<%BANNER%>

Using the Theory of Planned Behavior to Predict Extreme Ritualistic Alcohol Consumption on Game Day

Permanent Link: http://ufdc.ufl.edu/UFE0021591/00001

Material Information

Title: Using the Theory of Planned Behavior to Predict Extreme Ritualistic Alcohol Consumption on Game Day
Physical Description: 1 online resource (166 p.)
Language: english
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: alcohol, binge, campus, college, drinking, fan, football, game, students, theory
Health Education and Behavior -- Dissertations, Academic -- UF
Genre: Health and Human Performance thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Alcohol abuse remains a serious health issue for colleges and universities across the country. A particular area of concern involves the drinking that takes place before, during, and after college football games. This study examined this specific type of drinking behavior, termed 'Extreme Ritualistic Alcohol Consumption' (ERAC), and whether the Theory of Planned Behavior (TPB) explains drinking patterns on game day among college students at a large university in the southeastern United States. The definition of ERAC, based on a previously validated measure, consists of consuming 10 or more drinks on game day for a male and eight or more drinks for a female. The purposes of this study were to determine the prevalence of ERAC on game day, assess the extent to which the TPB predicted ERAC, and determine the causal relationships among the TPB variables. Data were collected from a random sample of 740 college students who completed an anonymous online survey. Survey items assessed participants' motivations for consuming alcohol and the total number of drinks consumed on game day. Sixteen percent of the respondents engaged in ERAC on game day. Male, Caucasian, Greek, and students of legal drinking age consumed alcohol at disproportionately high rates. With the exception of Perceived Behavioral Control (PBC), each of the TPB constructs was statistically significant in predicting ERAC. Behavioral Intentions to drink alcohol on game day predicted behavior. Intentions, in turn, were predicted by Attitude Toward the Behavior and Subjective Norm constructs. The TPB was useful in explaining alcohol use on game day with college students. However, the applicability of the PBC construct within the TPB model remains in question. Additional research with more effective PBC measures is needed before more definitive statements can be made concerning the TPB's efficacy in predicting college student alcohol consumption on game day. Alcohol use is common on game day, with a significant percentage of students placing themselves at risk by drinking large amounts of alcohol. To reduce alcohol abuse on college campuses, university officials need to implement and rigorously evaluate specific game day interventions.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: 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, 2008.
Local: Adviser: Rienzo, Barbara A.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0021591:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021591/00001

Material Information

Title: Using the Theory of Planned Behavior to Predict Extreme Ritualistic Alcohol Consumption on Game Day
Physical Description: 1 online resource (166 p.)
Language: english
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: alcohol, binge, campus, college, drinking, fan, football, game, students, theory
Health Education and Behavior -- Dissertations, Academic -- UF
Genre: Health and Human Performance thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Alcohol abuse remains a serious health issue for colleges and universities across the country. A particular area of concern involves the drinking that takes place before, during, and after college football games. This study examined this specific type of drinking behavior, termed 'Extreme Ritualistic Alcohol Consumption' (ERAC), and whether the Theory of Planned Behavior (TPB) explains drinking patterns on game day among college students at a large university in the southeastern United States. The definition of ERAC, based on a previously validated measure, consists of consuming 10 or more drinks on game day for a male and eight or more drinks for a female. The purposes of this study were to determine the prevalence of ERAC on game day, assess the extent to which the TPB predicted ERAC, and determine the causal relationships among the TPB variables. Data were collected from a random sample of 740 college students who completed an anonymous online survey. Survey items assessed participants' motivations for consuming alcohol and the total number of drinks consumed on game day. Sixteen percent of the respondents engaged in ERAC on game day. Male, Caucasian, Greek, and students of legal drinking age consumed alcohol at disproportionately high rates. With the exception of Perceived Behavioral Control (PBC), each of the TPB constructs was statistically significant in predicting ERAC. Behavioral Intentions to drink alcohol on game day predicted behavior. Intentions, in turn, were predicted by Attitude Toward the Behavior and Subjective Norm constructs. The TPB was useful in explaining alcohol use on game day with college students. However, the applicability of the PBC construct within the TPB model remains in question. Additional research with more effective PBC measures is needed before more definitive statements can be made concerning the TPB's efficacy in predicting college student alcohol consumption on game day. Alcohol use is common on game day, with a significant percentage of students placing themselves at risk by drinking large amounts of alcohol. To reduce alcohol abuse on college campuses, university officials need to implement and rigorously evaluate specific game day interventions.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: 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, 2008.
Local: Adviser: Rienzo, Barbara A.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0021591:00001


This item has the following downloads:


Full Text

PAGE 1

USING THE THEORY OF PLANNED BEHAVIOR TO PREDICT EXTREME RITUALISTIC ALCOHOL CONSUMPTION ON GAME DAY By TAVIS GLASSMAN 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 2008 1

PAGE 2

2008 Tavis Glassman 2

PAGE 3

Dedicated to the Gator Nation 3

PAGE 4

ACKNOWLEDGMENTS I wish to acknowledge my Father, David Glassman, who is no longer with us, for encouraging me to pursue my doctorate for as long as I can remember. I express profound gratitude to my Aunt, Deena Ohan a, for mentoring me through th e higher education process, as well as with other life events. I also apprecia te my friends and family members, including my extended family, for their ongoing support and love. I thank the members of my Di ssertation Committee for their valuable insight and patience in correcting both my stylistic and scientific erro rs. I am grateful to Dr. Wagenaar for always challenging me to do my best, for maintaining ri gorous standards, and for encouraging me to seek out innovative and meaningful research expe riences. I sincerely appreciate Dr. Sheu for his generosity with his time and guidance with the data analysis pro cessI learned a great deal from working with him. Words cannot express my appreciation and admiration for Dr. Dodd for helping me to design my research questions, in strument, and analyses from a theoretical and philosophical prospective. Finally, I thank Dr. Rienzo my Committee Chair, for always believing in and trusting me. She granted me autonomy while providing the necessary guidance to ensure that I grew personally and profe ssionally throughout my graduate program. My inspiration for this research was supported by GatorWell Health Promotion Services at the Student Health Care Center in conjuncti on with a United States Department of Education grant. I appreciate the Univers ity of Floridas support with this project, including the University Registrar for randomly selecting and providing me with participants. I am very appreciative of Dr. Varnes for allowing Dr. Dodd and myself to di sseminate the findings of this research at the Health Promotion Lecture Series I hope this research ultimate ly helps to reduce the excessive 4

PAGE 5

alcohol consumption and related consequences th at occur on game day at the University of Florida and at other campus es, throughout the country. I am forever indebted to all those who contributed to my ed ucation, including all the fine teachers and professors I have lear ned from throughout my schooling. Finally, I cannot thank Maureen Miller enough for her unwavering support, compassion, and love. 5

PAGE 6

TABLE OF CONTENTS Page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................9LIST OF FIGURES .......................................................................................................................11ABSTRACT ...................................................................................................................... .............12 CHAPTER 1 INTRODUCTION .................................................................................................................. 14Research Problem ...................................................................................................................16Purpose of the Study .......................................................................................................... .....18Significance of the Study ........................................................................................................19Research Questions ............................................................................................................ .....21Delimitations ................................................................................................................. ..........21Limitations ................................................................................................................... ...........22Assumptions ................................................................................................................... ........22Definition of Terms ................................................................................................................23Summary ....................................................................................................................... ..........252 REVIEW OF THE LITERATURE ........................................................................................26Factors Associated with 5+4+ Drinking .................................................................................26Personal Risk Factors ......................................................................................................27Environmental Risk Factors ............................................................................................29Consequences Associated With 5+4+ Drinking .....................................................................31Game Day Drinking Behaviors ..............................................................................................34Social Marketing and Alcohol Use .........................................................................................38Social Norm Campaigns ......................................................................................................... 40Theory of Reasoned Action (TRA) and Th eory of Planned Behavior (TPB) ........................43Theory of Planned Behavior and Alcohol Use .......................................................................48Alcohol Consumption Measures .............................................................................................52Summary ....................................................................................................................... ..........553 METHODS ....................................................................................................................... ......58Research Design .....................................................................................................................58Research Variables .................................................................................................................60Attitude Toward the Behavior .........................................................................................61Behavioral Beliefs ...........................................................................................................6 1Evaluation of Behavioral Outcomes ................................................................................61 6

PAGE 7

7 Subjective Norm ..............................................................................................................6 1Normative Beliefs ............................................................................................................6 2Motivation to Comply .....................................................................................................62Perceived Behavioral Control ..........................................................................................62Control Beliefs .................................................................................................................62Perceived Power ..............................................................................................................63Behavioral Intention ........................................................................................................63Behavior ...................................................................................................................... ....63Study Population .....................................................................................................................65Instrumentation ............................................................................................................... ........66Internal Consistency .......................................................................................................... .....71Data Collection .......................................................................................................................72Data Analysis ..........................................................................................................................75Research Question 1: .......................................................................................................76Research Question 2: .......................................................................................................76Research Question 3: .......................................................................................................78Research Question 4: .......................................................................................................78Research Question 5: .......................................................................................................78Research Question 6: .......................................................................................................79Summary ....................................................................................................................... ..........804 RESULTS ....................................................................................................................... ........95Participant Characteristics ................................................................................................... ...96Research Questions ............................................................................................................ .....97What is the Prevalence of Extreme Ritualistic Alcohol Consumption on a Typical Game Day for Fall 2006? .........................................................................................97How Much Variance Does the Combination of Constructs in the Theory of Planned Behavior Explain When Predicting Extreme Ritualistic Alcohol Consumption on Game Day? ..........................................................................................................99Which Constructs within th e Theory of Planned Behavior Account for the Largest Proportion of Variance when Predic ting Extreme Ritualistic Alcohol Consumption Behavior among Colle ge Students on Game Day? ..........................100Do the Constructs within the Theory of Planned Behavior Differ by Gender When Predicting Extreme Ritualistic Alcohol Consumption among College Students on Game Day? .............................................................................................................101Do the Constructs within the Theory of Planned Behavior Differ by Grade Classification When Predicting Extreme Ritualistic Alcohol Consumption among College Students on Game Day? ................................................................103What Are the Causal Effects in Predic ting Alcohol Consumption Rates Using the Constructs from the TPB? ......................................................................................106Summary ....................................................................................................................... ........1085 SUMMARY, DISCUSSION, IMPL ICATIONS, CONCLUSIONS ....................................116Summary ....................................................................................................................... ........116Limitations ................................................................................................................... .........119

PAGE 8

Discussion .................................................................................................................... .........123Implications .................................................................................................................. ........131Recommendations ............................................................................................................... ..132Future Research .............................................................................................................132Practice ...................................................................................................................... ....133Conclusion .................................................................................................................... ........134 APPENDIX A UF GAME DAY SURVEY ..................................................................................................137B GAME DAY SURVEY IRB ................................................................................................148C GAME DAY SURVEY E-MAIL INSTRUCTIONS ...........................................................150LIST OF REFERENCES .............................................................................................................151BIOGRAPHICAL SKETCH .......................................................................................................165 8

PAGE 9

9 LIST OF TABLES Table page 2-1 Summary of the TPB al cohol related articles ......................................................................57 3-1 Game day TPB Attitude Toward Behavi or direct and indirect measures ...........................813-2 Game day TPB Subjective Norm direct and indirect measures ..........................................823-3 Game day TPB Perceived Behavioral C ontrol direct and indirect measures ......................833-4 TPB Attitude Toward Behavior items and reliability values from the literature ...............843-5 TPB Subjective Norm items and relia bility values from the literature ...............................853-6 TPB Perceived Behavioral Control items a nd reliability values fr om the literature ...........863-7 Game day test-retest demographic items .............................................................................873-8 Game day test-retest prevention items ................................................................................873-9 Game day test-retest drinking items ....................................................................................883-10 Game day test-retest alcohol consequence items ................................................................883-11 Game day test-retest social norm items .............................................................................8893-12 Game day test-retest Attitude Toward th e Behavior indirect and direct items .................8893-13 Game day test-retest Subjective Norm indirect and direct items ........................................903-14 Game day test-retest Perceived Behavior al Control indirect and direct items ....................913-15 Game day test-retest Be havioral Intention items ................................................................913-16 Game day scale reliability ............................................................................................... ....92 4-1 Participant demographics compared to UF student population fall 2006 .........................1094-2 Chi-square analysis of ERAC rates by gender ..................................................................1094-3 Chi-square analysis of ERAC rates by classification ........................................................1094-4 Chi-square analysis of ERAC rates by ethnicity ...............................................................1104-5 Chi-square analysis of ERAC rates by greek status ..........................................................1104-6 Chi-square analysis of ER AC rates by legal drinking age ................................................110

PAGE 10

10 4-7 5+4+ drinking and ERAC rates by demographic ..............................................................1114-8 Correlation analysis of direct TPB composite measures and behavioral intention ...........1114-9 Logistic regression analysis TPB composite measuresERAC ......................................1114-10 Logistic regression analysis TPB composite measuresERAC women .........................1124-11 Logistic regression analysis TPB composite measuresERAC men ..............................1124-12 Logistic regression analysis TPB composite measuresERAC juniors ..........................112413 Logistic regression analysis TPB composite measuresERAC seniors ..........................1124-14 Logistic regression analysis TPB composite measuresERAC graduate and professional students .........................................................................................................1134-15 Logistic regression analysis TPB co mposite measuresERAC underage drinkers ........1134-16 Logistic regression analysis TPB com posite measuresERAC legal drinking age ........1134-17 Goodness of fit measures for TPB composite measures model tests ................................1134-18 Standardized effects on Behavioral Inte ntion and number of drinks consumed on game day....................................................................................................................... .....1134-19 Goodness of fit measures for TRA composite measures model tests ...............................113

PAGE 11

11 LIST OF FIGURES Figure page 2-1 TPB constructs .......................................................................................................................57 4-1 Path analysis from TPB composite measures ......................................................................115 4-2 Path analysis from TRA composite measures .....................................................................115

PAGE 12

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy USING THE THEORY OF PLANNED BEHAVIOR TO PREDICT EXTREME RITUALISTIC ALCOHOL CONSUMPTION ON GAME DAY By Tavis Glassman May 2008 Chair: Barbara A. Rienzo Major: Health & Human Performance Alcohol abuse remains a serious health i ssue for colleges and universities across the country. A particular area of concern involves the dr inking that takes pla ce before, during, and after college football games. This study examined this specific type of drinking behavior, termed Extreme Ritualistic Alcohol Consumption (ERAC), and whether the Theory of Planned Behavior (TPB) explains drinking patterns on game day among college students at a large university in the southeastern United States. The definition of ERAC, based on a previously validated measure, consists of consuming 10 or more drinks on ga me day for a male and eight or more drinks for a female. The purposes of this study were to dete rmine the prevalence of ERAC on game day, assess the extent to which the TPB predicted ERAC, and determine the causal relationships among the TPB variables. Data were collected from a random sample of 740 college students who completed an anonymous online survey. Su rvey items assessed participants motivations for consuming alcohol and the total number of drinks consumed on game day. Sixteen percent of the respondents engage d in ERAC on game day. Male, Caucasian, Greek, and students of legal drinking age consumed alcohol at disproportionately high rates. 12

PAGE 13

With the exception of Perceived Behavioral C ontrol (PBC), each of the TPB constructs was statistically significant in predic ting ERAC. Behavioral Intentions to drink alcohol on game day predicted behavior. Intentions, in turn, were predicted by Atti tude Toward the Behavior and Subjective Norm constructs. The TPB was useful in explaining alcohol use on game day with college students. However, the a pplicability of the PBC construc t within the TPB model remains in question. Additional research with more effective PBC measures is needed before more definitive statements can be made concerning th e TPBs efficacy in predicting college student alcohol consumption on game day. Alcohol use is common on game day, with a si gnificant percentage of students placing themselves at risk by drinking large amounts of alcohol. To reduce alcohol abuse on college campuses, university officials need to implemen t and rigorously evaluate specific game day interventions. 13

PAGE 14

CHAPTER 1 INTRODUCTION Alcohol misuse represents a serious health issue for colleges and universities across the country (Walters & Bennett, 2000). Surveys indicate that the majo rity of college students drink regardless of their age and that approximately two out of every five college students consume five or more drinks, four or more for a fema le, at least once in the last two weeks (Hingson, Heeren, Zakocs, Kopstein, & Wechsler, 2002). This type of drinking, commonly referred to as heavy episodic drinking (Wechsler, Lee, Kuo, Seibring, Nelson, & Lee, 2002), from this point forward will be referred to simply as + 4+ drinking. Consuming 5+4+ alcoholic drinks is associated with unintentional injury (e.g., motor vehicle crashes, falls, and drownings), sexually transmitted diseases, unintended pregnancy, se xual assault, violence, and poor academic performance (Kuo, Wechsler, Greenber, & L ee, 2003; Naimi, Brewer, Mokdad, Denny, Serdula, & Marks, 2003). The National Inst itute on Alcohol Abuse and Alcoholism (NIAAA) estimated that over 1,700 student deaths, 599,00 0 injuries, and 696,000 assaults annually are associated with consuming 5+4+ drinks (Hingso n, Heeren, Winter, & Wechsler, 2005). Those who consume 5+4+ drinks not only place themselves at increased risk for health consequences, they also affect others with their drinking. Wechsler and colleagues (2002) describe this phenomenon as secondhand drinki ng effects. They found in a national survey that a substantial percentage of students experienced the following negative effects from their peers drinking: 60% of students surveyed ha d their study or sleep in terrupted, 48% reported having to take care of a drunken student, and 29% reported being insulted or humiliated. In addition, 55% indicated that they experienced at leas t two secondhand effect s. These patterns have remained consistent over the la st decade (Wechsler et al., 2002). 14

PAGE 15

According to the Carnegie Foundation for the Advancement of Teaching, 5+4+ drinking represents the greatest single problem that America's univers ities must address (NIAAA, 2002a). Student death, injury, poor academic performa nce, property damage, vandalism, strained campus-community relations, and negative publicity are all issues that university presidents and other senior administration offi cials must manage because of alcohol abuse (Higher Education Center for Alcohol and Othe r Drug Prevention, 1997). Further, based on the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, nearly one-third of college students meet the criteria for a formal diagnosis of alcohol abuse, and one in 17 can be classified as alcohol dependent (Wechsler et al., 2002) Ironically, epidemiological studies show that college-bound high school students drank less th an their non-college-bound peers. However, upon arriving at institutions of higher education, college students drank more than their same age counterparts who did not attend college (Johnsto n, OMalley, & Bachman, 2001). National data indicate that cert ain groups of college students are at more risk than others. Overall, men are more likely to consume five or more drinks than women are to consume four or more drinks, 49% vs. 41% respectively. Caucas ians engaged in this behavior (50%) at higher rates than Hispanics (34%), Native American Indi ans (34%), Asians/Pacific Islanders (26%), and Blacks/African-Americans (22%). Students und er the age of 21 engaged in 5+4+ drinking slightly less (44%) than students ages 21-23 (50%); similar patt erns exist between underclassmen and upperclassmen. Finally, students who live onsite at a fraternity or sorority (75%), or who are members of a nonresidentia l fraternity or sorority (64%) report the highest rates of 5+4+ drinking (Wechsler et al., 2002). 15

PAGE 16

Research Problem The University of Florida, which annually ranks as a Top Party School (The Associated Press, August 23, 2005), has experienced a number of alcohol-related tragedies in recent years. According to the Gainesville Sun, from 2003-2005, there were six al cohol-related deaths among UF students, including incide nts on and off campus (Arndorfer, 2005; Word, 2005). In two separate events, UF students died as a result of falling from a height. Causes of the other four fatalities include a car crash, hit and run crash, suffocation, a nd another where a student was brutally beaten to death. In addition, the high-risk dr inking rate, defined as consuming five or more drinks on a single occasion during the pa st two weeks (Johnston, OMalley, Bachman, & Schulenberg, 2005), among UF students increased to an all time high of 57% in 2004. (Hereafter the term high-risk drinking will be referred to as 5+ drinks, with no gender distinction.) This marker represents a considerable difference from the Healthy Campus 2010 goal of reducing the 5+4+ drinking rate to 20% or lower by the ye ar 2010 (American College Health Association, 2002). The following statistics provided additional in dicators of problem drinking behaviors and consequences among UF students. In the fa ll of 2004, 473 UF students completed the Core Alcohol and Drug Survey Long Form (a standard ized instrument specifically designed for college students). While most st udents drink in moderati on or not at all, a si gnificant percentage experienced alcohol related cons equences in the past year: 38 % reported driving a car while under the influence, 44% reported missing a class, 25% performed poorly on a test or project, 42% had a blackout, 66% vomited, and 70% reported experiencing a hangover. Approximately, two-thirds (64%) of UF students believe that alcohol facilitates sexual oppor tunities. Despite the national minimum legal drinking age of 21, nearly 75% of students reported engaging in 16

PAGE 17

underage drinking. In addition, approximately 209 students received treatment in the campus mental health department for alcohol or ot her drug counseling. During the 2003-2004 school year, 179 students were found responsible for alcohol violations. Finally, in 2003 there were 56 DUI (Driving Under the Influence) arrests ma de on the University of Florida campus (UF Biennial Review, 2004). One of the most prominent, if not symbolic, challenges surroundi ng college prevention efforts involves game day (football) (Glassma n, Werch, Jobli, & Bian, 2007). A substantial number of alumni, students, and other football fans engage in 5+ drinking activities on game day. Most fans who consume alcohol will simply experience a minor consequence (such as a hangover), if any at all; however, for a select few, the results ca n be devastating. For example, when UF hosted the Tennessee Vo lunteers in the fall of 1999, an alcohol-related fight resulted in the death of two young adults (Swirko, 2000). In the fall of 2004 at the Florida/Georgia game held in Jacksonville (also known as the Worlds Largest Co cktail Party) (Arndorfer, 2005), a UF student fell to his death from a parking garage A year later at the same location, another UF student was brutally beaten to death af ter the football game (Arndorfer, 2005). Other Game Day Survey results substantiate the public health concern. Data from the UF Game Day Survey conducted in 2004 indica te that among those students who drank, 73% typically engage in 5+ alcohol consumption on ga me day. Further, over 70% of the students who drank reported experiencing a hangover due to drinking on game day, 29% vomited, 30% drove after drinking, 15% drove after having five or mo re drinks, 33% blacked out, and 21% got into a fight or an argument. Males re ported drinking more than females, and students drank more than nonstudents such as alumni and other fans (G lassman et al., 2007). These game day drinking patterns may increase the schools 5+ drinking rate for the fall semester. Cross sectional data 17

