs have a presence in Twitter TM the direct dialogues and communication s between the audiences and these media figures become more common which are mainly facilitated by the Internet relay chat function embedded in Twitter TM and i ts widely mobile applications. The third dimension, horizontal intimacy, measures the extent to which individual viewers emotionally respond to a television program and the affection of the viewers toward the branded content with other audiences in blogs and online discussion forums This dimension captures a deeper and more intimate connection between the viewers and the diegetic, narrative text dep icted in a program through peer to peer social media activities In addition, the horizontal intimacy dimen sion characterizes a mode of engagement that satisfies the viewer s surrounded by or submerged in television program ming Driven by the contextual and extratextual immersion, peer to peer behaviors such as expressing

PAGE 125

125 and responding to the perspectives from other viewers essentially demonstrate the The fourth dimension, horizontal influence, measures the degree of identification an d belonging, as well as the extent of meaningful influence in the direction or outcome of television program ming in a peer rel ated space like social networks Facilitated by the relationship focus and the identity nature of social networks like Facebook TM the desire show in this platform to some degree indicates that the viewer draws upon the program as part of his/her self and social identity In this way the viewer also add s meaning to his/her relationship with others. Moreover influential opportunities exist in relation to the shows when the particular content solicits the and program sharing in social networks. This has a pot ential impact on the individual s online fri ends, regardless of whether they are members of the program audiences or not. In this sense, the television viewer become s an ambassador on behalf of the television brand to advocate, recommend, and finally p ersonally promote certain television content. To assess scale reliability, this study used coefficient alpha to estimate internal consistency reliability for the overall social engagement scale and its four dimensions. The results illustrated t hat the scale with the fifteen sample items perform ed well in capturing the propo sed social engagement construct As a means of discriminant validity this study demonstrated that social engagement is conceptually and empirical ly different from attitude toward the program (program aff inity) and program involvement (Table 5 1)

PAGE 126

126 Table 5 1. Social engagement dimensions, definitions and scale items Dimension Definition Scale Items Vertical Involvement The dimension measures th e degree in which television viewers actively use a range of social media platforms to be involved with the core program ming content and its relevant information. s RSS feeds or podcasts. I have used my mobile phone to watch video clips, check photos and text alerts, or play games relevant to the program(s). I have used check in apps for the program(s) in Foursquare TM Miso, Philo, Starling, or GetGlue, etc.. s video clips or photos online. I have used social bookmarks (e.g., Digg TM and Delicious) to tag the program(s). I have uploaded or forwarded videos or photos relevant to the program(s). Diagonal Interaction The dimension measures the degree of social in teraction that viewers develop with characters, celebrities, and working staffs related to their favorite shows in microblogs such as Twitter TM I am a follower of the program(s) (including actors, writers, producers, etc.) in microblogs (e.g., Twitter TM ). s tweets (including actors, writers, producers, etc.,) in microblogs (e.g., Twitter TM ). s tweets (including actors, writers, producers, etc.) in microblogs (e.g., Twitter TM ). Horizontal Intimacy The dimension measures the extent to which individual viewers emotionally respond to a television program and the affection of the viewers toward the branded content with other audiences in blogs and online discussion forums. I have re ad blog posts relevant to the program(s). I have written or commented on blog posts relevant to the program(s). s posts in online discussion forums. s posts in online discuss ion forums. Horizontal Influence The dimension measures the degree of identification and belonging, as well as the extent of meaningful influence in the direction or outcome of television programming in a peer related space like social networks. I am a fa n of the program(s) and share them with my friends in social networks (e.g., Facebook TM and Myspace TM ). s posts in social networks ( e.g., Facebook TM and Myspace TM ).

PAGE 127

127 The social engagement construct with mul tiple dimensions was represented by two types of factor structures which were validated by CFA testing One was a four factor model and the other was a four factor model with on e higher order factor, w here the second order factor represented the overall co nstruct of social engagement While both the first order and the second order representation s of the social engagement construct fit the observed data adequately the comparison results indicated that the four factor model ( 2 = 377.785, df = 84, p = .000; CFI = .946, TLI = .933, RMSEA = .084; SRMR = .042) fit the observed data better than the four factor model with a second order factor ( 2 = 486.052, df = 86, p = .000; CFI = .927, TLI = .911, RMSEA = .097; SRMR = .063). Thus, this study prefer red the firs t four factor model by virt ue of its simplicity and better fit indices. To further examine the attribute of each social engagement dimension and their relationships, t he descriptive statistics from the main test revealed that the horizontal influence dim ension had the highest mean ( M = 3.35, SD = 1.24), followed by the horizontal intimacy dimension ( M = 3.00, SD = 1.22), diagonal interaction dimension ( M = 2.44, SD = 1.24), and vertical involvement dimension ( M = 2.24, SD = 1.10 ), respectively. Th ese test results social engagement behavior is more salient in sharing television viewing experience with friends, adding meaning to his/her relationship with others, and establishing connection s with media fi gures of the program, but less conspicuous in involvement with the core content and/or ancillary content of the program In addition, b y examining the intercorrelations of the four social engagement dimension s the results indicated that vertical involveme nt behavior is highly correlated with the diagonal interaction

PAGE 128

128 dimension but is least associated with horizontal influence. The correlation between the diagonal interaction dimension and the horizontal influence dimension is the weakest compared to the ot her intercorrelations among the four social engagement behaviors The descriptive statistics and correlations among the four social engagement dimensions are indicated in Table 5 2. Table 5 2. Descriptive statistics and correlations of four social en gagement dimensions Dimension M SD 1 2 3 4 1 Vertical Involvement 2.24 1.10 1.00 2 Diagonal Interaction 2.44 1.24 .910 1.00 3 Horizontal Intimacy 3.00 1.22 .721 .685 1.00 4 Horizontal Influence 3.35 1.24 .582 .536 .740 1.00 Note: p < .001 (two tailed) n = 494 Social Engagement with Different Television Program Genres The RQ 2 gauged how the four social engagement dimensions vary among the five different television genres, including drama s reality show s sitcom s game/t alk show s and animated comed ies This study employed multiple comparisons for one factor designs to answer the RQ2 by using the data analysis SAS (Version 9.2 ) software Specifically, the means of the five program genres were pairwise compared along each social engagement dimension. It should be noted that p lanned contrasts between any two means of television genres rather than Post hoc comparisons were performed. Thus, the familywise error was selected for control. The Shaffer Holm procedure was carrie d out so that the pairwise comparisons would be reasonably powerful and incorporate the omnibus test. Descriptive statistics for the five different television program genres along each social engagement dimension are presented in Table 5 3 Specifically, there were 209

PAGE 129

129 respondents who chose drama programs as their favorite s and evaluated their social engagement behavior related to th is type of television programs, followed by 125 respondents of reality shows, 73 respondents of animated comedies, 64 respon dents of sitcoms, and 23 respondents of game/talk shows, respectively. The results of planned comparisons revealed that there are no significant differences between any two television genres in terms of the diagonal interaction, horizontal intimacy, and ho rizontal influence dimensions, except for the vertical involvement dimension. As for the vertical involvement dimension, the planned comparison for game/talk shows versus dramas indicated that these two type s of television program s are significantly differ ent from each other in inducing the vertical involvement activities. A significant result was also apparent in the comparison of game/talk shows versus reality shows along the vertical involvement dimension The vertical involvement dimension characte rizes the participatory behavior demonstrated by social media users, who utilize d a range of social media platforms to connect with the core content and/or ancillary content of a program Such vertical involvement activities include : 1) using social bookma rks like Digg TM or Delicious to tag s video clips or photos online, 3) using check in apps for the program in several entertainment focused social networks such as Foursquare TM Miso, Philo, Starling, or G etGlue, etc., 4) using a mobile phone to watch video clips, check photos and text alerts, or play games relevant to the s RSS feeds or podcasts, and 6) uploading or forwarding videos or photos relevant to the program. When it comes to the vertical involv ement activities related to different television genres, the one way ANOVA

PAGE 130

130 omnibus test results showed that there is a significant difference among these five different types of program s ( F (4,489) = 4. 36 p = .0018 ) Specifically, it seem ed that the game/talk show was the most effective program genre that induces people to utilize an expand ing array of social media platforms to be involve d with television content ( M = 2.97 SD = 1.22 ). The stimulation factors of the ot her television genres were as follows: animated comed ies ( M = 2.46 SD = .93 ), sitcom s ( M = 2.29 SD = 1.29 ), reality show s ( M = 2.21 SD = 1.10 ), and drama s ( M = 2.10 SD = 1.03 ), respectively. The Shaffer Holm procedure was further carried out t o condu ct pairwise comparisons of any two television genre means in terms of the vertical involvement dimension Significant differences were found for these three types of program s : game/talk show s versus drama s ( t ( 489 ) = 3.69, p < .05 ) as well as game/talk sh ow s versus reality show s ( t (489) = 3.13, p < .05 ) The pairwise comparisons results s uggested that viewers tend to more actively engage in game/talk shows over reality programs and scripted dramas by involving a range of social media platforms to connect with the core content and/or ancillary information of the program, even though these five television genres could all stimulate different levels of vertical involvement activities surround ing this type of television program When it comes to the other thr ee dimensions in social engagement behavior surround ing the se five different types of program s the one way ANOVA test results indicated that there are no statistically significant differences among these five program genres in the measurement of diagonal interaction ( F (4, 489) = 1.44, p = .220 ), horizontal intimacy ( F (4, 489) = 1.35, p = .250 ) and horizontal influence ( F (4, 489) = 1.67, p = .156 ) More s pecifically, the omnibus test revealed that program genres

PAGE 131

131 regardless of reality shows or scripted dra ms, do not play a role when people use Twitter TM to interact with characters/celebrities o f the program n or do they influence audiences to use blogs and online discuss ion forums to read, write, or comment on the program. Likewise so cial engag ement activities in social networks (e.g., Facebook TM and Myspace TM ) such as indicating t hat they are fan s of the program shar ing the program with friends and s posts in social networks do not vary signifi cantly among these five television genres. Table 5 3. Descriptive statistics of different genres by social engagement dimensions Drama Reality Show Animated Comedy Sitcom Game/Talk Show n 209 125 73 64 23 Vertical Involvement M 2.10 a 2.21 b 2.4 6 2.29 2.97 a b SD 1.03 1.10 .93 1.29 1.22 Diagonal Interaction M 2.33 2.49 2.49 2.48 2.94 SD 1.19 1.29 1.14 1.33 1.33 Horizontal Intimacy M 2.89 3.13 3.05 2.89 3.32 SD 1.23 1.22 1.15 1.27 1.03 Horizontal Influence M 3.23 3.46 3.58 3.23 3.52 SD 1.24 1.24 1.12 1.34 1.26 Note: The mean difference that shares the same superscript is significant at the .05 level. Motivations behind Social Engagement The RQ 3 investigated what motives the audiences have for using social media to engage with telev ision content. To answer the research question, the EFA procedure rather than PCA was carried out to analyze the forty nine motive statements. Specifically, the factor analyses were performed on a polychoric correlation matrix using the maximum likelihood with mean and variance estimation procedure through an oblique Geomin rotation by Mplus (Version 6.0 ) program By analyzing the screen plots and goodness of fit indices, a series of models was estimated and compared, and a ten factor model showed the best fit ( 2 = 1999.91, df = 731, p =.000; CFI = .940, TLI =

PAGE 132

132 .903, RMSEA = .060, SRMR = .022). Thus, this study concluded that the ten factor solution best describe s the motive test. The factor loadings from the Mplus EFA procedure are displayed in Table 5 4 The cutoff value of .50 was selected in order to be conservative. The exploratory factor analysis yielded ten motives behind social engagement behavior corresponding to previous televisio n viewing motives (Rubin, 1983), the Internet use motives (Papacha rissi & Rubin, 2000), and YouTube video viewing motives (Haridakis & Hanson, 2009). The first factor R elaxation was comprised of three items related to a pleasant rest and relaxation driven motivation. The second factor, C ompanionship describe d a lonene ss relief as one of the reasons behind social engagement behavior. The third factor, P ass ing T ime describe d how television audiences use social media to interact with television content out of habit and to occupy time. The fourth factor, E ntertainment wa s comprised of three items illustrating the experience of social engagement with television content for amusement and enjoyment. The fifth factor, I nformation explain ed how the social engagement experience is derived from being informed. The sixth and sev enth factor s contain ed three items respectively, describ ing the A rousal and E scape motive s The eighth factor, Access measure d the use of social media to access television content because it is easier and a novel way of search ing for information and keep ing up with current issues. The ninth factor L earning reflected learning unknown and useful things as a motivation for social engagement behavior. The last factor, Interpersonal Utility was comprised of eight items related to using social media to be in volve d with television program s that measured belonging, inclusion, affection, social interaction, and expressive need s

PAGE 133

133 Based on the motive factor structure, this study further conducted reliability test ing coefficien t alpha. The acceptable value for Kline, 20 11; Nunnally, 1978). coefficient alpha values for the ten motives behind using social media to engage with television content ranged from .882 t o .937, suggesting that the ten motivation scales are reliable measures. In order to test discriminant validity, t his study next averaged the items with greater than .50 loadings on each motive factor in order to create composite motive variables. T he corr elations among the ten motive factors were all moderately as sociated with each other, ranging from .310 to .685, except for the correlation between Learning and Access ( r = .719). Overall, th e results demonstrated that each factor was distinct, with no sig nificant overlap and no additional factors present. To better understand which motive television audiences demonstrated most when they use d various social media platforms to engage with television content, this study further analyzed the descriptive stati stics of each motivation. The results revealed that the entertainment motive has the highest mean score ( M = 3.928, SD = .90), followed by the motivations of access ( M = 3.610, SD = .93) relaxation ( M = 3.525, SD = .98) learning ( M = 3. 468 SD = .9 8 3), a nd i nterpersonal utility ( M = 3.380, SD = .91) However, people us ing social media to connect with television program for the companionship need was the least notable motivation ( M = 2.742, SD = 1.10) Table 5 5 presents the descriptive statistics, Cronbac coefficient alpha, and correlations for the ten motive factors.

PAGE 134

134 Table 5 4 Exploratory factor analysis for motives behind social engagement behavior Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8 Factor 9 Factor 10 Because it relaxes me .794 .035 .010 .032 .066 .0 14 .003 .0 62 .041 .025 Because it allows me to unwind .842 .041 .016 .057 .013 .0 22 .020 .0 34 .052 .076 .751 .098 .013 .037 .008 .03 0 .016 .03 1 .052 .0 31 .152 .692 .003 .064 .030 .0 22 .087 .0 07 .023 .025 to or be with .010 .572 .254 .060 .134 .005 .034 .005 .004 .051 Because it makes me feel less lonely .001 .905 .055 .012 .001 .0 72 .001 .0 02 .004 .018 When I have nothing better to do .025 .092 .702 .035 .026 .0 74 .095 .0 58 .176 .065 Because it passes the time away, particularly when I am bored .003 .027 .914 .030 .078 .0 56 .006 .0 30 .057 .002 Because it gives me something to do to occupy my time .071 .005 .877 .091 .031 132 .002 014 .021 .055 Because it entertains me .021 .051 .004 .886 .037 .039 .002 .0 18 .009 .032 .114 .042 .035 862 .033 .0 53 .019 .0 16 .048 .012 Because it amuses me .044 .045 .031 .733 .060 .09 4 .097 .0 1 9 .008 .021 Because it helps me learn things about myself and others .033 .056 .043 .009 .676 .0 88 .001 .0 06 .179 .012 So I can learn how to do things .005 .018 .014 .014 .719 .004 .010 .00 7 .206 .028 So I can learn about what could happen to me .035 .081 .042 .077 .600 165 .105 039 .136 .008 .017 .001 .062 084 .179 .682 .006 .036 .067 .035 .018 .036 .014 .290 .022 .715 .014 .042 .024 .011 Because it peps me up .128 ,095 .019 .190 .057 .501 .069 .107 .046 .054

PAGE 135

135 Table 5 4 Continued Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8 Factor 9 Factor 10 So I can forget about school/work or other things .058 .021 .035 .106 .024 .124 .592 .037 .098 .000 So I can get away from the rest of the family or others .020 .009 .003 .001 .209 .007 .827 .035 .051 .020 So I can get away from what I am doing .025 .000 .033 .024 .001 .002 .847 .062 .070 .118 Because it easier to get information .038 .019 .011 .012 .020 .009 .026 .895 .015 .017 Because I can search for information .043 .029 .024 .002 .049 .009 .016 .815 .092 .038 Because I can get information for free .065 .052 .095 .022 .048 .055 .015 .782 .043 .084 So I can see what is out there .037 .012 .032 .017 .013 .049 .048 .234 .624 .036 So I can learn about useful things .030 .057 .053 .057 .180 .045 .005 .083 .721 .028 So I can learn about unknown things .002 .043 .001 .049 .019 .054 .034 .116 .775 .005 Because I want to show others encouragement .095 .085 .0 79 .039 .165 .002 .088 .056 .086 .530 Because I want to communicate with friends and family .065 .008 .055 .025 .042 .004 .035 .056 .082 .693 Because I want to belong to groups with the same interest as mine .040 .074 .056 .104 .102 .124 .032 .02 5 .001 .620 Because I want to let others know I care about their feelings .027 .171 .098 .054 .153 .039 .022 .022 .041 .662 Because I can express myself freely .110 .009 .036 .088 .031 .033 .003 .018 .064 .802 Because I enjoy answering uestions .005 .013 .008 .083 .060 .065 .016 .019 .042 .772 Because I can participate in discussion s .005 .017 .007 .024 .108 .077 .004 .148 .004 .820 Because I can meet new people .039 .029 .087 .035 .117 .063 .066 .031 .087 .777 Note: Fact or 1 = Relaxation, Factor 2 = Companionship, Factor 3 = Pass Time, Factor 4 = Entertainment, Factor 5 = Information, Factor 6 = Arousal, Factor 7 = Escape, Factor 8 = Access Factor 9 = Learning, Factor 10 = Interpersonal Utility Goodness of Fit Indices: 2 = 1999.91 ( df = 731, p =.00), CFI = .940, TLI = .903, RMSEA = .060, SRMR = .022. n = 494

PAGE 136

136 Table 5 5 Descriptive statistics internal consistency values, and intercorrelations for motive factors Factor Motive No. of Items Mean SD alpha 1 2 3 4 5 6 7 8 9 10 1 Relaxation 3 3.525 .98 .918 1 .00 2 Companionship 3 2.742 1.10 .891 485 1.00 3 Pass Time 3 3.301 1.07 .898 .310 .519 1.00 4 Entertainment 3 3.928 .90 .932 .650 .289 .318 1.00 5 Information 3 2.93 6 1.12 .899 .550 .651 .396 .370 1.00 6 Arousal 3 3.357 1.09 .905 .671 .481 .323 .651 .640 1.00 7 Escape 3 2.976 1.15 .882 .473 .536 .473 .377 .573 .547 1.00 8 Access 3 3.610 .93 .921 .451 .358 .400 .471 .450 .470 .386 1.00 9 Learning 3 3.468 .98 .911 .484 .427 .337 .497 .590 .523 .437 .719 1.00 10 Interpersonal Utility 8 3.380 .91 .937 .555 .541 .362 .447 .647 .585 .464 .635 .685 1.00 Note: alpha p < .001 (two tailed).

PAGE 137

137 Ante cedents and Consequences of Four Social Engagement Dimens ions The research questions (RQ4 to RQ7) and hypotheses (H1 through H11) were all related to the ante cedents and consequ ences of social engagement behavior. Specifically, two mo dels were tested to examine the antecedents and consequences of the social engagement experience in this study The first model tested the predictive power of antecedents for each social engagement d imension and their outcomes. This model was comprised of nineteen exogenous variables: 1) program genre preference, 2) program affinity, 3) program involvement, 4) perceived ease of use, 5) compatibility, 6) social presence, 7) relaxation, 8) companionshi p, 9) pass time, 10) entertainment, 11) information, 12) arousal, 13) escape, 14) access 15) learning, 16) interpersonal utility 17) innovativeness, 1 8 ) interpersonal interaction, and 19 ) social activity The eight endogenous variables were: 1) vertical involvement, 2) diagonal interaction, 3 ) horizontal intimacy, 4 ) horizontal influence 5 ) program attitudinal loyalty, 6 ) program behavioral loyalty, 7) audience satisfaction, and 8 ) product purchase likelihood. The second model tested the antecedents and consequences of the overall social engagement, which is treated as the second order construct. Therefore, for the second model testing, there were nineteen exogenous variables and five endogenous variables. Specifically, the antecedent and consequence var iables were the same in the two models except for the social engagement construct. The model testing on the four social engagement dimensions were presented in this section first, and the results of model testing with the overall social engagement were sho wn in the following section. The data analysis was conducted in two stages : 1) validation of the measurement sc ales and factor structure, and 2) use of simultaneous equation analysis (SEQM) to test a theoretical model through the Mplus (Version 6.0 ) pro gram To assess the

PAGE 138

138 the CFA procedure setting the measurement error variance to be zero, and further eliminated those items that did not meet the three criteria proposed by Jreskog and Srb om (1993). The SEQM approach was adopted for the cau s al structural models, and simultaneous path analyses were conducted using the maximum likelihood estimation instead of ordinal least square (OLS) estimation equation by equation because SEQM normally tu rns out more accurate estimates than step by step multiple regression which tends to inflate standard errors for path coefficients (Kline, 20 11 ). Measurement Model Assessment The purpose of measurement model testing is to specify which observed variable (indicators) define each construct (Kline, 20 11 ). By using the CFA procedure, twenty five variables with multiple items and two variables with single indicator were specified in the measurement model. To assess the model fit, the minimum fit function Chi s quare for the measurement model was 5758.373 ( df = 3 566 p < .001). It should be noted that the estimation of Chi square is sensitive to sample size, therefore the other goodness of fit indices were needed (e.g., CFI, TLI, RMSEA, and SRMR) There is a gene ral agreement on the effective measures of fit when CFI or TLI is greater than .90, RMSEA is below .06, and SRMR is less than .09 ( Hoyle & Duvall, 2004). Hu and Bentler (1999) were more conserva tive and recommended a cut off value of .95 or more for CFI and TLI, whereas the value of RMSEA should be close to .06. The goodness of fit indices for this measurement model were CFI = .9 2 7 TLI = .9 1 8 RMSEA = .0 35, and SRMR = .0 45, indicat ing that the measurement model with the four social engagement dimensions fi t the data adequately.