PAGE 18

collected annually at UF indicated that students engaged in 5+ dri nking at higher rates in the fall than they did in the spring (GatorWell Health Promotion Services, 2006). While there are a variety of possible explanations for this phenomenon, the fact th at there are no college football games in the spring has to be considered as a variable for this discrepancy. Game day represents a unique social event where, alcohol is c onsumed with greater intensity and for a longer time th an at other social events (Glassman et al., 2007). However, White, Kraus, and Swartzweider (2006) suggest that the standa rd 5+4+ drinking measure does not adequately capture how heavily people actually drink and c onsequently may not accurately indicate their risk. Instea d they recommend doubling the 5+4+ drinking threshold to identify more dangerous drinking patterns. While im plementing this recommendation, a customized term with previously identified measure (W hite et al., 2006) were developed for this investigation: Extreme Ritualistic Alcohol Consumption (ERAC), defined as consuming 10 or more drinks on game day for a male and eight or more drinks for a female. This study focuses on this term and measure, which are used to discus s the research questions and related analyses. Purpose of the Study The aim of this investigation was to examine alcohol consumption patterns among college students on game day. Specifically, the purposes were to (a) assess the prevalence of Extreme Ritualistic Alcohol Consumption among University of Florida male and female students ages 18-24 on college football Saturday; (b) determine the extent to which the Theory of Planned Behavior (TPB) can be used to pred ict ERAC rates among college students; and (c) assess the causal path among the TPB variables related to alcohol consumption on game day. 18

PAGE 19

Significance of the Study The national 5+4+ drinking rate has remained remarkably stable over the last decade despite increased attention and resources directed toward the issue (Wechsler et al., 2002). While there is considerable research on college st udents and 5+4+ drinking, little research exists on effective, non-policy, populationbased interventions. Dejong and colleagues (1998) contend that the key to reducing 5+4+ drinking centers on environmental management One of the hallmarks of environmental management involves the creation of a health-promoting normative environment. A number of schools have addressed this need by implementing a social norms campaign (Perkins, 2003). Social norm interventions rest on the assumption that the behaviors of people are influenced by their peers or at least their perceptions of their p eers. This concept is sometimes referred to as imaginary peers Because it is difficult to know the actual behaviors of peer group(s), people make generalizations and form perceptions about the groups behavior. Research indicates that college students grossly overestimate th e amount of alcohol their peers consume. Leading scholars in the field maintain that if those misperceptions can be corrected, the corresponding drinking rates will fall. Th e finding that students who overestimate the 5+ drinking rate are more likely to engage in th is behavior themselves has been well documented (Haines, 1998; Perkins, 2003; Thombs, et al., 2005). Social norm campaigns have become wide ly disseminated intervention strategies employed by universities throughout the country (Wechsler, Toben, Lee, Seibring, Lewis, & Keeling, 2003), although research a ssessing the effectiveness of thes e interventions is scarce and the results are in consistent. While the NIAAAs (2002b) Call to Action: Changing the Culture of Drinking at U.S. Colleges lists the use of a social norms ap proach to alcohol consumption as a 19

PAGE 20

promising practice, other prominent alcohol re searchers such as Wechsler (2002) are more skeptical. Thombs and colleagues (2005) study on social norms found that traditional social norms campaigns, which highlight the fact that M ost Students Have Zero to Four Drinks When They Party, failed to reduce the 5+ drinking rate and may not even have corrected the misperceptions regarding alcohol use. Thom bs and colleagues cautioned that while the conceptual underpinnings of the social norms theory may still have potential, there is a need to develop more effective applic ations of the model. The Theory of Planned Behavior (TPB) repres ents a comprehensive theory that includes multiple factors to explain health behavior, whereas social norm interventions simply attempt to correct the misperceptions people evoke concerning the perceive d prevalence of a behavior. Further, unlike social norm interventions, the Subjective Norm construct within TPB addresses the Motivation to Comply with normative belief s, which is extremely useful in shaping normative behavior. Another fundamental construc t incorporated into the Theory of Planned Behavior is the Attitude Toward the Behavior. Youth tend to have very favorable attitudes concerning alcohol, in part due to marketing efforts by alcohol producers and retailers (Chen, Grube, Bersamin, Waiters, & Keef e, 2005). Thus, simply correcting misperceptions concerning the 5+ drinking rate is likely to yield limited behavior change. The Theory of Reasoned Action and its extens ion the Theory of Planned Behavior, which includes perceived control, are useful when de signing interventions. To date, few alcohol messages targeting college students address perceived control. Furt her, very little information exists on specific game day drinking patterns, especially from a theo retical framework. Thus identifying which of the TPB constructs infl uence game day drinki ng among college students can produce more effective communication methods. 20

PAGE 21

Reducing the number of alcohol related inci dents associated with excessive alcohol consumption remains a public health priority. This is particularly important for game day, which represents a high-risk event, as well as a ne glected prevention venue. The findings from this study will be used to create health promoti on messages which can be implemented and later evaluated for their effectiveness. Research Questions The research questions for this study include: What is the prevalence of Extreme Ritualistic Alcohol C onsumption on a typical game day for fall 2006? How much variance does the combination of constructs in the Theory of Planned Behavior explain when predicting Extreme Ritualistic Alcohol Consumption on game day? Which constructs within the Theory of Planned Behavior (Subjective Norm, Attitude Toward the Behavior, Perceived Behavioral Control, and Behavioral Intention) account for the largest proportion of variance when predicting Extreme Ritualistic Alcohol Consumption behavior among colle ge students on game day? Do the constructs within the Theory of Planned Behavior differ by gender when predicting Extreme Ritualistic Alcohol Consumption among college students on game day? Do the constructs within the Theory of Pl anned Behavior differ by grade classification when predicting Extreme Ritualistic Alc ohol Consumption among college students on game day? What are the causal effects in predicting alc ohol consumption rates using the constructs from the Theory of Planned Behavior? Delimitations The following delimitations should be considered when interpreting th e results of this inquiry: Cross-sectional data were collected for this investigation utilizing a web-based survey. Participants in this study included college students, aged 18 to 24, enrolled at UF during the fall 2006 semester. 21

PAGE 22

A list of survey participants was randomly selected by the UF Registrar. Participants included students from the study population who agreed voluntarily to participate in the study. Findings from this study were based on self-re port data collected by using the Game Day Survey. The collection of data occurred at a single point in time afte r the last home football game of the season. Limitations The following limitations should be considered when interpreting the results from this investigation: Data collected from this cross-sectional survey design reflect responses from participants at a specific time; thus, caus ation cannot be established. Findings from this investigation cannot be generalized to other populations of college students. Students who did not voluntarily agree to partic ipate were excluded from the study. Variations in student partic ipation may have influenced the results of this study. The self-reported data collected for this investigation limits the ability to determine the extent of over-reporting or under-reporting data. Assumptions For the purposes of this investigation, the following assumptions were made: The registrar provided current and accurate student e-mail addresses. Students are assigned a UF e-mail account; however, select students may have dropped out of school or changed their e-mail addresses, thus pot entially influencing the response rate. Every participant had access to the internet. Th e University of Florida provides internet access to currently enrolled students at various venues throughout campus. The invitation to participate in the survey re ached the intended audience. E-mail software programs may automatically send mass e-mail so licitations to a junk e-mail folder. As a result, students may not have been aware of their invitation to partic ipate in the study. 22

PAGE 23

The students who participated in the study we re representative of the overall student population unless otherwise noted. The regist rar provided a random list of students to participate in the study. The students who agreed to participate in the stud y answered survey questions honestly. In the survey description, stude nts were assured of their anonymity and were encouraged to answer questions honestly. Definition of Terms Attitude: Overall evaluation of a behavior dete rmined by individuals beliefs concerning the outcomes or attributes of performing th e behavior (Glanz, Rimer, & Lewis, 2002). Behavioral Belief: An indirect measure of attitu de where individuals assess whether their actions or potential actions are associated with certain a ttributes or outcomes (Glanz et al., 2002). Behavioral Intention: Perceived likelihood of perfor ming the behavior of interest (Glanz et al., 2002). Heavy episodic drinking: Consumption of five or more drinks in a row for men and four or more drinks for women, at least once in the previous 2 weeks (Wechsler et al., 2002). College student: Random selection of individuals ages 18-24 years old who are registered to attend the University of Florida. Control Beliefs: An indirect measure of Percei ved Behavioral Control, where individuals assess the presence or absence of facilitators and barri ers to behavioral performance (Glanz et al., 2002). Drink: A standard drink equal to 12 ounces (oz) of beer, 12 oz of wine cooler, 5 oz of wine, or 1.25 oz of liquor ei ther straight or in a mi xed drink (White, Kraus, & Swartzwelder, 2006; Wech sler et al., 2002). Drunk: Consumption of alcohol to the point of impairing ones mental and physical abilities. In the state of Fl orida a person with a blood alcoho l level equal to or exceeding 0.08 mg/dl (milligrams per deciliter of blood) is considered legally intoxicated or drunk. Evaluation of Behavioral Outcomes: An indirect measure of attitude, where individuals assess the value attach ed to a belief or associated with an activity or attribute (Glanz et al., 2002). Extreme Ritualistic Alcohol Consumption (ERAC): Consumption of eight or more drinks for females and 10 or more drinks for males, in relation to a specific event such as 23

PAGE 24

game day, an event that is culturally associ ated with drinking patterns higher than those of typical social occasions. Game day: A typical home football game in cluding tailgating activities on or off campus, before, during, and after the game. High-risk drinking: Consumption of five or more dr inks on a single occasion during the previous two weeks (Jo hnston et al., 2005). Motivation to Comply: An indirect measure of the Subjective Norm where individuals assess their intrinsic drive to ac t in accordance with what they perceive their key referents deem appropriate concerning the behavior of interest (Gla nz et al., 2002). Normative Belief: An indirect measure of the Subjective Norm where individuals access the extent to which each referent approves or disapproves of the behavior of interest (Glanz et al., 2002). Perceived Behavioral Control: The degree to which indi viduals believe they have control over their actions (Glanz et al., 2002). Perceived Power: An indirect measure of Per ceived Behavioral Control, where individuals assess the impact of each factor in facilitating or inhibi ting the behavior of interest (Glanz et al., 2002). Place: Location and time at which the target au dience will perform the desired behavior, acquire any related tangible obj ects, and receive any relate d service. Conversely, place may represent the location used to make co mpeting behaviors seem less convenient i.e., reducing accessibility (Kotle r, Roberto, & Lee, 2002). Price: The cost that the target audience asso ciates with adopting the new behavior (Kotler et al., 2002). Product: The desired behavior a nd the associated benefits of that behavior or a complex bundle of benefits (Kotler et al., 2002). Promotion: Persuasive communicati on designed to ensure that the target audience knows about the stated benefits of the price, product, and pl ace. Promotion includes two major components: message and media (Kotler et al., 2002). Referent: Salient or influential person in the individuals life (Glanz et al., 2002). Social marketing: The use of messages intended to influence the target audience to voluntarily accept, reject, modif y, or abandon a behavior for the benefit of the individual, the larger group, or society as a whole (Kotler et al., 2002). 24

PAGE 25

Social norms marketing: A mass marketing campaign designed to correct misperceptions regarding the perceived prevalence of a behavior (Perkins, 2003). Subjective Norm: Belief about whether most people approve or disapprove of the behavior (Glanz et al., 2002). Summary This chapter provided the foundation for this investigation including a description of the research problem, purpose of the study, signi ficance, research questions, delimitations, limitations, assumptions, and definition of terms. College student drinking represents a serious public health problem. Every year, college students experience se vere alcohol related consequences, including death, sexual assault, fights, unplanned pregnancy, sexually transmitted infections, and poor academic perf ormance. The purpose of this study was to identify which of the TPB constructs to utilize when designing he alth promotion messages to discourage excessive alcohol use among college students on game day. This research may assist practitioners in reaching their goals of reducing alcohol consumption and the asso ciated consequences on game day. 25

PAGE 26

CHAPTER 2 REVIEW OF THE LITERATURE The purpose of this study was to determine the efficacy of the Theory of Planned Behavior in predicting alcohol use among college students on game day. The information contained in this chapter provides a review of the research that is relevant to this study. The initial section of this chapter provides an overv iew of the personal and environmental risk factors associated with 5+4+ drinking and the related health consequences. The next section presents a brief account of research related to game day drinking behavior. Finally, a theoretical description of social norms, social marketing, and the Theory of Planned Behavior will be presented. While considered a rite of passage by some, drinking in college is more pervasive and destructive than many people rec ognize. Too often todays headlin es bring news of yet another alcohol related tragedy involving a college student, whether it is a drinking and driving accident, homicide, sexual assault, or other misfortune. Th e traditions and the belie fs concerning alcohol are handed down from one generation to the next perpetuating the culture of alcohol abuse on college campuses. The issue is not simply that college students drink rather it is the way they drink that places them at high-risk for alc ohol related problems (NIAAA, 2006). Drinking games, dares, contests, and high volume drinki ng are common practices among college students who drink. Research indicates that 5+4+ drinking patterns are most prevalent during the late teens and early-to-mid-twenties (Naimi, Brewer, Mokdad, Denny, Serdula, & Marks, 2003). Factors Associated with 5+4+ Drinking Young adulthood represents a stage in life marked by change and self-discovery, especially from those entering a college away fr om home. Students move out of the homes of their parents and join their peers, and often live in residence halls or in an off-campus 26

PAGE 27

arrangement. In addition to managing course work, many students are employed and begin to form serious relationships. Typi cally, students explore w ho they want to be, what they want to do, and how they will fit into the world. Concur rently, parental authority weakens while peer influence increases. For the first time, many colle ge students are free to ma ke their own choices, including the decision to drink alcohol (NIAAA, 2006). Many people fail to recognize or appreciate thes e complexities. Indeed, the judgment of a young adult is influenced by a combination of personal and environmental factors. Key personal factors may include past alcohol use, family influences, genetics, and personality (Sher, Trull, Bartholow, & Vieth, 1999; Zucker, Fitzge rald, & Moses, 1995). In fact, a significant percentage of students arrive on campus with previously established drinking patterns. According to the data from 2005 Youth Risk Behavior Surveillance System (YRBSS), nearly half of high school students (43% ) reported consuming one or more drinks of alcohol in the last 30 days, and approximately one-fourth (26%) report ed consuming five or more drinks on one or more occasions in the previous 30 days (Cen ter for Disease Control and Prevention [CDC], 2006). Personal Risk Factors With regard to family influences, duri ng young adulthood parents influence on their childrens behavior wanes. Ne vertheless, the example parent s set with their own drinking patterns has been shown to make a lifelong impression on their children. Young people tend to model their own drinking patterns, including quanti ty, frequency, attitudes, and contexts of use, on the alcohol consumption practic es of their parents. Parent ing style, attachment, nurturing, bonding, abuse, neglect, conflict, discipline, a nd monitoring also influence alcohol use of young adults (White, Johnson, & Buyske, 2000). Parents, who do not actively mon itor their childrens 27

PAGE 28

behavior or do not remain consistently involved in their lives, place thei r children at increased risk for substance abuse (National Center on Addiction and Substance Abuse at Columbia University, 1999). Even while at college, parent al influence on alcohol issues may serve as a protective factor for some students (NIAAA, 2002b). Behavioral cues are not the only alcohol-related indicators parents pass along to their children. Alcoholism or alcohol problems seem to run in certain families (White et al., 2000). This family connection to alcohol abuse may be the result of genetics. However, the social influences of the family must also be consider ed. Not only might childr en model their drinking behavior after their parents, they may also emulate an older brother or siste rs drinking behavior. Also complicating the epidemiological trends ar e older siblings who may provide alcohol to younger siblings, because accessibility also represen ts a major influence in the decision to use and/or abuse alcohol. Neverthele ss, genetics should not be underestimated as a key contributing factor associated with alcohol abuse. Research indicates th at people with a family history of alcoholism are less likely to mature out of heavy alcohol use than those with no family history of alcoholism (Jackson, Sher, Gotham, & Wood, 2001). Personality also plays a major role in 5+ 4+ drinking and alcohol abuse. Studies document that people who are risk-takers, rebellious, impulsive, or sensationseeking tend to consume larger amounts of alcohol and drink more often than others (Arne tt, 2005). Impulsivity and sensation-seeking behavior are associated with deviant conduct and nonconformity, both of which are linked to heavy dri nking and the associated conseq uences (Baer, 2002). College students as a whole are also more likely to feel invincible. This l ack of vulnerability or optimistic bias may also result in increased drinking, b ecause students do not fear or expect negative consequences (Arnett, 2005). Finally, because college can be such a stressful time, some 28

PAGE 29

students drink to cope with ne gative moods, feelings of depr ession, or to relieve anxiety (Jackson, Sher, & Park, 2005). Environmental Risk Factors A variety of environmental factors influe nce drinking patterns among college students. Rates of 5+ drinking tend to be highest at co lleges and universities with large Greek systems and/or prominent athletic programs, as well as at schools located in the Northeast (Presley, Meilman, & Leichliter, 2002). In addition, local co mmunities that tolerate underage drinking and provide only minimal law enforcement efforts reinforce the culture of alcohol use and abuse (Toomey & Wagenaar, 2002). Other key environm ental factors include outlet density, alcohol pricing, and promotional practices Environmental interventions restructure the circumstances that lead to 5+4+ drinking occasions (Dejong, Vince-Whitman, Colthurst, Cretella, Gilbreath, Rosati, & Zweig, 1998). These strategies preclude the need to identify which individuals are most at risk or likely to bene fit from a program. Rather than screening indivi duals, populationbased interventions, such as creating new policie s or increasing enforcement efforts, are the focus of environmental management strategies For example, changing the minimum drinking age or drunken driving laws is arguably easier than targeting pote ntial youthful drinkers or drunk drivers individually. Researchers do not n ecessarily know which young persons behaviors changed due to the intervention, but they observed overall that drinking patterns and alcohol related fatalities decreased after states rais ed their minimum drinki ng age laws (Wagenaar, 1983). Several leading public hea lth experts assert that commun ities should focus their time and energies on environmental management strategi es rather than on traditional individual based education programs (Dejong et al., 1998; Wech sler & Wuethrich, 2002; Toomey & Wagenaar, 2002). 29

PAGE 30

Another method to reduce alc ohol access involves restricting the number of outlets that make alcohol available. In 2002 the NIAAA Task Force on College Drinking concluded that restricting outlet density is an effective appr oach for reducing alcohol-related problems among the general population (Holder et al., 2000; Gruenewald, Ponicki, & Holder, 1993; Gruenewald, Millar, & Roeper, 1996). Its eff ectiveness with a college population, however, has yet to be determined. Thus the strategy has been categor ized as a Tier 2 intervention by the NIAAA, meaning that it may be promising but additional re search is needed to determine its efficacy. Yet, Chaloupka and Wechsler (1996 ) reported that the density of al cohol outlets, especially those near campus, was related to 5+4+ drinking. In a subsequent investiga tion among eight diverse campuses, Weitzman, Folkman, Folkman, and Wech sler (2002) found that outlet density was associated with heavy drinking, frequent drinki ng, and drinking-related problems among college students but to varying degrees. For example, males with a lo ng history of 5+4+ drinking are less likely to be affected by ou tlet density. Environmental mana gement may be most effective for those students whose drinking patt erns are less well-established. The role of low alcohol prices and special promotions or dr ink specials represents another environmental risk factor for 5+4+ drinking. A number of studies have shown a relationship between the price of alcohol a nd consumption. Overall, as the price of alcohol decreases, the consumption rates increase (Kuo, Wechsler, Gr eenberg, & Lee, 2003). College students are more influenced by the cost of alcohol than other demographic groups (Chaloupka & Wechsler, 1996). Research by Gruenewald and colleagues (2006) indicates that the mo st effective pricing strategy may be to place minimum prices on each type of alcoholic beverage. The most promising price intervention to reduce 5+ drinking appears to be increasing the cost of inexpensive alcohol. People who can afford expensive types of al cohol are probably less 30

PAGE 31

sensitive to incremental pri ce increases (Gruenewald, Ponick i, Holder, & Romelsj, 2006). Conversely, price-conscious indi viduals such as college stude nts appear to be the most influenced by such economic policies. One of the limitations with environmental management strategies involves the difficulties in implementing them. The lobbying efforts of th e alcohol industry and others make it difficult to pass legislation at the local and state levels (Mosher, 1999). Elected authorities with a strong pro-business philosophy are often reluctant to impl ement policy proposals as well. According to Wagenaar and colleagues (2000), policy makers ha ve a disappointing record of passing and implementing policies which are introduced into the legislatures. Even when laws do pass, oftentimes they are weakened due to a series of compromises and/or th e various loopholes to avoid them. From a conceptual point of view, environmental interventions represent perhaps the most effective prevention strategies. Neverthe less, because of the limitations listed, it is important to balance policy efforts with complementary prevention initiatives. Consequences Associated With 5+4+ Drinking The need for effective alcohol interventions remains a top public health priority to curb the severe consequences caused by 5+4+ drinkin g. Sexual assault repres ents one of the most severe outcomes associated with 5+4+ drinking. Research indicates that a woman has between a one in four and a one in five chance of being sexually assaulted while at tending college (Fisher, Cullen, & Turner, 2000). Estimates of sexual assault and ot her related violence vary widely due to definitions, the underreporting of incidents, and the data collection methods used when conducting surveys. Nevertheless, data indicate that alcohol use is involved in at least 50% of all sexual assaults involving college women (Abbey, 2002). According to Presley (1997), after alcohol consumption, males feel more powerful, sexual, and aggressive, leading to intensified 31