PAGE 139

139 Convergen t validity is demonstrated when different items are used to measure the same construct. This study empirically assessed convergen t validity by examining factor loadings and the relevant p values. For acceptable construct vali dity, it is proposed that each item should have a minimum factor loading of .60 on its hypothesized latent factor (Nunnally, 1978). Following the three criteria proposed by Jreskog and Srbom (1993), three indicators of the program involvement construct w ere eliminated for the measurement model due to the lower standardized factor loadings ( Uninterested /Interested Superfluous/Vital and Nonessential/Essential ) As for the innovativeness construct, the three reversely coded items using negative statement s were converged as the first measurement scale while the other three positive statement items were clustered as the second scale of this construct. Because of the high correlation between the two scales of the innovativeness construct, this study retaine d the three positive items alone as the measurement scale of innovativeness in order to reduce overlap Regarding the two variables with single indicator program behavioral loyalty and audien ce satisfaction the Mplus program set the error variance fo r these two variables equal to zero in order to treat this type of observed variable as the pseudo latent variable in the mea surement model. However, the procedure used in the measurement model was not necessary when the stru ctural model was estimated late r. The desired convergen t validity was achieved for all constructs in the measurement model with factor loadings ranged from .6 18 to .9 5 3 ( p < .00 1 ), indicating that these items statistically measure the constructs as intended The standardized factor load ing results by the CFA procedure are displayed in Table 5 6

PAGE 140

140 Table 5 6 Confirmatory factor an alysis for measurement model with four dimensions Variable Item Standardized Factor Loading Social Engagement Vertical Involvement I have subscribed to s RSS feeds or podcasts. .78 5 I have used my mobile phone to watch video clips, check photos and text alerts, or play games relevant to the program(s). .732 I have uploaded or forwarded videos or photos relevant to the program(s). .665 I have used check in apps for the program(s) in Foursquare TM Miso, Philo, Starling, or GetGlue, etc.. .80 4 I have used social bookmarks (e.g., Digg TM and Delicious) to tag the program(s). .85 3 s video clips o r photos online. .854 Social Engagement Diagonal Interaction I am a follower of the program(s) (including actors, writers, producers, etc.) in microblogs (e.g., Twitter TM ). 704 s tweets (including actors, writers, producer s, etc.,) in microblogs (e.g., Twitter TM ). .869 s tweets (including actors, writers, producers, etc.) in microblogs (e.g., Twitter TM ). .929 Social Engagement Horizontal Intimacy I have read blog posts re levant to the program(s). .734 I have written or commented on blog posts relevant to the program(s). .898 s posts in online discussion forums. .80 5 s posts in online discussion for ums. .88 6 Social Engagement Horizontal Influence I am a fan of the program(s) and share them with my friends in social networks (e.g., Facebook TM and Myspace TM ). .698 s posts in social networks (e.g., Fac ebook TM and Myspace TM ). .898 Program Genre Preference The degree of attention paid when watching each of the following types of programs: reality shows, drama, game/talk shows, animated comedies, and sitcoms .935 The degree of enjoyment when watching e ach of the following types of programs: reality shows, drama, game/talk shows, animated comedies, and sitcoms. .884 Program Affinity I would feel lost without the program to watch. .857 .79 8 Watching the program is one of the most important things I do each day or each week. .804

PAGE 141

141 Table 5 6 Continued Variable Item Standardized Factor Loading Program Involvement Irrelevant (1) Relevant (5) .756 Means nothing to me (1) Means a lo t to me (5) .903 r (1) Matters to me (5) .868 Nonessential (1) Essential (5) .775 Perceived E ase of U se Learning to use social media to comment, post, watch, or read anything about television program s is easy for me. .911 It is easy for me to become skilled at using social media to comment, post, watch, or read anything about television program s .919 It is easy to use social media to comment, post, watch, or read anything about television program s .91 6 Compatibility Using socia l media to comment, post, watch, or read anything about television program s is compatible with most aspects of my television viewing. .84 2 Using social media to comment, post, watch, or read anything about television program s fits my lifestyle. .895 Usin g social media to comment, post, watch, or read anything about television program s fits well with the way I like to engage in television viewing. .911 Social Presence Unsociable (1) Sociable (5) .685 Impersonal (1) Personal (5) .759 Insensitive ( 1) Sensitive (5) .798 Cold (1) Warm (5) .796 Passive (1) Active (5) .680 Motive Relaxation Because it relaxes me .896 Because it allows me to unwind .916 .854 Motive Companionship e .850 .808 Because it makes me feel less lonely .924

PAGE 142

142 Table 5 6 Continued Variable Item Standardized Factor Loading Motive Pass Time When I have nothing better to do .747 Because it passes the time away, particularly when I am bored .943 Because it gives me something to do to occupy my time .909 Motive Entertainment Because it entertains me .941 .953 Because it amuses me .829 Motive Information Because it helps me learn things about myself and others .868 .853 So I can learn about what could happen to me .875 Motive Arousal .850 .914 Because it peps me up .860 Motive Escape So I can forget about school/work or other things .744 So I can get away from the rest of the family or others .902 So I can get away from what I am doing .902 Motive Access Because it easier to get information .917 Because I can search for information .905 Because I can get information for free .855 Motive Learning So I can see what is out there .831 So I can learn about useful things .902 So I can learn about unknown things .913 Motive Interpersonal Uti lity Because I want to show others encouragement .818 Because I want to communicate with friends and family .760 Because I want to belong to groups with the same interest as mine .789 Because I want to let others know I care about their feelings .83 7

PAGE 143

143 Table 5 6 Continued Variable Item Standardized Factor Loading Motive Interpersonal Utility Because I can express myself freely .801 .824 Because I can participate in the discussion .829 Because I ca n meet new people .796 Innovativeness If I heard that a new social media platform was available online, I would be interested enough to try it. .680 .829 I know more about new social media platforms before other people do. .788 Social Characteristics Interpersonal Interaction I g e t to see my friends as often as I would like. .706 I spend enough time communicating with my friends and family by telephone or mail. .618 I have ample opportunity for conversations with others. .801 I can always find someone to speak with when I need to talk. .730 Social Characteristics Social Activity I often travel, vacation, or take trips with others. .701 I often visit with friends, relatives, or neighbors in their homes. .679 I often participate in the meetings or activities of clubs, lodges, recreation centers, churches, or other organizations. .747 I often go places to socialize with others. .804 I often participate i n games, sports, or activities with others. .785 Program Behavioral Loyalty Over the past month, I have not missed any episodes of the program when they broadcast on television. 1.000 Program Attitudinal Loyalty I would recommend the program to ot hers. .821 I think of myself as a loyal viewer of the program. .869 I would be willing to watch the program rather than other shows. .795 Audience Satisfaction The overall satisfaction with watching the program 1.000

PAGE 144

144 Table 5 6 Continued Variable It em Standardized Factor Loading Product Purchase Likelihood The likelihood to purchase any of the following items when they were available in Memorabilia/merchandise of the television st ation/network .908 Memorabilia/merchandise of the television show or stars .920 Products shown in that television show .838 Note: All factor loadings are significant at p < .001 level (two tailed). Goodness of Fit Indices: 2 = 5758 373 ( df = 3566 p =.00 0 ), CFI = .92 7 TLI = .91 8 RMSEA = .0 35 SRMR = .0 45 n = 494 To assess scale reliability of all constructs included in the measurement model, coefficient alpha was employed to test the scale internal consistency reliability. The previo us study suggested that the criterion of the reliability alpha should exceed .70 (Nunally, 1978) The coefficient alpha values of all of the multi item scales ranged from .768 to .938 indicating that all constructs in the measurement model are r eliable. In addition, one alternative approach to check scale reliability was to consult the standardized factor loadings on their respective latent factor, in which the acceptable factor loading value was greater than .60 (Nunnally, 1978). The CFA result s demonstrated that the factor loadings of individual items in the measurement model were ranged from .6 18 to .953 ( p < .00 1). To sum up, b coefficient alpha and the standardized factor loadings indicated that all constructs with multiple ind icators included in the measurement model are reliable. Table 5 7 displays descriptive statistics with means and standard deviations as well as internal consistency values of the se multi item constructs in the measurement model.

PAGE 145

145 Table 5 7 Descriptive statistics and internal consistency values for constructs Construct No. of item alpha Mean SD Vertical involvement 6 .905 2.244 1.095 Diagonal interaction 3 .871 2.44 3 1.236 Horizontal intimacy 4 .902 2.995 1.218 Horizontal influence 2 .768 3.354 1.242 Program genre preference 2 .905 3.960 1.045 Program affinity 3 .858 3.304 1.093 Program involvement 4 .894 3.940 .843 Perceived ease of use 3 .938 3.949 .930 Compatibility 3 .914 3.523 .946 Social presence 5 .860 3.769 .822 Innov ativeness 3 .805 2.984 .949 Interpersonal interaction 4 .802 3.434 .823 Social activity 5 .858 3.044 .970 Relaxation 3 .918 3.525 .980 Companionship 3 .891 2.742 1.103 Pass time 3 .898 3.301 1.066 Entertainment 3 .932 3.928 .900 Information 3 .8 99 2.936 1.124 Arousal 3 .905 3.357 1.086 Escape 3 .882 2.976 1.149 Access 3 .921 3.610 .930 Learning 3 .911 3.468 .977 Interpersonal utility 8 .937 3.380 .912 Program attitudinal loyalty 3 .867 4.205 .830 Product purchase likelihood 3 .917 2. 860 1.089 Structural Model Testing To examine the causal relationships proposed in the structural model for the antecedent and consequence test s this study next carried out the simultaneous equation analysis. Before implementing the path analysis, inte rcorrelations among the antecedent variables were tested in order to check the multicollinearity, as a

PAGE 146

146 multicollinearity problem could influence the results of SEM as it does in regression analysis (Hair, And erson, Tatham, & Black, 199 5 ). As shown in Table 5 8 some antecedent variables were significantly correlated. The inter correlations among the correlated antecedent variables were all moderate rang ing from 092 to .6 85 except for the correlation between the two motives of l earning and access ( r = .719 ). However, in most cases, correlations exceeding .80 can be treated as indicator s of a multicollinearity problem (Hair, Anderson, Tatham, & Black, 199 5 ). Thus, there was no multicollinearity problem for the antecedent variables in the structural model. T he RQ6 and RQ7 pertained to the antecedents and consequences of the four social engagement dimensions, i.e., vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influence. This study next examined the cau s al relationship s repres ented in the structural model to answer these two research questions. Before carrying out the path analysis, the structural model with the four social engagement dimensions was estimated first. The minimum fit function Chi square 2 for the structural mo del was 7266.911 ( df = 3887, p < .001 ). The goodness of fit indices (CFI = .90 6 ; TLI = 897 ; RMSEA = .04 3 ; SRMR = .047) indicated that the structural model fit the data somewhat adequately Thus, modifications indices were reviewed with an attempt to impro ve the model. Based on the most apparent modification indices, this study added a structural residual covariance for any two social engagement dimensions, resulting in a reasonably good fit ( 2 = 6869.780 df = 3 640 p < .001; CFI = .9 11 TLI = .90 2 RMSEA = .042, SRMR = .047 )

PAGE 147

147 Table 5 8 Correlations matrix for antecedent variable s 1 2 3 4 5 6 7 8 9 10 1 Relaxation 1.00 2 Companionship .485** 1.00 3 Pass t ime .310** .519** 1.00 4 Entertainment .650** .289** .318** 1.00 5 Information .550** .651** .396** .370** 1.00 6 Arousal .671** .481** .323** .651** .640** 1.00 7 Escape .473** .536** .473** .377** .573** .547** 1.00 8 Access .451** .358** .400** .471** .450** .470** .386** 1.00 9 Learning .484* .427** .337** .497** .590** .523** .437** .719** 1.00 10 Interp ersonal u tility .555** .541** .362** .477** .647** .585** .464** .635** .685** 1.00 11 Genre p reference .168** .048 .197** .229** .027 .094* .114* .158** .092* .072 12 Program a ffinity .4 65** .371** .233** .370** .414** .422** .325** .432** .410** .501** 13 Program i nvolvement .412** .256** .191** .380** .289** .338** .247** .370** .346** .412** 14 Perceived e ase of u se .264** .083 .163** .331** .025 .185** .163** .313** .237** .191** 1 5 Compatibility .510** .371** .299** .510** .420** .484** .394** .431** .461** .497** 16 Social p resence .473** .352** .218** .456** .424** .485** .334** .353** .377** .522** 17 Innovativeness .459** .458** .330** .338** .521** .464** .459** .389** .418* .519** 18 Interpersonal i nteraction .390** .281** .219** .334** .426** .402** .297** .304** .359** .353** 19 Social a ctivity .403** .359** .220** .251** .501** .421** .395** .297** .332** .422**

PAGE 148

148 Table 5 8 Continued 11 12 13 14 15 16 17 18 19 1 Relaxation 2 Companionship 3 Pass time 4 Entertainment 5 Information 6 Arousal 7 Escape 8 Access 9 Learning 10 Interpersonal utility 11 Genre preference 1 .00 12 Program affinity .245** 1.00 13 Program involvement .297** .642** 1.00 14 Perceived ease of use .063 .120** .117** 1.00 15 Compatibility .139** .378** .256** .506** 1.00 16 Social presence .071 .350** .453** .248** .431** 1.00 17 Innovativeness .085 .292** .168** .330** .514** .390** 1.00 18 Interpersonal interaction .068 .239** .166** .199** .315** .305** .490** 1.00 19 Social activity .009 .200** .096* .177** .391** .309** .619** .579** 1.00 Note: p < .05, ** p < .01 (two tailed)

PAGE 149

149 Antecedents to four social engagement dimensions The RQ 6 asked about the antecedents to the four social engagement dimensions i.e., ding vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influ ence Specifically, the first set of variables examine d the predictive power of perceptions o f television program s including program genre preference, program affinity, and program involvement. The second set of variables address ed the relationship s betwe en perceived characteristics of social media and the different dimensions of social engagement behavior. The third set of variables investigate d the impact s of audience attributes, including motives, innovativeness, and offline social characteristics (i.e. interpersonal in teraction and social activity). To test the cau s al relationships, this study employed a two tailed test in the structural model to emphasize the significant results of the antecedents to the four social engagement dimensions For the first social engagement dimension vertical involvement, the significant antecedents included innovativeness ( = .521, p < .001), offline social activity ( = .222, p < .01), program affinity ( = .21 0 p < .01), and perceived ease of use ( = .1 88 p < .0 0 1) The results first indicated that the construct of innovativeness exhibits the strongest positive influence on the involvement tendencies compared to other predictors. Specifically, the innovativeness construct portrays the individuals as the among the adopter category proposed by Rogers (1995). Thus, the results suggested that the more innovative tendencies the individuals demonstrate; the more likely they are to engage with various social media platforms to connect w ith television content. The second powerful predictor was social activity ies in their real lives. It appeared that viewers who keep more social networks and social activities in their real lives als o

PAGE 150

150 tend to be active online, using various online social media platforms to interact with television content. Program affinity is the third determinant, suggesting that audiences who like the television program and perceive it important in their daily lives are more likely to use a range of social media platforms to connect with core program content and /or ancillary content of the program In addition, perceived ease of use of the general social media system negatively influence d the vertical involvement act ivity. It seemed that the lack of diversity in technology proficiency in these social media platforms might play a role in predicting th e vertical involvement behavior Regarding the second social engagement dimension diagonal interaction, the significan t antecedents were innovativeness ( = .501, p < .001), social activity ( = .272, p < .01), the motive of inter personal utility ( = 232 p < .01), the offline interpersonal interaction ( = .1 53 p < .0 5 ), and program genre preference ( = .089, p < .0 5). Besides the stronger predictive ability of innovativeness and social activity as shown in vertical involvement, the results indicated that both the motive of interpersonal utility and the offline interpersonal interaction exhibit negative im pacts on diagonal interaction behavior. This suggested that if the individuals want to meet their interpersonal utility needs, such as belonging to groups with same interest, showing encouragement and care to others, seeking to communicate with others and express people, they are less likely to engage with Twitter TM to dialogue with those media figures. Likewise it seemed logical that interpersonal interaction, one of the indiv social characteristics in their real lives, is negatively predictive of the diagonal interaction behavior. This result further suggested that audiences who have ample

PAGE 151

151 opportunities to interpersonally communicate with friends, family, relatives, or others in their real lives tend to avoid the communication opportunities with media figures of the program through Twitter TM In addition, the results revealed that program affinity is one of predictors, suggest ing that the audiences who have preferences f or certain type s of television program s are more likely to use the microblog Twitter TM to interact with the characters, celebrities, and relate d working staff s of the program. Moreover, the predictive ability of audience attributes such as the individual innovative tendency and social activit y in their real lives are all positively salient in relation to diagonal interaction In terms of the third social engagement dimension horizontal intimacy, the significant antecedents included = .264, p .220, p p p < .05), and .129, p < .05). The results first illustrated that the two perceived social media characteristics co mpatibility and social presence exhibit opposite influences on the horizontal intimacy experience. Specifically, perceived compatibility was found to positively predict the social engagement behavior, whereas perceived social presence of the general soci al media system is negatively predictive of the horizontal intimacy dimension. The findings suggested that audiences who see using blogs/online discussion forums to read, post, or comment on the program as compatible with their lifestyle tend to develop a stronger intimate connection with the branded television program. However, the viewers who regard the general social media system as sociable, personal, sensitive, warm, active, and open tend to be less likely to engage in peer to peer behaviors in blogs/ opinions and responding to the perspectives from other viewers related to television

PAGE 152

152 content. It could be possible that the discrepancy in perceptions between the general social media concept and the specific soc ial media platform (i.e., blogs and online discussion boards) play a role in the impact In addition, the more affinity of the audiences have for the television content, the more likely they are to engage with blogs and online forums to develop a deeper an d more intimate connection with other audience members through peer to peer social media activities surrounding television content. nnovative tendenc ies are both positively predictive of the horizontal intima cy connection the predictive power of innovativeness is not so prominent as its influences on vertical involvement and diagonal interaction. With respect to the last dimension of social engagement horizontal influence, the salient antecedents were program affinity ( = .24 0 p < .0 1), innovativeness ( = .225, p < .05), compatibility ( = .1 94 p < .01), social activity ( = .168, p < .05), and the motive of inter personal utility ( = .1 58 p < .0 5) S pecifically, the findings showed that the program affinity is the most prominent predictor of horizontal influence dimension compared to other determinates This meant that viewers king and affinity for the programs are more strongly drive them s update their status, shar e the program, and read or post comments related to the program in a peer related space like social networks. In addition, to meet the ir inter personal utility needs ( such as belonging to groups with the same interest, seeking to communicate with others and express themselves freely, or seeking to meet new people ) the viewers are more likely to engage with the horizontal influence act ivities in Facebook TM Furthermore, the individuals who possess higher innovative tendencies

PAGE 153

153 and exhibit active social intera ctions in their real lives tend to be active online using social networks to update their status or input opinions in relation to the program as well as share the program with friends. As discussed above the perceived compatibility of the general social media system was a lso a significant antecedent to horizontal influence. In s ummary, the salient antecedent s to the four social engagement dimensions consisted of program affinity, genre preference, perceived ease of use, compatibility, social presence, the motivations of inter personal utility, and different f acets of audience attributes. In particular both innovativeness and offline social activity demonstrated the universal predictive power for all four social engagement dimensions Th e fact s uggested innovative tendencies plus their networking and socializing abilities in their real lives are most likely to influence their social engagement behavior with television content P rogram affinity was found to be a significant predictor fo r t hree social engagement dimensions except for the diagonal interaction behavior while compatibility and inter personal utility influence d on two social engagement dimensions. The antecedents which were only predictive of a single social engagement dimens ion were genre preference (to diagonal interaction) perceived ease of use (to vertical involvement ) social presence (to horizontal intimacy) and interpersonal interaction (to diagonal interaction) Thus it appeared that audience attributes especially the play ed a more significant role than the program perception variables and perceived characteristics of social media in predict ing the four social engagement dimensions The causal relationships with standardized path co efficient and standard error supported by this study are displayed in Table 5 9

PAGE 154

154 Table 5 9 Antecedents to the four social engagement dimensions Antecedents Vertical Involvement Diagonal Interaction Horizontal Intimacy Horizontal Influence Standardi zed path coefficient SE Standardized path coefficient SE Standardized path coefficient SE Standardized path coefficient SE Genre preference .002 .039 .08 9 .044 .0 2 0 .047 .0 44 .045 Program affinity .21 0 ** .071 .135 .078 .1 65 .082 .24 0 ** .077 Program involvement .1 0 1 .06 7 .0 65 .074 .012 .077 099 .07 3 Perceived ease of use .1 88 ** .050 .085 .05 4 .087 .05 5 .0 65 .05 4 Compatibility .101 .059 .043 .065 200 ** .066 .1 94 ** .06 2 Social presence .00 3 .055 .009 .060 .1 29 .06 1 .104 .058 I nn ovativeness .5 21 *** .08 1 .5 01 *** .0 88 .2 64 .09 1 .225 .0 88 Relaxation .0 01 .06 5 .048 .07 2 .092 .07 2 .0 84 .0 68 Companionship 116 .06 2 .0 11 .0 68 .058 .0 69 .0 76 .0 6 7 Pass time .042 .047 .035 .052 .10 4 .053 .0 5 6 .052 Entertainment .0 57 .06 6 .1 32 .073 .0 68 .07 4 105 .07 0 Information .133 .091 .19 5 .100 .08 9 .10 1 .0 66 .09 6 Arousal .0 47 .076 .1 57 .084 .061 .087 .080 .085 Escape .03 1 .05 3 .1 02 .059 .0 50 .06 0 .014 .0 59 Access .0 41 .06 8 .111 .076 .0 40 .07 7 .0 42 .074 Learning .020 .079 .05 2 .087 .041 .08 8 .038 .08 3 Interpersonal utility 025 .07 8 .2 32 ** .08 1 .1 30 .08 2 .1 58 .0 79 Interpersonal interaction .0 63 .06 0 .1 53 .06 7 114 .0 67 .0 2 4 .06 6 Social activity 222 ** .072 .27 2 ** .0 79 .22 0 ** .082 .1 68 .078 Note: p < .05, ** p < .01, *** p < .001 (two tailed). Goodness of Fit Indices: 2 = 6869.780 ( df = 3640, p < .001); CFI = .911, TLI = .902, RMSEA = .042, SRMR = .047. n = 494.