PAGE 32

sexual expectations. Additionally, students often blame alcohol use as justification for engaging in inappropriate behaviors (Abbey, 2002). Some fe minists and social activists believe that the social climate tends to excuse males for their behaviors when they drink, whereas females are often blamed and victimized for their choices when they consume alcohol (Katz & Kilbourne, 2004). Overall, women who attend colleges with hi gh rates of 5+4+ drinking are at an increased risk of sexual assault (Kuo, Dowdall, Koss, & Wechsler, 2004). Other alcohol-related violent acts include interpersonal violence, physical assault, homicide, and suicide. Violen ce includes behaviors that are th reatening, hostile, or damaging (NIAAA, 1997). Results from one nationally repr esentative survey of college students indicated that approximately 17% of students experienced some form of alcohol related violence or harassment during the previous year (Langford, 2006). Also, Smith, Branas, and Miller (1999), reported that the perpetrator had been drinking in 37% of assaults and 86% of homicides. In addition, approximately one in four (23%) suicide fatalities are at tributable to alcohol. Violence prevention represents a prominent public health issue, given that the second and third leading causes of death for college students are homicide and suicide, respectivel y (Barrios, Everett, Simon, & Brener, 2000). Excessive alcohol use may also result in deleterious sexual he alth and reproductive consequences for college students and others. People who abuse alcoho l are more likely to engage in unprotected sex, report having more se x partners, and use intravenous drugs (NIAAA, 2002c). Hingson (2002) estimated that during th e past year, 400,000 American college students between the ages of 18 and 24 years particip ated in unprotected se x after drinking, and 100,000 engaged in sex when intoxicated and unable to consent. The spread of sexually transmitted diseases represents an ongoing public health struggle. The United States e xhibits the highest rate 32

PAGE 33

of sexually transmitted diseases in the devel oped world (National Center on Addiction and Substance Abuse at Columbia University, 1999). Effective alcohol prevention may result in the reduction of unintended pregnancies and sexually transmitted infections, including HIV. Alcohol use during pregnancy also remains a serious probl em, as 13% of women age 18 to 44 years report drinking alcohol at least once during their pregnancy, and 3% reporting 5+4+ drinking (Centers for Disease Control and Pr evention, 2002). Maternal alcohol use during pregnancy contributes to a va riety of effects on exposed o ffspring, including hyperactivity, attention problems, learning problems, memory de ficits, and problems with social and emotional development. The most severe consequence of maternal drinking during pregnancy involves fetal alcohol syndrome (FAS), which results in a distinctive set of faci al anomalies, growth retardation, significant learning problems, and behavioral cha llenges. Drinking during the preconception period is strongly associated with unintended pregnancies, as well as other adverse maternal and pediatric health outcome s (Naimi, Lipscomb, Brewer, & Gilbert, 2003). No safe level of drinking exists due to the i nherent risks asso ciated with alcohol consumption during pregnancy (NIAAA, 2004b). Several studies have demonstrated that high-risk drinking co mpromises academic performance. The NIAAA (2002b) estimated that 25% of college students reported academic difficulties caused by alcohol use, such as earning lower grades, doing poorly on tests or papers, and falling behind. According to a 2002 national survey, approx imately one-third of students missed a class in the previous year due to alcohol use (Core Institute, 2005). Research also indicates an association between the number of drinks students consume and their academic performance. Presley and colleagues (1997) found th at third-year students with an A average consume about four drinks a week, B students consume six drinks a week, C students consume 33

PAGE 34

eight drinks a week, and students who receive Ds or Fs average almost 10 drinks a week. Even more problematic than poor grades studies suggest that college drinking is a major factor in students dropping out of school (Sulli van & Risler, 2002; Perkins, 2002). Another alcohol-related issue college students and societ y endure involves the monetary costs associated with 5+4+ drinking. Research ers estimated that college students spend about 5.5 billion dollars annually on alcohol. Ironica lly, this amount exceeds the money they spend on books, soda, coffee, and juice combined (Eigan, 1991). Precise estimat es on how much 5+4+ drinking costs society do not currently exist. However, a wealth of information exists on the costs of underage drinking. A study released in 2006 indicated that und erage drinking costs Americans 62 billion dollars every year. This figure surpasses what the federal government spent on relief for Hurricanes Katrina and Rita combined. Deaths, injuries, and lost work time, car crashes, violent crime, highrisk sex, and addiction treatment compose the vast majority of the expenses associated with underage drinki ng (Miller, Levy, Spi cer, & Taylor, 2006). Game Day Drinking Behaviors College football game day represents a unique public health challenge for prevention advocates, university administrators, and city le aders. College football games present not only high-risk drinking situations, but also sym bolic events for the university and surrounding community (Glassman et al, 2007). For roughly six home games a year, tens of thousands of fans come to campus and tailgate while open cont ainer laws are typically ignored. Perceived and real pressure from alumni, students, and other fans make the game day culture highly resistant to change. Research indicates that sports fans are less likely to abst ain from alcohol and more likely to drink excessively than others (Nelson & Wech sler, 2002). Alcohol consumption is considered 34

PAGE 35

to be a contributing factor in fan/spectat or aggression (Coons, Howard-Hamilton, & Waryold, 1995). Riots, stampedes, fights, and fatal beatings caused by rowdy spectators under the influence of alcohol have occurred at alarming rates (National Collegiat e Athletic Association, 2003). Inappropriate fan behavior not only diminishes the spectator experience and creates a poor public image but costs institutions and co mmunities thousands of dollars in extra police enforcement, cleanup and restoration expenses. Equally problematic for universities attempting to improve their prevention efforts are the mixed messages that excessive drinking while tailgating (drinking and partying on cam pus before and after athletic events), sends to students. Finally, the injuries and or fatalities related to game day alcohol cons umption place universities at increased risk for litigation. Although previous studies examined alcohol us age and the associated negative outcomes related to 5+ drinking, little is known about the specific drinki ng behaviors of college football fans on game day. Neighbors and colleagues (200 6) found that 77% of undergraduates consumed alcohol while tailgating a nd on average had 3.8 drinks. In this study, students underestimated the percentage of tailgaters who consumed alc ohol but overestimated how much they drank. Students who overestimated the nu mber of drinks their peers c onsumed while tailgating tended to be more likely themselves to drink and e xperience negative alcohol related consequences on game day. Neal and Fromme (2007) conducted related research for two years by examining the drinking behaviors of freshmen who attended the University of Texas. They implemented a web-based alcohol and drug survey semiannually and tracked students with a 30-day daily selfmonitoring instrument at randomly assigned start dates. Relative to non-game day Saturdays, student alcohol use was greater during both home and away games. The authors surmised that 35

PAGE 36

increased alcohol consumption occurred regard less of whether students attended the game. However, the increased drinking resulted only wh en school was in session, not during holiday or semester breaks (i.e., bowl ga mes). High profile games agains t national and conference rivals resulted in more drinking than games against less competitive teams. On average, more drinking occurred during away games than home games. The authors speculated that the game itself serves as a protective factor, because student s cannot drink while in side the stadium. In its inaugural year of 2003, Haun and colleagues administered the Game Day Survey on site at the University of Florida campus in Gaines ville to 497 participants. Findings revealed that while males drink more alcohol on game day, fema les suffer more alcohol related consequences associated with college football events. Adve rse consequences include experiencing a hangover, vomiting, driving under the influence, blacki ng out, suffering an injury, fighting, being victimized sexually, or getting into trouble with police. These findings are consistent with national trends which indicate th at; overall, females suffer di sproportionately high rates of negative consequences related to alcohol consumption (Clarke, 1995; Schafter, 2002; Korcuska & Thombs, 2003). Haun and colleagues (2007) r ecommended designing specific game day messages targeting women. The following year Glassman and colleagues ad ministered the revised Game Day Survey. This study included 762 participan ts who were randomly selected to complete an online survey assessing game day alcohol relate d behaviors and attitudes. Th e purpose of the study was to (a) determine if college football fans drink more on ga me day than they typically do when they party or socialize, and (b) ascertain if drinking status and fan demogr aphics (student vs. non-student) alter support for game day prevention initiative s. Somewhat surprisingl y, the result s indicated that over half of college footba ll fans surveyed reported that th ey do not typically drink on game 36

PAGE 37

day. Fans who did drink, reported that they dr ank significantly more on game day than they did the last time they partied or socialized. Overa ll, males drank considerably more than females on game day (Glassman et al., 2007). In addition, the more fans drink the less likely they are to support ga me day interventions. Nondrinkers were the most supportive of game day interventions followed by moderate drinkers; whereas, heavy drinkers showed the least support. Compared to non-students, students were more supportive of designated tailgating areas where open containers are permitted. Not surprisingly, students were less supportive of increasing underag e drinking enforcement efforts than non-students. The authors suggested that significant public support exists for prevention initiatives and that decreasing alcohol consump tion on game day is fundamental to reducing the schools 5+ drinking rate during th e fall semester. On average th ere is a home football game once every 2 weeks during the months of Augus t (last 2 weeks of th e month) September, October, and November. Thus, game day drin king behaviors of student s are likely to be reflected in their responses to the standard 5+ drinking item which asks During the past 2 weeks did you consume five or more drinks in one sittin g? As a result game day is a mediator for 5+ drinking during the fall semester, but not in the sp ring when there are no college football games. Furthermore, Dodd and Glassman (2006) conducted focus groups among UF first year students and found that students described game day drinking as different from regular drinking, in that game day drin king begins earlier in the day, sometimes as early as 9:00 a.m., and often continues until late at night. Student drinkers described game day as two separate events with drinking occurring before and after the game. Tailgating typically begins on campus parking lots before the game starts. During th e game university security and law enforcement officials strictly enforce the no alcohol policy wi thin the stadium (excluding the luxury boxes). 37

PAGE 38

Students who are visibly intoxicat ed and pose a health threat or are otherwise belligerent are asked to leave the stadium. In addition, a nyone who leaves the stadium is not allowed to reenter. After the game some students go to bars or house parties and cont inue drinking. This level of alcohol consumption a ppears to be more prevalent during the fall semester; although, game day is not the only ritual in which ex treme drinking occurs. Spring break, weddings, holidays and other events foster social environm ents in which excessive drinking is normalized and often expected. Social Marketing and Alcohol Use Given the financial burden and human suffe ring that alcohol causes, the need for evidence-based, cost-effective alcohol interventi ons on college campuses is apparent. One way to reach large numbers of people is through the us e of social marketing campaigns. While many definitions for social marketing exist, all co ntain the basic premise that social marketing employs commercial marketing techniques to influe nce a target audience to voluntarily accept, reject, modify, or abandon a behavior for the be nefit of individuals, groups, or society as a whole (Kotler, Roberto, & Lee, 2002, p.5). Philip Kotler and Gerald Zaltman first introduced the concept of social marketing ov er 30 years ago when they published: Social Marketing: An Approach to Planned Social Change (1971). Since then a growing interest in the use of social marketing concepts, tools, and practices has em erged and the field has evolved from simply designing messages to addressing complex envi ronmental and community issues (Kotler, Roberto, & Lee, 2002). The Centers for Disease Control and Preventi on (2006) encourages programs to apply the principles of social marketing to public health problems to increase the effectiveness of interventions. The primary objectiv e of social marketing is to in fluence the behavior of select 38

PAGE 39

audience members for the purpose of getting them to voluntarily change, maintain, or prevent behaviors. Social marketing programs do this by offering members of the target audience attractive benefits, and/or by re ducing barriers, which might prevent or otherwise discourage them from engaging in the desired behavi ors (Glanz, et al., 2002). Effective social marketing efforts offer produc ts, services, or ideas that members of the target audience perceive to be in their own best interests. Ideally, this is achieved through gaining a thorough understanding of the target population. The package of benefits must resonate among the intended audience. Perceive d barriers to the desire d behavior must be alleviated or lessened, and the al ternative behaviors need to provide more compelling benefits (Glanz, et al., 2002). The key is to determine what advantages people see in performing their current behavior. For example, college student s may perceive that drinking alcohol will help facilitate dating and romantic relationships. Modifying alcohol expectancies or attitudes represents a fundamental step toward changing social norms (W all, Hinson, & McKee, 1998). Conversely, it is also necessary to determine wh at negative effects people may see in their current behavior. Continuing with the exampl e of college student drinking, students may determine that a hangover represents a cost that is so severe that they decide to moderate their drinking or quit all together. Only through extensive formative research can these motivational cues be determined. Utilizing social marketing strategies to address 5+ drinking among college students is an emerging practice in the field of college health. For example, Brower and colleagues used a social marketing approach at the University of Wisconsin to curb alcohol consumption among students (Brower, Ceglarek, & Crowley, 2001 cite d in Glanz, et al., 2002, p. 455). Their needs 39

PAGE 40

assessment indicated that freshmen drank at higher rates than the rest of the student population. Consequently they decided to segment students base d on year in school and re adiness to change. The researchers targeted 5+4+ drinkers who reported relatively minimal drinking in high school, as opposed to 5+4+ drinkers with a relatively long history of dri nking, based upon their belief that the latter group would be le ss resistant to intervention. Their findings indicated that students engaged in 5+4+ drinking because they perceive d that there was nothing else to do. The first year students felt disconnected upon arriving on campus, while at the same time they wanted to assert their newly found independence. As a resu lt, the school developed an attractive series of late night activities and marketed them based on student feedback. Student organizations were promoted through advertisements, which each began with a pun to draw students in and concluded by mentioning that there were over 645 student orga nizations available to them on campus. These ads did not addres s alcohol issues, but instead focused on the needs of students to be social and active. While outcome results fr om this project are not available, process data indicated that as a result of the social mark eting campaign more student s were taking advantage of non-alcohol activities on campus (Brower, Cegl arek, & Crowley, 2001 cited in Glanz, et al., 2002, p. 59). Social Norm Campaigns Perhaps the most common social marketi ng campaign used across college campuses during the late 90s and the early turn of the millennium were social norms interventions. The social norms approach began in the mid-80s w ith the research of We sley Perkins and Alan Berkowitz (Linkenbach, 1999). They, and subsequently other researchers, found that adolescents and young adults quite often overestimated the alcohol consumption rates of their peers (Prentice & Miller, 1993; Graham, Mark s, & Hansen, 1991; Hansen & Graham, 1991; 40

PAGE 41

Perkins & Berkowitz, 1986). Students mistakenly believe that their peers consume more alcohol than they actually do. As a result, college student s drink more in an attempt to match the norm, albeit a false norm (Haines, 1998; Prentice & Miller, 1993). Accordi ng to Linkenbach (1999), students are more concerned with what they percei ve as normative, than with what they discern as healthy. In the late eighties health educators at Northern Illinois University (NIU) developed and implemented a social norm marketing campaign designed to curb alcohol abuse among students attending the university. NIU, st udents mistakenly believed that 70% of the population engaged in 5+ drinking, yet at the time of the assessment, 43% of the students took part in this behavior (Carson, 1995). A campus-wide media campaign was implemented which exposed students to the following fact: Most NIU students drink zero to five drinks when they party (Haines, 1998). Classified advertisements, a weekly co lumn, posters, and leaf lets were used to communicate the true norm (Haines, 1998). Five years after the onset of the social norms marketing intervention, the perceive d 5+ drinking rate had fallen and the actual 5+ drinking rate had been reduced to 28% (Carson, 1995). Over a nine-year period, Northe rn Illinois University reported a 44% reduction in their 5+ drinking rate (Haines, 1998). Not surprisingly, the negative consequences associated with heavy drinking fell as well. For example, there was a 31% reduction in reported alcohol-relate d injuries to the individual and a 54% reduction of alcoholrelated injuries to others (Linkenbach, 1999). A major limitation of this research project involved the non-experime ntal methods used conduct the evaluation (Haines & Spear, 1996). Rather than randomly selecting participants, surveys were disseminated through undergraduat e general education cl asses, which were reflective of the overall NIU undergraduate population. The response rate for the questionnaire 41

PAGE 42

averaged 89% over a 5-year period. In any given year at least 600 students participated in the survey in a school that usually enrolls a bout 23,000 students (Haines & Spear, 1996). In addition, no control groups were used for this study, consequently any change in the drinking rates may be the result of external influences. Other college institutions have benefited fr om this type of social norms marketing campaign. The University of Arizona also placed ads in the school newspaper, stating, % of U of Arizona Students have four or fewer drinks when they party. According to Linkenbach (1999), as a result of the social norming marke ting campaign, the University of Arizonas 5+ drinking rate fell 29% in 3 year s. Hobart and William Smith Colleges reported a 12% reduction in 5+ drinking over 2 years and Western Washingt on University had an 8% decrease in the 5+ drinking rate over a 1-year peri od using similar social norms marketing strategies (Haines, 1998). During the summer of 2001, all 23 campuses in the California Stat e University System initiated social norms campa igns (Frauenfelder, 2001). While social norm campaigns have been used widely, empirical da ta supporting social norms interventions have been lacking. Werch and colleagues (2000) conducted a randomized control study on social norms interventions a nd found no differences between the intervention and control groups on any alcohol use or risk measures. Several year s later, a national assessment of social norms interventions revealed that almost half of the schools surveyed employed a social norms intervention, yet no decreases were found on the seven alcohol measures for the study (Wechsler, Nelson, Lee, Se ibring, Lewis, & Keeling, 2003). In fact, there were increases in two of the five measures: al cohol use in the previous month and consumption of 20 or more drinks in the previous month. In addition, sc hools which implemented social norms interventions reported so me increases in lower level drinking, a result anticipated by the 42

PAGE 43

original theorists, Perkins and Berkowitz. Soci al norm proponents maintain that practitioners do not truly understand social norms theory and, as a result, they compromise fidelity when implementing this type of intervention. C onversely, opponents suggest that any documented success claimed by researchers is riddled with methodological flaws. Research by Campo and colleagues (2003) s uggested that drinking behavior relates positively to perceptions of friends drinking as in dicated by the Theory of Planned Behavior. In contrast to what social norms th eory predicts, they found that drinking behavior is not related to the perception of a typical drinking behavior of st udents. Male and female students in this study were more influenced by the perception of a male friends drinking th an by a female friends drinking. This finding might help explain why there has been a rapid increase in drinking among teenage girls, perhaps teenage girls are trying to keep pace with their male friends. In their recommendations for future research, Campo a nd colleagues suggested emphasizing other types of normative behavior, such as pacing, monito ring blood alcohol cont ent, and other common harm reduction practices. They also suggested when utilizing the Theory of Planned Behavior to focus on the social network of students rather th an inferring social comp arisons to the student body at large. Theory of Reasoned Action (TRA) and Theory of Planned Behavior (TPB) The Theory of Reasoned Action (TRA) and its extension, the Theory of Planned Behavior, involve the relations between beliefs, attitudes, intentions, and behavior (Ajzen & Fishbein, 1980). The major assumption of the TRA is that attitude (Attitude Toward the Behavior) and perceived acceptance of a beha vior (Subjective Norm) influence a persons intention (Behavioral Intention). Behavioral Inte ntion in turn, influences a persons decision to perform (or refrain from) the behavior of inte rest. Ajzen and collea gues (1991) later expanded 43

PAGE 44

the theory by adding Perceived Behavioral Control as an additional construct. This enhancement allowed researchers to examine behaviors that are not totally volitiona l by taking into account factors such as resources and opportunities (K uther, 2002). The various TRA/TPB constructs will be explained in the following sections. According to TRA, the construct Attitude Toward the Behavior represents an individuals beliefs about the behavior coupled with the we ighted evaluations of those outcomes. Thus, people who believe drinking alcohol will make them more social, and also value being social, are more likely to have favorable attitudes toward alcohol. Conversely, people who believe that alcohol makes them feel sick and who value thei r health are more likely to have unfavorable attitudes toward alcohol. Obvi ously, the more favorable attitudes toward the behavior, the more likely people are to engage in the behavior. The two fundamental indirect measures (sub-scales which link related concepts to the direct m easures) which comprise Attitude Toward the Behavior (ATB) include Behavioral Beliefs and Evaluation of Beha vioral Outcomes. Behavioral Beliefs consist of a peoples attitude towards performing a be havior; where as, Evaluation of Behavioral Outcomes concerns the relative import ance or assessment of engaging in the behavior of interest. The Subjective Norm represents peoples beli efs about whether most referents approve or disapprove of their behavior, and how motivated they are to comply with what these key referents think. A referent represents an influe ntial person such as a family member, best friend, or spouse. Most young adults tend to be motivated by the opinions of their close peers. Inherent in the Subjective Norm are the two constructs (indirect measures), Normative Beliefs and Motivation to Comply. Normative Beliefs in clude whether specific referents approve or disapprove of the behavior, whereas Motivation to Comply involve s whether the individual cares 44

PAGE 45

what the specific referent thinks. For example, a male college student may perceive that his best friend approves of his drinking heavily and may be very motivated to comply with this belief. However, this same student may have a girlfr iend who disapproves of his drinking heavily, but he may not be motivated by this referent. The Subjective Norm concerns peoples overall assessments of their key referent s approval of the behavior of in terest and their Motivation to Comply with these referents. Finally, the most important determinant of be havior is Behavioral Intention, a persons likelihood of performing the behavior. In gene ral Behavioral Inten tion characterizes an individuals plan or probability of performing a behavior. Attitude, norms, and perceived control over the behavior each shape Behavioral Intention. The amount of influence each of these constructs has on Behavioral Intention differs among various populations and behaviors. Behavioral Beliefs and Eval uation of Behavioral Outcom es are typically assessed by using bipolar scales scored -3 to +3 (Monta no, Kasprzyk, & Taplin, 1997). The outcome values from this calculation are called indirect attitudes (Albarracin, Johnson, Fishbein, & Muellerleile, 2001). Ajzen (2002) proposed an alternate wa y to measure a persons attitude toward a behavior. In essence, people n eed to determine their overall ev aluation of performing a specific behavior. For example, For me to cut back on my drinking is: Bad -3 -2 -1 0 1 2 3 Good extremely quite slightly neutral slightly quite extremely A score of indicates that the respondent believes that cutti ng back on drinking is extremely good, and therefore has a positive attitude toward cutting back on drinking. The more positive the attitude is to the behavior the stronger the intention to perf orm the behavior. The Subjective 45