PAGE 155

155 Consequences of four social engagement dimensions The RQ7 pertained to the possible consequences o f the four social engagement dimensions i.e., p rogram behavioral loyalty, program attitudinal loyalty, audience satisfaction, a nd product purchase likelihood. As shown in Table 5 10 the diagonal interaction behavior did not appear to be related to any of the proposed consequences However, the horizontal intimacy and horizontal influence dimensions seemed to pre dict four outcomes. More specifically, the horizontal intimacy dimension demonstrated the negative effects on all consequences: program behavioral loyalty ( = 2. 592 p < .001), program attitudinal loyalty ( = 2. 724 p < .001), audience satisfaction ( = 2. 149 p < .001), and product purchase likelihood ( = 1.4 12 p < .001). By contrast, the horizontal influence activities exhibited positive influences on all four proposed outcomes: program behavioral loyalty ( = 2. 415 p < .001), program attitudinal loyalty ( = 3. 182 p < .001), audience satisfaction ( = 2. 567 p < .001), and product purchase likelihood ( = 1. 473 p < .001). In terms of the vertical involvement behavior it appeared to be significantly related to product purchase likelihood alone ( = .7 5 4, p < .05), but it had no significant effects on the other three proposed consequences. These results would appear to provide somewhat mixed suppor t for the importance of the social engagement behavior. It should be noted that this study added a structural residual covariance for any two of the four social engagement dimensions according to modification indices to improv e the structural model. All residual covariance w ere statistically significant at the level of p < .001, suggesting that the net effects o f the four social engagement dimensions outside the model are significantly correlated with each other. The following figures display the schemati c representation of the significant antecedents and consequences.

PAGE 156

156 Table 5 10 Consequences of the four social engagement dimensions Social Engagement Dimension Consequences Program Behavioral Loyalty Program Attitudinal Loyalty Audience Satisfacti on Product Purchase Likelihood Standardized path coefficient SE Standardized path coefficient SE Standardized path coefficient SE Standardized path coefficient SE Vertical Involvement .24 6 .49 2 .765 580 721 483 .7 54 .3 04 Diagonal Interact ion 136 .47 8 .4 30 .5 53 419 .4 61 .2 11 .29 0 Horizontal Intimacy 2. 592 *** .6 59 2. 724 *** .6 40 2. 149 *** .5 27 1.4 12 *** .3 35 Horizontal Influence 2. 415 *** .53 8 3. 182 *** .6 09 2. 567 *** 499 1. 473 *** .3 18 Note: p < .05, ** p < .01, *** p < .001 (two tai led). Goodness of Fit Indices: 2 = 6869.780 ( df = 3640, p < .001); CFI = .911, TLI = .902, RMSEA = .042, SRMR = .047. n = 494.

PAGE 157

157 Figure 5 1. Antecedents and consequences of the ve rtical involvement dimension

PAGE 158

158 Figure 5 2. Antecedents and consequences of the diagonal interaction dimension

PAGE 159

159 Figure 5 3. Antecedents and consequences of the horizontal intimacy dimension

PAGE 160

160 Figure 5 4 Anteceden ts and consequences of the horizontal in fluence dimension

PAGE 161

161 Antecedents and Consequences of the Overall Social Engagement The hypotheses (H1 through H11) and research questions (RQ4 to RQ5) pertained to the antecedents and consequences of the overall socia l engagement. There were nineteen exogenous variables and five endogenous variables in this model. Specifically, the antecedent and consequence variables were the same as the model with the four social engagement dimensions except for the social engagement construct, which was treated as the second order variable The data analysis was also conducted in two stages: measurement model assessment and structural model testing. Measurement Model Assessment Through the CFA proced ure using Mplus (Version 6.0 ) p rogram twenty six variables with multiple items and two variables with single indicator were specified in the measurement model. To assess the model fit, the minimum fit function Chi square for the measurement model was 6 060 .7 03 ( df = 3 637 p < .001). The goodness of fit indices for the measurement model with the overall social engagement construct were desirably above or below their recommended thresholds ( CFI = .91 9 TLI = .9 11 RMSEA = .03 7 and SRMR = .0 53), suggesting that the measurement model fit th e data adequately Comparatively, the fit indices showed that the measurement model with the four social engagement dimensions ( CFI = .927, TLI = .918, RMSEA = .035, and SRMR = .045 ) fit the data better than the measurement model with the overall social en gagement construct. As the procedure described in the above section, t his study next empirically assessed convergen t validity by examining factor loadings and the relevant p values. For acceptable construct validity, it is proposed that each item should h ave a minimum factor loading of .60 on its hypothesized latent factor (Nunnally, 1978). The desired

PAGE 162

162 convergen t validity was achieved for all constructs in the measurement model with factor loadings rang ing from .6 1 8 to .9 7 7 ( p < .001), indicating these ite ms statistically measure the constructs as intended. Particularly, the standardized factor loadings of the four dimensions on the second order variable the overall social engagement were: vertical involvement ( = .97 7 p < .001 ), diagonal interaction ( = .920, p < .001) horizontal intimacy ( = .7 60 p < .001) and horizontal influence ( = .6 2 1, p < .001). To further assess the scale reliability of all constructs included in th is measurement model, alpha was employed to test sca le internal consistency reliability. were identical i n these two measurement models. T social engagement dimensions and the overall social engag ement were .905, .871, .902, .768, and .941, respectively, indicating that all constructs in the measurement model with the overall social engagement are reliable. The mean of the overall social engagement with the fifteen items was 2.632, rang ing from 2.0 26 to 3.565. Structural Model Testing This study next examined the cau s al relationship s suggested in the structural model to answer the research questions (RQ4 to RQ5) and test the hypotheses (H1 through H11). The minimum fit function Chi square 2 fo r the structural model was 7 315.180 ( df = 3 708 p < .001). The goodness of fit indices were CFI = 901, TLI = .89 3 RMSEA = .044, and SRMR = .078. Specifically, TLI = .893 was a little bit below the cutoff values of .90, while other goodness of fit indices were desirably above (CFI) or below (RMSEA and SRMR) their recommended thresholds. By referring to modifications indices this study added a structural residual covariance for the overall

PAGE 163

163 social engagement with its four consequences, but no significant imp rovement was detected for the structural model. As there were no more theoretically plausible modifications that should be included in the structural model after the scrutiny this study continued to interpret the path coefficients with caution. Anteceden ts to the overall social engagement The hypotheses (H1 through H 7 ) and research questions (R Q4 and RQ5) were related to the antecedents to the overall social engagement. Specifically, the hypotheses (H1 to H3) posited that television program related varia bles including program genre preference, program affinity, and program involvement predict the social engagement behavior. The hypotheses (H4 to H6) addressed the relationship s between perceived characteristics of social media and the overall social engage ment. The hypothesis (H7) and research questions (RQ4 and RQ5) investigated the impact s of audience attributes, includin g motives, innovativeness, and social characteristics (i.e., interpersonal interaction and social activity). Although the directions of some hypotheses were specified based on the review of literature, this study employed a two tailed test in the structural model to emphasize the significant results of antecedents to the overall social engagement. The significant antecedents for the ov erall social engagement included program affinity ( = .207, p < .001), program involvement ( = .163, p < .001), program genre preference ( = .06 6 p < .0 1), the motive of pass ing time ( = .06 4 p < .0 5), innovativeness ( = .1 56 p < .01), and interpersonal interaction ( = .0 99 p < .0 1 ). Specificall y, the results first indicated that all program related variables such as program affinity, program involvement, and program genre preference, are predictive of the overall social engagement behavior. This suggested that viewers who possess stronger

PAGE 164

164 prefer ence for a specific type of program, show more affinity toward the program, and perceive it more important and relevant in their daily lives tend to actively utilize various social media platforms to comment, post, watch, or read anything about this televi sion show. Furthermore, it appeared that the more innovative tendencies that the individuals demonstrate; the more likely they are to employ different social media platforms to obtain information of the program, to interact with celebrities and characters of the program through Twitter TM to form in timate connection s with other viewers and the diegetic, narrative text dep icted in a program through peer to peer activities in blo gs/online discussion forums, as well as Facebo ok TM or Myspace TM With respect to the predictive power of in their real lives interpersonal interaction rather than social activity appeared to be significantly predictive of the overall social engagement experien ce. The results suggested that even though the individuals have ample opportunities to interpersonally communicate with friends, family, relatives, or others in their real lives, they desired to further engag e in their communication with other audience mem bers with different levels submerge virtual space y did not exhibit any influences on the social engagement tendenc y which was originally found to be the most salient antecedent along with innovativeness to all four social engagement dimensions. When it comes to the motives behind the overall social engagement th e result s showed passing time to be the only significant motivatio n but it had a negative impact

PAGE 165

165 on the overall social engagement. It seemed that people driven by the motivation of passing time tend to be less likely to us e various online social media platforms to interact with television content, indicating a distinct s cenario with traditional television viewing In addition, perceived characteristics of the general social media system did not play a role in predicting the overall social engagement experience. To sum up, perceptions of program and audience characteristic s rather than the perceived attributes of social media appeared to be significant predictors of the overall social engagement. The causal relationships are presented in Table 5 1 1 Table 5 1 1 Antecedents to the overall social engagement Antecedents T he Overall Social Engagement Results Standardized path coefficient SE H1 Genre preference .06 6 .02 2 Supported H2 Program affinity .2 07 .042 Supported H3 Program involvement .1 63 .037 Supported H4 Perceived ease of use .00 6 .029 Not supported H5 Compatibility .0 28 .032 Not supported H6 Social presence .009 .030 Not supported H7 Innovativeness .1 56 .05 3 Supported RQ4 Relaxation .007 .035 Not significant RQ4 Companionship .042 .03 4 Not significant RQ4 Pass time .06 4 .02 6 Significant RQ4 Entertainment .0 48 .036 Not significant RQ4 Information .0 69 .050 Not significant RQ4 Arousal .0 28 .041 Not significant RQ4 Escape .0 30 .029 Not significant RQ4 Access .0 53 .038 Not significant RQ4 Learning .018 .043 Not signific ant RQ4 Interpersonal utility .0 0 1 .04 0 Not significant RQ5 Interpersonal interaction .0 99 .03 3 Significant RQ5 Social activity .0 19 .041 Not significant Note: p < .05, ** p < .01, *** p < .001 (two tailed). The goodness of fit indices: 2 = 7315 .180 ( df = 3708, p < .001); CFI = .901, TLI = .893, RMSEA = .044, SRMR = .078.

PAGE 166

166 Consequences of the overall social engagement The hypotheses (H8 to H11) pertained to the possible consequences of the overall social engagement, including program behavior al loyalty, program attitudinal loyalty, audience satisfaction, and product purchase likelihood. As shown in Table 5 1 2 the overall social engagement was a significant and substantial predictor for all four following outcomes. More specifically, the overa ll social engagement demonstrated the positive predictive power on all proposed consequences: program behavioral loyalty ( = 9 18 p < .001), program attitudinal loyalty ( = 1.03 0 p < .001), audience satisfaction ( = 82 4 p < .001), and product purchase likelihood ( = 1. 0 87 p < .001). C omparatively, the predictive ability of the overall social engagement on product pur chase likelihood was most salient, followed by program attitudinal loyalty, program behavioral loyalty, and audience satisfaction, respectively. The results would appear to provide a definite support for the importance of the social engagement behavior. It should be noted that this study added a structural residual covariance for the overall social engagement with its four proposed consequences individually according to modification ind ices to improv e the structural model. All of the residual covariance w ere stati sti cally significant at the level of p < .001, suggesting that the net effects on the overall social engagement outside the model are significantly correlated with the net effects on these four proposed consequences Figure 5 5 is a schematic repr esentation of those variables which appear ed to be significant antec edents and consequences of the overall social engagement behavior Figure 5 6, 5 7, 5 8, 5 9, and 5 10 summarize the salient antecedents and consequences of each dimension and the overall social engagement under the proposed research framework.

PAGE 167

167 Table 5 1 2 Consequences of the overall social engagement Note: p < .05, ** p < .01, *** p < .001 (two tailed). The goodness of fit indices: 2 = 7315.180 ( df = 3708, p < .001); CFI = .901, TLI = .893, RMSEA = .044, SRMR = .078. Consequences The Overall Social Engagement Results H8 Program Behavioral Loyalty Standardized path coefficient .9 18 Supported SE .10 5 H9 Program Attitudinal Loyalty Standardized path coefficient 1.03 0 Supported SE 0.14 7 H10 Audience Satisfaction Standardized path coefficient .82 4 Supported SE .12 7 H11 Product Purchase Likelihood Standardized path coeffi cient 1.0 8 7 Supported SE .079

PAGE 168

168 Figure 5 5 Antecedents and consequences of the overall social engagement

PAGE 169

169 Figure 5 6. Visual depiction of the salient results of the vertical involvement dimension

PAGE 170

170 Figure 5 7. Visual depiction of the sa lient results of the diagonal interaction dimension

PAGE 171

171 Figure 5 8. Visual depiction of the salient results of the horizontal intimacy dimension

PAGE 172

172 Figure 5 9. Visual depiction of the salient results o f the horizontal influence dimension

PAGE 173

173 Figure 5 10. Visual depiction of the salient results of the overall social engagement

PAGE 174

174 CHAPTER 6 DISCUSSION AND CONCL USION This study adopts the active audience behaviorist approach and ai ms to achieve two research purpose s : 1) build an active audience behavior model to decipher the emerging multimedia television consumption pattern, and 2) examine the social media practices currently implemented in some media organizations by v alidating a set of social engagement scale s consequences. More specifically, t he topics discussed in this section include: 1) the implications and contributions of the four social engagement dimensions, 2) the var iance of social engagement with different television program genres, 3) the comparison on the predictive power of various antecedents to social engagement (i.e., program perceptions, perceived social media characteristics and audience attributes ) and 4) the predictive effects of social engagement on the proposed consequences (i.e., program behavioral loyalty, program attitudinal loyalty, audience satisfaction, and product purchase likelihood) In this chapter, the findings are summarized first by each to pic followed by the theoretical an d practical implications of research questions and hypotheses. Finally, limitations of this investigation and future research prospects are discussed. Summary of Findings : Social Engagement The conceptualization and o perationalization of social engagement in t his study contribute s to our understanding of the consumption of primetime network programming in a social media context. In p articular this investigation introduces the social engagement concept and proposes fou r underlying dimensions in its measure ment scale Social engagement refers to the degree of intensity or types of connections that

PAGE 175

175 audiences develop with television content through social media platforms over time. The term television content covers a broa der scope, including television programming and its relevant information, the characters or celebrities related to the program, and the professional working staff s such as producers or directors of the show. Social media here consist of an expanding array of online media applications which can facilitate information sharing, knowledge distribution, and opinion exchanges. This investigation proposes the four dimensions in social engagement i.e., v ertical involvement, diagonal interaction, horizontal intimac y, and horizontal influence. The four social engagement dimensions extend beyond the traditional, passive television viewing pattern, representing both active behavioral engagement and emotional connec tion that viewer s develop with television characters or program cont extual settings through social med ia platforms. Specifically, vertical involvement characterizes the degree to which television viewers actively use a range of social media platforms to be involved with the core program content and its relevan t information. Diagonal interaction depicts the degree of social interaction that viewers develop with the characters, celebrities, and working staffs related to their favorite shows in microblogs such as Twitter TM Horizontal intimacy indicates the extent to which individual viewers emotionally respond to a television program and the affection of the viewers toward the branded content with other audiences in blogs and online discussion forums. Horizontal influence describes the degree of identification and belonging, as well as the extent of meaningful influence in the direction or outcome of television programming in a peer related space like social networks (e.g., Facebook TM and My space TM ).

PAGE 176

176 Th e descriptive statistics from the main test reveal that the pr opensity in the social engagement activities is currently best represented b y horizontal influence experience ( M = 3.35, SD = 1.24), followed by horizontal intimacy ( M = 3 00 SD = 1.2 2 ), diagonal interaction ( M = 2.44, SD = 1.2 4 ), and vertical involvement ( M = 2.24, SD = 1. 10 ), respectively. Th e findings suggest that engagement behavior is more salient in sharing television viewing experience with friends, adding meaning to their relationship s with others, a nd building intimate connection s with other audience members surround ing television program s However, the social engagement experience is less potent in establishing interaction with the media figures of the program searching for program relevant informa tion or being involv ed with program content alone The se findings essentially correspond to several academic research and industry reports, which found that people utilizing social media to interact with television are mainly driven by the ir social and ps ychological needs such as sharing (television) experiences, recognition (by what they are watching), and being heard (through contributions) ( Harris Interactive, 2011 ). By conducting a three stage research pro cess this study introduces and tests a relia ble scale comprised of fifteen items to measure the social engagement construct. While both a first order and second order factor structure fit the observed data adequately this investigation prefers adopt ing the first order representation by virtue of si mplicity and better fit. Further, t o assess scale reliability, this study use s coefficient alpha to estimate internal consistency reliability for the overall social engagement scale and its four dimensions. The results demonstrate that the scale with the fifteen sample items perform ed well in capturing the proposed social engagement

PAGE 177

177 successfully illustrates that social engagement is conceptually and empiricall y different from attitude toward the program (program affinity) and program involvement T he participants in the pilot test and the main test cover a b r oad cross section of television viewers and social media users indicating that the social engagement co nstruct is relevant and applicable to different demographic groups Summary of Findings : Social Engagement with Different Program Genres Television audience social engagement experiences could vary greatly with different types of television programmin g The findings from p rior studies demonstrate d that preferences of different types of content would stimulate diverse social viewing experiences and communication patterns surrounding certain program genres In particular, different television ge nre s c ould influence the way viewers talk, chat, or interact with each other while watching television or afterward s (Geerts, Cesar, & Bulterman, 2008 ; Simmons, 2008). The current study provides additional evidence show ing that there are significant v aria social engagement experiences with dramas, reality shows, sitcoms, animated comedies, and game/talk shows in the online world through diverse social media activities When examining the ability of each program genre to inspire various social engagement activities the present study finds that game/talk show s are most leading program genre whose viewers utilize the most in an expanding array of social media platforms to be involve d with television content ( M = 2.97, SD = 1.22), to inter act with media figures of the program in Twitter TM ( M = 2.94, SD = 1.33), and to build intimate connections with other audience members related to the program in blogs/online forums ( M = 3.32, SD = 1.03) However, the viewers for the genre of animated come d y display

PAGE 178

178 the highest level of engagement for the horizontal influence dimension ( M = 3.58, SD = 112). When it comes to the tendency in each social engagement experience surrounding these five different types of program ming th is study finds that there are no significant differences among these five program genres in the measurement of diagonal interaction, horizontal intimacy, or horizontal influence. However, t he current study does discover that there is a significant difference among these five progra m genres along the vertical involvement dimension ( F (4,489) = 4. 36 p = .0018 ) In particular, s ignificant differences are found for the followings : game/talk shows versus dramas ( t (489) = 3.69, p < .05 ) and game/talk shows versus reality shows ( t (489) = 3.13, p < .05 ) The pairwise comparison findings suggest that viewers tend to be more aggressively engage d in game/talk shows than reality programs and scripted dramas through the use of a n expand ing array of social media platforms to connect with the c ore content programming and/or ancillary information of the program Summary of Findings : Antecedents to Social Engagement This study presents an active audience behavioral model which integrates the theory of television program choice, technology accep tance model, innovation diffusion theory, social presence theory, and the uses and gratifications approach. The ma in purpose of the integration is to offer a comprehensive framework to better understand why television audiences are actively involved with v arious social media platforms to connect with television content, the characters/celebrities, and other audience members Specifically, the current study identifies the determinants of the four social engagement activities and the overall social engagement behavior from the perspectives of television program perceptions, social media characteristics, and

PAGE 179

179 audience attributes. T he various sets of findings provide a good basis for comparing the strengths and weaknesses of each theoretical branch that forms the research framework. In terms of the four social engagement dimensions the salient predictors consist of program affinity and genre preference all perceived social media characteristics (i.e., perceived ease of use, compatibility, social presence ) and different aspects of audience attributes (i.e., the motivations of interpersonal utility, innovativeness, interpersonal interaction, and social activity). Comparatively, both innovativeness and offline social activity demonstrate the universal ly predictive power for all four social engagement dimensions Program affinity is found to be a significant predictor for three social engagement dimensions except for the diagonal interaction behavior, while compatibility and the motive of interpersonal utility ha ve effects on two social engagement dimensions. The antecedents to the single social engagement dimension include : g enre preference ( to diagonal interaction ) perceived ease of use ( to vertical involvement ) social presence ( to horizontal intimacy ) and inter personal interaction ( to diagonal interaction ) In general, it appears innovative tendencies and social activities play a relatively more significant role than the program perception variabl es and social media characteristics in predicting the social engagement tendency regarding specific types of engagement behavior. With respect to the significant antecedents to the overall social engagement, this study identifies the following determinants: program af finity program involvement, program genre preference, the motive of pass ing time, innovativeness, and interpersonal interaction In p articular, the current study discovers that all program

PAGE 180

180 related variables especially program affinity, are strongly predi ctive of the overall social engagement behavior. With respect to the predictive power of audience attributes, the motive of passing time is found to be the salient motivation alone but displays a negative impact on the overall social engagement. This inves tigation further reveals that the more innovat ive the individuals are ; the more likely they are to utilize various social media platforms to read, watch, post, or comment on something about television programming In addition, one of ne social characteristics interpersonal interaction appears to be a significant positive predictor of the overall social engagement experience. However, t offline social activities in their real lives are found to be in significant alth ough they were originally salient in predicting the four social engagement behaviors along with innovativeness. Finally, t he perceived characteristics of the general social media system s do not exhibit any impacts on the overall social engagement experienc e (i.e., perceived ease of use, compatibility, and social presence) Summary of Findings : Consequences of Social Engagement To investigate the predictive effects of social engagement, this study further proposes four consequences of the social engagemen t experience : program behavioral loyalty, program attitudinal loyalty, audience satisfaction, and product purchase likelihood. The current study, however, finds inconsistent predictive effects on the four proposed outcomes when examining the four dimension s and the overall social engagement. Specifically, f or each social engagement dimension, the current study discovers that both horizontal intimacy and ho rizontal influence are strongly predict ive of all of the four proposed outcomes, but display opposite e ffects. T he findings show that horizontal intimacy demonstrates the negative effects on all consequences; while

PAGE 181

181 horizontal influence activities exhibit positive influences on these four outcomes In addition, vertical involvement is found to be significant ly related to product purchase likelihood alone, but it is not salient in predicting the other three proposed consequences. Furthermore, th e present study finds no significant associations between the diagonal interaction behavior and any of the proposed c on sequences. These findings provide somewhat mixed support for the importance of each social engagement behavior. When it comes to the overall social engagement, the findings illustrate that the overall social engagement is a significant and substantial pr edictor for all proposed consequences Comparatively, the predictive effects of the overall social engagement on product purchase likelihood is the most salient, followed by program attitudinal loyalty, program behavioral loyalty, and audience satisfaction respectively. The se results appear to provide a definite and strong support for the importance of the overall social engagement behavior. Theoretical Implications Benefits of the Integrated Framework for Active Audience Behavior The active audience pe rspectives are now achieving premier status in the behaviorist research domain due to the continuously e volving media environment and audience media consumption patterns. In particular, with the prosperity of online social media platforms relevant to telev ision programming, one way, traditional television viewing pattern is gradually shift ing to the cross media, multitasking consumption mode. In contrast to the television, the viewers who utilize an expanding array of social media platforms to connect with television content represent more initiatives, freedom, and autonomy in the process of media choice. The process indicat es that media use today is purposive and

PAGE 182

182 planned; med ia content is selecte d; and viewing experience is engaged In short, the contemporary television audiences (also social media users) could interact with television content anytime and anywhere in the platform s that best suit their needs. To further decipher this emerging multi media t elevision consumption pattern, the present study proposes an integrated framework incorporating various theoretical threads of mass communication, marketing, and information processing. The first benefit of the integrated approach for active audienc e behavior i s that it provide s a comprehensive picture to better understand the social engagement process. As suggested in the audience behaviorist research tradition, audiences are variably active across several qualitative dimensions and along the tempor al dimension before, during, and after media exposure (Rubin, 19 87a 19 87b ). Thus, th e present study first approaches the multimedia television consumption pattern by validating the qualitative dimensions in social engagement as vertical involvement, diago nal interaction, horizontal intimacy, and horizontal influence. Th is study then examines the temporal dimension of before, during, and after the social engagement experience. In particular, the antecedents and consequences are examined for each social enga gement dimension and the overall social engagement behavior. Another benefit of the integrated framework i s that it present s a good basis for comparing the strengths and weaknesses of each theoretical branch that forms the active audience behavior model Specifically, this inve stigation i dentifies three categories of explanatory factors to predict the social viewing experience from the perspectives of media content (i.e., perceptions o f television programs) media channel (i.e., perceived characteristics of social media) and media user (i.e., audience

PAGE 183

18 3 attributes) The predictive ability of each factor is tested and compared in predict ing different social engagement patterns At the same time, the predictive effects of social engagement on the following pr oposed consequences are evaluated including program behavioral and attitudinal loyalty, audience satisfaction, and product purchase likelihood. In particular, b y assessing the strength s or weaknesses of different determinants through path analys e s, the cu rrent study could identify which predictor s play more significant role s for each social engagement behavior or the overall social viewing experience. M o st important ly the finding s of the dissertation demonstrate that the social engagement process is a com posite result which is determined by multiple components jointly under the integrated framework of active audience behavior Contributions of Social Engagement and I ts Measurement Scale Th is investigation on social engagement construct and its measuremen t scale makes a variety of important contributions toward s advancing knowledge within the behaviorist audience research domain. Starting with the theory specific contributions, the current study introduces a new social engagement construct that essential ly ex tends the television audience research scope. While the research on the attributes of engagement and its effective measurement scale have received much attention in television and advertising related literature this is one of the first studies that fo cu s on the television engaging experience in a new social media context Further more the conceptualization and operationalization of social engagement with television content synthesize several bodies of literature to help validate an active audience beha vior theory and establish a more comprehensive framework for the viewer engagement process in a social media context. Finally, given that most research in the audience behaviorist tradition has been limited by the separate examination of engagement in

PAGE 184

184 diff erent media platform contexts and the lack of valid measurement indicators in an integrated fashion, th e findings from the current s tudy are indicative that the contexts As a behavioral oriented and qualitative focused measure of audiences, social engagement deciphers the social viewing experience happening simultaneously at both the content and the platform level s Specifically, the measurement instruments offer systematic metrics to pro file active audience behavior, incorporating prior perspectives of viewer engagement (Askwith, 2007; Epps, 2009; Russell, Norman, & Heckler, 2004a) and platform engagement (Calder, Malthouse, & Schaedel, 2009; Haven, 2007; Takashi, 2010; Yanga & Kangb, 200 9). T he current study ad a pts the three relation ship s in social interaction (i.e., viewer program, viewer characters / celebrities, and viewer viewer) proposed by prior television connectedness or engagement studies ( Askwith, 2007; Russell, Norman, & Heckler, 2004a ). At the same time, this investigation identifies the four distinct, prominent attributes in the engaging process as involvement, interaction, intimacy, and influence The four attributes synthesize the previous point s of view of engagement with dif ferent media platforms including traditional mass media (Kilger & Romer, 2007), the Internet ( Calder, Malthouse, & Schaedel, 2009 ), blogs ( Yanga & Kangb, 2009 ), and social networks ( Takashi, 2010 ). Most important ly t he two dimensional analysis approach undertaken in this study depicts a n aggregate, vivid profile of multimedia television consumers, who have already developed ideas about what type of content fits best on which social media platforms. Given that t he innate characteristics of the specific so cial medium definitely shape the type of content and the model of communication on that platform, television

PAGE 185

185 audiences have take n advantage of the capability of each platform to approach diverse television content and best reflect t heir consumption needs For example, audiences have recognized Twitter TM as one of the best communication channel s that bridge the conversation s between viewers and characters / celebrities, while viewing social networks (e.g., Facebook TM and My s p a ce TM ) more as a peer related space for sharing content/information and garnering r ecognition through contributions. In addition to demonstrating th at social engagement experience should be viewed at both the content and media levels simultaneously, t his study is unique in that it identifie s the four underlying dimensions in social engagement behavior They are vertical involvement, diagonal interaction, horizontal intimacy, and horizontal influence. The multi dimensional social engagement experience is particularly relevant, given the drama tic changes in media technologies like alternative video platforms and the increasing fragmentation of television audiences with decreas ing loyalty. The four social engagement dimensions demonstrate a continuum in which audience social media activities sur rounding television content range from the lowe st level (i.e., vertical involvement) to the high est level (i.e., horizontal influence) and the resultant various impacts In particular the vertical involvement dimension seems to present the lowest degree of this social viewing experience reflect ing the participatory behavior in relation to the core content and/or ancillary information of a program. Wh ereas vertical involvement depicts various s ocial media touchpoints, the dimension of diagonal interactio n represents communications and dialogues between audiences and media figures through microblogs like Twitter TM The horizontal intimacy dimension goes

PAGE 186

186 beyond diagonal interaction to characterize the affection or sentiment that audience s possess for televi sion program ming, measuring peer to peer activities such as s in blogs and online forums. This is somewhat supported by prior study, which suggested that television audiences are no t really engaged unless they are talking about plot and characters rather than hype s and actors (Haven, 2007). The horizontal influence dimension goes and social identification, and further rep resents the individual audience influential potential on his/her friends in social networks like Facebook TM through sharing and recommendation of the program. In other words, the social engagement experience starts with the relationship between audiences and the branded television content and continues to extend that relationship to other audience members. Furthermore the current study empirically validates that the higher levels of social engagement behaviors could yield more salient predictive effects than lower levels on audience satisfaction, program loyalty, and product purchase likelihood. In particular, t he findings regarding the consequences of each social engagement dimension indicate that the higher levels of social engagement behaviors (i.e., h orizontal intimacy and horizontal influence ) have significant effects on audience satisfaction, program altitudinal and behavioral loyalty, and product purchase likelihood simultaneously. However, the impacts of the lower levels of social engagement behavi ors are only apparent on product purchase likelihood alone (vertical involveme nt ) or not consequential at all ( diagonal interaction ) Accordingly, iden tifying the television audience along the four social engagement dimensions instead of audience size and

PAGE 187

187 viewing frequency in the new media environments is significantly meaningful to the behaviorist audience research domain. Further more the theoretical implications of social engagement could contribute to other relevant streams of research, including advert ising, marketing, and audience measurement. Different Tendenc ies in Social Engagement with Television Program Genres There is a long tradition of studying program genres in the television research domain. Although some genre studies were conducted from audience behavior perspectives, they mainly examined what constitutes a genre, how certain television programs fit into a genre, and especially how audiences engage with different genre s to understand and enjoy programs (Bignell, 2003). T he sociability an d communication pattern surrounding different television genres have yet to be explicitly and fully investigated. The current study attempts to bridge the gap and systematically examine the sociability of different television program genres in a social med ia context. In general, the study finds evidence for the different levels of sociability and communication patterns surrounding these five program genres (i.e., dramas, reality shows, sitcoms, animated comedies, and game/talk shows) T he genre talk/game shows appears to be most effective in inducing viewers to utilize an expanding array of social media platforms to be involved with television content, to interact with media persona of the program in Twitter TM and to build intimate connections with other audience members related to the program s in blogs/online forums. In addition to demonstrating diverse degrees of socia l engagement ability of these five genres, the investigation further illustrates that viewers tend to be more aggressively engaged in tal k/ game shows than reality programs and scripted dramas in the vertical involvement

PAGE 188

188 dimension, involving a range of social media touchpoints to access the core content and/or ancillary information of the program. So what is so different about the genre of t alk /game shows ? P rior stud ies claimed that talk shows conflict and confrontation among guests and studio audience m 1997, p. 106). Peck (1995) noted that hos ts play a major role in these conflicts and s with the hosts. Moreover, t alk shows are found to be appealing to certain television audience s because the topics discussed tend t o center on issues of high relevan ce such as famil ies, sexual topics, dating, and relationships ( Greeber g Sherry, Busselle, Hnilo, & Smith, 1997 ) In particular, television viewers are drawn to television talk shows that deal with personal and relational topics (Rubin & Step, 1997). It could be argued that talk show viewers appear be more socially active in multimedia behavior than viewers of other programming genres due to the personal focus and participatory nature in talk shows. The findings in the current study resonate well with o ne industry engagement research, reporting that talk shows like The Oprah Winfrey Show exhibit multimedia synergy in terms of word of mouth. In particular, this type of talk show is most effective in inducing their audien ces to engage in cross media activities, such as watching the shows on television, reading The Oprah Magazine or visiting the Oprah.com we bsite ( Fetto, 20 10 ). In addition, t alk show viewers are found to show higher interests in a parasocial relationship w ith the hosts as well as informative, realistically perceived content, mostly focusing on celebrities and families (Rubin, Haridakis, & Eyal, 2003). The se academic and industry findings to some degree

PAGE 189

189 support the conclusion that talk/game shows are most eff ective in inspir ing audiences to search for relevant information or access programming content therefore representing a higher engagement tendency in the vertical involvement dimension compared to other program genres This investigation, however, doe s not find statistical differences for the program genres in the other three social engagement behaviors, i.e., diagonal interaction, horizontal intimacy, and horizontal influence. The findings suggest that program genres do not play a role when people use Twitter TM to interact with characters/celebrities of the program, or influence audiences to use blogs and online discuss ion forums to read, write, or comment on the program. Likewise, engagement activities in social networks such as sharing the program video in Facebook TM do not vary significantly among these five programming genres. It appears that the innate characteristics of Twitter TM blogs/online discussion forums, and social networks primarily shape the type of content and the model of communication on that platform. Given the strong preference and affinity possessed by specific program genre viewers, expertise on one specific platform such as Twitter TM Blogs, or Facebook TM may not vary greatly resu lting in similar levels of interactive, intimate, and influential engagement behavior among these five program genres. By contrast, the discrepancy of a range of social media platforms (i.e., RSS feeds, widgets, podcasts, mobile applications, and social bookmarks ) may lead to significant variances in the vertical involvement activities happening around different programming genres.

PAGE 190

190 Audience Innovativeness as a Critical Determinant of Social Engagement This study asses s es the strengths and weaknesses of each theoretical branch that predicts the social engagement experience The a udience dispositional factor innovativeness is found to be the most salient determinant of the overall social engagement behavior and the fou r social engagement dimensions In particular, the current study finds that the degree of a udience innovativeness essentially influence s viewers to utiliz e a n expanding array of social media to be involved with the program content and its ancillary informa tion. Likewise compared to the predictive ability of the other antecedents, innovativeness exhibits the strongest influence on the propensity of people using Twitter TM to communicate with characters, celebrities, and other professional working staffs of t he program. tional trait innovativeness is purported to contribute to his o r her cognitive response towards making an innovation adoption decision (Lin, 2004 p. 447 ). The degree of innovativeness, novelty seeking, and creat ive ability displayed in single out those who ha ve a greater propensity for early adoption of an innovation (Hirschman, 1980). Among the five categories of innovation adopters (i.e., innovators, early adopters, early majo rity, late majority, laggards) i nnovators or early adopters were regarded to possess a higher degree of personal innovativeness (Roger s 1995). Similarly, recent studies on innovative attributes and W eb based technology adoption generally support the effe cts of innovativeness on innovation adoption. In particular, prior studies found that the more innovativeness an individual possesses the higher the level of Internet use (Busselle, Reagan, Pinkleton, & Jackson, 1999). Likewise, Lin revealed that an indivi significant predictor for personal computer adoption (1998) as well as webcasting

PAGE 191

191 adoption (2004). The significant role of personal innovativeness seems to hold true in the context of social engagement as well. At prese nt, the consumption pattern of utilizing various social media to interact with television content is still prevalent among a small propor tion of the television population. A recent industry research reported that among online U.S. adults, two in five have gone online or utilized social media to comment, post, watch, or read something about television programming (Harris Interactive, 2011). With reg ard to using social media in relation to television content, this study f ound that among the total of 1,314 onl ine social media users, 41% of them ha ve utilized social media to interact with television shows or programs. Furthermore, t he social media employed to interact with television programming are mainly concentrated on th e mo st popular platforms, such as Face book TM Twitter TM Myspace TM and mobile texting. S everal entertainment oriented social networks like GetGlue are still not well known by the majority of online users, even though these platforms provide the check in applications specific for television pr ogramming Accordingly, to plunge oneself into becoming a socially engaged audience in all dimensions the online user has to solicit a certain level of curiosity, initiative, and skills in exploring this relatively new and somewhat challenging digital com munication mode It is logical that at this stage of social television diffusion, the social media experience with television programming will be comprised of those online users who represent more innovative ness in their social engagement pattern s Im portance of Audience Social Characteristics as Predictors of Social Engagement In predicting the four social engagement behaviors and the overall social engagement experience, the current study identifies audience social characteristics

PAGE 192

192 (i.e., impersonal interaction and social activity) in their real lives as important predictors Although prior stud ies concluded personal attributes like interpersonal interaction and social activities have effects on their online media use behavior (DiMaggio, Hargittai, Neuman, & Robinson, 2001), the predictive effects of the social characteristics were in consistent (Hampton & Wellman, 2003; Kraut, Patterson, Lundmark, Kiesler, Mukopadhyay, & Scherlis, 1998; Papacharissi & Rubin, 2002). This investi gation also yields mixed results regarding the predictive effects of the two social characteristics social activity and interpersonal interaction Turning to the study reveal that social activity is one of the most salient determinants predicting all four social engagement patterns It appears that viewers who maintain more social networks and social activities in their real lives also tend to be active in the online world using various online so cial media platforms to engage with television content. The results are not unexpected As a number of studies focused on the relation ship s between personal traits and s ocial media use illustrate, the level of extroversion personality trait has the greates Amiel & Sargent, 200 0 ; Haridakis & Hanson, 2009 ). More specifically, sociability, social contact, and a preference for companionship are likely to be pursued with particular intensity by those hi gh in extr oversion (Amiel & Sargent, 2004). T he extroverts who have more social networks and social activities in their real lives are found to be heavier social media users interacting in Facebook TM etc. ( Hall 200 5 ). Most importantly, this study resona tes well with one prior study focused on social viewing experience in YouTube (Haridakis & Hanson, 2009). The

PAGE 193

193 authors empirically discovered that socially active people are more likely to turn to YouTube for the social interaction and co viewing purposes taking YouTube as a way of sharing online activities with family and friends with whom they have existing social ties (Haridakis & Hanson, 2009). A main reason driv ing the extroverted individuals to be more active online i s that online environment i s a place where individuals can supplement their offline relationships and further solidify their established contacts in the real world For example, Hampton and Wellman (2003) considered the potential of the Internet as a social medium that can supplemen t and Th is logic is based on the social enhancement premise, which states that the extroverted and outgoing persons are motivated to add online contacts to their established large network of offline fri ends (Zywic a & Danowski, 2008). Another probable explanation is that the extroverted individuals possess more successful experiences in their real social lives, and they are more likely to experience more successful social interaction in the online environment (Zywic a & Danowski, 2008) It seems plausible that the extroverted successful social interaction skills and experience s may drive them to be more engaged in social media activities to interact with other audience members related to televisi on programming. The findings here may challenge results from those previous studies based on media compensation hypotheses suggesting that people who are less sociable and dissatisfied with face to face interaction are more likely to use media as compensa tion (e.g., Papacharissi & Rubin 2000) As for the second audience social characteristic, interpersonal interaction, t he empirical validation of the predictive ability of interpersonal interaction to social

PAGE 194

194 engagement is particularly interesting. I nterpe rsonal interaction exhibits the predictive effects but in opposite directions on the diagonal interaction dimension and the overall social engagement. For the four social engagement dimensions, the findings reveal that the predictive ability of interperson al interaction is salient on diagonal interaction alone in a negative way Th e findings suggest that audiences who have ample opportunities to interpersonally communicate with friends, family, relatives, or others in their real lives tend to avoid the comm unication opportunities with media persona of the program through Twitter TM However, the predictive effect of interpersonal interaction is significant and positive on the overall social engagement behavior. Th e behavioral discovery implies that audiences, who do have ample opportunities or are satisfied with their interpersonal communication in their own lives, would still be inclined to utilize various platforms to engage in different levels of social media activities surrounding television programming. Specific to the diagonal interaction dimension the social engagement experience characterizes the degree of social interaction that viewers develop with characters, celebrities, and working staffs related to their favorite shows in microblogs like Twitter TM The social interaction between viewers and media figures to some degree is a type of parasocial interaction, in which viewers believe they know the media persona as they do a friend treat ing the interaction as an interpersonal relationship. Thus, peop le who are lonely or find reduced satisfaction in their face to face encounters tend to search for an alternative means of communicati on such as television talk shows or other media channels ( e.g., Papacharissi & Rubin, 2000; Perse & Rubin, 1990; Rubin, H aridakis, & Eyal, 2003). By contrast, if the individuals have ample opportunities to interpersonally

PAGE 195

195 communicate with friends, family, relatives, or others in their real lives, they may be less likely to engage in dialog s or communication s with media perso na through Twitter TM In a sense, t he empirical findings provide the evidence in support of the functional alternatives theory which relie s on the premise be satisfied in more than one way, and different habits, pract ices, and acts can fulfill the ojerback, & Hedinsson, 1986, p. 48). Impact s of Program Affinity, Involvement, and Genre Preference on Social Engagement This investigation identifies three program rela ted factors as predictors of social engagement, i.e., program affinity, program involvement, and genre preference. The findings show that program affinity plays a critical role in predicting three social engagement dimensions except for diagonal interactio n, whereas genre preference displays the predictive effect on the diagonal interaction behavior. Program involvement is found to be associated with none of the four social engagement dimensions. Further, when examining the overall social engagement experie nce, all program related factors are strongly predictive of the social viewing experience, especially in terms of the predictive power of program affinity on the overall social engagement. Television program affinity is defined as an attitudinal construc t, measuring the It appears that the audiences who possess higher levels of program affinity tend to utilize a range of social media platforms to be involved with the core program content, and to establish intimate connections with other audience members in blogs/online forums. Further, audience program affinity is the most critical determinant in predicting

PAGE 196

196 the audience behavior al tendency in signaling their fan status or recommending the program to friends in social networks. However, audience program affinity has no association with diagonal interaction, which involves more interpersonal interactions between television viewers and media persona in Twitter TM The f indings are not surprising as the deepest engagement usually happens at the level of media content but the media channels ( Epps, 2009 ). The vertical involvement dimension signifies that audiences utilize multiple social media touchpoints to access televi sion content and it s relevant information. The horizontal intimacy dimension is also content driven, capturing a deeper and more intimate connection between the viewers and the diegetic, narrative text depicted in a program through peer to peer posting or commenting activities Furthermore, in terms of horizontal influence, the desire to signal identification and influential tendency about a particular show in a peer related space like Facebook TM requires him/her to demonstrate strong er aff inity towards the program content Thus, audience program affinity is salient in predicting the three social engagement behaviors mainly driven by programming content On the other hand, the diagonal interaction behavior is mainly motivated by the paras ocial relationship developed between audiences and media persona. While audience p rogram affinity m ight help maintain and intensify the parasocial relationship, media persona lities, who have ample opportunities to directly interact with audiences online or offline seem to play a more critical role in the process. Moreover, the media persona in different television program genres may solicit different degrees of para social interaction. For example, the talk show genre is found to be one of the most

PAGE 197

197 effectiv e types of programming to foster such parasocial interaction ( Rubin, Haridakis, & Eyal, 2003 ) Thus, driven by the strong preference for specific program genres in which media persona may inspire diverse levels of parasocial interaction, audiences tend to illustrate significant variances in diagonal interaction tendency In a sense, whereas the programming content is critical, the media figures seem to play a more essential role in the diagonal interaction process When it comes to the overall social engagement experience, all program r elated factors exhibit the strong est predictive abilities compared to the other variables related to social media characteristics and individual audience attributes. The findings once again are indicative of the value o f content imply ing still multimedia television consumption environments. In particular, in the contemporary, interactive video consumption networks like the Internet and social media plat forms, the definition of television co ntent expands to a broad er scope, which includ es the core programming content, the characters, celebrities, and other media persona of the program, and even various media acti vities. Accordingly, the deepest level of social engagement is primarily driven b y the quality of content, regardless of which content formats and social media platforms are used Thus, how to develop the best strategy to foster viewer affinity towards the specific television content and to fu rther promot e involvement with the program become the most critical issues when examining audience social engagement tendency Different Influences of the Perceived Social Media Characteristics as Predictors of Social Engagement When approaching the social engagement experience from the perspec tive of media characteristics this study integrates the technology diffusion theory, technology