PAGE 46

Norm can also be measured using a semantic di fferential scale (Ajzen, 2002). If people believe that their key referents would a pprove of them engaging in a ce rtain behavior the higher the score and the greater their Beha vioral Intention. Strong Behavi oral Intention increases the likelihood of carrying out the behavior. The constructs that make up Perceived Beha vior Control (PBC) include Control Beliefs and Perceived Power (both indirect measures). Control Belief is the assessment one makes about the presence or absence of facilitators and barriers to performing the behavior. Perceived Power is the evaluation of each condition making the behavi or more or less difficult. Both constructs are typically scored -3 to +3 or 1 to 7. The construct of PBC is the overall assessment of ones power or control to perform or discontinue the behavior. According to Bandura (1977), the PBC construct is analogous to self-e fficacys contribution to the Soci al Cognitive Theory. Studies, however, indicate that PBC inadequately predicts behavior. A meta-analysis revealed that only one study demonstrated a signifi cant relationship between inte ntion and PBC (Ajzen, 1991). Controlling for intention further weakened the st atistical support for this construct (Reinecke, Schmidt, & Ajzen, 1996). Figure 2-1 illustrates TP B constructs, including direct and indirect measures. A critical step in utilizing the TRA/TPB theory is to conduct an elic itation study to create a pool of items for the inst rument (Montano, Kasprzyk, & Taplin, 1997). An elicitation interview should be conducted with at least 15 to 20 people. The sample should include both those people who intend to perform the behavior and those w ho do not. The TRA constructs necessitate asking respondents to describe posit ive and/or negative attr ibutes or outcomes of performing the behavior. Participants should also be asked to describe any individuals or groups to whom they might listen, who are either in favor of or opposed to their performing the 46

PAGE 47

behavior. This information subsequently provides the theoretical foundation for the interventions and survey item development. A variety of approaches exist to develop questionnaires, including personal interviews (Michels & Kugler, 1998), focus groups (JenningsDozier, 1999), and open-ended surveys (Bosompra, 2001). Some instruments are simply based on a related literature review or suggestions provided by Fishbein and Ajzen (Goksen, 2002). While researchers continue to successfully use the TRA, and its extension the TPB, in a variety of applications since the theory was first introduced by Fishbein in 1967, according to Glanz, Rimer, & Lewis (2002) the theory has not escaped criticism. Goksen (2002) pointed out that the Subjective Norm construc t may be too complicated to me asure using one dimension. In addition, the construct of Behavior al Intention could be removed fr om the theory altogether to make the theory more parsimonious. Conversely, in an attempt to make the theory more explanatory, researchers have s uggested adding a variety of constr ucts to the theory, such as moral norms, self-identity, temporal stabilit y, and past behavior (Conner & Armitage, 1998; Sheeran & Abraham, 2003). Nevertheless, this theory provides a fram ework for identifying key behavioral and Normative Beliefs affecting behavi or. The use of this informa tion can contribute to message development designed to target and change identi fied beliefs or values, ultimately leading to a change in behavior. TRA/TPB assume that underlying reasons determine peoples motivation for performing a behavior. These reasons ultimately influence peoples Attitude Toward the Behavior and Subjective Norm, regardless of whether these beliefs are ra tional, logical, or correct (Glanz, et al., 2002). This framework wa s used to analyze what motivates students to engage in 5+4+ drinking and what can mo tivate them to alter this behavior. 47

PAGE 48

Theory of Planned Behavior and Alcohol Use In a cross-sectional study, Wall, Hinson, and McKee (1998) used the Theory of Planned Behavior to predict alcohol consumption among undergraduate students of legal drinking age at the University of Western Ontario. Consistent with Schlegels research (1992), they found that the Theory of Planned Behavior was superior to the Theory of Reasoned Action in predicting problem drinking. Attitudes (good/ba d, wise/foolish, pleasant/unpleasant, favorable/unfavorable) emerged as the strongest predictor of the intentions of undergraduates to drink too much (R2 males = 0.34 and R2 females = 0.34). In general the more favorable the attitudes of the participants, the more likely thei r intentions to drink increased. The Subjective Norm was less predictive. The intentions of fe males to drink in excess did not appear to be influenced by the Subjective Norm at all, a nd the construct only acc ounted for 1.6% of the variance among the male population. The R2 increased by 4% and 17.3% for males and females, respectively when the Perceived Behavioral C ontrol construct was added to the model. The work of Wall and colleagues (1998) ra ised several issues. When assessing the Subjective Norm, participants were asked to rate the likelihood that four specific sets of people expect them to drink too much (1 unlikely to 5 likely). These groups included most societal relationships: family, friends, and boyfriend/girl friend. The category most people in society does not match well with the strict definition of a referent, who is someone influential. In addition, most college students are less influenc ed by their family members than they were formerly. The reliability measure was so low for this scale that the rese archers had to deviate from their original methods. Instead, they decide d to assess the Subjective Norm with a single item which measured the motivation of participants to comply with perceived general societal pressure to engage in excessive consumption. Again, this measure is inconsistent with the 48

PAGE 49

fundamental underpinnings of the Subjective Norm and influential referents. This study may have produced more valid results if the Subjectiv e Norm were measured using the following sets of individuals: best friend, significant other and close frie nds. Finally, the survey asked participants if their referents expect them to drink too much. Since the Subjective Norm construct is not about expectati ons per se, the question should ha ve focused on what participants perceive referents would think about them engaging in the behavior of inte rest. A better item, found on the Core Alcohol and Drug Survey Long Form, asks participants the extent to which their close friends would disapprove of their having five or more drinks in one setting. The responses included do not disapprove, disapprove, and strongly disapprove. While these responses may be somewhat biased and limited, they appeared to be much more appropriate than those used by Wall and colleagues (1998). Conversely, in a series of three studies w ith undergraduates Trafimow (1996), found that Attitudes Toward the Behavior were better predictors of drinking Intentions than Subjective Norms or Perceived Behavioral Cont rol. He suggested that it is not necessary to spend time and resources measuring actual behavior, because inte ntions do such an effective job of predicting behaviors. Moreover, he stressed the importance of focusing on the type of drinking measure utilized. In his study, participants were asked to indicate their attitudes, norms and intentions toward (a) avoiding drinking, (b) drinking enough to get a slight buzz (a term used by the authors), or (c) drinking enough to get drunk. This study did not adequately identify 5+4+ drinkers. A slight buzz for a heavy drinker may involve enough alcohol to get a moderate drinker intoxicated. Specific comparisons be tween drinking intensity among the various participants were lack ing. Trafimows findings concurred with those of Budd and Spencer (1984) as well as his earlier research (Trafimow & Finlay, 1996). He c oncluded that Subjective 49

PAGE 50

Norms and Perceived Behavioral Control were relatively unimportant pr edictors of drinking among undergraduate students intenti ons to get drunk. Other research concerning college students a nd alcohol use also i ndicated that TPB did not significantly add predictive powers above and beyond the TRA (OCalla ghan, Chant, Callan, & Baglioni, 1997). OCallaghan and colleagues sp eculated that because most respondents were non-problem drinkers they experi enced high perceived self-control over their drinking. Too few problem drinkers with low Percei ved Behavioral Control were in the study for the construct to elicit significant influence. One limitation of the OCallaghan study was the authors focused on intentions to drink alcohol rather than intentions to get drunk. However, this study did reveal a significant link between past beha vior and intentions. The res earchers suggested that alcohol consumption may be a mindless or unthinking process for some college students. Thus, drinking may not be motivated by attitudes, but me rely by having consumed alcohol on previous occasions (habit effect). OCallaghan and co lleagues believed that more accurate measures of attitude are needed to identify it s role in behavior prediction over and above the influence of past behavior. Indeed, prior research indicated that attitudes may be generated from past behavior (Fazio & Zanna, 1981). While OCallaghan and colleague s found the construct PBC to be negligible in predicting drinking patterns among college students, Norman, Bennett, and Lewis (1 998) reported just the opposite. In a study conducted at a Welsh Univ ersity, 136 undergraduates were asked about their drinking behaviors as they relate to the TPB. However, the definition of heavy episodic drinking used in this study was quite different from the version used in studies with U.S. students. In Wales, the heavy episodic drin king criterion for males wa s 10 drinks in a single session, seven for females, within the previous week. 50

PAGE 51

Norman and colleagues (1998) created a series of indirect measures for each construct as outlined in the TPB. Positive and negative Behavioral Beliefs, Normative Beliefs, and Control Beliefs were utilized. The regre ssion analysis revealed that gende r explained nine percent of the variance. Together, the variab les under consideration explaine d 38% of the variance in the frequency of heavy episodic drinking. Only Pe rceived Behavioral Cont rol and positive Control Beliefs were significant i ndependent predictors. The most recent study found on TPB and college student drinking was conducted by Johnston and White (2004). The sample for this study included 139 first-year female undergraduate students enrolled in an introdu ctory psychology class at a large Australian university. The authors conducted an elicitation study to develop th e indirect measures for each of the TPB constructs. The aim of the study was to examine the range of beliefs that differentiated 5+4+ drinkers from non-5+4+ drinkers. More specifically, the researchers assessed the beliefs of 5+4+ drinking as they re lated to perceived costs and benefits, beliefs concerning controllability, and be liefs about how others influen ced an individuals decision to drink. Multiple logistic regres sion analyses were utilized to determine which of the indirect measures best predicted 5+4+ drinking. The results indicated only three statistica lly significant measures predicted 5+4+ drinking, including beliefs about the costs associated with 5+4+ drinking (having a hangover/feeling sick, damaging hea lth, behaving embarrassingly), Ev aluation of the Benefits of drinking (relaxing/unwinding, havi ng fun/socializing, reducing inhibitions ), and Normative Beliefs (how key referents thi nk about engaging in 5+4+ drinking). They concluded that messages targeting college women should downplay the perceived benefits of 5+4+ drinking and should highlight the perceived costs. Findings from this st udy indicated that women who 51

PAGE 52

engaged in 5+4+ drinking believed that their ke y referents approved of their consuming four or more drinks. Finally, the effect of control factors revealed no significant differences between 5+4+ and non-5+4+ drinkers. Overall, these studies indicated that the TPB represents an appropriate model for examining drinking patterns among college students. However, th e literature contains mixed reports on the extent to which Perceived Behavior al Control influences 5+ 4+ drinking behavior among college students. There are also discre pancies concerning whether Attitude Toward the Behavior or Subjective Norm re presents the most powerful cons truct in predicting Behavioral Intention with this population and behavior. Different ques tionnaires, items, and methods produce disparate results even among the same constructs, behavior, and population group. Table 2-1 summarizes key findings from the publ ished studies and denot es which constructs were significant in predicting 5+4+ drinking among college students. Additional research needs to be conducted in this area. Alcohol Consumption Measures A great deal of debate exists on how best to measure high-risk drinking as well as how to label this behavior. For example, the use of the term binge drinking concerns some university officials, prevention specialists, and substance abuse professionals. Inconsistent meaning and definition of the term yields diffe rent implications. Clinicians dealing with addiction define binge as a multiple day bender (NIAAA Newsletter, 2004a), whereas Wechsler and Nelson (2001) described a binge as consuming five or more drinks in one sitting within the previous two weeks. Wechsler and Nelson (2001) concluded that risk for alcohol related consequences increases significantly afte r men consume five or more drinks (four or more drinks for a female) and st ate that the term binge is justified based on such scientific 52

PAGE 53

evidence. Some professionals believe the use of the expression binge ex aggerates or distorts the problem. The use of this e xpression may also contribute to th e misperception that everyone is drinking heavily, thereby perpet uating a false norm (Perkins, 2003). The media continue to use the term bing e in newspaper headlin es across the country (Vicary & Karshin, 2002). Ho wever the editors of the Journal of Studies on Alcohol discourage the use of this term in thei r journal (Thombs, Olds, & S nyder, 2003). The professional consensus to describe this be havior includes the use of te rms like high-risk or heavy episodic. All of these terms, including binge, utilize esse ntially the same 5+/4+ drinks measurement criteria (four or more drinks in one sitting for female s within the previous 2 weeks, five or more drinks for males) (Higher Edu cation Center for Alcohol & Other Drug Abuse and Violence Prevention, 2000). Other researchers are concer ned that the conventional 5+/4+ drinks measurement criterion fails to identify episodes of intoxi cation accurately. Thombs and colleagues (2003) used breathalyzers to measure BAC (Blood Alcohol Concentration) levels in college students returning to their residence halls after socializ ing on Wednesday, Thursday, Friday, and Saturday nights. In most instances they found that those students who reported having 5+/4+ drinks did not have BACs above the legal limit. This findi ng is consistent with Lange and Voas (2001) research, in which they found that the 4+/5+ measure predicted relatively low BAC rates among young adults crossing the U.S. border at Tijuana, Mexico. Because the 4+/5+ measure fails to include body weight and time parameters, the measure provides high sensitivity but little information about the degree of into xication (Thombs et al., 2003). 53

PAGE 54

Researchers and governmental organizations co ntinue to work toward a more accurate measurement of problematic drinking. Th e NIAAA (2004a) National Advisory Council attempted to clarify the i ssue by defining a binge as: a pattern of drinking alcohol that brings blood alcohol concentration (BAC) to 0.08 gram percent or above. For the typi cal adult, this pattern corres ponds to consuming 5 or more drinks (male) or 4 or more dri nks (female) in about 2 hours. One limitation to this definition is that it may be difficult for peopl e to recall accurately the amount of time in which they consumed a certain quantity of alcoholic drinks. White, Kraus, and Swartzwelder (2006) contend that while the 5+/4+ measure is helpful in identifying high-risk drinking, it fails to capture data regard ing how heavily people actually drink. To remedy this issue they measured not on ly the standard binge drinking rate (5+/4+), but also assessed double (10+/8+) and triple (15+/12+ ) the binge drinking thre sholds using a national sample of college students. They found that near ly one-fifth of males consumed 10+ drinks at least once during the previous 2 weeks. Further, roughly half of all males categorized as binge drinkers regularly drank at tw ice the binge drinking threshold. Clearly, the probability of experiencing catastrophic consequences increases substantially when drinking occurs at these very high levels. Drinking at these high levels is thought to be precipitated by events which are culturally associated with heavy drinking, such as New Years Eve, St. Patricks Day, Spring Break, 21st Birthday, Cinco de Mayo, weddings, and game da y among others. These ritualistic events represent unique circumstances for consuming alcohol, occasi ons where overindulgence seems more acceptable and even expected. Listiak (a s cited in Voas, Furr-Holden, Laurer, Bright, Johnson, & Miller, 2006) referred to this phenom enon as a time out period, in which social control is relaxed and deviant acts are legitimized by an attitude of tolerance, a time out from the status quo. Time out events are associated w ith heavy drinking, illegal drinking, fighting, sexual 54

PAGE 55

activity, and overall impulsive beha vior (Voas, et al., 2006). Ther e is little research about the drinking patterns associated with context-specific ritualistic events (Nei ghbors et al., 2006). Yet law enforcement efforts increase substantially fo r these major events, suggesting that there is a need for intervention. Given the increased numbe r of hours spent drinking and the number of drinks consumed at these events, a specific inst rument that can measur e the unusually high rates of alcohol consumption at these events appears warranted. Summary Alcohol abuse among college stude nts represents one of the most prominent health issues for colleges and universities. Desp ite increased attention and resources directed toward this issue there has been very little improvement over the last decade either nationally or locally. An abundance of published research exists analyzing th e health risks associated with alcohol use. While no clear formula for determining the conflu ence of factors that cause 5+4+ drinking, the presence of multiple risk factors such as pe rsonality, family, genetics, and environmental influences increases the probability of persons experiencing negative consequences due to excessive alcohol consumption. A particular area of concern regarding 5+4+ drinking on college campuses involves college football games, also known as Game Day. Alcohol consumption is considered to be a contributing factor in fan/spectator aggression. Riots, stampedes, fights, and fatal beatings caused by rowdy spectators under th e influence of alcohol warrant intervention. In general, game day represents a ritualistic event wher e overindulgence seems more acceptable and even expected. This phenomenon sometimes referred to as a time-out period can be characterized by lax social cont rol and deviant acts which are le gitimized by an attitude of 55

PAGE 56

tolerance. Additional research is needed to better understand how to prevent or minimize this extreme and ritualistic context-sp ecific drinking behavior. Social marketing constitutes a cost-effec tive method for reaching large numbers of people. Changing the social no rms and mores concerning alcohol represents a fundamental step toward modifying the social environment. Utilizing the Theory of Planned Behavior as a framework for developing effective social mark eting messages shows promise. However, its overall applicability to college student 5+4+ drinking remains in question, given the mixed results present in the literature. In addition, the issue of how most effectively to assess and label high volume drinking further comp licates alcohol research. 56

PAGE 57

Behavioral beliefs Evaluations of behavioral outcomes Normative beliefs Motivation to comply Control Beliefs Attitude toward behavior Subjective norm Behavioral Intention Behavior Perceived behavioral control Perceived Power Figure 2-1. Theory of Pla nned Behavior Constructs Table 2-1. Summary of the TPB Alcohol Related Articles Author Year published Attitude Subjective Norm Perceived Behavioral Control Johnston & White 2004 X X Norman, et al. 1998 X* Wall, et al. 1998 X X OCallaghan, et al. 1997 X X Trafimow 1996 X Schlegel 1992 X X Budd & Spencer 1984 X X Note : X = statistically significant predictor = construct served as the strongest predictor = indirect measure Normative Belief = indirect measure Control Belief 57

PAGE 58

CHAPTER 3 METHODS This chapter provides information on the data collection methods employed for this study. The purpose of this inves tigation was to determine the Th eory of Planned Behaviors (TPB) efficacy in predicting Extreme Ritualis tic Alcohol Consumption (ERAC) within a population of undergraduates at a large unive rsity on game day. ERAC is defined as consuming 10 or more drinks on ga me day for a male and eight or more drinks for a female on game day. Gender and grade clas sification variables were also analyzed to determine their respective influence on the TPB c onstructs as they relate to ERAC. This chapter includes a description of the research design, research variables, st udy population, instrumentation, participants, item development, data collec tion procedures, and data analysis methods. Research Design A cross-sectional survey design with randomly selected participants was utilized to answer the research questions proposed for this investigation. Cross-sect ional surveys are often used in social science research to provide a single point in ti me examination of the sample population (Dooley, 2001). However, several disadva ntages exist when using cross-sectional study design including: it only includes those people who completed the questionnaire, it may represent only those who have the disease or engage in the behavior, and it may not be effective if the disease or behavior is rare (Timmreck, 1994). Very impor tantly, a cross-sectional design does not allow for analyzing cause and effect relationships. Causal relationships can only be established by using experimental de sign (Cottrell & McKenzie, 2005). Nonresponse bias is another important issu e to consider when implementing a crosssectional design (Timmreck, 1994,). Nonresponse bias accounts for differences between those participants who complete the study and those who do not. If selected participants randomly 58

PAGE 59

complete or fail to complete the survey, the response bias is negligible. The decision by participants to complete the survey depends on their characteristics, attitudes, opinions, and interests with regard to the research agenda. C onsequently, some classes of individuals are likely to be over, or under-represented, and therefore ma y bias the research findings (Alreck & Settle, 1995). General principles can be employed to estima te the degree of non-response bias. Those who are typically interested or involved with the topic are much more likely to participate in a survey than those who are not. This includes individuals who have strong feelings about the issue, whether they are negative or positive. Ov erall, participants who feel apathy or who are inexperienced with the topic tend to be less likely to complete th e survey. In addition, persons who are very busy are less likely to respond to a survey than are people with fewer time constraints. Generally, resear chers should estimate the degree to which non-respondents will affect the results of the survey. It is recommen ded that in cases in whic h nonresponse is likely to be significant, researchers should consider collecting th e data in another manner, such as via personal interviews (Alreck & Settle, 1995). Using cross-sectional design provides some inhe rent advantages when compared to other research methods. It is less expensive and more expedient to conduct cr oss-sectional research, because data are collected once rather than multiple times. In addition, cross-sectional research supplies useful information for designing interven tions (Friis & Sellers, 1996), provides a means of exploring the interrelatedness of attributes of disease or conditions within a group, and is based on a sample of a population, not on individuals who present themselves for medical treatment or policy violations. When compared to longitudinal studies, cro ss-sectional data often yield similar results (Timmreck, 1994). 59

PAGE 60

Given the advantages of cross-sectional research a nd limited resources for this investigation, the Game Day Survey was administer ed at a single point in time. These methods were consistent with Game Day Survey da ta collection procedures implemented by the University of Florida Student Health Care Cent er for the past three years. This enables University officials to compare key indicators over time. The results from prior Game Day Survey analyses reveal that approximately a third of the sample participated in the survey (Glassman et al., 2007), and simila r response rates were expected in the this investigation. In addition, participant demographics were matche d to the overall student population to assess response bias. Although, persona l interviews represent an in formative method for obtaining information on alcohol issues w ith college students (Dodd & Gla ssman, 2006) this strategy was not feasible due to the limited resources available for this investigation. Research Variables In this study multiple observatory variables from the Theory of Planned Behavior (TPB) as they relate to extreme ritualistic alcohol consumption were examined. These included Subjective Norm, Attitude Toward the Behavior Perceived Behavioral Control, and their corresponding indirect measures, Behavioral Beliefs, Evaluation of Behavioral Outcomes, Normative Beliefs, Motivation to Comply, C ontrol Beliefs, and Perceived Power. A description of how the TPB constructs for the Game Day Survey were created and how each measure was assessed is requisite to unde rstanding the research design. The format for constructing the Game Day Survey items was mo deled after the method developed by Ajzen and Fishbein (1980). The theory suggests that a persons feelings, perceived acceptance, and control determine a persons intention to engage in a be havior. Subsequently, the individuals intention predicts whether the person performs the behavior The direct measures (Attitude Toward the 60