PAGE 198

198 acceptance model, and social presence theory to examine the predictive effects in the context of the perceived characteristics of social media. In particular, t his investigation illustrates that perceived ease of use, compatibility, and social presence of the general social media system are statistically significant predictors but yield mixed results. Comparatively, compatibility is the strongest determinant, e xhibiting the positive effects on the tendencies in horizontal intimacy and horizontal influence. Perceived ease of use and social presence are found to be negatively associated with vertical involvement and horizontal intimacy, respectively. Nevertheless, none of the three perceived social media characteristics are found to be related to the overall social engagement tendency. As for the strongest predictor, c ompatibility is used to assess whether using social media to interact with television programmin g is compatible with most aspects of the horizontal intimacy and horizontal influence engagement behaviors, the findings here suggest that audiences who feel utilizing s ocial media to interact with television content is compatible with their television viewing experience tend to build intimate connections with other audience members through blog/electronic board posting and commenting activities Likewise, if the individu al audiences perceive that using social media platforms to interact with television programming fit s their lifestyle, they are also more likely to engage in sharing program s and making recommendation s in social networks like Facebook TM The predictive powe r of compatibility on the two social engagement behaviors, i.e., horizontal intimacy and horizontal influence, are noteworthy. The findings first resonate

PAGE 199

199 well with several previous studies o n innovative attribute s and online services adoption. For example Tornatzky and Klein (1982) employed a meta analysis approach to examine innovation characteristics and innovation adoption implementation and concluded that there is a positive, though not always statistically significant, relationship between the compat ibility of an innovation and its adoption. Wu and Wang (2005) discovered that compatibility is the strongest factor predicting the intention to use mobile commerce, compared to other predictors such as perceived usefulness, perceived ease of use, cost, an d perceived risk. Lin (2001) also discovered that compatibility would adoption decision s regarding Internet based services since the author found that online service adoption is not compatible with non W eb based adoption rates The different diffusion rates of various s ocial media platforms may contribute to the fact that the predictive ability of compatibility is only salient on the social engagement behaviors related to Facebook TM Myspace TM blogs, and online forums Compar atively, this study finds that the social med ia involving the dimensions of horizontal influence and horizontal intimacy are most widely used (i.e., Facebook TM Myspace TM and blogs/online forums ), while the platforms employed in vertical involve ment behav ior are less recogniz able or even utilized ( i.e., RSS feeds, podcasts, social bookmarks, widgets, and mobile applications). Furthermore, w hile the phenomen a of social engagement attract much attention from the television and advertising industries, the pra ctice is still in its nascent stage and has not yet evolved into a common television consumption pattern for the majority of television viewers Therefore, it may be hard for th e viewers to develop definite perceptions o f th ose social media platforms that they

PAGE 200

200 have rarely used and to further estimate their actual social engagement behavior. Until the majority of television audiences are actively involved with a range of social media platforms to engage with television content on a regular basis, the predict ive power of compatibility of various social media platforms will be more salient o n the social engagement tendency The significance of perceive d ease of use in predicting vertical involvement is also dictated by the current different diffusion levels of various social media platforms Prior study argued that the effects of perceived ease of use on new technology adoption become weak and even disappeared, if a critical mass has already adopted the technology (Lin, 2004). For example, the perceived ease of use was found to have no effects on the cable television adoption once the ca ble penetration rate reached 80% (Lin, 2004). According to the latest industry database, the penetration rate of Facebook TM in the United States is 50.28%, and the ratio is even higher in the online population, reaching 65.20% ( www.socialbakers.com ). Given the comparatively high adoption rate of Facebook TM Myspace TM blogs/online forums, and Twitter TM it is not surprising that there ar e no significant effects of perceived ease of use on the three social engagement activities happening around those platforms (i.e., diagonal interaction, horizontal intimacy, and horizontal influence) On the other hand, the factor of perceived ease of u se appears to play a different role in predicting the vertical involvement pattern in a negative way T he vertical involvement activities facilitated by these social media platforms (i.e., RSS feeds, podcasts, social bookmarks, widgets, and mobile applicat ions) seem to require more effort and overall know how on the part of the viewers. For example, using social

PAGE 201

201 online involves searching, storing, and uplo ading the video co ntent they want Such content driven social media activities are more laborious and time consuming for the inexperienced viewers, compared to other social media activities such as chatting with friends about a program o n Facebook TM or following a character in Twitter TM Thus, the perceived difficulty and actual complexity encountered by social media users may deter them from further engagement in this type of social engagem ent behavior. In addition, the significant variances in the knowledge and skills indi cated by social me dia users for these multiple platforms (i.e., RSS feeds, podcasts, social bookmarks, widgets, and mobile applications) may result in great er variations in vertical involvement behavior In alignment with the predictive effects of compat ibility and perceived ease of use, the findings in this study impl y that certain perceived characteristics (e.g., perceived ease of use) of social media platforms m ay become less pertinent when consumers become more technologically proficient through incr eas ed exposure to new platforms and thus decreasing learning curves Moreover, the social media characteristics that are more relevant to audience lifestyles and television viewing habits (e.g., c ompatibility), combined with skills or technology competency factors may play a larger role in the social engagement process. When it comes to the predictive ability of social presence of the general social system on the social engagement pattern the negative effects of social presence are detected in the horizo ntal intimacy behavior. The perceptional discrepancy in social presence between specific social media platform s and the general social media system m ay attribute to the unexpected results. Social presence refers to the ability of

PAGE 202

202 computer mediated communi cation media to transmit interpersonal ly oriented content effectively used to help differentiate between informational and interpersonal uses of the Internet ( p. 181 ). Prior study concluded t hat online audiences, who perceive the Internet as warm, social, and active, use it primarily to satisfy many needs such as interpersonal utility and entertainment desire (Papacharissi & Rubin, 2000). However, the current study suggests that audiences wh o perceive the general social media system as social, personal, and sensitive tend to avoid engaging in blog posting or commenting activities to build intimate connection s with other audience members. Specifically, this study examines the audience presence perceptions o f the context of the overall social media systems, noting that the horizontal intimacy dimension involves specific social media platforms like blogs and online discussion forums. It is plausible that audiences p erce ive the ability of the general social media s ystem to transmit interpersonal type of content to be dissimilar with the capability of a specific social media platform like blogs or online discussion forum Thus, future research on the predictive ability of social presence re garding a specific social media platform may yield more robust and valid results for social engagement behavior s Instrumental Motivations behind Social Engagement This investigation develops a systematic scale of social engagement motive by integratin g the previous motivation s of traditional t elevision viewing (Rubin, 1983), the Internet use (Papacharissi & Rubin, 2000), and YouTube video viewing (Haridak is & Hanson, 2009). Through scale development and validation process this study locates ten prima ry motives behind social engagement, including relaxation, companionship, passing time, entertainment, information, arousal, escape, access, learning, and

PAGE 203

203 interpersonal utility. In particular, three motivations, access, learning, and interpersonal utility, are derived from new media uses like the Internet and YouTube T he two motives, interpersonal utility and passing time, are found to be related to the social engagement experience. Given that the act of social engagement combines television consumption w ith the use of s ocial media platforms it is not surprising that the motive s rel evant to both new media uses ( interpersonal utility ) and traditional television viewing (passing time) are identified in the current study simultaneously It is interesting th at t he two social engagement patterns diagonal interaction and horizontal influence, are reflected in the interpersonal utility motivation but in opposite directions The motive of i nterpersonal utility found in this study is ver y similar to the construct which characterizes the degree of inclusion, affection, social interaction, expressive need, and surveillance in the context of the Internet uses The interpersonal utility motive behind diagonal interaction suggests that if audiences are driven by the inclusion, affection, and social interaction purposes, they tend to avoid communicating or having dialog s with media persona in Twitter TM By contrast if audiences are motivated by interpersonal utility needs, they are more likely to share their television experiences and seek recognition through participations or contributions i n Facebook TM In addition, passing time is the single salient motive that predict s a negative effect on the overall social engagement behavior. The salient effects of the motives interpersonal utility and passing time on social engagement are interesting but not unexpected. For the motive of interpersonal utility, the logic behind its negative effect on diagonal interaction behavior i s akin to the predictor of interpersonal interaction one of the audience social characteristics. The

PAGE 204

204 nature of b oth predictors impl ies that th e social interaction experience with media characters via a platform like Twitter TM cannot really satisfy audienc interpersonal communication needs or substitute for the interpersonal interaction activities in their real lives. On the other hand, if the audiences are deeply motivated by the inclusion, affection, and social interaction needs, they are more likely to go to a space that maximizes peer interactions like Facebook TM to satisfy these needs through such activities as sharing contributi ng, and making recommendation s Finally, the negative effect of passing time on the ov erall social engagement provides th e evidence in support of the notion tha t social media uses in the context of television consumption are more driven by instrumental than ritualized needs The two motivations i.e., interpersonal utility and passing time behind social engagement can be s een as either instrumental or ritualized oriented, a notion p roposed and examined by many uses and gratifications scholars. S tudies in the tradition suggest that these media orientations reflect the amount and type of media use, media attitude and expect ation (Rubin, 2009 ). Specifically, ritualized orientation means using media more habitually to consume time and for diversion. I t entails greater exposure to and affinity with the medium ( Rubin, 2009, p. 172 ) I nstrumental orientation focuses on seeking certain message content for information reasons 172). Instrumental use is active and purposive, suggest ing utility, intention, selectivity, and involvement (Rubin, 2009) The motive of interpersonal utility behind s ocial engagement behavior reflects that social media uses in the context of television consumption are instrumental ly orient ed, which features the utility selectivity, and involvement elements. On the other hand, a ritualized orientation p assing time present s

PAGE 205

205 While audiences may actively engage in various social media activities in relation to television programming to fill the free ti me they have it appears that the y do not consider using social media to interact with television programming as the ideal way to pass time for a diversion Salient Effects of Social Engagement on Program Loyalty, Audience Satisfaction, and Product Purchase Likelihood To investigate the predictive effects of social engagement, this study further proposes four consequences of the social engagement experience : program behavioral loyalty, program attitudinal loyalty, audience satisfaction, and product purchase likelihood. T he overall social engagement is found to have significant effects on audience satisfaction, program loyalty, and product purchase intention. Consistent with the findings here, prior research from academia and industry discovered increasing cross platform, multitasking media consumption patterns would help promote program attitudinal and behavioral loyalty (Ha & Chan Olmsted, 20 04; Lu & Lo, 2007) enhance audience satisfaction (Lin, 1993; Palmgree & Rayburn, 1985; Perse & Rubin, 1998) and improve the likelihood of product purchase (Kilger & Romer, 2007). In particular, the overall social engagement behavior bears the strongest positive relationship with the likelihood of purchasing products that have been advertised on the program websites. These products include memorabilia / merchandise of television stations or networks, memorabilia/merchandise of televis ion shows, and the products shown in that television program. A s suggested by prior study, however, it is still challenging for television managers who plan to utilize their website as a platform to conduct e commerce sts and experience in televis ion e commerce (Ha & Chan Olmsted, 2004). Similarly, o ne recent

PAGE 206

206 study examined the effects of media usage and discovered that cross media involvement with televis ed programs could improve the program loyalty, and further promote product purchase intention ( Lin & Cho, 2010 ). Nevertheless, the authors of this study also admitted that television e commerce and underutilized by the current online users (Lin & Cho, 2010). While the overall social engagement pattern could yield the salient predictive effects on all proposed consequence s i.e., audience satisfaction, program attitudinal and behavioral loyalty, and product purchas e likelihood it is worthwhile to investigate what specific social engagement dimension may play a more critical role in the process. As discussed earlier, it appears that the higher levels of social engagement dimensions such as horizontal intimacy and ho rizontal influence exhibit stronger significant effects on more after viewing consequences than the lower levels of social engagement behaviors like vertical involvement and diagonal interaction. Specifically, horizontal influence behavior bears strong p ositive relationships with program attitudinal and behavioral loyalty, audience satisfaction, and product purchase likelihood, while horizontal intimacy behavior displays the negative effects on the four consequences. The different directional effects may be attributed to the innate characteristics of each social media platform that shape the type of content and the model of communication on that platform. Due to the anonymous nature and core n most online discussion f orums and blogs, people are more likely to express both positive and negative feelings on these two platforms freely By contrast, the identity and persona in

PAGE 207

207 social networks like Facebook TM may to some degree diminish the t endency for negative expressions (e.g., there is no dislike button in Facebook TM ). Finally, the diagonal interaction behavior bears no associations with any outcomes implying that focusing on the para social relationship driven engagement experience alon e may not be an optimal strategy for program marketing and promotion. Practical Implications Implications for the Audience Research Industry This investigation highlights the contribution of the social engagement construct and its measurement scale in overcom ing the limited nature of traditional audience measures based on reach and frequency The television ratings system as a traditional audience measure has been a useful metric to advertisers and broadcasters for a long time, but the system mainly emp hasize s audience size and volume of viewing As audience television consumption habits continue to fragment across devices and social media platforms it is now critical to intentionality and engagement towards a televisi on program through multiple media platforms. In this context, social engagement, defined as the social viewing experience via the multiple media platform s is meaningful to the audience research industry when examin ing the quality and quantity of audiences for commercial purposes The introduction of the social engagement construct and its validation empirically point to the utility of developing a social television rating s system in the audience research industry Although the ownership of television sets in the U. S. households has dropped for the first time in twenty years, the on demand video platforms such as playback device DVRs and online streaming, as well as cross media, multitasking television consumption continue to grow Thus traditional me ans of measuring

PAGE 208

208 television audience are insufficient in capturing th e evolving video viewing behavior It might be valuable to develop a social television ratings system that can incorporate the social engagement viewing activities via multiple platforms. The social television ratings may aim at aggregating publicly available social commentary and filtering and normalizing this data fr o m disparate sources (e. g. Facebook TM Myspace TM Twitter TM and event based social networks like GetGlue, Miso, and Fours quare TM etc.) to further assess the underlying sentiment of a broader range of online users (Calic, 2011) In addition, t h e metric may provide a more complete view of the engagement associated with television programs a cross diverse social media platforms in real time as well as beyond the initial airing time slot of each episode T hese findings might be just the data set necessary to become the de facto social television rating to rival Nielsen Calic, November 7, 2011 ) However, it should be noted tha t a social television ratings system is not meant to replace the current ratings systems for television advertisers In fact, it should function as a n active complement for the present passive audience measures based on audience size or volume of viewing. Because the additional qualitative audience data are highly valuable for the current advertisers and broadcasters to estimate advertising effectiveness and expense allocation the development and implementation of actual engagement related metrics could b e the next phase of audience research agenda. In fact, the world largest audience research company, Nielsen, has already jumped on the bandwagon to video usage a cross the three screens TV, Internet, and mobile ( e.g., A2/M2 Three Screen Report ) but the company has not embedded the

PAGE 209

209 social media platforms into their analytics This study envision s that the social television ratings system in alignment with traditi onal passive metrics will play a more critical role in the social viewing environments in the near future Implications for the Television Industry From the managerial perspective, as the television industries increasingly compete against alternative dist ribution platforms while facing a fragmented audience with decreasing loyalty, it is vital for television organizations to develop a more long term relationship with their viewers through cross platform strategies. This study, therefore, empirically addres ses the issue of whether television broadcasters / advertisers should devote resources to develop a social engagement strategy and how they should approach it. This is important as the careful deployment of resources is most essential in a competitive enviro nment. This study anticipates that there will be more partnerships between the television and social media industries in the near future, which will significantly impact all stakeholders in the television industry, including program producer, cable/broadca st networks, local stations, television service providers (like satellite broadcasters and IPTV providers), and advertisers. Th e current study has several practical implications for programming producers and b roadcasters in the television industry In p articular, t he four social engagement dimensions identified in the social viewing experience are essentially valuable to the television broadcasters, program producers, and advertisers B ecause the four dimensions depict a social engagement profile, an agg regate description of the types and levels of engagement the audiences exhibit in a social media context. Based on the social engagement profiles, programming broadcasters and producers can deliver differentiated marketing strategies utilizing specific soc ial media platforms to meet

PAGE 210

210 diverse marketing objectives from television shows, broadcast and cable networks, and local stations. Given that the lower levels of social engagement activities (i.e., vertical involvement and diagonal interaction) happen pri marily around a range of program related social media and microblogs like Twitter TM programming broadcasters and producers may employ these social media platforms to create program awareness and enhance audience viewership of the program ming For example, some event based social networks, such as BuddyTV, GetGlue, Miso and Tunerfish, have partnered with television service providers verse and Direct TV) to improve programming awareness through their check in services allow ing fans to connec t through mobile and online platforms and share their opinions about certain shows across their social profiles In addition, broadcasters may adopt real time platforms like Twitter TM to drive tune in and take advantage of the growing trend of simultaneous Web TV usage by dispensing online information in tandem with the airing of the programs (Leavy, 2010). ant for all February 23, 2010). Turning to the higher levels of social engagement behaviors ( i.e., horizontal intimacy and horizontal influence ) programming broad casters and producers may explore the capabilities of blogs, online discussion forums social networks and similar social media platforms to promote affinity for the brand and eventually enhance their loyalty to the program. In p articula r, establishing a presence for certain

PAGE 211

211 programs on social networks, such as Facebook TM and Myspace TM could build affinity for the program brand by providing a platform for discussion am ong devoted fans (Leavy, 2010). When viewers share an active connectio social network like Facebook TM the viral nature of social networks in syndicating and reposting content would be an effective way of driving audience awareness, involvement, and hopefully loyalty. In a sense, the use of soc ial media to enhance audience engagement has tremendous marketing potentials as the socially engaged viewers are more likely to stick with a show, talk about the show, and spread word of ough real time interaction can humanize broadcasters, which enables them to listen to, affirm, and amplify the opinions of their fans (Leavy, 2010). When it comes to the salient predictors of the social engagement activities, t he findings are partially ill ustrative of innovative tendency and extroverted personality. It is advisable for programming broadcasters and producers to design more innovative functions online and social viewing applications related to their televisi on programming o n diverse social media platforms. Further more these online innovative functions and social viewing applications should be compatible with the majority of social media usage habits and lifestyle especially in terms of privacy an d cost consideration In addition, social engagement viewers are mainly driven by instrumental motives, whi ch mean s these viewers are purposeful in seek ing certain television content for inclusion, affections, social interaction, and free expression needs. Th e instrumental orientation essentially implies that current television audiences are more attached to content than to the medium per se; thus, programming

PAGE 212

212 broadcasters and producers should develop and implement an optimal cross media strategy to repurpo se and leverage their content to maximize More specifically, the screen television strategy. On one hand, the sig nificance of all program related factors for the overall social engagement behavior the multimedia television consumption environments. On the other hand, the individual au dience s innate characteristics of the specific social medium definitely shape the type of content and the model of communication on that platform, implying and platfor m is Accordingly, delivering platform t o suit an is the most critical consideration for program broadcasters and producers when implementing a successful multi screen television strategy. In fact, have already developed clear ideas about what type of content fits best on which social media platforms For example, social media users tend to watch full length television shows in the content sharing communities while keepi ng updated and receiv ing news and information about programs via Twitter TM and social networks. Finally, when it comes to the major player, advertiser s in the television industry, the findings in this study highlight the growing importance of cross platf orm advertising campaigns in reaching targeted and involved audiences It is no doubt that, at present, the social engagement behavior has not evolved into a common consumption pattern among television viewers and the demographic structure of social engag ement viewers

PAGE 213

213 is not representative of the overall U.S. population. Nonetheless, the evolving social viewing pattern increasingly adopted by online users provides a great opportunity for advertisers because they can target these television audiences in a highly engaged environment by extending their television advertising for particular shows to the equivalent social media channels and mobile devices. In particular, from the perspectives of media planning, there are great opportunities for advertisers and merchandisers, who leverage consumer behavior to create unique and useful social viewing experiences along the temporal dimension before, during, and after television exposure. This study and one recent industry survey ( Harris Interactive, 2011 ) found that audiences are prone to engage in different social engagement patterns before, while, and after they are watching television. For media planning purposes, t he three stage social television viewing experience provides diverse opportunities for advertisers. For example, if the marketers and advertisers are more interested in the social conversations happening across all of these viewer engagements, and check in services offered by some entertainment focus social networks like GetGlue and social television gui I t is also valuable to assess the effectiveness of social engagement on product purchase intentions and actual purchase behaviors. The present study provides empirical evidence in support of the predictive effects of social engagement on the likelihood of purchasing program product purchase intentions into actual purchase behaviors is an important area of investigation for marketers who inten t to fully utilize cross media platforms to boost the value of advertising. For example, a good strategy for multi platform advertising

PAGE 214

214 campaign s is to convert the socially engaged television viewers to be the advertisers advocates on social media platfor ms. Th e notion is based on the fact that social ly engage d viewers tend to show a higher propensity in sharing their interests in television shows on various online social venues. In addition, audiences trust the opinions of those they know or share similar interests with on their online and offline networks. Further more the multiple platform campaigns should not only focus on advertising space on social media platforms but also tap into the demographics that actually talk about certain products and televis ion programs. According to a recent industry analysis, people between the ages of 35 49 make up the highest percentage of online discussion of television shows at 30% compared to other demographic sectors (Nielsen, 2011). Therefore, advertisers may tap int o this consumer group to maximize the effectiveness of their multi platform advertising campaigns Implications for the Social Media Industry This investigation also has several practical implications for the social media industry. First, considering t he symbolize relationship between television and social media, social media companies may establish partnerships with entertainment brands including programming producer and content distributors to add television viewing to social media experience s In par ticular social media companies should incorporate specific online initiatives and applications related to social media consumption behaviors, allowing fans to interact with one another as well as with the shows and their stars thus incr easing the value and enhance the experience of that platform For example, Facebook TM recently partnered with Direct TV, Netflix and Hulu to track Facebook TM time programming watching activities. The social media

PAGE 215

215 platform also summarizes the u put top movies and episodes on display for his/her friends to view. M ore impor tantly, considering the volume and sentiment of chatter occurring online about entertainment brands and shows across various social media platform s, social media companies should develop analytic s tools to integrate television shows related conversations across the Internet To some degree, to have access to such d ata exclusively increase the value of the social media platforms for advertising and m edia sources (Calic, 2011) One typical example of this type of database is Trendrr.tv index (http://www.trendrr.tv/), which incorporates Twitter TM mentions, public Facebook TM posts, Miso and GetGlue check in to analyze engagement around television and brands by processing real time activities across a variety of social media platforms. Limitations This study highlights some valuable f indings related to utilizing social media t o engage with television content over time However, there are several l imitations that should be taken into account when evaluating the results of the research and interpreting the conclusions. While the use of online consumer panels sampled from the rea l online population for most of the hypotheses and research questions helps enhance the external validity of the findings, these results should not be generalized to all online users. Given that the hypotheses and research questions in this study necessita ted the use of a purposive sample of online users who possess certain social media experiences related to television content consumption these findings are not necessarily applicable to all online consumers or social media users. In addition, considering that

PAGE 216

216 sample frame is consisted of solely U.S. based online consumers, the relationships identified in the results may not be applicable to the social media users who possess the social engagement experiences in other countries. It should b e noted that the online instruments employed in the main test is somewhat long since they examine four dimensions of social engagement behaviors and a branch of antecedents and consequences of the social engagement pattern. It is possible that the particip ants of the online survey become fatigue when they filled out the questionnaire, thus further influencing the quality of data. The low incident rates in the main test (34.6%) and the pilot test (37.0%) to some degree reflect this fact. It is no doubt that there are some missing data, but the percentage of missing data is negligible (0.1% 0.2%). In addition, one consideration is related to the television program sample used in this study ly, t he current study construct s a specific program list by referring to an online Social Television Charts database ( http://trendrr.tv/ ), which includes various social media activities surrounding television shows, such as public Facebook TM posts, Twitter TM mentions, GetGlue check ins, and Miso check ins. The specific program list of this study was comprised of twenty programs/shows with the highest degree of those social media activities during August 29 th to September 4 th of 2011, the week before the main survey was implemented However, it is possible that the degree of social engagement inspired by the episode of the program may not be the same as the following episode after a week. It is recommended to consult more, l atest social television data sources to improve the validity of the online survey instrument.