PAGE 61

Behavior, Subjective Norm, and Perceived Behavi oral Control) are t ypically more strongly associated with intention and behavior than are the indirect measures. However, the indirect measures (Behavioral Belief, Evaluation of Behavioral Outcomes, Normative Beliefs, Motivation to Comply, Control Be liefs, and Perceived Power) may be more beneficial in helping researchers and practitioners deve lop interventions because the information is more specific in terms of what motivates behavior (Glanz, et al., 2002). Note: Tables 3-1, 3-2, and 3-3 provide the items related to the u nderlying TPB constructs. Attitude Toward the Behavior Attitudes about inebriation on game day were assessed using six different seven-point semantic differential scales (good/bad, bene ficial/ harmful, enjoyable/unenjoyable, healthy/unhealthy, favorable/unfavorable, and wise /foolish; direct measure, Table 3-1). See Appendix A for the survey instrument. Behavioral Beliefs Behavioral Beliefs items were measured by asking participants their beliefs concerning game day drinking (indirect meas ure, Table 3-1). For example, participants indicated the likelihood that I would have more fun if I got drunk on game day. Evaluation of Behavioral Outcomes Evaluation of Behavioral Outcom es (indirect measure, Tabl e 3-1) items were measured by asking participants the value th ey attach to game day drinki ng. For example, participants indicated their level of agreement that Havi ng fun on game day is important to me. Subjective Norm Subjective Norm (direct measure, Table 3-2) was measured by asking participants to rate the extent to which individuals, whom they value, would approve or disapprove of their drinking 61

PAGE 62

alcohol to the point of int oxication on game day (home football game). For example, participants were asked the extent to which th ey agreed with the following statement: Most people I hang out with get drunk on game day. Normative Beliefs Behavioral Belief (indirect measure, Table 3-2) items were measured by asking participants if specific key referents would approve of their alcohol use on game day. For example, participants indicated their level of agreement that My best friend would approve of me getting drunk on game day. Motivation to Comply Motivation to Comply items (indirect meas ure, Table 3-2) were measured by asking participants how motivated they were to meet th e expectations of specific key referents. For example, participants indicated their level of motivation to comply by answering the following question. When it comes to dri nking alcohol, how motivated are you to meet the expectations of your best friend? Perceived Behavioral Control Perceived Behavioral Control (direct meas ure, Table 3-3) items were measured by asking participants to indicate their level of ag reement concerning personal drinking behavior. For example, participants responded to statements such as I am confident that I can limit my alcohol consumption on game day. Control Beliefs Control Beliefs (indirect measure, Table 33) measured the per ceived opportunities and barriers of participants concerni ng alcohol consumption on game da y. For example, participants indicated how often they Use a designated driver or safe transportation on game day? 62

PAGE 63

Perceived Power Perceived Power items (direct measure, Table 3-3) assessed the likelihood that opportunities and barriers would infl uence the alcohol consumption of participants on game day. For example, they indicated the extent to which Having a designated driver or safe transportation influences my deci sion to get drunk on game day? Behavioral Intention Finally, participants were asked to disclose their level of intention to drink alcohol on Game Day. For example, they rated their agreemen t with the statement I intend to get drunk at the next Gator home football game I attend. Behavior The alcohol consumption questions were measured by asking students how many alcoholic beverages they typica lly consume on game day. Recall bias represents an important limitation when designing an item which asks part icipants to summarize their drinking over an extended time period (Dawson, 2003). The colleg e football season starts in early August and ends in late November, a span of 4 months. As king participants to think back over that time period and determine what is typical may be prob lematic for some individuals. Nevertheless, it is not uncommon to find items on alcohol-related surveys which ask partic ipants to report their usual drinking behavior over the previous year. For example, Dawson (2003) used the following item to measure this outcome: On the days when you drank beer in the last 12 months, about how many (cans/bottles/glasses) did you US UALLY drink in a single occasion? The Game Day Survey also included an exact recall item which asked participants to report the total number of alcoho lic drinks they consumed before, during, and after the final Gator home football game of the 2006 season (Uni versity of Florida ve rsus Western Carolina 63

PAGE 64

University (WCU) November 18, 2006). Since the survey was distributed on November 20, 2006 this item minimized recall bi as because it reduced the time period between the behavior and the data collection. It is mu ch easier for participants to reflect back on the most recent Gator home football game than it is for them to synthe size their typical or aver age home football game experience over the course of the entire season. There are other variables to consider regard ing home college football games as well. These include the time of day the game is play ed and the weather. Thus exact recall questions are limited in that they may represent an atypical scenario which is not reflective of the usual drinking patterns of participants. In add ition, exact recall questions may not capture the consequences associated with certain behaviors because they occur infr equently (Dawson, 2003). For example, the probability of a participant vom iting as a result of their alcohol consumption over the course of an entire football season is much more likely than vomiting after one particular home game. The item asking for the number of alcohol dr inks typically consumed on game day was included in an attempt to be consistent with other standardized alcohol measures. The AUDIT [Alcohol Use Disorders Identification Test] (B abor, Higgins-Biddle, Saunders, & Monteiro, 2001), BASICS [Brief Alcohol Screening Instrument College Students] (Dimeff, Baer, Kivlahan, & Marlatt, 1999) and the first article written on game day drinking (Neighbors et al., 2006), all use the word typical when asking the respondent s to summarize their drinking behavior. In addition, utilizing this item format matches the three previous Game Day Surveys implemented at this university, allowing co mparisons over time and reducing the need for recurrent assessment of reliabili ty (Dawson, 1998). 64

PAGE 65

One way to increase reliability and validity on items involving alc ohol is to include questions on the drinking cont ext (Dawson, 1998). The Game Day Survey has two items which specifically included contextual cues. One of th ese asked participants to identify where they drink before, during, and after the game. Another question asked participan ts to report the total number of alcoholic drinks they consumed dur ing the two hours before the game, during the game itself, and the two hours after the game. This measure is consistent with NIAAAs (2004a) recommendation that a time peri od be designated to assess 5+4+ drinking more accurately. These various items complement one another becau se no single item alone can adequately assess risk. As Dawson (1998) states Given researchers varying object ives and the lack of a single set of measures that works best toward all possible ends, it is not surpri sing that there is not real consensus as to what are the best measures of consumption (p. 965). The Game Day Survey was approved by the Institutional Review Board (IRB, 2006-U604) (Appendix B). All data were collected anon ymously. Participants reviewed the informed consent process and voluntarily accepted the terms before logging on to the survey. Study Population The study population for this investigation included approximately 50,000 University of Florida students who were enrolled for classes during the fall semester of 2006. A total of 2,000 students, ages 18-24, were randomly selected by the registrar to participate in the survey. The registrar used a Statistical Analysis Software (S AS) program to identify students and their e-mail addresses for the survey. For the purposes of this inquiry, standardized tables indicated that at least 381 randomly selected subjects needed to participate in the survey for it to be considered a representative sample of the overall student population (Krejcie & Morgan, 1970). A power analysis conducted using Raosoft, Inc., web survey software corrobo rated the sample number provided in the table. 65

PAGE 66

Given the population size, a margin of error of 5%, a confidence level of 95%, and an estimated response rate of 40% (with incen tives), at least 3 67 participants were needed for this investigation. Previous surveys administered at this inst itution yielded modest response rates. For example, the 2004 Game Day Survey had a 35 % response rate without using incentives. Other alcohol and drug surveys implemented at UF yielded less, approximately 20%-25%. Thus, a sample of 2,000 students was util ized for this investigation. Instrumentation The items from this study were based on th e Game Day Survey created by Glassman and colleagues (2003, 2004). In 2003, Glassman and co lleagues developed the Game Day Survey specifically designed to measure the alcohol c onsumption of fans, on th e day (or night) of college football events (Haun, Glassman, & Dod d, 2007). The survey items were modified and adapted from the standardized Core Alc ohol and Drug Survey Long Form developed by Dr. Presley and colleagues (Core Institute at Southern Illinois University, 1994). Other nationally recognized college alc ohol instruments were also revi ewed and utilized including the National College Health Assessment and the Coll ege Alcohol Survey (Wechsler et al., 2002). The survey was designed by the institutions Co ordinator of Alcohol & Other Drug Prevention (Glassman). It was analyzed by a variety of expe rts from the University of Florida, including a college health promotion specialist, a student affairs administrator, an alcohol and drug researcher, and a Distinguished Professor in the College of Pharmacy. A ll expert reviewers had at least ten years of experience in their respectiv e fields. The Game Day Survey was pilot-tested six times, with at least 15 UF college students completing the survey each time. Reliability coefficients, using Cronbachs alpha, for each of the scales based on the 2003 Survey were as 66

PAGE 67

follows: alcohol consumption items, 0.86; conse quences from alcohol consumption items, 0.82; and social norm items, 0.84. The original Game Day Survey consisted of 30 items. Alcohol consumption items assessed the total number of drinks consumed on game day as well as the number of the drinks consumed before, during, and after the game. Nine items measured alcohol related consequences, utilizing dichotomous response opti ons (yes/no). Three items included a 7-point categorical scale related to ge ographic settings for alcohol co nsumption (home, friends home, restaurant, bar, tailgate area, other). In additi on, there were three 5-poin t Likert interval items related to social norms concerni ng the perceived alcohol consumpti on of others; and four 5-point Likert ordinal items related to attitudes about game day drinking preventi on initiatives, including support for policy change. Finally, three demogr aphic items assessed participant age, gender, and attendee status (undergradua te, graduate, employee, UF Alumnus, Gator fan, visiting fan, other). The Game Day Survey has been substantiall y revised since its inception in 2003. In 2004, survey items involving alcohol consumption were changed from ordinal responses, listing a range of drinks consumed, to c ontinuous responses where particip ants could simply indicate the number of drinks they consumed up to 12+ dr inks. In addition, the 2004 version of the Game Day Survey was administered electronically, ra ther than the onsite paper/pencil questionnaire utilized in 2003. In 2006, additional questions were added to the survey to measure the constructs associated with the Theory of Planned Behavior as shown in Tables 3-1, 3-2, and 3-3. These 55 items included six items assessing Attitude Toward the Behavior, six items assessing Behavioral Beliefs, six items assessing Evaluation of Behavi oral Outcomes, four items assessing Subjective 67

PAGE 68

Norm, four items assessing Normative Beliefs, si x items assessing Motivation to Comply, five items assessing Perceived Behavioral Control, six items assessing Control Beliefs, six items assessing Perceived Power, and f our items assessing Behavioral Intention. For the indirect measures of the Attitude Toward the Behavior and Perceived Behavioral Control, the six items consisted of three cost (nega tive behavioral and Control Belie fs) and three be nefit (positive behavioral and Control Beliefs) items associated with getting drunk on game day. These items were developed based on a literature review (Wall et al., 1998; Ajzen, 2002; Trafimow, 1996; Norman, Bennett, & Lewis, 1998). Ta bles 3-4-6 list items found in the literature review and their respective reliability scores. A test-retest assessment was conducted to determ ine the reliability of the revised version of Game Day Survey items. These test-retest measures were completed in two Personal and Family Health (HSC 2100) classes taught at the University of Flor ida The first was administered 9 days after the final UF home football game (November 27, 2006) and the retest was conducted 9 days later (December 6, 2006). Students were instruct ed to provide the last four digits of their student identification numb ers on the top right hand corner of th e paper/pencil survey. A total of 120 surveys were administered. Ei ghty-nine of these were returned with the information needed to match results, yielding a 74% response rate. To determine the test-retest values for th e categorical items and for the continuous variables, Spearman and Pearson correlation coefficients were calculated, respectively. Overall the test-retest results were reliable. Each item analysis was statistically significant at the 0.05 alpha level. According to Portney and Watkins (2000), correlation values of 0.00 to 0.25 indicate little or no relationship; values of 0.25 to 0.50 suggest a fair degree of relationship; values of 0.50 to 0.75 are considered moderate to good, and values above 0.75 are classified as 68

PAGE 69

good to excellent. The results shown in tables 3-7-15 indicate at l east a fair relationship between items and most correlations were moderate to good. Table 3-7 illustrates that the seven demographic items had th e highest correlation values, ranging from 1.00 to 0.97 with an average of 0.99. The six prevention items had modest correlation values ranging from 0.62 to 0.42 with an average of 0.53, as noted in Table 3-8. The two foil questions (GatorLight tip sheet and GatorHealth Guide) had the lowest correlation values in the prevention scale, 0.42 and 0.46, respectively. The foil items utilized for this survey were fictitious, meaning no participant should have indicated that they had seen the prevention initiative because it did not exist. The foil it ems represent a measure to assess the participants social desirability when answering items. If the two foil questions were removed, the range for the prevention items would be slightly improved to 0.62 to 0.56 and the average would increase to 0.57. Overall the test-retest results for the game day drinking items (Table 3-9), consequence items (Table 3-10) and social norms items (Tab les 3-11) indicated good te st-retest reliability (Portney & Watkins, 2000). Th e range for the game day drinking items was 0.87 to 0.53 with a good average of 0.77. The item with the 0.87 value was, What is the total number of drinks that you typically consume before, during, and after a Gator home football game? This item represented the dependent or endoge nous variable for this study. The item with the lowest value in this scale (0.53) asked partic ipants where they typically sp end the majority of their time drinking alcoholic beverages before the game. Th is lower value may be due to the fact that students may participate in this activity in varied settings from one game to next. The range for the 11 consequence items was 1.00 to 0.29 with a good average of 0.73. The question with the 0.29 value asked participants if they vomited as a result of their game day related drinking over 69

PAGE 70

the course of the 2006 football season. This item is consistent with other standardized surveys such as the Core Alcohol and Drug Survey (Presley et al., 1997). The three social norm items ranged from 0.85 to 0.75 with a good average of 0.79. Overall, the test-retest reliability of Theory of Planned Behavior measures was moderate to good. The range of the 18 items assessing Attitude Toward the Behavior (direct and indirect measures, Table 3-12) was 0.88 to 0.41 with a moderate average of 0.71. The Attitude Toward the Behavior direct measures proved more reliab le than the indirect measures with correlation averages of 0.80 and 0.66, respectively. The range of the 16 items which assessed the Subjective Norm (direct and indirect measures, Table 313) was 0.85 to 0.56 with a moderate average of 0.69. The Subjective Norm direct measures were mo re reliable than the indirect measures with correlation averages of 0.71 and 0.67, respectively. The range of the 17 items which assessed the Perceived Behavioral Control (direct and indire ct measures, Table 3-14) was .85 to .54 with a moderate average of .68. The Perceived Behavior al Control direct measures were slightly less reliable than the indirect measures with aver ages of .62 and .70, respectiv ely. Finally, the range of the four items measuring Be havioral Intention (Table 3-15) was 0.78 to 0.53 with a moderate average of .70. The item with the lowest value in the range (.53) was, I intend to not drink any alcoholic beverages at the next Gator home football game I attend. Perhaps, if this item had been written more clearly the corr elation would have been higher. For example, this item could have been rewritten as follows: I intend to abst ain from drinking any alco holic beverages at the next Gator home football game I attend. While most of the Game Day Survey items de monstrated adequate reliability, those with lower values may have been compromised from repeated exposure. For example, respondents had time between testing to think about issues in areas that had not been previously considered 70

PAGE 71

and may have changed responses on certain items. The desire of the respondent to present socially desirable responses may al so explain lower reliability coe fficients. In addition, perhaps survey fatigue also compromised the reliability of these results. Nevertheless, the large majority of items yielded moderate to excellent correl ation values, indicating stability over time. Internal Consistency To determine survey item reliability, Cronbach s alpha scores were calculated on each of the 14 Game Day Survey scales, and the results are presented in Table 3-16. As previously noted, correlation values of 0.50 to 0.75 are cons idered moderate to good, and values above 0.75 are classified as good to excellent (Portney & Watkins, 2000). Ov erall, the internal consistency values of the scales were acceptable, rangi ng from moderate to excellent. The non-TPB scales had good reliability scores. The alcohol consumption, consequences, and the social norms reliability coefficients had good to excellent values ranging from 0.861, 0.841, and 0.689, respectively. The five prevention items generated a moderate reliability coefficient value of 0.608. This modera te level may be due to the relative visibility and heterogeneity of the prevention initiatives. For example, the advertisement in which the head football coach encourages students to be responsible is likely much more conspicuous to students than the Code of Conduct Moreover, up to one-fifth of the participants indicated observing a prevention initiative that did not exist. Such incons istencies may have lowered the Cronbachs alpha score. The TPB scales produced a fairly wide range of reliability coefficients. The Attitude Toward the Behavior scale yielde d the highest reliability score with a value of 0.955. However, the indirect measures of this construct produced lower values of 0.559 for Behavioral Beliefs and 0.685 for the Evaluations of Behavioral Outcomes. The Attitude Toward the Behavior scale was 71

PAGE 72

the only TPB scale which did not include a matrix format; instead each of the six questions was asked independently. The Subjective Norm, Norm ative Beliefs, and the Motivation to Comply scales all had excellent relia bility coefficients of 0.908, 0.941, and 0.949, respectively. The Perceived Behavioral Control, Control Beliefs, and Perceived Power yielded moderate to good reliability coefficients of 0.740, 0.668, and 0.793, respec tively. The Behavioral Intentions scale produced a good reliability coefficient of 0.829. Table 3-16 also includes the Cronbachs alpha va lues for each of the survey scales when the item of interest was removed from the analysis. Williams and colleagues (as cited by Komro, Perry, Munson, Stigler, & Farbakhsh, 2004) reco mmend removing an item if the Cronbachs value for a scale improves by 0.02 when the item is deleted. For example, the Cronbachs alpha value for the Beh avioral Beliefs scale on the Ga me Day Survey is 0.559. When item number 21.4, asking about the lik elihood of feeling hungover from drinking excessively on game day, was removed from the survey the Cronbachs alpha for the scale improves to.704. During future administrations of the game day survey strong consideration should be given to removing this item due to the increased Cronbachs alpha value for the scale. In addition, deleting items 21.3 concerning the embarrassment associated with becoming drunk, and 22.3 regarding the perceived importance of having sex with someone on game day, would increase the reliability of the Behavioral Beliefs and Evaluations of th e Behavioral Outcomes scales, respectively. Data Collection The data for this investigation were collected electronically. El ectronic or web-based surveys provide a means for improving survey participation, particularly among persons who have access to the internet (e .g., college students) (Schmidt, 1997; Couper, 2000). In a study where college students were randomly assigned to either a mail or a web-based survey, the 72

PAGE 73

response rates were 40% and 63%, respectively (McCabe, Boyd, Couper, Crawford, & DArcy, 2002). Web-based data collection strategies have proliferated in a wide range of subject areas (Dillman, 2000) including alcohol and drug use (McC abe et al., 2002). The benefits of utilizing web-based surveys include reduced implementation costs, faster data collection, improved formatting, elimination of data entry, and reduced processing costs (Witmer, Colman, & Katzman, 1999, as cited by Kypri & Gallagher, 2003). To further increase the response ra te, Dillman (2000) recommended sending prenotification messages and multiple reminders. One week preceding the implementation of the survey, on November 13, 2006, approximately 2,000 randomly selected UF students received a prenotification message via e-mail. Utilizing a prenotification message is especially important when conducting e-mail surveys because of the ease with which an electronic survey can be deleted, particularly if a person is not aware of or ready to take the survey (Dillman, 2000). Participants received the prenot ification e-mail explaining the pur pose of the research, what the survey entailed, and how the data would be utilize d. Participants were inst ructed to monitor their email for further information regarding the surv ey. The Game Day Survey was administered on November 20, 2006, the Monday after the final home football game. An e-mail was sent to the sample population with the link to the Web-base d survey (Appendix C). This e-mail explained the rationale for the survey, th e deadline for completing the su rvey, participant rights, and benefits of participation. Subj ects who clicked on the hyperlink provided in the e-mail received general instructions on how to complete the surve y. Participants selected response options using a left-click of the mouse. On November 27, December 4, and December 11, 2006 participants received electronic reminders to take the survey. 73

PAGE 74

Survey incentives. Kypri and Gallagher (2003) studied strategies to increase web-based survey participation rates utiliz ing an alcohol-related survey. Th eir initial response rate was only 40%, but through multiple reminders (up to eight contacts) via mail, e-mail, and telephone contact, an 85% response rate was later obtaine d. Of the 150 college st udents eligible to participate in their study, 128 co mpleted the survey. Both pape r and pencil surveys were made available, but the large majority of participants chose to complete the survey electronically (n=123). Focus groups conducted prior to the st udy revealed that, wit hout at least a token incentive, students would not likely complete a surve y. This finding is consis tent with the results obtained by Edwards and colleagues (2002). Based on their findings, Kypri and Gallagher recommended giving each participant a nonconditional token incentive. Since only limited incentive monies existed for this study, it was not feasible to offer all 2,000 students a token incentive. Therefore, an alternate incentive st rategy was developed. While the IRB (Institutional Review Board) at the University of Flor ida prohibits random distribution of incentives, performance based in centives are permitted. Thus, the incentive strategies used for this study were based more on logistics than on best pract ices. At the end of the survey, participants were instructed to send an e-mail to a specified address indicating their interest in receiving the incentive. This e-mail provided the contact inform ation for participants. Because no identifiers were linked to participant survey responses, anonymity was maintained. The first three, the middle three, and the last three students completing the incentive protocol received a $50.00 gift card to the UF Bookstore. A participan t did not have to answer any questions to be eligible for the incentive. In centives were awarded approximately one week following the close of the survey. Students were notified of their selectio n via e-mail and picked 74

PAGE 75

up the gift card in the Department of Health Ed ucation and Behavior in the College of Health and Human Performance at UF. Data Analysis Survey Monkey, a commercial internet survey software program, was used to store the Game Day Survey data. Data were subsequently entered into SPSS (Statistical Package for the Social Sciences) statistical software package, versi on 14.0. All statistical analyses for this investigation were performed a ssuming a Type I error rate of = 0.05. Each question on the survey was coded numerically to facilitate the an alysis of the aggregated data. All TPB items were coded 3 to -3 except the Motivation to Comply items, which were coded 1 to 7 as recommended by Ajzen and Fishbein (1980). Th is procedure is recommended to capture the psychology of a double negative, in which a belief that a behavior will no t result in a negative outcome contributes positively to the individuals attitude (Glanz et al., 2002). Two Behavioral Beliefs, three Evaluation of Behavioral Outcomes, three Control Beliefs, three Perceived Power, three Perceived Behavioral Control items, and one Behavioral Intention item were reverse-coded to ensure that the anticipated participant respon ses would follow a consistent pattern within each of the respective scales. Descriptive statistics such as the mean, medi an, standard deviation, and percentage were calculated to describe the demogr aphic characteristics of the respondents. Descriptive statistics were also used to report the responses to the items which incl uded drinking prevalence on game day, Attitude Toward the Behavior (drinking on game day), Subjective Norms, and Perceived Behavioral Control. Participan t demographics were compared to the overall student population to determine the extent to which the sample wa s representative of the target population. These data were also used to help assess ho w respondents differed fr om nonrespondents. 75