PAGE 217

217 As discussed previously this study identifies three major exploratory factors of social engagement from the perspectives of media content, media platform s, and audience attributes. Thus, the theoretical and practical implications of this investigation also center on these aspects. T here are other external factors such as market and economic issues that might impact the adoption process In addition, it is plausible that there might be some net effects of external factors in combin ation with the social engagement behaviors to influence the four proposed consequences. Therefore, it is necessary to take these external factors into account when in terpreting the social engagement process. A limitation of this study lies on its use of an online survey as a method of data collection; engagement experience. Prior studies suggested that user en gagement is indicated as a process, including the point of engagement (and reengagement), engagement, and television content also experiences the cycling stages, including dec ision making ( at the point of engagement ) watching (while engagement), and reviewing (reengagement). It may be valuable to employ the experiment research method to capture the whole process of social engagement behavior in a social media context. There ar e also some statistical considerations when examining the social engagement factor structure and some scale s By adopting a three stage research process, this study introduces and tests a reliable scale comprised of fifteen items to measure the social engagement construct. By analyzing the first order factor structure loaded from the CFA procedure, there are different numbers of

PAGE 218

218 indicators involving each social engagement dimension. However, based on the general practice in scale deve lopment and validation it appears to be a better choice for the multiple item scale to includ e three or more indicators in order to enhance this On a related note, while this study identifies that both a first order and second order factor structure fi t the observed data adequately, this investigation prefers the first order with four factors representation by virtue of simplicity and better fit of indices Nevertheless, it should be noted that there are higher correlations among the four social engagement dimensions, especially in terms of the correlation between the dimension s of vertical involvement and diagonal interaction. Therefore, it is suggested to focus more on the overall social engagement instead of the four dimensions In addition, this study acknowledges that the measurement scale of the four dimensions is typically behavioral not very strong conceptually yet To incorporate the social media use depth and breadth dimensions for each measurement item may improve the scale validity. T he operationalization of some variables in this study needs modification and verification In particular, this study employs a single item to measure program behavioral loyalty and audience satisfaction In building measurement models, multipl e indicator measurement models have been preferred since Gerbing, 1988). In addition, when examining the effects of social media characteristics on social eng of the general social media system rather than a specific social media platform, which may lead to discrepant views of social media characteristics. Thus, these caveats

PAGE 219

219 should be take n into account when assessing the effects of the variables in structural equation modeling. Another statistical consideration is related to variance explained in factor analysis by using the Mplus (Version 6.0 ) program Basically, variance explained in a set of variables by a factor is not given in the Mplus program due to two reasons proposed by acted in which case the concept of variance explained by a factor is not clear Therefore, this study does not report explained variance data when utilizing the Mplus (Version 6.0 ) program to conduct factor analysis and structural equa tion modeling. Future Research The integrated theoretical framework and empirical findings provided by the present study should serve as a good start for future research. This study first recognizes that validity testing of a newly established construct is an ongoing effort. This investigation offers the evidence that the social engagement construct is unique from other constructs. But, th is study also acknowledges the need for further discriminant and nomological validity testing, particularly to fully differentiate social engagement from involvement, attitude, and connectedness. Thus, more robust statistical testing of the validity of social engagement requires a wider range of audience attitude and involvement levels, and viewer connectedness. Further m ore the present study employs twenty television programs and shows within five program genres for the scale validation and the antecedents and consequences testing. Given the salient social engagement power inspired by talk/game shows, the number of this type of programs included in the main test is comparatively small compared to other genre programming

PAGE 220

220 Thus, this investigation may include a broader array of online users, programs, and genres to fully validate the social engagement construct and its pred ictors and outcomes The investigation on the relationships among the four social engagement dimensions is a promising direction for future research on social engagement behavior. A lthough this study discovers that the four social engagement dimensions co ver a spectrum of social viewing behavior ranging from the lower to the higher level s I t still requires more empirical evidence to validate the cau s al relationships among the four dimensions through the structural equation modeling approach. In particula r, further research may focus on whether the lower levels of dimensions predict the higher levels or whether there are reciprocal relationships among these four social engagement dimensions. Th e investigator anticipates that the examination of the relation ships among the four dimension s could further advance our understanding of the social engag ement experience with television programming in a social media context One fruitful approach would be for future studies to address the cycling process of soc ial engagement experience and its resultant effects As discussed earlier, the social engagement viewing may experience three stages, i.e., point of engagement (reengagement), engagement, and, disengagement. The point of social engagement may happen at any points during the social interaction when viewers actively search for information or advice for television content. For example, Facebook TM and Twitter TM may influence people to decide what to read, what videos to watch, and want news stories to follow. O nce the viewers engage in television programming, their attention and interest must maintain. It is also suggested that the intensity of engagement is

PAGE 221

221 varied among different program genres. Drama and action shows are found to be low social programming, whi le reality games and sports shows are both high in social engagement Accordingly, to further investigate the cycling process of social engagement and its ensuing effects may highlight the different attributes represented in the different stage s of social engagement. Future research may also measure the social engagement activities associated with a specific program genre. Whereas the present study does investigate the exploratory factors and consequences of social engagement with different television pro gramming it is worth pointing out the possible predictors of social engagement with program genre s and its predictive effects on the proposed consequences. Given that different program genres could stimulate diverse levels of sociability and communication patterns surrounding that program genre, it is valuable to further investigate and compare the predictive power of a set of exploratory factor s such as program related variables, perceived social media characteristics, and audience attributes, and then d iscover which variables play a more critical role regarding different program genre engagement In addition, it could also be interesting to find out the predictive effects of social engagement with different program genres, and compare these effects on pr ogram loyalty, audience satisfaction, and product purchase likelihood. An important point in the audience behaviorist research is to identify the As a starting point, this study incorporate s thre e sets of exploratory factors and four possible after viewing consequences to examine the relationships between social engagement and its antecedents and consequences. However, these causal relationships validated in the

PAGE 222

222 current results may be moderated or mediated by several demographic factors. level of reward) variable that effects the direction and/or strength of the relationship between an independent or predict ive variable and a dependent or criterion variable (Baron & Kenny, 1986, p. 1174) T he mediator variables may 1176). Therefore, examin ing the functionality of demographic variables in testing the antecedents and consequences of social engagement behavior is a promising direction that points to further study as to whether these variables function as mediators or moderators Another fruitful a pproach for future research is to compar e social engagement behavior among different demographic sectors. innovativeness and social characteristics play the most important roles in predicting social engagement behavior, it is nece ssary to further investigate the different levels of social engagement patterns among different age, gender, and ethnic groups. According to prior literature about innovation diffusion regarding the Internet, the earlier adopters of the Internet are more l ikely to be male, of the ethnic majority, younger, better educated, and more affluent than the general population (Bonfadelli, 2002; Chen & Wellman, 2004). Thus, it is plausible that the use of various social media platforms to engage wit h television progr amming may vary differently across diverse demographic groups and further influence its predictors and consequences. Turning to the predictors of social engagement patterns, the theory of technology fluidity, in particular, the Internet fluidity, should b e explored as a potential determinant

PAGE 223

223 perceptions o f the Internet fluidity significantly influence the adoption of webcasting (online video streaming) (Lin, 2004). The I nternet is recognized as a fluid medium due to its ability of transforming into text, graphic, audio, voice, or visual modalities, or a combination of these communication platforms. Thus, the perceived fluidity of the Internet and even different social med ia platforms may p lay a role when people utilize social media platform s to in teract with television content. By the same token, the more predictive ability to influence the social engagement experience. Thus, the future depth and breadth use experience. In addition, further modification, development, and verification of the scales that measur e perceived characteristics of each social media platform may help advance our understanding of the predictive effects of social media. Finally, future research may analyze more possible after viewing consequences of social engagement behavior. In particul ar, the evidence supporting the relationship between viewer behavior and television content calls for future study with a focus on the program context effects and advertising effectiveness. As suggested by prior scholars, viewer connectedness as an alterna tive to traditional mood manipulations could directly impact the effectiveness of advertising (Russell, Norman, Heckler, 2004b). Likewise, the current study anticipates that the multi platform engagement behavior in relation to television content will infl uence how the advertising is evaluated. In addition, exploring the predictive effects of social engagement on the effectiveness of product placements could product valuable information, especially in terms of interactive online product

PAGE 224

224 placement. Prior res media usage could lead to better website loyalty, which further improves the use of interactive online product placement (Lin & Cho, 2010). Accordingly, inclusion of more consequences related to the social e ngagement experience may help inform the practice of advertisers and broadcasters as well as contribute to our understating of the evolving cross platform, multitasking television consumption pattern.

PAGE 225

225 A PPENDIX A LITERATURES ON ENGAG EMENT AND MEASUREMEN T SCALES Sources Purpose Context Attributes Measurement Scales/Description Russell, Norman & Heckler (2004a) To measure the parasocial relationship between television viewers with television programs and characters in those programs TV Escape: defined th e cathartic element that connects a viewer to a television program. Watching ___ is an escape for me. problems. If I am in a bad mood, watching ___ puts me in a better mood. Modeling: measures a social learning process by capturing the degree to which individuals relate their lives to the lives of characters. I learn how to handle real life situation by watching ____. I get ideas from ____ about how to interact in my own life. I relate what happens in ____ to my own life. Fashion: represents the extent to which a viewer is influenced appurtenance. I like the clothes they wear on ____. I like the hairstyles on _____. I often buy clothing style that I have seen in ___. Imitation: charact erizes the inclination to imitate the patterns. I imitate the gestures and facial expressions from the characters in _. I find myself saying phrases from ___ when I interact with other people. I try to speak like the charact ers in ____. Aspiration: identifies how people aspiring to actually be on the show or meet with the characters. I would love to be an actor in ____. I would love to meet the characters of ____. Paraphernalia: measures the degree to which people co llect items to bring the show into their real world. I have objectives that relate to ___ (badge, book, picture, etc.). I read books if they are related to ___.

PAGE 226

226 Kilger & Romer (2007) To explore the effects of media engagement on product purchase likel ihood TV magazines the Internet Inspiration I have inspired by this program, magazine, or Internet site; I have an emotional connection to this program, magazine, or site. Trustworthy I trust that this program, magazine, or website tells the truth and do not sensationalize things. I also feel safe giving this website my personal information. Life enhancing I am always learning about new things and places from this program, magazine, or website things that help me make better decision in my life. Social involvement This program, magazine, or website constantly provides fodder for conversation that I have with friends and family. Personal timeout This program, magazine, or Internet site is special to me the time I spent with this media element is enjoyable and considered Advertising attention receptivity I am open to viewing/reading advertising on this program, magazine, or Internet site because it is interesting and relevant to me. Television Pers onal connection I have a personal association with the characters/situations in this vehicle, and I would sign up to receive a newsletter or products offered relating to this vehicle. Near and Dear This program is part of my regular schedule and I dev ote my full attention to it. The Internet Interactivity/community I enjoy and benefit from the feedback of other users of this site Enjoyment/attraction This site piques my curiosity. I really enjoy visiting this site.

PAGE 227

227 Calder, Malthouse & Schaedel (2009) To examine the relationship of online engagement with advertising effectiveness The Internet Stimulation and Inspiration It inspires me in my own life. This site makes me think of things in new ways. This site stimulates my thinking about lots of different topics. This site makes me a more interesting person. Some stories on this site touch me deep down. Social Facilitation I bring up things I have seen on this site in conversations with many other people. This site often gives me something to talk about. I use things from this site in discussions or arguments with people I know. Temporal This is one of the sites I always go to anytime I am surfing the web. I use it as a big part of getting my news for the day. It helps me to get my day started in the morning. Self Esteem and Civic Mindedness Using this site makes me feel like a better citizen. Using this site makes a difference in my life. This site reflects my values. It makes me more a part of my community I am a better person for using this site. Intrinsic Enjoyment Going to this site improves my mood, makes me happier. I like to kick back and wind down with it.

PAGE 228

228 I like to go to this site when I am eating or taking a break. Whil e I am on this site, I don't think about other sites I might go to. Utilitarian This site helps me make good purchase decisions. You learn how to improve yourself from this site. This site provides information that helps me make important decisions. This site helps me better manage my money. I give advice and tips to people I know based Participation and Socializing I do quite a bit of socializing on this site. I contribute to the conversation on this site. I o ften feel guilty about the amount of time I spend on this site socializing. I should probably cut back on the amount of time I spend on this site socializing. Community I'm as interested in input from other users as I am in the regular content on this site. A big reason I like this site is what I get from other users. This site does a good job of getting its visitors to contribute or provide feedback. visit this site. se Overall, the visitors to this site are pretty knowledgeable about the topics it covers so you can learn from them.

PAGE 229

229 Takahashi (2010) To measure Japanese engagement with social networking sites Social netwo rking sites (e.g., Myspace and Mixi) Information seeking activity and selectivity Seeking, collecting and sharing information relevant to daily life of close friends, school life or general interests, news, and events given by Mixi news and transnational i ssues, virtual authenticity. Connectivity Forming a connection to people or groups, transnational and trans age connectivity, disembedding from immediate locale, multi uchis connectivity via phones, and connectivity with soto, uchi creation and recreat ion. Bricolage from different communities nationally and impression management with profiles, maimiku communities. Participation Lack of political participation through fear and disbelief in efficacy, participation in taste community and transnational community. Yanga & Kangb (2009) To validate blog engagement Blogs Interactivity How interested you were in reading the How comfortable you would feel if they were asked to interact with the blogger. ideas and thoughts How likely you would be to link to the r blog if you have one. Self company connection Company __ reflect who I am. I can identify with Company __. I feel a personal connection to Company _.

PAGE 230

230 I can use Company ___ to communicate who I am to other people. I think Company __ could help me become the type of person I want to be. ___ suits me well. Company attitude Reputable/unrepeatable, Responsible/irresponsible Financially stable/unstable, Established/fly by0night Long run oriented/short run or iented Word of Mouth communication intention I would encourage friends to buy products from Company___. I would encourage family members or relatives to buy products from Company ___. I would recommend Company ___ products to someone who asked my ad vice. I would say positive things about Company ___ and its products to other people. Epps (2009) Haven (2007) To validate online engagement Social media and rich Internet application Involvement: the presence of a person at various brand touchpoints. Visitor to a site or applications Page views or page view equivalents per visitor Time spent per session or per application Repeat visitors Frequency of visit Subscriptions (to publications, email, RSS, or other services) TV viewership Mobile applicatio n use Content consumption via syndicated partners Interaction: the interaction of a person at various brand Actions within a page Videos played

PAGE 231

231 touchpoints Community contributions Ratings, reviews, and votes submitted Photo or video uploaded Text messa ges sent Quizzes taken Content saved or tagged Subscriptions renewed Intimacy: the affection of a person for a brand Sentiment measured in blog posts, blog comments, and discussion forums Call center feedback Search traffic that come from a branded s earch term Influence: the likelihood of a person to advocate on behalf of the brand Forwarded content Tagged content Widget and video embeds Friends and fans in social networks The rate at which content spreads over time Satisfaction rating Gift subscr iptions

PAGE 232

232 APPENDIX B ASSENT SCRIPT Assent Script July XX, 2011 I am a researcher from the University of Florida who is conducting a study to learn about how television audiences use various social media to engage with television content. Specifically you will be asked to share your opinions and perceptions about your overall television viewing experience, social media experience, and engagement with television content in a social media context. The online survey will take no more than 30 minutes to complete. All your responses will be kept confidential within reasonable limits. Only people directly involved with this project will have access to the surveys. Your participation is completely voluntary and there is no penalty for not participating. You have the right to withdraw from the study at anytime without consequences. If you have any questions or concerns about completing the questionnaire or about participating in this study, you may contact me at (352) 846 5415 or at miaoguo@ufl.edu. Research at the UF is overseen by the Institutional Review Board 02 UFIRB #2011 U 0097. Questions regarding your rights as a participant should be addressed to: (352) 392 0433 or irb2@ufl.edu. Thank you in advance for your time! Sincerely, M. Melissa Guo Mass Communication Doctoral Program P. O. Box 118400 University of Florida Gainesville, FL 32611 2250

PAGE 233

233 APPENDIX C QUESTIONNAIRE QUESTIONNAIRE Q 1. Which of the following social media and online applications have you used before? (Plea se select all that apply) Facebook Twitter Tagged Flickr Miso Myspace Tumblr YouTube Gowalla Philo Bebo Reddit StumbleUpon Podcasts Starling Friendster Digg Foursquare RSS Feeds GetGlue Hi5 (5) Delicious Vimeo Widgets Ning Blogs (e.g., Xanga WordPress) Online Discussion Forums FunnyOrDie Mobile texting and applications None of them Q2. How often do you usually use each of the following social media and onli ne applications? Very Frequently Frequently Occasionally Rarely Very Rarely Facebook 5 4 3 2 1 Myspace 5 4 3 2 1 Bebo 5 4 3 2 1 Blogs (e.g., Xanga WordPress) 5 4 3 2 1 Twitter 5 4 3 2 1 Tumblr 5 4 3 2 1 Reddit 5 4 3 2 1 Digg 5 4 3 2 1 Delicious 5 4 3 2 1 Online Discussion Forums 5 4 3 2 1 Tagged 5 4 3 2 1 YouTube 5 4 3 2 1 StumbleUpon 5 4 3 2 1 Foursquare 5 4 3 2 1 Vimeo 5 4 3 2 1 FunnyOrDie 5 4 3 2 1

PAGE 234

234 Flickr 5 4 3 2 1 Gowalla 5 4 3 2 1 Podcasts 5 4 3 2 1 RSS Feeds 5 4 3 2 1 Widgets 5 4 3 2 1 Mobile texting and applications 5 4 3 2 1 Miso 5 4 3 2 1 Philo 5 4 3 2 1 Starling 5 4 3 2 1 GetGlue 5 4 3 2 1 Ning 5 4 3 2 1 Friendster 5 4 3 2 1 Hi5 5 4 3 2 1 None of them Q 3 Have you ever used soci al media to comment, post, watch, or read anything about the following television programs? (Please select all that apply) Glee Pretty Little Liars True Blood The Simpsons Big Brother Gossip Girls South Park Family Guy Criminal Minds Jer sey Shore Conan Teen Mom NCIS Monday Night Raw The Office The Vampire Diaries How I Met Your Mother The Big Bang Theory Talent Keeping Up With the Kardashians None of them of Block

PAGE 235

235 Q4. When do you typically use social media to comment, post, watch, or read anything about each of the following television programs? Before I watch the program on TV While I watch the program on TV After I watch the program on TV Glee Big Brother Criminal Minds NCIS How I Met Your Mother Pretty Little Liars Gossip Girls Jersey Shore Monday Night Raw The Big Bang Theory True Blood South Park Conan The Office Ameri The Simpsons Family Guy Teen Mom The Vampire Diaries Keeping Up With the Kardashians Q5. Which one is your most favorite among the following television programs you have chosen before? (Please select ON LY one) Glee Pretty Little Liars True Blood The Simpsons Big Brother Gossip Girls South Park Family Guy Criminal Minds Jersey Shore Conan Teen Mom NCIS Monday Night Raw The Office The Vampire Diaries How I Met Your Mother The Big Bang T heory Talent Keeping Up With the Kardashians None of them

PAGE 236

236 Q 6 The following statements describe how you use social media to interact with your favorite show. Please indicate how much you agree with each of the following statements. Strongly Disagree Disagree Neutral Agree Strongly Agree I am a follower of the program (including actors, writers, producers, etc.) in Twitter. 1 2 3 4 5 RSS feeds or podcasts. 1 2 3 4 5 I have used my mobile phone to watch video clips, check photos and text alerts, or play games relevant to the program. 1 2 3 4 5 I have uploaded or forwarded videos or photos relevant to the program. 1 2 3 4 5 I have used check in apps for the program in GetGlue, Foursquare, Miso, Philo, or Starling, etc.. 1 2 3 4 5 I have read blog posts relevant to the program. 1 2 3 4 5 I have written or commented on blog posts relevant to the program. 1 2 3 4 5 online discussion forums. 1 2 3 4 5 I have written or commented on the forums. 1 2 3 4 5 microblogs (e.g., Twitter). 1 2 3 4 5 I have written or commented on the (e.g., Twit ter). 1 2 3 4 5 I have used social bookmarks (e.g. Digg and Delicious) to tag the program. 1 2 3 4 5 I am a f an of the program and share it with my friends in social networks (e.g., Facebook and Myspace ). 1 2 3 4 5 I have written or commented on the (e.g., Facebook and Myspace ). 1 2 3 4 5 I have used widgets to embed the program's video clips or photos online. 1 2 3 4 5

PAGE 237

237 Q 7 Please tell us how much you agree with each of the following statements about you r favorite show. Strongly Disagree Disagree Neutral Agree Strongly Agree I would feel lost without the program to watch. 1 2 3 4 5 program, I really miss it. 1 2 3 4 5 Watching the program is one of the most import ant things I do each day or each week. 1 2 3 4 5 Q 8 Please use the following adjectives to tell us how you feel about your favorite show. Irrelevant 1 2 3_ 4 5 Relevant Means nothing to me 1 2 3_ 4 5 Means a lot to me 1 2 3_ 4 5 Matters to me Uninterested 1 2 3_ 4 5 Intere sted Superfluous 1 2 3_ 4 5 Vital Nonessential 1 2 3_ 4 5 Essential Q9. When you watch each of the following types of progra ms, how much attention do you typically pay each types of programs? Extremely Very Moderately Slight Not at all Reality Shows (e.g., Big Brother, Jersey Shore) 5 4 3 2 1 Game/Talk Shows (e.g., Monday Night Raw, Conan) 5 4 3 2 1 Animated Comed ies (e.g., The Simpsons, Family Guy) 5 4 3 2 1 Drams (e.g., NCIS, Glee) 5 4 3 2 1 Sitcoms (e.g., How I Met Your Mother, The Office) 5 4 3 2 1

PAGE 238

238 Q10. How much do you enjoy watching each of the following types of programs? Extremely Very Moderately Sl ight Not at all Reality Shows (e.g., Big Brother, Jersey Shore) 5 4 3 2 1 Game/Talk Shows (e.g., Monday Night Raw, Conan) 5 4 3 2 1 Animated Comedies (e.g., The Simpsons, Family Guy) 5 4 3 2 1 Drams (e.g., NCIS, Glee) 5 4 3 2 1 Sitcoms (e.g., How I Met Your Mother, The Office) 5 4 3 2 1 Q11. Please indicate how much you agree with each of the following statements about your favorite show. Strongly Disagree Disagree Neutral Agree Strongly Agree Over the past month, I have not miss ed any episodes of the program when they broadcast on television. 1 2 3 4 5 I would recommend the program to others. 1 2 3 4 5 I think of myself as a loyal viewer of the program. 1 2 3 4 5 I would be willing to watch the program rather than other shows. 1 2 3 4 5 Q12. How satisfied were you with watching your favorite TV show? Not at all satisfied 1 2 3_ 4 5 Very satisfied after watching the program, would you be more likely to buy them? Definitely Probably Be unsure Probably not Definitely not Memorabilia/merchandise of the TV station/network 5 4 3 2 1 Memorabilia/merchandise of the TV show or TV stars 5 4 3 2 1 Products shown in that TV show 5 4 3 2 1

PAGE 239

239 Q14. The following statements describe your social media use experience. Please indicate how much you agree with each of the following statements. Strongly Disagree Disagree Neutral Agree Stron gly Agree Learning to use social media to comment, post, watch, or read anything about television program is easy for me. 1 2 3 4 5 It is easy for me to become skilled at using social media to comment, post, watch, or read anything about television prog ram. 1 2 3 4 5 It is easy to use social media to comment, post, watch, or read anything about television program. 1 2 3 4 5 Q15. Please indicate how much you agree with each of the following statements. Strongly Disagree Disagree Neutral Agree Str ongly Agree Using social media to comment, post, watch, or read anything about television program is compatible with most aspects of my television viewing. 1 2 3 4 5 Using social media to comment, post, watch, or read anything about television program f its my lifestyle. 1 2 3 4 5 Using social media to comment, post, watch, or read anything about television program fits well with the way I like to engage in television viewing. 1 2 3 4 5 Q16. Please use the following adjectives to indicate how you fe el about social media in general. Unsociable 1 2 3_ 4 5 Sociable Impersonal 1 2 3_ 4 5 Personal Insensitive 1 2 3_ 4 5 Sensit ive Cold 1 2 3_ 4 5 Warm Passive 1 2 3_ 4 5 Active

PAGE 240

240 Q17. The following statements describe the REASONS why you use social media to interact with the television show. Please indicate how much you agree with each of the following statements. Strongly Disagree Disagree Neutral Agree Strongly Agree Because it relaxes me 1 2 3 4 5 Because it allows me to unwind 1 2 3 4 5 Because i a pleasant rest 1 2 3 4 5 1 2 3 4 5 or be with 1 2 3 4 5 Because it makes me feel less lonely 1 2 3 4 5 When I have nothing better to do 1 2 3 4 5 Because it passes the time away particularity when I am bored 1 2 3 4 5 Because it gives me something to do to occupy my time 1 2 3 4 5 Because it entertains me 1 2 3 4 5 1 2 3 4 5 Because it amuses me 1 2 3 4 5 Because it helps me learn things about my self and others 1 2 3 4 5 So I can learn how to do things 1 2 3 4 5 So I could learn about what could happen to me 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 Because it peps me up 1 2 3 4 5 So I can forget about school/work or other things 1 2 3 4 5 So I can get away from the rest of the family or others 1 2 3 4 5 1 2 3 4 5

PAGE 241

241 Q18. The following statements describe the REASONS why you use social med ia to interact with the television show. Please indicate how much you agree with each of the following statements. Strongly Disagree Disagree Neutral Agree Strongly Agree Because it is easier to get information 1 2 3 4 5 Because I can search for information 1 2 3 4 5 Because I can get information for free 1 2 3 4 5 So I can see what is out there 1 2 3 4 5 So I can learn about useful things 1 2 3 4 5 So I can learn about unknown things 1 2 3 4 5 Because I want to show other s encouragement 1 2 3 4 5 Because I want to communicate with friends and family 1 2 3 4 5 Because I want to belong to groups with the same interest as mine 1 2 3 4 5 Because I want to let others know I care about their feelings 1 2 3 4 5 Because I can express myself freely 1 2 3 4 5 Because I enjoy answering others' questions 1 2 3 4 5 Because I can participate in discussion 1 2 3 4 5 Because I can meet new people 1 2 3 4 5

PAGE 242

242 Q19. The next set of questions is designed to understand y our personal characteristics. Please tell us how much you agree with each of following the statements. Strongly Disagree Disagree Neutral Agree Strongly Agree In general, I am among the last in my circle of friends to use a new social media platform when it appears. 1 2 3 4 5 If I heard that a new social media platform was available online, I would be interested enough to try it. 1 2 3 4 5 Compared to my friends, I use few social media platforms. 1 2 3 4 5 I will use a new social media yet. 1 2 3 4 5 In general, I am the last in my circle of friends to know the names of the latest social media platforms. 1 2 3 4 5 I know more about new social media platforms before other people do. 1 2 3 4 5 Q20. Please tell us how much you agree with each of the following statements. Strongly Disagree Disagree Neutral Agree Strongly Agree I get to see my friends as often as I would like. 1 2 3 4 5 I spend enough time communicating with my friends and family by telephone or mail. 1 2 3 4 5 I have ample opportunity for conversations with others. 1 2 3 4 5 I can always find someone to speak with when I need to talk. 1 2 3 4 5

PAGE 243

243 Q21. Please tell us how much you agree with each of the following s tatements. Strongly Disagree Disagree Neutral Agree Strongly Agree I often travel, vacation, or take trips with others. 1 2 3 4 5 I often visit with friends, relatives, or neighbors in their homes. 1 2 3 4 5 I often participate in the meetings or activities of clubs, lodges, recreation centers, churches, or other organizations. 1 2 3 4 5 I often go places to socialize with others. 1 2 3 4 5 I often participate in games, sports, or activities with others. Q22. Which of the following new technology equipment do you have at home? (Please select all that apply) Blue ray player Video On Demand (VOD) DVD player Tablet (e.g., iPad, etc.) DVR/TiVo Cellular Phone iPod or other portable MP3 Player E reader (e.g., Kindle, Nook, e tc.) HDTV Videogame System (e.g., Nintendo Wii, Playstation 3, Xbox 360, etc.) DVD burner/recorder Video iPod or other portable video player Computer

PAGE 244

244 Q23. Please check the subscription type(s) of your television service at home. Over the air onl y Basic and expanded basic cable Satellite (e.g., Dish Network, Direct TV, etc.) IPTV (e.g., U verse TV, FiOS, etc.) Others Q24. Please check the subscription type of your Internet connection in your house. High speed Dial up No Internet connectio n Q25. What is your gender? Male Female Q26. What is your age? ______________ Q27. What is the last grade of school you completed? Less than high school graduate High school graduate Some college College graduate or more Q28. For statistical p urposes, please estimate your total yearly household income (from all sources) before taxes. Under $30,000 $30,000 to just under $50,000 $50,000 to just under $75,000 $75,000 to just under $100,000 $100,000 or more

PAGE 245

245 Q29. What is your racia l or ethnic background? White African American Asian Latino, Latina, Hispanic Other Q30. What is your current employment status? Employed outside the home full time (30 hours or more per week) Employed outside the home part time (1 to 29 hours per week) Doing income producing work at home Temporarily unemployed Full time student Going to school part time Retired Full time homemaker Q31. What is your marital status? Single, never married Married Other

PAGE 246

246 LIST OF REFERENCES Aaker, D. A. (1991 ). Managing brand equity: Capitalizing on the value of brand name New York, NY: The Free Press. Ahuja, M., Carley, K. ( 1998 ). Network structures in virtual organizations Journal of Computer Mediated Communication 3 (4). Amiel, T. & Sargent, S. L. (200 4). Individual differences in Internet usage motives. Computers in Human Behavior, 20 (6), 711 726. Anderson, J. C., & Gerbing, D. W. (1988). Structural equations modeling in practice: A review and recommended two step approach. Psychological Bulletin, 10 3 (3), 411 423. ARF. (2006). Defining engagement initiative. Retrieved December 12, 2010, from http://www.thearf.org/research arf initiatives defining engagement.php Askwith, I. D. (2007). Television 2.0: Reconceptualizing TV as an engagement medium. Retrieved September 12, 2010, from http://cms.mit.edu/research/theses/IvanAskwith2007.pdf Atkin, D, J., Jeffries, L. W. & Neuendorf, K. A. (1998). Understanding in ternet adoption as telecommunications behavior. Journal of Broadcasting & Electronic Media 42 (4): 475 90. Baron, R. M., & Kenny, D. A. (1986). The moderator mediator variable distinction in social psychological research: Conceptual, strategic, and statisti cal consideration. Journal of Personality and Social Psychology, 51 (6), 1173 1182. Barwise, T. P., & Ehrenberg, A. S. C. (1987). The liking and viewing of regular TV series. Journal of Consumer Research, 14 (1), 63 70. Bielby, W. T., & Bielby, D. D. (19 Making and the rhetoric of network prime time program development. American Journal of Sociology, 99 (5), 1287 1313. Bignell, J. (2003). An introduction to television studies New York, NY: Routledge Biocca, F. A. (1988). Opposing conceptions of the audience: The active and passive hemispheres of mass communication theory. In J. A. Anderson (Ed.), Communication yearbook (Vol. 11, pp. 51 80). Newbury Park, CA: S age. Biocca, F., & Harms, C. (2002). Wha t is social presence? In F. Gouveia & F. Biocca (Eds.), Presence 2002 p roceedings Porto, Portugal: University of Fernando Pessoa Press.

PAGE 247

247 Biocca, F., Harma, C., & Burgoon, J. K. (2003). Towards a more robust theory and measure of social presence: Review and suggested criteria. Presence: Teleoperators & Virtual Environments, 12 (5), 456 480. Blumler, J. G. (1979). The role of theory in uses and gratifications studies. C o mmunication Research, 6 (1), 51 80. Bollen, K. A. (1989), Structural e quations wit h l atent v ariables New York NY : John Wiley & Sons, Inc. Boneva, B., Kraut, R., & Frohlich, D. (2001) Using email for personal relationships: T he difference gender makes. American Behavioral Scientist, 45 (3) 530 549. Bonfadelli, H. (2002). The Intern et and knowledge gaps. European Journal of Communication, 17 (1), 65 84. Brosius,H.B., Wober, M., & Weimann, G. (1992). The loyalty of television viewing: How consistent is TV viewing behavior? Journal of Broadcasting & Electronic Media, 36 (3), 321 335. Browne, M. W. (2001). An overview of analytic rotation in exploratory factor analysis. Multivariate Behavioral Research, 36 (1) 111 150. Busselle, R., Reagan, J., Pinkleton, B., & Jackson K. (1999). Factors affecting I nternet use in a saturated access population. Telematics & Informatics, 16 (1/2), 45 68. Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming Mahwah, NJ: Lawrence Erlbaum Association, Inc.. Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1997). Beyond bipolar c onceptualization and measures: The case of attitudes and evaluative space. Personality and Social Psychology Review, 1 (1), 3 25. Calder, B. J., Malthouse, E. C., & Schaedel, U. (2009). An experimental study of the r elation ship between online engagement and advertising effectiveness. Journal of Interactive Marketing, 23 (4), 321 331. Calic, A. (2011, November 7). Time for television ratings t o get social. Message posted to http://www.readwriteweb.com/archives/time_for_television_ratings_to_get_social. php Campo, K., Gijsbrechts, E., & Nisol, P. (2000). Toward understanding consumer response to stock outs. Journal of Retailing, 76 (2), 219 242.

PAGE 248

248 Cattell, R. B., Eber, H. W., & Tatsuoka, M. M. (1970). Handbook for the s ixteen p ersonality f actor q uestionnaire (16 PF ). Champaign, IL: Institute for Personality and Ability Testing. Chan Olmsted, S. M., & Chang, B H. (2006). Audience knowledge, perceptions, and f actors affecting the adoption intent terrestrial digital television. New Media & Society, 8 (5), 773 780. Chen, L., Gillenson, M. L., & Sherell, D. L. (2002). Enticing online consumers: E xtended technology acceptance perspective. Inform ation & Management, 39 (8), 705 719. Chen, W., & Wellman, B. (2004). The global digital divide within and between countries. IT & Society, 1 (7), 39 45. Cheong, J. H., & Park, M. (2005). Mobile internet acceptance in Korea. Internet R esearch, 12 (5), 125 140 Churcchill, G. A., Jr. (1979). A paradigm for development better measures of marketing constructs Journal of Marketing Research, 16 (1), 64 73. Cohen, J. (2002). Television viewing preferences: Programs, schedules, and the structure of viewing choices made by Israeli adults. Journal of Broadcasting & Electronic Media 46 (2), 204 221. Comrey, A. L. & Lee, H. B. (1992). A first course in factor analysis Hillsdale, NJ: Erlbaum. Cooper, R. (1996). The status and future of audience duplication research: A n assessment of ratings based theories of audience behavior. Journal of Broadcasting & Electronic Media, 40 (1), 96 111. Cunningham, S. M. (1966). Brand loyalty w hat, where, how much? Harvard Business Review, 34 (1 2), 116 128. Daft, R. L. & Lengel, R. H (1984). Information richness: a new approach to managerial behavior and organizational design. In: Cummings, L. L. & Staw, B. M. (Eds.), Research in organizational behavior 6, ( pp. 191 233 ) Homewood, IL: JAI Press. Davis, F. D. (1989). Perceived usefuln ess, perceived ease of use, and end user acceptance of information technology. MIS Quarterly, 13 (3), 319 339. Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User acceptance of computer technology: A comparative of two theoretical models. Managemen t Science, 35 (8), 982 1003.

PAGE 249

249 DeCoster, J. (1998). Overview of Factor Analysis. Retrieved September 12, 2010, from http://www.stat help.com/notes.html DeVellis, R. F. (1991). Scale development : T heory and ap plications (Applied Social Research Methods Series, Vol. 26). Newbury Park, CA: Sage. Dick, A. S., & Basu, K. (1994). Customer loyalty: Towards an integrated conceptual framework. Journal of the Academy of Marketing Science, 22 (2), 99 113. Dillon, C., Keo gh, E., & Freeman, J. (2002, October). It's been emotional: Affect, physiology and presence. Paper presented at the 5th International Workshop on Presence, Porto, Portugal. DiMaggio, P., Hargittai, E., Neuman, W. R., & Robinson, J. P. (2001). Social impli cations of the Internet. Annual Review of Sociology, 27 (1), 307 336. Dimmick, J. W., Jaspreet, S., & Patterson, S. J. (1994). The gratificatio ns of the household telephone: S ociability, instrumentality, and reassurance. Communication Research, 21 (5), 64 3 661. Epps, S. R. (2009). What engagement means for media companies. Retrieved December 12, 2010, from http://www.slideshare.net/ad_crystal/forrester what engagement means for media companies Everitt, B. S. (1975). Multivariate analysis: The need for data, and other problems. British Journal of Psychiatry. 126 257 240. Feltham, T. S., & Arnold, S. J. (1994). Program involvement and Ad/program consistency as moderator of program context effects Journal of Consumer Psychology, 3 (1), 51 77. Ferguson, D. A., & Perse, E. M. (2004). Audience satisfaction among TiVo and replay TV users. Journal of Interactive Advertising, 4 (2), 1 8. Fetto, J. (2010, February 5). Multi media engagement study findings. Message posted to http://www.experian.com/blogs/marketing forward/2010/02/05/multi media engagement study findings/ Flanaqin A. J., & Metzger, M. J. ( 2001 ). Internet use in the contemporary media environment. Human Communication Research, 27 (1), 153 181. Floyd, F. J., & Widaman, K. F. (1995). Factor analysis in the development and refinement of clinical assessment instruments. Psych ological Assessment, 7 (3), 286 299.

PAGE 250

250 Fulk, J., Steinfield, C., Schmitz, J., & Power, J. (1987). A social information processing model of media use in organizations. Communicati on Research, 14 ( 5), 529 552. Ganesh, J., Arnold, M. J., & Reynolds, K. E. (200 0). Understanding the customer base of service providers: An examination of the differences between stayers and switchers. Journal of Marketing, 64 (3), 65 87. Gardner, D. G., Cummings, L. L., Dunham, R. B., & Pierce, J. L. (1998). Single item versus mult iple item measurement scales: An empirical comparison. Educational and Psychological Measurement, 58 (6), 898 915. Garramone, G. M., Harris, A. C., & Anderson, R. (1986). Uses of political computer bulletin boards. Journal of Broadcasting & Electronic Medi a 30 (3), 325 339. Geerts, D., Cesar, P., & Bulterman, D. (2008). The implications of program genres for the design of social television systems, In Proceedings of the International Conference on Designing Interactive User Experiences for TV and Video (UX TV 2008), Mountain View, USA, 71 80. Gitlin T. (1983). Inside prime time. New York NY : Pantheon Books. Goffman, E. (1974). Frame analysis Cambridge: Harvard University Press. Goldsmith, R. E., & Hofacker, C. F. (1991). Measuring consumer innovative ness. Journal of the Academy of Marketing Research, 19 (3), 209 221. Goldsmith, R. E., & Flynn, L. R. (1992). Identifying innovators in consumer product market. European Journal of Marketing, 26 (12), 42 55. Gorsuch, R. L. (1983). Factor analysis (2 nd e d.). Hillsdale, NJ: Erlbaum. Green, M. C. & Brock, T. C. (2000). The role of transportation in the persuasiveness of public narratives. Journal of Personality and Social Psychology, 79 (5), 701 721. Green, M. C., Brock, T. C., & Kaufman, G. F. (2004). Und erstanding media enjoyment: The role of transportation into narrative worlds. Communication Theory, 14 (4), 311 327. Greenberg, B. S., Sherry, J. L., Busselle, R. W., Hnilo, L. R., & Smith, S. W. (1997). Daytime television talk shows: Guests, content, and interactions. Journal of Broadcasting & Electronic Media, 41 (4), 412 426. Guest, L. (1994). A study of brand loyalty. Journal of Applied Psychology, 28 (1), 16 27. Guest, L. (1995). Brand loyalty twenty years later. Journal of Applied Psychology, 39 (6), 405 408.

PAGE 251

251 Guo, M. (2011, August). The Impact of alternative video distribution platforms on traditional television viewing: How motives, affinity, consumption patterns, and perceived characteristics affect substitution. Paper presented at the 2011 Annual Convention of the Association for Education in Journalism & Mass Communication (AEJMC), St. Louis, MO. Ha, L., & Chan p erception of enhanced TV features and TV commerce on broadcast network Web sites. The International Journal on Media Management, 3 (4), 201 213. Ha, L., & Chan Olmsted, S. M. (2004). Cross media use in electronic media: The role of cable television web sites in cable television network branding and viewership. Journal of B roadcasting & Electronic Media, 48 (4), 620 645. Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1995). Multivariate data analysis (4 th ed.). New York, NY: Macmillan Publishing Company. Hall, A. (2005). Audience personality and the selection o f media and media genre. Media Psychology, 7 (4), 377 398. Hall, A. (2009). Perception of the authenticity of reality programs and their relationships to audience involvement, enjoyment, and perceived learning. Journal of Broadcasting & Electronic Media, 53 (4), 515 531. Hamilton, N. F., & Rubin, A. M. (1992). The influence of religiosity on television use. Journalism Quarterly 69 (3) 667 678. Hampton, K., & Wellman, B. (2003). Neighboring in netville: How the Internet supports community and social capital in a wired suburb City & Community, 2 (4), 277 311. Haridakis, P. M. (2006). Men, women, and television violence: Predicting viewer aggression in male and female television viewers. Communication Quarterly 54 (2) 227 255. Haridakis, P., & Hanson, G. (20 09). Social interaction and co viewing with YouTube: Blending mass communication reception and social connection. Journal of Broadcasting & Electronic Media, 53 (2), 317 335. Harris Interactive. (2008, June 25). The next eave in media measurement and enga gement: Multi screen engagement study finds greatest advertising value among audiences that Cross platforms and are the most engaged. Retrieved from http://www.businessw ire.com/news/home/20080625006000/en/Wave Media Measurement Engagement Multi Screen Engagement Study

PAGE 252

252 Harris Interactive. (2011, June 25). Are the online marketing efforts of TV shows and programs worthwhile? Retrieved from http://www.harrisinteractive.com/vault/HI Harris Po ll 24.7 Social Media TV 2011 03 30.pdf Harrison, C., & Amento, B. (2007). Collaborate TV : Using asynchronous communication to make TV social again. Retrieved December 12, 2010, from http://www2.research.att.com/~brian/papers/2007EUROITV CTV.pdf Harwood, J. (1999). Age identification, social identity, gratifications, and television viewing. Journal of Broadcasting & Electronic Media, 43 (1), 123 136. 2011, from http://www.adobe.com/engagement/pdfs/marketings_new_key_metric_engagem ent.pdf Hawkes, P. (1994). Building bran d loyalty and commitment. Journal of Brand Management, 1 (6), 337 347 Hawkins, R., Pingree, S., Hitchon, J., Gorham, B. W., Ka nnaovakun, P., Gillligan, E., Schmidt, T. (2001). Predicting selection and activity in television genre viewing. Media Psycholo gy, 3 (2), 237 263. Hazemi, R., & Hailes, S. (2001). The digital university: Building a learning community New York NY : Springer. Hecht, M. L. (1978). The conceptualization and measurement of interpersonal communication satisfaction. Human Communication Research, 4 (3), 253 264. Heeter, C. (1985). Program selection with abundance of choice: A process model. Human Communication Research, 12 (1), 126 152. Hendrickson, A. R., & Collins, M. R. (1996). An assessment of structure and causation of is usage. The D ata Base for Advances in Information Systems, 27 (2), 61 67. Hinkin, T. R. (1995). A review of scale development practices in the study of organizations. Journal of Management, 21 (5), 967 988. Hirschman, E. C. (1980). Innovativeness, novelty seeking, and consumer creativity. Journal of Consumer Research, 7 (3), 283 1 95. Hobman, E. V., Bordia, P., Irmer, B., & Chang, A. (2002). The expression of conflict in computer mediated and face to face groups. Small Group Research, 33 (4), 439 465.