PAGE 76

The following statistical procedures were ut ilized to answer each of the research questions for this study. Research Question 1: What is the prevalence of Extreme Ritu alistic Alcohol Consumption on game day for fall 2006? Frequencies were calculated to determine the prevalence of ERAC utilizing the Game Day Survey item which asked respondents How many drinks do you typically consume on game day? Responses to this item were recoded dichotomously, separating the participants engaging in ERAC on game day from those who do not. This recoded variable is the outcome measure for research questions two through five In addition, ERAC rates were compared by ethnicity, year in school, gender, Greek status, and by legal drinking age (under/over 21) utilizing Chi-square analyses. Research Question 2: How much variance does the combination of constructs in the Theory of Planned Behavior explain when predicting Extreme Ri tualistic Alcohol Consumption on game day? Before conducting any multivariate analysis it was necessary to determine the relationship between the indirect measures, direct measures, be havioral intention, and be havior (Glanz et al., 2002). Zero order correlations were computed am ong each of the TPB constructs and reported in an inter-correlation matrix. To correctly utilize the TPB, Ajzen and Fishbein (1980) recommend that each indirect measure should be multiplied by its corresponding item found in the complimentary indirect scale, and then those products should be summed together to create a mathematical or composite version of the direct measure. For example, the score for the Behavioral Belief item which states, I would have more fun on game day if I got drunk on game day, was multiplied by the 76

PAGE 77

scores of its matching Evaluati on of Behavioral Outcomes item, Having fun on game day is important to me. The remaining five paired item s were also calculated in this manner. Then all six product scores were added together to create a composite of the direct measure, Attitude Toward the Behavior. This procedure was also conducted with the Subjective Norm and Perceived Behavioral Control i ndirect measures as well. A multiple logistic regression analysis was us ed to determine the amount of variance the TPB constructs account for when predicti ng ERAC (dichotomous variable) among college students on game day (Bobko, 2001). Logistic regression represents a specific type of regression analysis in which the outcome variable (categorical or discrete) is binary and the observatory or independent variable(s) can be of any type. The outcome or dependent variable for this analysis is the response to the questi on: How many drinks do you ty pically consume on Game Day? The results were recoded into a dichotomy, with th e participants engaging in ERAC (10 or more drinks for males and eight or more drinks for females) coded as "yes," and those who do not coded as "no." The observed va riables included the direct and indirect measures of the TPB constructs: Subjective Norm, Attit ude Toward the Behavior, Perceived Behavioral Control, and Behavioral Intention (see Game Day Survey in Appendix A). The values for the pseudo R2 statistics (Cox & Snell, and Nagelkerke statistics specific to multiple logistic regression) were calculated to determine the strength of the association between the outcome variable and collective set of observatory vari ables. The Cox & Snell and Na gelkerke statistics represent estimates of R2 and are used to assess overall model fit (Mertler & Vannatta, 2002). 77

PAGE 78

Research Question 3: Which constructs within the Theory of Planned Behavior account for the largest proportion of variance when predicting Extr eme Ritualistic Alcohol Consumption behavior among college students on game day? The odds ratio results from the multiple logistic regression analysis indicated which of the obser vatory variables were most influential in predicting ERAC behavior on game day (Mertler & Vannatta, 2002). Research Question 4: Do the constructs within the Theory of Planned Behavior differ by gender when predicting Extreme Ritualistic Alcohol Cons umption among college students on game day? Initially, in order to explore po ssible gender differences, a chi-s quare cross tabulation analysis was conducted with the ERAC (yes/no) and gend er variables (Portney & Watkins, 2000). A subsequent multiple logistic regression analysis was computed to determine the differences between males and females as it relates to predicting ERAC on game day. Once again the outcome variable for this analys is was the response to the que stion: How many drinks do you typically consume on Game Day? which was recoded into a dichotomy, separating those participants who engaged in ERAC on game day from those who do not. The observatory variables also remained the same as in the previous analysis. However, separate multiple logistic regression analyses were conducte d independently by gender. Research Question 5: Do the constructs within the Theory of Planned Behavior differ by grade classification when predicting Extreme Ri tualistic Alcohol Consumption among college students on game day? A chi-square cross tabulation anal ysis was performed with the ERAC and grade classification variable s to determine the differences between the groups (Portney & 78

PAGE 79

Watkins, 2000). Separate multiple logistic regression analyses were conducted with each of the respective grade classifications to obtain informa tion regarding the predictability of each of the TPB constructs among the academic designations. Research Question 6: What are the causal effects in predicti ng alcohol consumption rates using the constructs from the TPB? Analysis of Moment Structures (AMOS) 6.0 was used to conduct the path analysis to determine the causal links between the variables. This type of modeling technique provides information on whether the pa ttern of inter-correlati ons among the variables is consistent with the TPB as it relates to alcohol consumption on game day. The endogenous variable for this analys is is the response to the question: How ma ny drinks do you typically consume on Game Day? The exogenous variables included the composite measures of the TPB constructs: Subjective Norm, Attit ude Toward the Behavior, Perceived Behavioral Control, and Behavioral Intention. A series of multiple lo gistic regression analyses were conducted to determine the variable relationshi ps specified in the path model. The causal model was depicted in a path diagram, which will be described in Chapter 4. Goodness-of-fit indices were calculated in orde r to determine the extent to which the correlations observed in the data match those produced by the path analysis. This is analogous to determining the degree to which observed values correspond to predicted values in a simple linear regression equation. If the observed distribution of vari ables and the values generated by the statistical model correspond w ith one another, then the model fits the data. However, because of the varying characterist ics of data sets, no consensus exists on the single best measure for determining model fit. Thus, in structural equation modeling a variet y of model fit indices exist. In the present study the following goodness-of-fit indices we re calculated: Chi-square, 79

PAGE 80

Standardized Root Mean Square Residual (SRM R), relative fit index (RFI), Comparative Fit Index (CFI) and Root Mean Square Error of Approximation (RMSEA). Summary This chapter provided a description of the methods associated with this study, including research design, research variables, study population, instrumentation, human subjects permission form, item development, data collecti on, and analyses. A cross-sectional research design utilizing randomly selected college students at the University of Florida was employed for this investigation. Items for the survey were based on a literature review and other standardized instruments. The survey results contain inform ation to determine which of the TPB constructs account(s) for the largest proporti on of variance when predicti ng ERAC on game day. In addition, a path analysis was c onducted to determine the causal effects in predicting alcohol consumption rates on game utilizing the TPB constructs. 80

PAGE 81

Table 3-1. Game Day TPB Attitude Toward Behavior Direct and Indirect Measures Construct Item Anchors Attitude Toward Behavior 1. For me to get drunk on game day is: Good Bad (Direct Measures) 2. For me to get drun k on game day is: Beneficial Harmful 3. For me to get drunk on game day is: Enjoyable Unenjoyable 4. For me to get drunk on game day is: Healthy Unhealthy 5. For me to get drunk on game day is: Favorable Unfavorable. 6. For me to get drunk on game day is: Wise Foolish. Behavioral Beliefs (Indirect Measures) 1. I would have more fun if I got drunk on game day. Extremely Likely Extremely Unlikely 2. I would be more social if I got drunk on game day. 3. My chances of hooking up with someone (having sex) would increase if I got drunk on game day. 4. I would have a hangover if I got drunk on game day. 5. I would enjoy watching the game less if I got drunk on game day. 6. I would embarrass myself if I got drunk on game day. Evaluations of Behavioral Outcomes (Indirect Measures) 1. Having fun on game day is important to me. Strongly Agree Strongly Disagree 2. Being social on game day is important to me. 3. Meeting someone and hooking up (having sex) with them on game day is important to me. 4. Having a hangover on game day is a concern of mine. 5. Watching the game is important to me. 6. Embarrassing myself due to my drinking on game day is a concern of mine. Note: All anchors are on a seven point scale. 81

PAGE 82

Table 3-2.Game Day TPB Subjective Norm Direct and Indirect Measures Construct Item Anchors Subjective Norm (Direct Measures) 1. The people in my life whom I value get drunk on game day. Strongly Agree Strongly Disagree 2. The people in my life whom I value would approve of me getting drunk on game day. 3. Most people I hang out with get drunk on game day. 4. The people in my life whom I value encourage me to get drunk on game day. Normative Beliefs (Indirect Measures) 1. My best friend would approve of me getting drunk on game day. Strongly Agree Strongly Disagree, N/A 2. My close friends would approve of me getting drunk on game day. 3. My mother (legal guardian) would approve of me getting drunk on game day. 4. My father (legal guardian) would approve of me getting drunk on game day. 5. My current partner would approve of me getting drunk on game day. 6. My ideal future partner would approve of me getting drunk on game day. Motivation to Comply (Indirect Measures) 1. When it comes to drinking alcohol, how motivated are you to meet the expectations of your best friend? Very Motivated Not Motivated At All, N/A 2. When it comes to drinking alcohol, how motivated are you to meet the expectations of your close friends? 3. When it comes to drinking alcohol, how motivated are you to meet the expectations of your mother (legal guardian)? 4. When it comes to drinking alcohol, how motivated are you to meet the expectations of your father (legal guardian)? 5. When it comes to drinking alcohol, how motivated are you to meet the expectations of your current partner? 6. When it comes to drinking alcohol, how motivated are you to meet the expectations of your ideal future partner? Note: All anchors are on a seven point scale. 82

PAGE 83

Table 3-3. Game Day TPB Percei ved Behavioral Control Direct and Indirect Measures Construct Item Anchors Perceived Behavioral Control (Direct Measures) 1. I am confident that I can limit my alcohol consumption on game day. Strongly Agree Strongly Disagree 2. I can resist pressure from friends to consume alcohol on game day. 3. It is difficult for me to drink moderately on game day. 4. As I get drunk, I start to lose control over the number of drinks I consume. 5. Its difficult for me to refuse free alcoholic drinks on game day. Control Beliefs (Indirect Measures) 1. Use a designated driver or safe transportation on game day. Always Never 2. Attend pre-game tailgating activities on game day. 3. Are given free alcoholic drinks on game day. 4. Notice the police on game day. 5. Consider the financial costs associated with consuming alcoholic beverages on game day. 6. Feel hungover from drinking alcohol on the day after game day. Perceived Power (Indirect Measures) 1. Having a designated driver or safe transportation influences my decision to get drunk on game day. Extremely LikelyExtremely Unlikely 2. Attending pre-game tailgating opportunities influences my decision to get drunk on game day. 3. People offering me free alcoholic drinks influence my decision to get drunk on game day. 4. The presence of police deters me from getting drunk on game day. 5. The financial costs associated with alcoholic beverages deter me from getting drunk on game day. 6. Having a hangover the day after a game, deters me from getting drunk on game day. Note: All anchors are on a seven point scale. 83

PAGE 84

Table 3-4. TPB Attitude Toward Behavior Items and Reliability Values from the Literature Construct Item Cronbachs Alpha Citation Attitude Toward Behavior (Direct Measures) Drinking alcohol to get drunk is: Good Bad. 0.84 1 Drinking alcohol to get drunk is: Beneficial Harmful. Drinking alcohol to get drunk is: Enjoyable Unenjoyable. Drinking alcohol to get drunk is: Healthy Unhealthy. Drinking alcohol to get drunk is: Favorable Unfavorable. Drinking alcohol to get drunk is: Wise Foolish. Behavioral Beliefs (Indirect Measures) Drinking 3.5 pints 7 shorts in a session would make me feel more confident. 0.64 2 Drinking 3.5 pints 7 shorts in a session would make me feel happy. Drinking 3.5 pints 7 shorts in a session would make me suffer from a hangover later on/the next day. Relaxing/unwinding N/A 3 Having a hangover/feeling sick Having fun/socializing Damaging your health Reducing inhibitions Behaving embarrassingly Evaluations of Behavioral Feeling more confident would be ...good/bad 0.64 2 Outcomes (Indirect Measures) How pleasant or unpleasant are the consequences from binge drinking N/A 3 Relaxing/unwinding Having a hangover/feeling sick Having fun/socializing Damaging your health Reducing inhibitions Behaving embarrassingly (Citations for the scales are list ed as follows: 1 = Schlegel, D Avernas, Zanna, Decourville, & Manske, 1992; 2 = Norman, Bennett, & Lewis, 1998; 3 = Johnston & White, 2004; 4= Wall, Hinson, & McKee, 1998; 5 = OCallagha n, Chant, Callan, & Baglioni, 1997). 84

PAGE 85

Table 3-5. TPB Subjective Norm Items and Re liability Values from the Literature Construct Item Cronbachs Alpha Citation Subjective Norm (Direct Measures) Perceived general societal pressure to engage in excessive alcohol consumption N/A 4 How likely or unlikely is it that most people who are important to you think you should drink alcohol in the next month 0.76 5 Normative Beliefs (Indirect Measures) How likely is it that your spouse thinks it is OK for you to get drunk? 0.84 1 How likely is it that your mother thinks it is OK for you to get drunk? How likely is it that your father thinks it is OK for you to get drunk? How likely is it that your close friends think it is OK for you to get drunk? How likely is it that your close friends at work think it is OK for you to get drunk? Motivation to Comply (Indirect Measures) With regard to my drinking, I want to do what my friends think I should. 0.62 (positive index) 0.76 (negative index) 2 Friends N/A 3 Parents Other family members Partner Work colleagues (Citations for the scales are list ed as follows: 1 = Schlegel, D Avernas, Zanna, Decourville, & Manske, 1992; 2 = Norman, Bennett, & Lewis, 1998; 3 = Johnston & White, 2004; 4= Wall, Hinson, & McKee, 1998; 5 = OCallagha n, Chant, Callan, & Baglioni, 1997). 85

PAGE 86

Table 3-6. TPB Perceived Behavioral Control Items and Reliability Values from the Literature Construct Item Cronbachs Alpha Citation Perceived Behavioral Control Rate the amount of control you have over your drinking behavior 0.81 4 (Direct Measures) Rate the ease you have of controlling your drinking behavior Could you limit your alcohol consumption to the point where you just begin to feel intoxicated? How much control do you have over the whether you do or do not drink alcohol in the next month? 0.76 5 For me to drink alcohol in the next month is... If I wanted to I could easily drink alcohol in the next month Control Beliefs (Indirect Measures) Celebrating an event would make me more/less likely to drink 3.5 pints/7 shorts in a session 0.81 2 Being at a club or party would make me more/less likely to drink 3.5 pints/7 shorts in a session Having to be up early the next day would make me more/less likely to drink 3.5 pints/7 shorts in a session 0.52 2 Lack of money/cost of drinking N/A 3 Transportation difficulties Work/study commitments Taking prescription medication Lack of opportunity Perceived Power Lack of money/cost of drinking N/A 3 (Indirect Measures) Transportation difficulties Work/study commitments Taking prescription medication Lack of opportunity (Citations for the scales are list ed as follows: 1 = Schlegel, D Avernas, Zanna, Decourville, & Manske, 1992; 2 = Norman, Bennett, & Lewis, 1998; 3 = Johnston & White, 2004; 4= Wall, Hinson, & McKee, 1998; 5 = OCallagha n, Chant, Callan, & Baglioni, 1997). 86

PAGE 87

Table 3-7. Game Day Test-R etest Demographic Items Item r Which football games did you attend in 2006? 0.971 What is your sex? 1.000 What is your cl assification? 0.974 How do you describe yourself? 1.000 Are you currently a member of a fraternity? 0.968 How old are you? 0.991 How much do you weigh? 0.999 Note: All items were statis tically significant at 0.01 level, two tailed test Pearson Correlation conducted Table 3-8. Game Day Test-Retest Prevention Items Item r Ad in the Alligator with Coach Meyers picture encouraging Gators to be responsible fans? 0.555 GatorLight tip sheet about drinking in moderation? 0.463 The Gator Fans Code of Conduct? 0.564 GatorHealth Guide? 0.422 The public service announcement Gators Se t the Standard Respect the Swamp? 0.618 The public service announcement Gators Se t the Standard Respect the Swamp concerns which of the following: 0.562 Note: All items were statis tically significant at 0.01 level, two tailed test 87

PAGE 88

Table 3-9. Game Day Test-Retes t Game Day Drinking Items Item r Do you typically drink alcohol on Game Day? 0.748 Where do you spend the majority of your time drinking alcoholic beverages before the game? 0.813 Where do you spend the majority of your time drinking alcoholic beverages during the game? 0.530 Where do you spend the majority of your time drinking alcoholic beverages after the game? 0.814 What is the total number of alcoholic dr inks that you typically consume before, during, and after a Gator home football game? 0.865 How many alcoholic drinks do you typical ly consume during the two hours before a Gator home football game? 0.827 How many alcoholic drinks do you typi cally consume during a Gator home football game? 0.704 How many alcoholic drinks do you typically consume during the two hours after a Gator home football game? 0.667 What is the total number of alcoholic drinks that you consumed before, during, and after the Gator home football game against Western Carolina? 0.855 What is the total number of hours that you t ypically spend drinking on Game Day? 0.830 Not including Game Day, how many alcohol ic drinks did you have the last time you partied/socialized? 0.861 Note: All items were statis tically significant at 0.01 level, two tailed test Pearson Correlation conducted Table 3-10. Game Day Test-Retest Alcohol Consequence Items Item r Had a hangover 0.699 Vomited 0.292 Drove after drinking alcohol 0.717 Drove after having 5 or more drinks 0.554 Had a memory loss (black out) 0.772 Was hurt or injured 0.792 Got into a fight or argument 0.700 Got reprimanded by the police 1.000 Arrested/ticketed by the police 1.000 Took advantage of someone sexually 1.000 Had been taken advantage of sexually 0.491 Note: All items were statis tically significant at 0.01 level, two tailed test 88

PAGE 89

Table 3-11. Game Day Test-R etest Social Norm Items Item r What percentage of your friends do you be lieve typically get drunk on game day? 0.845 What percentage of current UF students do you believe typically get drunk on game day? 0.746 What percentage of Gator Fans (who are not current students) do you believe typically get drunk on game day? 0.780 Note: All items were statis tically significant at 0.01 level, two tailed test Table 3-12. Game Day Test-Retest Attitude Toward the Behavior Indirect and Direct Items Note: All items were statis tically significant at 0.01 level, two tailed test Item r I would have more fun if I got drunk on game day. 0.818 I would be more social if I got drunk on game day. 0.791 My chances of hooking up with someone (hav ing sex) would increase if I got drunk on game day. 0.762 I would have a hangover if I got drunk on game day. 0.495 I would enjoy watching the game less if I got drunk on game day. 0.408 I would embarrass myself if I got drunk on game day. 0.521 Having fun on game day is important to me. 0.711 Being social on game day is important to me. 0.596 Meeting someone and hooking up (having sex) with them on game day is important to me. 0.818 Having a hangover on game day is a concern of mine. 0.650 Watching the game is important to me. 0.842 Embarrassing myself due to my drinking on game day is a concern of mine. 0.518 For me to get drunk on game day is: good/bad 0.877 For me to get drunk on game day is: beneficial/harmful 0.756 For me to get drunk on game day is: enjoyable/unenjoyable 0.793 For me to get drunk on game day is: healthy/unhealthy 0.663 For me to get drunk on game day is: favorable/unfavorable 0.828 For me to get drunk on game day is: wise/foolish 0.886 89

PAGE 90

Table 3-13. Game Day Test-retest Subject ive Norm Indirect and Direct Items Item r My best friend would approve of me getting drunk on game day. 0.828 My close friends would approve of me getting drunk on game day. 0.849 My mother (legal guardian) would appr ove of me getting drunk on game day. 0.776 My father (legal guardian) would appr ove of me getting drunk on game day. 0.676 My current partner would approve of me getting drunk on game day. 0.727 My ideal future partner would approv e of me getting drunk on game day. 0.729 When it comes to drinking alcohol, how mo tivated are you to meet the expectations of your best friend? 0.578 When it comes to drinking alcohol, how mo tivated are you to meet the expectations of your close friends? 0.619 When it comes to drinking alcohol, how mo tivated are you to meet the expectations of your mother (l egal guardian)? 0.573 When it comes to drinking alcohol, how mo tivated are you to meet the expectations of your father (legal guardian)? 0.589 When it comes to drinking alcohol, how mo tivated are you to meet the expectations of your current partner? 0.612 When it comes to drinking alcohol, how mo tivated are you to meet the expectations of your ideal future partner? 0.559 The people in my life whom I value get drunk on game day. 0.590 The people in my life whom I value woul d approve of me getting drunk on game day. 0.816 Most people I hang out with get drunk on game day. 0.783 The people in my life whom I value encourage me to get drunk on game day. 0.685 Note: All items were statis tically significant at 0.01 level, two tailed test 90

PAGE 91

Table 3-14. Game Day Test-Retest Perceived Behavioral Contro l Indirect and Direct Items Item r Use a designated driver or safe transportation on game day. 0.634 Attend pregame tailgating ac tivities on game day. 0.739 Are given free alcoholic drinks on game day. 0.758 Notice the police on game day. 0.613 Consider the financial costs associated with consuming alcoholic beverages on game day. 0.630 Feel hung over from drinking alcohol on the day after game day. 0.688 Having a designated driver or safe trans portation influences my decision to get drunk on game day. 0.582 Attending pre-game tailgati ng opportunities influences my decision to get drunk on game day. 0.821 People offering me free alcoholic drinks in fluence my decision to get drunk on game day. 0.826 The presence of police deters me from getting drunk on game day. 0.695 The financial costs associated with alc oholic beverages deter me from getting drunk on game day. 0.762 Having a hangover the day after a game, deters me from getting drunk on game day. 0.673 I am confident that I can limit my alcohol consumption on game day. 0.612 I can resist pressure from friends to consume alcohol on game day. 0.566 Its difficult for me to dri nk moderately on game day. 0.689 As I get drunk, I start to lose control over the number of drinks I consume. 0.539 Its difficult for me to refuse fr ee alcoholic drinks on game day. 0.706 Note: All items were statis tically significant at 0.01 level, two tailed test Table 3-15. Game Day Test-retes t Behavioral Intention Items Item r I intend to get drunk at the next Gator home football game I attend: 0.769 I intend to drink in moderation at the next Gator home football game I attend: 0.775 I intend to not drink any alcoholic beverages at the next Gator home football game I attend: 0.531 I intend to get drunk at every Gator home football game I attend: 0.727 Note: All items were statis tically significant at 0.01 level, two tailed test 91

PAGE 92

Table 3-16. Game Day Survey Scale Reliability Scale variable Type of variable Number of response options Cronbachs Alpha if Item Deleted Reliability coefficient Prevention Items Continuous 2 0.608 Coach Meyers ad Q8.1 0.574 GatorLite tips (foil) Q8.2 0.570 Code of conduct Q8.3 0.524 Health guide (foil) Q8.4 0.521 PA Q8.5 0.576 Alcohol Consumption Continuous Openended 0.861 Total # Q12 0.804 Before # Q13 0.817 During # Q14 0.875 After # Q15 0.829 UWC Q16 0.854 Last party Q18 0.826 Consequences Continuous 5 0.841 Hangover Q19.1 0.851 Vomited Q19.2 0.820 Drove after drink Q19.3 0.830 Drove after 5 Q19.4 0.823 Blackout Q19.5 0.815 Hurt Q19.6 0.815 Fight Q19.7 0.822 Reprimanded Q19.8 0.830 Ticketed Q19.9 0.830 Took advantage Q19.10 0.832 Taken advantage Q19.11 0.834 Social Norms Continuous 3 0.689 Your friends Q20.1 0.716 Current UF Q20.2 0.475 Gator fans Q20.3 0.484 Attitude Toward Behavior Continuous 7 0.955 Good-Bad Q23 0.941 Beneficial-Harmful Q24 0.943 Enjoyable-Unenjoyable Q25 0.942 Healthy-Unhealthy Q26 0.965 Favorable-Unfavorable Q27 0.938 Wise foolish Q28 0.944 Behavioral Beliefs Continuous 7 0.559 Fun Q21.1 0.310 Social Q21.2 0.337 Sex Q21.3 0.466 92

PAGE 93

Table 3-16. Continued Scale variable Type of variable Number of response options Cronbachs Alpha if Item Deleted Reliability coefficient Hangover Q21.4 0.704 Enjoy Q21.5 0.534 Embarrass Q21.6 0.583 Evaluations of Behavioral Outcomes Continuous 7 0.685 Fun important Q22.1 0.615 Social important Q22.2 0.611 Sex important Q22.3 0.703 Hangover important Q22.4 0.645 Watching important Q22.5 0.636 Embarrassing important Q22.6 0.648 Subjective Norm Continuous 7 0.908 Value getting drunk Q31 0.836 Approve of getting drunk Q32 0.874 Hang out with drunk people Q33 0.897 Encourage getting drunk Q34 0.883 Normative Beliefs Continuous 8 0.941 Best friend approve Q29.1 0.929 Close friends approve Q29.2 0.927 Mother approve Q29.3 0.940 Father approve Q29.4 0.938 Current partner approve Q29.5 0.924 Future partner approve Q29.6 0.920 Motivation to Comply Continuous 8 0.949 Friend Q30.1 0.940 Close friends Q30.2 0.942 Mother Q30.3 0.940 Father Q30.4 0.941 Current Partner Q30.5 0.936 Future Partner Q30.6 0.936 Perceived Behavioral Control Continuous 7 0.740 Limit alcohol Q34.1 0.728 Peer pressure Q34.2 0.717 Drink moderately Q34.3 0.660 Loss of control Q34.4 0.666 Free drinks Q34.5 0.684 Control Beliefs Continuous 7 0.668 Use designated driver Q32.1 0.626 Attend tailgating Q32.2 0.563 Given free alcoholic drinks Q32.3 0.580 Notice police Q32.4 0.664 93

PAGE 94

Table 3-16. Continued Scale variable Type of variable Number of response options Cronbachs Alpha if Item Deleted Reliability coefficient Consider financial costs Q32.5 0.655 Feel hungover after Q32.6 0.646 Perceived Power Continuous 7 0.793 Designated driver influences Q33.1 0.750 Tailgating influences Q33.2 0.754 Free drinks influence Q33.3 0.754 Police influence Q33.4 0.768 Cost influence Q33.5 0.761 Hangover influence Q33.6 0.778 Behavioral Intentions Continuous 8 0.829 Drunk next game Q35.1 0.726 Moderation next game Q35.2 0.814 Abstain next game Q35.3 0.805 Drunk at every game Q35.4 0.783 94

PAGE 95

CHAPTER 4 RESULTS Self-reported rates of alcohol consumption on college football game day were examined in this study. More specifically, the Theory of Planned Behavior (TPB) was utilized in a survey to predict Extreme Ritualistic Alcohol Consumption (ERAC) among a random sample of college students at the University of Florida. ERAC, a concept deve loped for this study, identifies drinking practices which occur on game day. ERAC was defined as consuming 10 or more drinks on game day for a male, and eight or mo re drinks for a female. The ERAC measure provides more specificity than the standard 5+4+ drinking measure in identifying at-risk drinkers. The results of this investig ation are organized to addre ss six research questions: What is the prevalence of Extreme Ritualistic Alcohol C onsumption on a typical game day for fall 2006? How much variance does the combination of constructs in the Theory of Planned Behavior explain when predicting Extreme Ritualistic Alcohol Consumption on game day? Which constructs within the Theory of Planned Behavior (Subjective Norm, Attitude Toward the Behavior, Perceived Behavioral Control, and Behavioral Intention) account for the largest proportion of variance when predicting Extreme Ritualistic Alcohol Consumption behavior among colle ge students on game day? Do the constructs within the Theory of Planned Behavior differ by gender when predicting Extreme Ritualistic Alcohol Consumption among college students on game day? Do the constructs within the Theory of Pl anned Behavior differ by grade classification when predicting Extreme Ritualistic Alc ohol Consumption among college students on game day? What are the causal effects in predicting alc ohol consumption rates using the constructs from the TPB? 95

PAGE 96

Participant Characteristics The target population for this study was college students, ages 18 to 24 years, attending the University of Florida. Of the 2,096 participan ts sent a survey, 15 e-mails were returned as "undeliverable," and one participant was removed fr om the analysis due to incomplete data. A total of 740 students out of the 2,080 eligible participants completed the Game Day Survey yielding a 36% response rate. Table 14-1 contains the demographic data fo r the participants, as well as the overall student population. Of the respondents the ma jority were Caucasian (73.6%) followed by, Hispanic or Latino (11.6%), Asian or Pacific Islander (6.9%), Africa n-American (6.2%), American Indian/Alaskan Native (0.3%), and othe r ethnic groups (1.4%). Students were broadly distributed by grade classification (20.4% freshmen, 18.9% sophomore, 10.2%, juniors, 27.2% seniors, and 23.2% graduate/professional students), and except for freshmen and graduate/professional students, th ese percentages also correspond to those for the UF student body as a whole. Females composed 60.8% of the sample, a somewhat higher value than the percentage of females on campus. Approximately one out every five participants (19.4%) was a member of a fraternity or sorori ty (referred to as part of the campus Greek system). The mean age of the sample was 20.30 years ( SD = 1.66), with ages ranging from 18-24 years. The sample demographics corresponded to the overall stud ent demographic population with the exception of a slight overrepresentation of females and Caucasians. In addition to the demographic background of the participants, the descriptive statistics provided fundamental information concerning the game day habits of college students. For example, approximately four out of five (80.1%) participants attended at least one home game during the 2006 college football season. Approximately half (50.8%) of all survey respondents 96

PAGE 97

(58.1% of males and 46% of female s) indicated that they typical ly drank on game day, but many drank in moderation with the mean number of drinks consumed equaling 3.9 ( SD = 5.06). Finally, while participants on aver age spent two and a half (2.55; SD =3.25) hours drinking on game day; one in five (21.4%) participants reported drinking for five or more hours. Research Questions What is the Prevalence of Extreme Ritualis tic Alcohol Consumption on a Typical Game Day for Fall 2006? A frequency analysis was performed on the Game Day Survey item How many drinks do you typically consume on game day? Responses to this item were recoded to categorize those participants who engaged in Extreme Ritualistic Alcohol Consumption (ERAC) on game day from those who do not. Male participants who consumed 10 or more drinks on game day and females who consumed eight or more dr inks were coded into the ERAC group, and participants who did not meet these criteria we re coded into the non-ER AC group. As Table 4-2 indicates 15.7% (116) of the sample engaged in ERAC on game day. When analyzed by gender, males engaged in ERAC (23.4%) at more than twice the rate of females (10.7%). The Chisquare analysis revealed that males were appr oximately 2.5 times more likely to engage in ERAC on game day than females (OR = 2.565, p < 0.001). ERAC rates were also analyzed by grade classification. The Chi-square results ( p < .001), found in Table 4-3, illustrate that stud ents enrolled in the more advanced grades (upperclassmen) engaged in ERAC at higher rates than their counterparts (underclassmen). The ERAC rate for freshmen on game day was 6.0% for sophomores 10.7%, for juniors 15.7%, and for seniors 24.9%. In addition, 18.7% of the gra duate/professional students engaged in ERAC, a rate, second only to th at of seniors. 97

PAGE 98

Table 4-4 shows the Chi-square results ( p < 0.001) for ERAC subdivided by ethnicity. Caucasians drank at disproportiona tely high rates compared to the other ethnicities. Nearly onefifth of Caucasians (18.0%) e ngaged in ERAC on game day followed by Hispanics/Latinos (15.1%). Blacks/African Americans and Asia n/Pacific Islanders engaged in ERAC at substantially lower rate s, 2.2% and 2.0%, respectively. The Other category included American Indians/Alaskan Natives and other participants who did not classify themselves into any of the aforementioned ethnic groups. The ERAC rate for this group was 25%. Note the Other category includes only 1.7% (n=12) of the total sample, and generalizing the findings associated with this particular grou p is not recommended. The ERAC rates were also analyzed by Greek status and legal drinki ng age. As Table 4-5 illustrates, nearly a quar ter (24.5%) of Greeks e ngaged in ERAC on game day compared to 13.6% of non-Greeks. The Chi-square analysis re vealed that Greeks were approximately twice as likely, to engage in ERAC on game day than non-Greeks (OR = 2.060, p < .001). In addition, 23.5% of the participants, age 21 or over, indicat ed that they engage in ERAC on game day, compared to 9.7% of students under the legal dr inking age. The results indicated that the participants of legal dr inking age were nearly three times more likely to engage in ERAC on game day than underage participants (OR = 2.865, p < .001). See Table 4-6 for the distribution rates of ERAC by legal drinking age. The more conventional 5+4+ drinking measure was also assessed in this study. Table 4-7 summarizes the ERAC rates and 5+4+ drinking rates on game day by demographic status. Overall, the 5+4+ drinking rates on game day were at least twice as high as the ERAC rates (36.2% 5+4+ compared to 15.7% ERAC). Males took part in this beha vior (43.1%) at higher rates than females (31.8%). More than ha lf of the seniors and two-fifths of the 98

PAGE 99

graduate/professional students in this sample reported 5+4+ drinking on game day. Caucasians engaged in 5+4+ drinking (40.6%) at substantially higher rates than the other ethnicities. Over two-fifths Greeks (46.9%) engaged in 5+4+ drinki ng; whereas, approximately one-third (33.7%) of non-Greeks participated in this behavior on game day. Fi nally, 52.5% of those participants 21 years or older and 23.7% of those under the le gal drinking age, engaged in 5+4+ drinking on game day. How Much Variance Does the Combination of Constructs in the Theory of Planned Behavior Explain When Predicting Extreme Ritualistic Alcohol Consumption on Game Day? Table 4-8 illustrates the zero order corre lations computed among each of the TPB constructs. The inter-correlation matrix reveal s a statistically significant relationship between each of the TPB direct measures and Behavioral Intention. Th e bivariate relationship between Behavioral Intention and Attitude Toward the Behavior was 0.597 which indicates a moderate relationship at the = 0.01 significance level. The rela tionship between Behavioral Intention and Subjective Norm was slightly stronger at 0.652 and was also statistica lly significant at the = 0.01 significance level. The relationship between Behavior al Intention and Perceived Behavioral Control was weak at 0.107 but was statistically significant at the = 0.05 significance level. Overall the correlation values indicate statistically significant relationships between the variables, but are not highly correlated. The TPB variables are independent from one another, thus lim iting concerns of multicolinearity. Once the relationship between the direct measures and behavioral intention was established, a multiple logistic regression analysis was conducted to determine the amount of variance in predicting ERAC attributable to the TPB constructs. The results from the logistic regression (n=318) reveal the Cox and Snell pseudo R2 value was 0.39 and the Nagelkerke 99

PAGE 100

pseudo R2 value was 0.61 and both were statistically significant at the 0.05 level. Overall, the model correctly classified 88.7% of the cases. As shown in Table 4-9, the predictor set had a significant effect on ERAC (Model Chi-Square = 156.08, df = 4, p < 0.001). Depending on the Cox and Snell or Nagelkerke R2 estimates, the TPB construc ts accounted for 39% or 61%, respectively, of the Extreme Ritu alistic Alcohol Consumption that occurs on a typical game day. In addition to determining th e prevalence of alcohol use on game day, this investigation sought to determine the amount of variance explained by the combin ation of constructs in the Theory of Planned Behavior. The two pseudo-R2 statistics (Cox & Snell and Nagelkerke statistics specific to multiple logistic regres sion) revealed that, as a whole, the TPB model accounted for 39% and 61%, respectively, of the variance in explaining ERAC on game day. The discrepancy between the two R2 estimates result from the methods of calculation. The simpler Cox & Snell formula is based on the estimated model likeli hood (L) value. More specifically, the L for an intercep t-only model is divided by the L for the full model, and the n-th root of the resulting quotient is then subtra cted from one. The more complex Nagelkerke statistic takes the Cox and Snell result and divides it by a Cox and Snell value having a full model likelihood of one. The Nagelkerke statistic takes into account that a perfectly fitting full model will not have a correct Cox and Snell va lue of 1.0. Conversely, a full model that is equivalent to an intercept-only model will not ha ve a correct Nagelkerke value of 0.0 (Hair, et al., 2006). Which Constructs within the Theory of Planned Behavior Account for the Largest Proportion of Variance when Predicting Extreme Ritualistic Alcohol Consumption Behavior among College Students on Game Day? The odds ratio results from the multiple logist ic regression analysis shown in Table 4-9, indicated Behavioral Intention (OR = 1.40, p < 0.001) was the strongest predictor of ERAC on 100

PAGE 101

game day. A one unit change in this variables score was associated with a 40% increase in the odds of engaging in ERAC. The odds ratio results from the Attitude Toward the Behavior variable (OR = 1.04, p = 0.025) and the Subjective Norm variable (OR = 1.02, p = 0.010) were much more modest. Perceived Behavioral Cont rol also predicted ERAC at a statistically significant level (OR = .97, p= 0.028). However, a one unit change in this variables score was associated with a 3% decrease in the odds of e ngaging in ERAC. In summary, while each of the TPB constructs was statistically significant, Be havioral Intention was the only construct which yielded a substantial od ds ratio value. Do the Constructs within the Theory of Planned Behavior Differ by Gender When Predicting Extreme Ritualistic Alcohol Co nsumption among College Students on Game Day? As shown previously in Tabl e 4-2, the Chi-square analysis conducted with the variables ERAC and gender revealed that males engage in th is behavior at higher rates than females. Separate multiple logistic regression analyses were conducted with females and males (Tables 4-10 and 4-11, respectively) to determine if the motivations for engaging in ERAC, based on the TPB, differ by gender. The multiple logistic regression results indicate that Behavioral Intention and Perceived Behavioral Control were the onl y two statistically significant TPB variables which predicted ERAC among females (Chi-Square = 67.92, df = 4, p < 0.001). The model correctly classified 87.2% of the cases. The Cox and Snell pseudo R2 value was 0.29 and the Nagelkerke pseudo R2 value was 0.55, and both were statistically significant at p < 0.001. The odds ratio results for Behavioral Intention (OR = 1.42, p < 0.001) revealed that it was th e strongest pred ictor of ERAC on game day among females. A one unit change in this variables score was associated with a 42% increase in the odds of engaging in ERAC. The odds ratio for Perceived Behavioral 101

PAGE 102

Control predicted ERAC at a statistically significant level (OR = 0.95, p = 0.031). A one unit change in this variables score was associated with a 5% decrease in the odds of engaging in ERAC. Thus, the more perceived control females re port, the less likely they were to engage in ERAC. Table 4-11 illustrates the multiple logistic regression results conducted with males. Attitude Toward the Behavior and Behavioral Intention accounted for 48% to 68% of the Extreme Ritualistic Alcohol Cons umption, respectively. The model correctly classified 86.2% of the cases. Behavioral Intention and Attitude Toward the Behavior represented the only two statistically significant TPB variables which predicted ERAC among males (Chi-Square = 80.51, df = 4, p < 0.001). The odds ratio results for Behavioral Intention reveal ed that it was the strongest predictor of ERAC (OR = 1.42, p < 0.001). A one unit change in the score of this variable was associated with a 42% increase in the odds of enga ging in ERAC. The odds ratio for Attitude Toward the Behavior also predicted ERAC at a statistically significant level (OR = 1.06, p < 0.031). A one unit change in the score of this variable was associated with a 6% increase in the odds of engaging in ERAC. Ther efore, the more positive male attitudes were toward alcohol, the more likely they were to engage in ERAC. In summary, Behavioral Inte ntion was the stronge st predictor of ERAC among males and females, while the Subjective Norm construct was not predictive of ERAC with either gender. Perceived Behavioral Control wa s a moderate predictor of ERAC among females, but not with males. Attitude Toward the Behavior was a mo derate predictor of ERAC among males but not with females. Thus, male ERAC rates on ga me day are explained in part by their positive expectancies associated with alcohol use on ga me day, whereas female ERAC rates decline as their perceived control increases. 102

PAGE 103

Do the Constructs within the Theory of Planned Behavior Differ by Grade Classification When Predicting Extreme Ritualistic Alcoho l Consumption among College Students on Game Day? As previously shown in Table 4-3, the Chisquare analysis conducted on the variables ERAC and grade classification revealed that th ere are statistically si gnificant ERAC patterns among the grade classifications, with more advanced grades (juniors, seniors, and graduates/professional students) tending to engage in ERAC at higher rates than the less advanced grade classifications (freshmen and sophomores). Separate multiple logistic regression analyses were conducted with each of the five grade classifications independently to determine their respective influence on predicting ERAC. The results revealed no statistically significan t findings among freshmen or sophomores. However, with juniors (Table 4-12 ) both the Subjective Norm (OR = 1.04, p = 0.025) and Behavioral Intention (OR = 1.41, p = 0.025) odds ratios were statis tically significant. The model correctly classified 79.5% of the cases (Chi-Square = 45.61, df = 4, p < 0.001). The Cox and Snell pseudo R2 value was 0.44 and Nagelkerke pseudo R2 value was 0.69, and both were statistically significant at the 0.05 level. The multiple logistic regression results conduct ed with seniors, shown in Table 4-13, indicated the overall model of one predictor (Behavioral Intention) was statistically reliable in distinguishing between those who engage in ERAC from those who do not (Chi-Square = 44.75, df = 4, p < .001). The model correctly classified 70.8% of the cases. Regression coefficients are presented in Table 4-13. The odds ratio re sults for Behavioral Intention (OR = 1.31, p = 0.018) revealed that it was the only statistically sign ificant TPB predictor of ERAC on game day among seniors. A one unit change in the score of this va riable was associated with a 31% increase in the 103

PAGE 104

odds of engaging in ERAC. The Cox and Snell pseudo R2 value was 0.37 and the Nagelkerke pseudo R2 value was 0.53, and both were statistically significant at the 0.05 level. Finally, Table 4-14 illustrates the multiple lo gistic regression results conducted with graduate/professional students, which show that the TPB accounted for 49% to 73% of ERAC that occurs on a typical game day. The model correctly classified 89.5% of the cases. Behavioral Intention was the only construct statistically reliable in distinguishing between those who engage in ERAC from thos e who do not (Chi-Square = 25.33, df = 4, p < 0.001). The odds ratio results for Behavi oral Intention (OR = 2.23, p = 0.021) revealed it was a strong predictor of ERAC on game day among seniors. A one unit change in the score of this variable was associated with nearly a two-and-a-half-fold incr ease in the odds of engaging in ERAC on game day. In summary, Behavioral In tention was a strong predicto r of ERAC among juniors, seniors, and graduate/professional students. None of the other TP B constructs were statistically significant in predicting ERAC am ong the various grade classifications, except for Subjective Norm, which was a moderate predictor of ERAC among juniors. There were no statistically significant predictors among the freshmen and sophomore classifications. While the logistic regression results from the grade classification differences were minimal from a pragmatic perspective, the findi ngs from the Chi-square analysis on ERAC and legal drinking age, shown in Table 4-16 warranted further inquiry. Indeed, the odds ratio for the legal drinking age was 2.865 (1.892 4.340); ( p < 0.001) which indicates that participants 21 and over were nearly three times more likely to engage in ERAC on game day than those under the legal drinking age. Thus, an additional logistic regression analysis was conducted to determine if the TPB constructs differ by legal drinking age st atus in predicting ERAC on game day. 104