PAGE 253

253 Holland, J. H., Ho lyyoak, K. J., Nisbett, R. E., & Thagard, P. R. (1989). Processes of inference, learning, and discovery Cambridge: MIT Press. Hoyle, R. H., & Duvall, J. L. (2004). Determining the number of factors in exploratory and confirmatory factor analysis. In D. Kaplan (Ed.). The SAGE handbook of quantitative methodology for social sciences (pp. 301 315). Thousand Oaks, CA: Sage. Hsu, C., & Lu, H. (2004). Why do people play online game s? An extended TAM with social influence and flow experience. Information and M anagement, 41 (7), 853 868. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6 (1), 1 55. Jacoby, J (1971). A model of multi brand loyalty. Journal of Advertising Research, 11 (3), 25 31. Jacoby, J., & Kyner, D. B. (1973). Brand loyalty vs. repeat purchasing behavior. Journal of Marketing Research, 10 (1), 1 9. Jacoby, J., & Chestnut, R. (1978). Brand l oyalty: Measurement and management NY: John Wiley and Sons. in a web based course environment. International Journal of Educational Telecommunications, 6 (4), 317 3 38. Jreskog, K. G. (1969). A general approach to confirmatory maximum likelihood factor analysis. Psychometrika 34 (2) 183 202. Jreskog, K. G., & Srbom, D. (1993). Testing str ucture equation modeling. In K. A. Bollen & J. S. Long (Eds.). Testing stru ctural equation models (pp. 294 317). Newbury Park, CA: Sage. Kaasines, E., Kivinen, T., Kulju, M., Lindroos, L., Oksman, V., Kronlund, J., et al. (2008). FinPilotz final report User acceptance of mobile TV services Retrieved November 1, 2008, form http://www.finnishmobiletv.c om/press/FinPilot2_Final_Report_20080529.pdf Kahn, B. E., Kalwani, M. U., & Morrison, D. G. (1986). Measuring variety seeking and reinforcement behaviors using panel data. Journal of Marketing Research, 23(2), 89 100.

PAGE 254

254 Katz, E., Blumler, J. G., & Gurevitch, M. (1974). Uses and gratifications research. The Public Opinion Quarterly, 37 (4), 509 523. Kaye, B. & Johnson, T. (2003). From here to obscurity? Media substitution theory and traditional media in an on line world. Journal o f the American Society for Information Science & Technology, 54 (3), 264 274. Kilger, M., & Romer, E. (2007). Does measures of media engagement correlate with product purchase likelihood? Journal of Advertising Research, 47 (3), 313 325. Kline, R. B. (201 1). Principles and practice of structural equation modeling (3 rd ed.). New York, NY: The Guilford Press. Kraut, R., Patterson, M., Lundmark, V., Kiesler, S., Mukopadhyay, T., & Scherlis, W. (1998). Internet paradox: A social technology that reduces socia l involvement and psychological well being. American Psychologist, 53 (9), 1017 1032. Kwon, H. S., & Chidambaram, L. (2000). A test of the technology acceptance model: The case of cellular phone adoption. In Proceedings of the 33rd Hawaii International Co nference on System Science Hawaii, the United States, 1 10. Leavy, D. (2010, October 14). Television and social media: A marriage made in heaven? Message posted to http://www.navigatenewmedia.com/Home/Article.aspx?100029 Levy, M. R., & Windahl, S. (1984 ). Audience activity and gratification: A conceptual clarification and exploration. Communication Research, 11 (1), 51 78. Lin, C. A. (1993). Modeling the gratification seeking process of television viewing. Human Communication Research, 20 (2), 224 244. Lin, C. A. (1998). Exploring personal computer adoption dynamics. Journal of Broadcasting & Electronic Media, 42 (1), 95 112. Lin, C. A. (2001). Audience attributes, media supplementation, and likely online service adoption. Mass Communication & Society, 4 (1), 19 38. Lin, C. A (2004). Webcasting adoption: T echnology fluidity, user innovativeness, and media substitution. Journal of Broadcasting & Electronic Media, 48 (3), 446 465. Lin, C. A. (2006). Predicting webcasting adoption via personal innovativenes s and perceived utilities. Journal of Advertising Research, 46 (2), 228 238. Lin, J., & Cho, C. (2010). Antecedents and consequences of cross media usage: A Journal of Broadcasting & Electronic Media, 54 (2), 316 3 36.

PAGE 255

255 Lombard, M., & Ditton, T. (1997). At the heart of it all: The concept of presence. Journal of Computer Mediated Communication, 3 (2). Lu, X., & Lo, H. (2007). Television audience satisfaction: Antecedents and consequences. Journal of Advertising Resear ch, 47 (3), 354 363. MacCallum, R. C., Widaman, K. F., Zhang, S., & Hong, S. (1999). Sample size in factor analysis. Psychological Methods, 4 (1), 84 99. Macintosh, G., & Lockshin, L. S. (1997). Retail relationships and store loyalty: A multi level perspec tive. International Journal of Research in Marketing, 14 (5), 487 497. Massey, W. F., Montgomery, D. B., & Morrison, D. G. (1970). Stochastic models of buyer behavior Cambridge: MIT Press. McBride, J. Z. (2010, August 11). TV networks use social media to drive loyalty. Message posted to http://www.customerthink.com/blog/tv_networks_use_social_media_to_drive_loy alty platform marketing to brand affinity among Web users. AdWeek Retrie ved from http://www.adweek.com/aw/content_display/news/media/e3i26f1bfd408799a20d a8594be e639d74c?pn=2 McQuail, D. (2006). McQua (5 th ed.). London: Sage Publications. Mellens, M., Dekimpe, M. G., & Steenkamp, J. E. M. (1996). A review of brand loyalty measures in marketing. Tijdschrift voor Economie en management, 41 507 533. Midgley, D. F., & Dowling, G. R. (1978). Innovativeness: T he concept and its measurement. Journal of Consumer Research 4 (4), 229 242. Moore, G. C. & Benbasat, I. (1991). Development of an instrument to measure perceptions of adopting an information technology innovation. Information Systems Research 2 (3), 192 222. Moorman, M., Neijens, P.C., & Smit, E. G. (2007). The effect of program involvement on commercial exposure and recall in a naturalistic setting Journal of Advertising, 36 (1), 121 137. Moyer Gus, E. (2010). Preference for television programs about sexual risk: The role of program genre and perceived message intent. Media Psychology, 13 (2), 180 199.

PAGE 256

256 Murry, J. P., Lastovicka, J. L., & Singh, S. N. (1992). Feeling and liking responses to television programs: An examination of two explanations for medi a context effects. Journal of Consumer Research, 18 (4), 441 451. Muthen, B., Kaplan, D., & Hollis, M. (1987). On structural equation modeling with data that are not missing completely at random Psychometrika, 52 (3), 431 462. Muthen, B. (2008, June 11). Calculating the % variance explain. Messaged posted to http://www.statmodel.com/discussion/messages/8/46.html?1321051784 Myers, J. M., & Wells, A. D. (2003). Research design and statistical analysis (2 nd ed.). Mahwah, NJ: Erlbaum. Networked Insights. (201 0). Social sense TV: Networking ratings report. Retrieved December 10, 2010, from http://www.networkedinsights.com/NI01tv17979d53 a071 491e a37a 9a8346c2e491.pdf Nielsen. (2006). Engagem ent. Retrieved December 12, 2010, from http://www.nielsen.com/us/en/measurement/television measurement.html Nielsen. (2010a, March 22). Americans using TV and Internet together 35% more than a year ago. Message posted to http://blog.nielsen.com/nielsenwire/online_mobile/three screen report q409/ Nielsen. (2010b, August 2). What Americans do online: social media and games dominate activity. Message posted to http://blog.nielsen.com/nielsenwire/online_mobile/what americans do online social media and games dominate activity Nielsen. (2011, October 11). Social media and TV abo ut? Message posted to http://blog.nielsen.com/nielsenwire/global/social media and tv whos talking when and what about/ Nunnally, J. C. (1978). Psychometric theory (2 nd ed.). New York: McGraw Hill. Norberg, M. M., Wetterneck, C. T., Sass, D. A., & Kanter, J. (2011). Development and psychometric evaluation of the Milwaukee psychotherapy expectations questionnaire. Journal of Clinical Psychology, 67 (6), 574 590. O framework for defin ing user engagement with technology. Journal of the American Society for Information Sciences and Technology, 59 (6), 938 955. measure user engagement. Journal of the Amer ican Society for Information Sciences and Technology, 61 (1), 50 69.

PAGE 257

257 O'Hara, K., Mitchell, A. S., & Vorbau, A. (2007). Consuming video on mobile devices. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems 857 866. Oliver, R. L. (1997). Satisfaction: A behavioral perspective on the consumer NY: McGraw Hill. Oliver, R. L. (1999). Whence consumer loyalty? Journal of Marketing, 63 (4), 33 44. Olsen, S. O. (2007). Repurchase loyalty: The role of involvement and satisfaction. Psycholog y & Marketing, 24 (4), 315 341. Owen, B. M., & Wildman, S. W. (1992). Video economics Cambridge: Harvard University Press. Palmgreen, P., & Rayburn, J. D. (1979). Uses and gratifications and exposure to public television: A discrepancy approach. Communica tions Research, 6 (2), 155 180. Palmgreen, P. (1984). Uses and gratifications: a theoretical perspective. In R. N. Bostrom (Ed.), Communication yearbook 8 (pp. 20 55). Beverly Hills, CA: Sage. P almgreen, P., & Rayburn, J. D. (19 85 ). A comparison of grati fication models of media satisfaction. Communication Monographs, 52 (4), 334 346. Palmgreen, P., Wenner, L. A., & Rosengren, K. E. (1985). Uses and gratifications research: the past ten years. In K. E. Rosengren, L. A. Wenner, & P. Palmgreen (Eds.), Media gratifications research: C urrent perspectives (pp. 11 37). Beverly Hills, CA: Sage. Papacharissi, Z., & Rubin, A. M. (2000). Predictors of Internet use. Journal of Broadcasting & Electronic Media, 44 (2), 175 196. Park, C. W., & McClung, G. W. (1986). The effect of TV program involvement on involvement with commercials. In R. J. Lutz (Ed.), Advances in consumer research (Vol. 13, pp. 544 548). Provo, UT. Association for Consumer Research. Patwardhan, P. (2004). Exposure, involvement and satisfaction with online activities. Gazette: The International Journal for Communication Studies, 66 (5), 411 436. Peck, J. (1995). TV talk shows as therapeutic discourse: The ideological labor of the televised talking cure Communication Theory, 5 (1), 58 81. Perse, M. E (1990). Media involvement and local news effects Journal of Broadcasting & Electronic Media, 34 (1), 17 36.

PAGE 258

258 Perse, E. M., & Rubin, A. M. (1990). Chronic loneliness and television use. Journal of Broadcasting & Electronic Media 34 (1) 37 53. Perse, M. E ., & Ferguson, A. D. (1993). The impact of the newer television technologies on television satisfaction. Journalism Quarterly, 70 (4), 843 853. Perse, E. M., & Courtright, J. A. (1993). Normative images of communication media: Mass and interpersonal channel s in the new media environment. Human Communication Research, 19 (4), 485 503. Perse, M. E., & Rubin, M. A. (1988). Audience activity and satisfaction with favorite television soap opera. Journalism Quarterly 65 (2), 368 375. Prior, M. (2005). News vs. en tertainment: How increasing media choice widens gaps in political knowledge and turnout. American Journal of Political Science 49 (3), 577 592. Rice, R. E. (1993). Media appropriateness: Using social presence theory to compare traditional and new organiza tional media. Human Communication Research, 19 (4), 451 484. Rice, R. E., & Katz, J. E. (2001). The Internet and h ealth c ommunication: Experiences and e xpectations. Thousand Oaks, CA: Sage. Rogers, E.M. (1995). Diffusion of Innovations (4 th ed.). New Yor k, NY: The Free Press. Rogers, E.M. (2003). Diffusion of Innovations (5 th ed.). New York, NY: The Free Press. Rubin, A. M., & Rubin, R. B. (1982). Contextual age: Development of a life position index. Paper presented at the Annual Meeting of the Speech C ommunication Association. Rubin, A. M. (1983). Television uses and gratifications: The interactions of viewing patterns and motivations. Journal of Broadcasting, 27 (1), 37 51. Rubin, A. M. (1986). Television, aging and information seeking. Journal of Lang uage & Communication, 6 (1/2), 125 137. Rubin, A. M., & Perse, E. M. (1987a). Audience activities and television news gratifications. Communication Research, 14 (1), 58 84. Rubin, A. M., & Perse, E. M. (1987b). Audience activities and soap opera involvemen t: A uses and effects investigation. Human C ommunication Research, 14 (2), 246 268.

PAGE 259

259 Rubin, A. M., & Rubin, R. B. (1989). Social and psychological antecedents of VCR use. In M. R. Levy (Ed.), The VCR age: Home video and mass communication (pp. 92 111). Beve rly Hills, CA: Sage. Rubin, A. M., & Step, M. M. (1997). Viewing television talk shows. Communication Research Reports, 14 (1), 106 115. Rubin, A. M., Haridakis, P. M., & Eyal, K. (2003). Viewer aggression and attraction to television talk shows. Media Psy chology, 5 (4), 331 362. R ubin, A. M. (2009). Uses and gratifications perspective on media effects. In J. Bryant & D. Zillmann (Eds.) Media effects: Advances in theory and research (pp. 165 184). New York, NY: Routledge. Ruggiero, J. E. (2000). Uses an d gratifications theory in the 21 st century. Mass Communication & Society, 3 (1), 3 37. Rundle Thiele, S., & Mackay, M. M. (2001). Assessing the performance of brand loyalty measures. Journal of Service Marketing, 15 (7), 529 546. Russell, C. A., Norman, A. T., & Heckler S. E. (2004a). The consumption of television programming: Developing and validation of the connectedness scale. Journal of Consumer Research, 31 (2), 150 161. Russell, C. A., Norman, A. T., & Heckler S. E. (2004b). People and their telev ision shows: An overview of television connectedness. In L. J. Shrum (Ed.), The Psychology of Entertainment Media: Blurring the Lines between Entertainment and Persuasion (pp. 275 290). Mahwah, NJ: Lawrence Erlbaum Association, Inc.. Rust, R. T., Kamakura, W. A., & Alpert, M. I. (1992). Viewer preference segmentation and viewing choice models of network television. Journal of Advertising, 21 (1), 1 18. Saade, R., & Bahli, B. (2005). The impact of cognitive absorption on perceived usefulness and perceived e ase of use in online learning: A n extension of the technology acceptance model. Information & Management, 42 (1), 317 327. Saba vala, D. J. & Morrison, D. G. (1977). A model of TV show loyalty. Journal of Advertising Research, 17 (6), 35 43. Sass, D. A., & Schmitt, T. A. (2010). A comparative investigation of rotation criteria within exploratory factor analysis. Multivariate Behavioral Research, 45 (1) 1 33.

PAGE 260

260 Save, E., Guazzelli, A., & Poucet, B. (2001). Dissociation of the effects of bilateral lesions of the dorsal hippocampus and parietal cortex on path integration in the rat. Behavioral Neuroscience, 115 (6), 1212 1223. Schepers, J. & Wetzels, M. (2007). A meta analysis of the technology acceptance model: Investigating subjective norm and moderation effe cts. Information & Management, 44 (1), 90 103. Selm, M. V. & Jankowski, N. W. (2006). Conducting online surveys. Quality & Quantity, 40 (3), 435 456. Shen, K. N., & Khalifa, M. (2008). Exploring multi dimensional conceptualization of social presence in t he context of online communities. International Journal of Human Computer Interaction, 24 (7), 722 784. Short, J., Williams, E., & Christie, B. (1976). The social psychology of te lecommunications London: John Wiley & Sons. Simmons. (2008). Multi media en gagement study. Retrieved December 12, 2010, from http://s3.amazonaws.com/thearf org aux ssets/downloads/cnc/engagement/2008 08 20_ARF_Eng_Simmons.pdf Stafford, T. F., Stafford, M. R., & Schkade, L. L. (2004). Determining uses and gratifications for the Internet. Decision Science, 35 (2), 259 287. Stanton, J. M. (1998). An empirical assessment of data collection using the Internet. Personnel Psychology, 51 (3), 709 7 25. Steeples, C., & Jones, C. (2002). Networked learning: Perspectives and issues London: Springer. Steiner, P. O. (1952). Program patterns and preferences, and the workability of competition in radio broadcasting The Quarterly Journal of Economics, 66 ( 2), 194 223. Stelter, B. (2010, February 23). Water New York Times Retrieved from http://www.nytimes.com/2010/02/24/business/media/24cooler.html Stephen, A. T., & Galak, J. (2009). The complementary roles of traditional and social media in driving marketing performance Retrieved October 11, 2009, from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1480088 Suh, J., & Yi, Y. (2006). Wh en brand attitude affect the customer satisfaction loyalty relation: The moderating role of product involvement. Journal of Consumer Psychology, 16 (2), 145 155.

PAGE 261

261 Sun, T., Youn, S., Wu. G., & Kuntaraporn, M. (2006). Online word of mouth (or mouse): An explor ation of its antecedents and consequences. Journal of Computer Mediated Communication, 11 (4), 1104 1127. Takahashi, T. (2010). Myspace or Mixi? Japanese engagement with SNS (social networking sites) in the global age. New Media & Society, 12 (3), 453 475. Thompson, B. (1992).A partial test distribution for cosines among factors across samples. In B. Thompson (Ed.), Advances in social science methodology (Vol. 2, pp. 81 97). Greenwich, CT: JAI. Tornatzky, L. G., & Klein, K. J. (1982). Innovation character istics and innovation adoption implementation: A meta analysis of findings. IEEE Transactions on Engineering Management, 29 (1), 28 45. Toy, T. (2010, July 26). How social media is affecting our television viewing habits and dictating what we watch. Message posted to http://tommytoy.typepad.com/tommy toy pbt consultin/2010/07/ Venkatesh, V., & Davis, F. D. (1996). A model of the antecedents of perceived ease of use: Developm ent and test. Decision Science, 27 (3), 451 481. Webster, J. G., & Wakshlag, J. J. (1983). A theory of television program choice. Communication Research, 10 (4) 430 446. Webster, J. G. (1998). The audience. Journal of Broadcasting & Electronic Media, 42 (2), 190 207. Webster, J. G., Phalen, P. F., & Lichty, L. W. (2006). Rating analysis: The theory and practice of audience research (3 rd ed.). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Weiming, S., Kremer, R., Ulieru, M., & Norrie, D. (2003). A collaborati ve agent based infrastructure for Internet enabled collaboration enterprise. International Journal of Production Research, 41 (8), 1621 1638. Windahl, S., Hojerback, I., & Hedinsson, E. (1986). Adolescents without television: A study in media deprivation. J ournal of Broadcasting & Electronic Media, 30 (1), 47 63. Wimmer, R. D. & Dominick, J. R. (2006). Mass media research: An introduction (8 th ed.). Belmont, CA: Thomson Wadsworth. Wober, J. M., & Gunter, B. (1986). Exploring the entertainment experience. In J. Bryant & D. Zillmann (Eds.), Perspectives on media effects Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

PAGE 262

262 Wright, K. B. (2005). Researching Internet based populations: Advantages and disadvantages of online survey research, online questionnaire aut horing software packages, and web survey services. Journal of Computer Mediated Communication, 10 (3). Wu, J H., & Wang, S C. (2005). What drives mobile commerce? An empirical evaluation of the revised technology acceptance model. Information & Management, 42 719 729. Yang, K. C. C., & Kang, Y. (2006). Exploring factors influencing Internet users' adoption of Internet television in Taiwan. First Monday, 11 (3). Retrieve November 14, 2008, from http://www.firstmonday.org/issues/issue11_3/yang/index.html Yan ga, S., & Kangb, M. (2009). Measuring blog engagement: Testing a four d imensional scale. Public Relations Review, 35 (3), 323 324. Youn, S. (1994). Program type preference and program choice in a multichannel situation. Journal of Broadcasting & Electron ic Media, 38 (4), 465 476. Zaichkowsky, J. L. (1985). Measuring the involvement construct. Journal of Consumer Research,12 (3), 341 352. Zywica, J., & Danowski, J. (2008). The faces of Facebook ers: Investigating social enhancement and social compensati on hypotheses; predicting Facebook TM and offline popularity from sociability and self esteem, and mapping the meanings of popularity with semantic networks. Journal of Computer Mediated Communication. 14 (1), 1 34.

PAGE 263

263 BIOGRAPHICAL SKETCH Miao Guo receive d her Ph.D. degree in M ass C ommunication from the University of Florida in the spring of 2012 Her teaching and research interests include social media, audience analysis, media effects, and online branding and marketing. Miao has assisted with and taught Telecommunication Planning and Operations, Perspectives of Radio and Television and the online course Writing and Reporting for Interactive Media at universities. She taught the course Telecommunication Research at the University of Florida, earning posit ive reviews from her students. She also has strong interests in incorporating social media into the classroom and has participated in online course design and development. Miao has already demonstrated herself as a productive scholar, with sole authored an d collaborative research producing five scholarly publications and fifteen conference presentations at the meetings of the Association for Education in Journalism and Mass Communication (AEJMC), the International Communication Association (ICA), the Broadc ast Education Association (BEA), and the World Media Economics and Management Conference. Miao has worked as research assistant on various grant projects and the International Journal on Media Management Prior to pursuing her doctoral studies, Miao receiv ed a master's degree in radio, television, and film from the University of North Texas in 2007, as well as a communication master's degree from Tsinghua University in China in 2004. She has served on the faculty of Beijing Normal University in Zhuhai, Chin a.