PAGE 105

The multiple logistic regression results conducted with underage participants (Table 4-15) indicated the overall model of two predictors (Subjective No rm and Behavioral Intention) was statistically reliable in di stinguishing between those who e ngage in ERAC from those who do not (Chi-Square = 71.99, df = 4, p < 0.001). The model correctly classified 93.7% of the cases. The odds ratio for Subjective Norm predic ted ERAC at a statistically significant level (OR = 1.03, p = 0.018). A one unit change in the score of this variable was associated with a 3% increase in the odds of engaging in ERAC. The o dds ratio results for Behavioral Intention (OR = 1.37, p = 0.006) revealed that it was the strongest predictor of ERAC on game day among underage participants. A one unit change in the sc ore of this variable was associated with a 37% increase in the odds of engaging in ERAC. The Cox and Snell pseudo R2 value was 0.37 and the Nagelkerke pseudo R2 value was 0.70, and both were statistically significant at the 0.05 level. Table 4-16 illustrates the multip le logistic regression results conducted with participants of legal drinking age, which show that the TP B constructs accounted for 37% to 54% of the ERAC that occurs on a typical game day. The model correctly classified 83.6% of the cases. Behavioral Intention was the only construct statistically reliable in distinguishing between those who engage in ERAC from thos e who do not (Chi-Square = 74.17, df = 4, p < 0.001). The odds ratio results for Behavi oral Intention (OR = 1.41, p < 0.001) revealed that it was a strong predictor of ERAC on game day among participants of legal drinking age. A one unit change in the score of this variable was associated with a 41% increase in the odds of engaging in ERAC on game day. In summary, the TPB proved to be slightly more effective in predicting ERAC among underage participants. For underage participants Subjective Norm and Behavioral Intention were predictive of ERAC, whereas for participants of legal drinking age only Be havioral Intention was 105

PAGE 106

predictive of ERAC. As with pr evious logistic regression analys es Behavioral Intention was the strongest predictor of ERAC. What Are the Causal Effects in Predicti ng Alcohol Consumption Rates Using the Constructs from the TPB? Table 4-17 illustrates that, ove rall, the goodness-of-fit measures were acceptable. The Chisquare test of model fit wa s statistically significant, 2 (2) = 18.352, p < 0.001. The Standardized Root Mean Square Residual (SRMR) of 0.036, the relative fit index (RFI) of 0.844, and the Comparative Fit Index (CFI) of 0.972 all fall with in the acceptable range s for their respective indices (Hair, Black, Babin, Ande rson, & Tatham, 2006). Since the Root Mean Square Error of Approximation (RMSEA) is a goodness-of-fit index with a known statistical distribution, the calculated value yields a p-value and a confiden ce interval for the population RMSEA. Hair and colleagues (2006) recommend an RMSEA less than .08 for studies with fewer than 13 observed variables and fewer than 250 observations. This analysis yielded an RMSEA of 0.161, p < 0.001 with a 90% confidence interval of (0.099, 0.223) sugge sting an inadequate m odel fit. While the high RMSEA value warrants caution when interp reting the results from the path analysis, goodness-of-fit measures may vary from acceptable to unacceptable depending on the index used (Hair, et al., 2006). Overall, th e goodness-of-fit measures indicate that the TPB model represents an acceptable model for explaining alcohol consumption on game day. Figure 4-1 illustrates the sta ndardized causal links between the TPB variables when used to predict alcohol consumption among college students on game day. In the path analysis, correlations between variables ar e represented by double headed a rrows, and a straight line with a single arrowhead denotes a direct causal e ffect (Mertler & Vannatta 2002). There was a modest correlation between A ttitude Toward the Behavior and Subjective Norm (r = 0.51, p< 0.001) (Portney & Watkins, 2000), but there were no statistically significa nt correlations found 106

PAGE 107

between PBC and the other TPB constructs. Furthe r, intentions (Behavioral Intention) to drink alcohol on game day pred icted actual behavior ( R2 = 0.52) with a standardized path coefficient of 0.72. Intentions, in turn, were predicted by Atti tude Toward the Behavior and Subjective Norm constructs ( R2 = 0.54), with standardized path coefficients of 0.39 and 0.46, respectively. In general, positive expectancies concerning alc ohol use (Attitude Towa rd the Behavior) and perceived acceptance of drinking (Subjective Norms) predicted inte ntions (Behavioral Intention) to get drunk on game day. The path analysis i ndicated that Perceived Behavioral Control (PBC) did not elicit any statistically si gnificant path coefficients between Behavioral Intentions or the behavior (number of drinks consumed on game day). Table 4-18 provides information on the direct an d indirect effects on Behavioral Intention and behavior (number of drinks consumed on game day). The indirect effects on the number of drinks consumed for the constructs Attitude Toward the Behavior and Subjective Norm were 0.278 and 0.330 p< 0.001, respectively. Behavioral Intentio n served as the strongest predictor for the number of drinks consumed on game day, and to some extent it mediated the impact of the other constructs. Nevertheless, Attitude Toward the Behavior and Subjective Norm variables contributed significantly to the explanatory capacity of the TPB model, while the Perceived Behavioral Control construct failed to elicit any statistically significan t direct or indi rect effects. Figure 4-2 reveals that the path coefficients do not change when the PBC construct is removed from the TPB, nor do th e goodness-of-fit measures vary substantially (Table 4-19). The original Theory of Reas oned Action (TRA) does not include the PBC construct; thus its removal has precedent. A Chi-square differen ce test was conducted between the TPB and the TRA path analyses to determine the extent to wh ich these two models differ. The results were not statistically significant, 2(5) = .906, p = .97, indicating the more complex model (TPB with 107

PAGE 108

PBC construct) does not enhance model fit. The simpler TRA model sufficiently reproduces the observed correlations, consequently the TRA constitutes a more parsimonious model when explaining alcohol consumption on game day. Summary This chapter contains the findings for the re search questions rais ed in this inquiry, including the results from the descriptive stat istics, multiple logistic regression and path analyses. Overall, the sample population matched the UF population with the slight overrepresentation of Caucasian and female studen ts. The descriptive fi ndings revealed that students who are Caucasian, male, Greek, and over the age of 21 engage in ERAC at disproportionately high rates. The results from the logistic regression an alyses indicated that each of the TPB constructs were statistically significant in predicting ERAC, although the odds ratio values were modest. Overa ll, the Behavioral Intention func tioned as the strongest predictor of ERAC. Males appeared to be more influenced by the Attitude Toward the Behavior construct, whereas the Perceived Behavioral Control c onstruct was more robust among females. The Subjective Norm construct was pr edictive with underage partic ipants but not among those of legal drinking age. Finally, the findings from the path analysis demonstrated that, with the exception of the Perceived Behavioral Control (P BC) construct each of the TPB variables in the model were statistically significant. A Chi-s quare difference test re vealed that the TRA represents a more appropriate model than th e TPB for explaining alcohol use on game day. 108

PAGE 109

Table 4-1. Participant demographics co mpared to UF student population fall 2006 Characteristic UF # UF % Game day # Game day % Gender Male 24,322 47.2 290 39.2 Female 27,195 52.8 450 60.8 Grade Freshman 5,636 10.9 151 20.4 Sophomore 7,899 15.3 140 18.9 Junior 10,012 19.4 172 23.2 Senior 11,152 21.6 201 27.2 Graduate/professional/ 10,828 21.0 76 10.2 Ethnicity Caucasian 33,315 64.7 544 73.6 Hispanic 5,749 11.2 86 11.6 Black/African American 4,092 7.9 46 6.2 Asian/Pacific Islander 3,592 7.0 51 6.9 Amer./Alaskan Native 171 0.3 2 0.3 Others 3,213 6.2 10 1.4 Greek Status Member of a fraternity or sorority 5,240 14.0 143 19.4 *Note : Only persons age 18-24 were included in the Game Day Survey sample. Table 4-2. Chi-square analysis of ERAC rates by gender Gender ERAC # Total # ERAC % Male 68 290 23.4 Female 48 450 10.7 Total 116 740 15.7 OR = 2.565, p < 0.001, 95% C.I.=1.712, 3.843 Table 4-3. Chi-square analysis of ERAC rates by classification Classification ERAC # Total # ERAC % Freshmen 9 151 6.0 Sophomores 15 140 10.7 Juniors 27 172 15.7 Seniors 50 201 24.9 Graduates/Professionals 14 75 18.7 Total 115 739 15.6 p < 0.001 109

PAGE 110

Table 4-4. Chi-square analysis of ERAC rates by ethnicity Ethnicity ERAC # Total # ERAC % Asian/Pacific Islander 1 51 2.0 Black/African American 1 46 2.2 Hispanic or Latino 13 86 15.1 Caucasian 98 544 18.0 Other 3 12 25.0 Total 116 739 15.7 p < 0.001 Table 4-5. Chi-square analysis of ERAC rates by Greek status Greek status ERAC # Total # ERAC % Non-Greek 81 597 13.6 Greek 35 143 24.5 Total 116 740 15.7 OR = 2.060, p < 0.001, 95% C.I.= 1.317, 3.223 Table 4-6. Chi-square analysis of ERAC rates by legal drinking age Legal drinking age ERAC # Total # ERAC % Under 21 40 414 9.7 21 and over 76 324 23.5 Total 116 738 15.7 OR = 2.865, p < 0.001, 95% C.I.= 1.892, 4.34 110

PAGE 111

Table 4-7. 5+4+ drinking and ER AC rates by demographic 5+4+ ERAC Demographic n Percentage n Percentage Gender Male 125 43.1% 68 23.4% Female 143 31.8% 48 10.7% Grade classification Freshmen 21 13.9% 9 6.0% Sophomores 42 30.0% 15 10.7% Juniors 62 36.0% 27 15.7% Seniors 107 53.2% 50 24.9% Graduates/professionals 35 46.7% 14 18.7% Ethnicity Asian/Pacific Islander 6 11.8% 1 2.0% Black/African American 5 10.9% 1 2.2% Hispanic or Latino 32 37.2% 13 15.1% Caucasian 221 40.6% 98 18.0% Other 4 33.3% 3 25.0% Greek Status Non-Greek 201 33.7% 81 13.6% Greek 67 46.9% 35 24.5% Legal drinking age Under 21 98 23.7% 40 9.7% 21 and over 170 52.5% 76 23.5% Total 268 36.3% 116 15.7% Table 4-8. Correlation analysis of direct TPB composite measures and Behavioral Intention TPB ATB SN PBC BI ATB 1.00 SN 0.509** 1.00 PBC 0.105** 0.027 1.00 BI 0.597** 0.652** 0.107* 1.00 Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). Note : ATB = Attitude Toward the Behavior, SN = Subjective Norm, PBC = Perceived Behavioral Control, and BI = Behavioral Intention. Table 4-9. Logistic regression anal ysis TPB composite measuresERAC p Variables in the model OR (95% CI) Wald Attitude Toward the Behavior 1.04 (1.01 1.08) 5.01 0.025 Subjective Norm 1.02 (1.00 1.03) 6.57 0.010 Perceived Behavioral Control 0.97 (0.94 1.00) 4.85 0.028 Behavioral Intention 1.40 (1.23 1.60) 24.35 <0.001 Model Chi-Square = 156.08, df = 4, p < 0.001. Total N = 318 Cox and Snell pseudo R2 = 0.39 Nagelkerke pseudo R2 = 0.61 111

PAGE 112

Table 4-10. Logistic regression analysis TPB composite measuresERAC women p Variables in the model OR (95% CI) Wald Attitude Toward the Behavior 1.01 (0.96 1.07) 0.28 ns Subjective Norm 1.02 (0.99 1.04) 3.01 ns Perceived Behavioral Control 0.95 (0.91 .995) 4.67 0.031 Behavioral Intention 1.42 (1.18 1.71) 13.93 <0.001 Model Chi-Square = 67.92, df = 4, p < 0.001. Total N = 195 Cox and Snell pseudo R2 = 0.29 Nagelkerke pseudo R2 = 0.55 Table 4-11. Logistic regression analys is TPB composite measuresERAC men p Variables in the model OR (95% CI) Wald Attitude Toward the Behavior 1.06 (1.01 1.12) 4.67 0.031 Subjective Norm 1.00 (1.00 1.04) 3.26 ns Perceived Behavioral Control 0.98 (0.94 1.03) 0.60 ns Behavioral Intention 1.42 (1.16 1.76) 11.01 <0.001 Model Chi-Square = 80.51, df = 4, p < 0.001. Total N = 123 Cox and Snell pseudo R2 = 0.48 Nagelkerke pseudo R2 = 0.68 Table 4-12. Logistic Regression Analysis TPB composite measuresERAC juniors p Variables in the model OR (95% CI) Wald Attitude Toward the Behavior 1.04 (0.93 1.16) 0.24 ns Subjective Norm 1.04 (1.01 1.08) 4.99 0.025 Perceived Behavioral Control 0.95 (0.86 1.04) 1.45 ns Behavioral Intention 1.41 (1.00 1.97) 3.92 0.048 Model Chi-Square = 45.61, df = 4, p < 0.001. Total N = 78 Cox and Snell pseudo R2 = 0.44 Nagelkerke pseudo R2 = 0.69 Table 4-13. Logistic regression analysis TPB composite measuresERAC seniors p Variables in the Model OR (95% CI) Wald Attitude Toward the Behavior 1.04 (0.98 1.10) 1.56 ns Subjective Norm 1.01 (0.10 1.03) 2.37 ns Perceived Behavioral Control 1.01 (0.96 1.06) 0.26 ns Behavioral Intention 1.31 (1.05 1.63) 5.58 0.018 Model Chi-Square = 44.75, df = 4, p < 0.001. Total N = 96 Cox and Snell pseudo R2 = 0.37 Nagelkerke pseudo R2 = 0.53 112

PAGE 113

Table 4-14. Logistic regression an alysis TPB composite measuresERAC graduate and professional students p Variables in the model OR (95% CI) Wald Attitude Toward the Behavior 1.00 (0.87 1.15) 0.00 ns Subjective Norm 0.94 (0.10 1.07) 0.02 ns Perceived Behavioral Control 0.91 (0.81 1.02) 2.68 ns Behavioral Intention 2.23 (1.13 4.42) 5.32 0.021 Model Chi-Square = 25.33, df = 4, p < 0.001. Total N = 38 Cox and Snell pseudo R2 = 0.49 Nagelkerke pseudo R2 = 0.73 Table 4-15. Logistic regressi on analysis TPB composite m easuresERAC underage drinkers p Variables in the model OR (95% CI) Wald Attitude Toward the Behavior 1.06 (1.00 1.12) 3.203 ns Subjective Norm 1.03 (1.01 1.05) 5.507 0.018 Perceived Behavioral Control 0.96 (0.91 1.01) 2.797 ns Behavioral Intention 1.37 (1.10 1.71) 7.678 0.006 Model Chi-Square = 71.99, df = 4, p < 0.001. Total N = 158 Cox and Snell pseudo R2 = 0.37 Nagelkerke pseudo R2 = 0.70 Table 4-16. Logistic regressi on analysis TPB composite meas uresERAC legal drinking age p Variables in the model OR (95% CI) Wald Attitude Toward the Behavior 1.04 (0.99 1.09) 2.105 ns Subjective Norm 1.01 (1.00 1.03) 2.040 ns Perceived Behavioral Control 0.98 (0.94 1.02) 1.459 ns Behavioral Intention 1.41 (1.19 1.67) 15.211 <0.001 Model Chi-Square = 74.17, df = 4, p < 0.001. Total N = 159 Cox and Snell pseudo R2 = 0.37 Nagelkerke pseudo R2 = 0.54 Table 4-17. Goodnessof-fit measures for TPB composite measures model tests 2 p df SRMR CFI RFI RMSEA Model TPB 18.352 <0.001 2 0.036 0.972 0.844 0.161 Note: TPB = Theory of Planned Behavior, SRMR = Standardized Root Mean Square Residual, CFI = Comparative Fit Index, RFI = Relative Fit Index, RMSEA = Root Mean Square Error of Approximation. 113

PAGE 114

Table 4-18. Standardized effects on Behavioral Intention and number of drinks consumed on game day r Effect Direct Indirect Total Correlation On BI Unspecified Correlation 0.619 ** 0.386 *** 0.000a0.386 *** Of ATB 0.233 0.654 ** 0.458 *** 0.000a 0.458 *** Of SN 0.196 Of PBC 0.049 0.002 0.000a 0.002 0.047 On # drinks Spurious Correlation Of ATB 0.543 ** 0.000a 0.278 *** 0.278 *** 0.265 Of SN 0.560 ** 0.000a 0.330 *** 0.330 *** 0.230 Of PBC -0.031 -0.066 0.001 -0.065 -0.034 Of BI 0.717 ** 0.720 *** 0.000a 0.720 *** -0.003 Notes: a effect is 0.000 due to prespecified constraint; ** p < 0.01; *** p < 0.001. BI = Behavioral Intention, ATB = Attitude Toward the Behavior, SN = Subjective Norm, PBC = Perceived Behavioral Control. Table 4-19. Goodness-of-fit measures for TRA composite measures model tests Model 2 p df SRMR CFI RFI RMSEA TPB 17.419 <0.001 2 0.043 0.973 0.910 0.156 Note: TRA = Theory of Reasoned Action; SRMR = St andardized Root Mean Square Residual; CFI = Comparative Fit Index; RFI = Relative Fit Index; RMSEA = Root Mean Square Error of Approximation. 114

PAGE 115

* statistically significant at p < 0.001 alpha level. Figure 4-1. Path analysis from TPB composite measures Attitude Toward the Behavior Perceived Behavioral Control Subjective Norm R2 = 0.54* Behavioral Intention R 2 = 0.52* Number of Drinks Consumed on Game Day 0.39* 0.46* -0.07 0.72* 0.51* 0.04 0.08 0.00 Attitude Toward the Behavior Subjective Norm R2 = 0.54 Behavioral Intention R 2 = 0.51 Number of Drinks Consumed on Game Day 0.39 0.46 0.72 0.51 All statistically significant at p < 0.001 alpha level. Figure 4-2. Path analysis from TRA composite measures 115

PAGE 116

CHAPTER 5 SUMMARY, DISCUSSION, IMPLICATIONS, CONCLUSIONS Summary This chapter provides a summary of the study purpose, methods, and results, as well as discussion, implications, and conclusions. The implications incorporate recommendations for future research, preven tion strategies, and social marketing practices. This inquiry examined college student alcohol consumption on game day, which includes drinking that occurs before, during, an d after the game, both on or off campus. While the literature encompasses considerable research on college students and 5+4+ drinking, a dearth of information exists on game day drinking behaviors. The goal of this exploratory research was to utilize the Theory of Planned Behavior (TPB) to gain a better understanding of the motivational factors associated with college student alcohol consumption on game day in an attempt to guide future prevention efforts. Specifical ly, the purposes of this research were to (a) assess the rates of Extreme R itualistic Alcohol Consumption (ERAC) among college students; (b) test the effectiveness of th e Theory of Planned Behavior (TPB) in predicting ERAC rates; and (c) determine which, if any, of the TPB constructs to utilize when designing interventions to decrease ERAC by college students. This investigation incorporated a cross-sec tional survey design with randomly selected participants. The Game Day Survey instrument used in this study was modified and adapted from previous research by Haun and colleagues ( 2007). TPB items were added to the Game Day Survey based on information obta ined from a review of the pr ofessional literature. Survey content validity was established through review by an expert panel, including University of Florida professionals in college health promo tion and student affairs, an alcohol and drug researcher, and a Distinguished Professor in the College of Pharmacy. The feedback from this 116

PAGE 117

panel resulted in the minor modification of certain survey items. The Cronbachs alpha and testretest values indicate that the Game Day Survey is a reliable instrument. The study population included approximately 51,520 UF students enrolled for classes during the 2006 fall semester. A randomly selected list of e-mail addresses for 2,083 students ages 18-24 was provided by the Registrar. The Game Day Survey was administered electronically on November 20, 2006, the Monday af ter the final home football game of the season. The last possible date for students to complete the survey was December 18, 2006. During the four-week study duration three reminders were sent to participants. A total of 740 students responded to the anonymou s electronic survey, yielding a response rate of 36%. Overall, the sample matched the student population with the exception of a slight overrepresentation of females and Caucasians. The study was designed to answer the following research questions: What is the prevalence of Extreme Ritualistic Alcohol C onsumption on a typical game day for fall 2006? How much variance does the combination of constructs in the Theory of Planned Behavior explain when predicting Extreme Ritualistic Alcohol Consumption on game day? Which constructs within the Theory of Planned Behavior (Subjective Norm, Attitude Toward the Behavior, Perceived Behavioral Control, and Behavioral Intention) account for the largest proportion of variance when predicting Extreme Ritualistic Alcohol Consumption behavior among colle ge students on game day? Do the constructs within the Theory of Planned Behavior differ by gender when predicting Extreme Ritualistic Alcohol Consumption among college students on game day? Do the constructs within the Theory of Pl anned Behavior differ by grade classification when predicting Extreme Ritualistic Alc ohol Consumption among college students on game day? What are the causal effects in predicting alc ohol consumption rates using the constructs from the TPB? 117

PAGE 118

Inferential statistics were used to examine th e rate of ERAC on game day and to assess how this behavior differs by vari ous demographics. Multiple logi stic regression analyses were conducted to determine the amount of variance TP B contributed when used to predict ERAC. Multiple logistic regression analysis yielded odds ratios revealing which of the TPB constructs were most influential in predicting ERAC. Sepa rate multiple logistic regression analyses were conducted to ascertain differences influenced by gender, grade classification, and the legal drinking age when predicting ERAC. Path anal ysis was utilized to determine if the TPB provided a valid framework for explaini ng alcohol consumption on game day. The results from the multiple logistic regr ession analyses revealed that the TPB model correctly classified 88.7% of th e cases. While each of the TPB constructs was statistically significant, Behavioral Intenti on was the strongest predictor of ERAC. The findings indicated that the construct Attitude Toward the Behavior was predictive of ERAC with males, whereas the construct Perceived Behavior al Control was the strongest predictor among females. The differences in predicting ERAC among grade classifications were mi nimal, but noteworthy distinctions existed based on legal drinking ag e. Among underage students the Subjective Norm construct was useful in predicting ERAC, but not among college students 21 and older. The path analysis revealed that, with the exception of Pe rceived Behavioral Cont rol, each of the TPB constructs contributed significant explanatory power in predicting intention and behavior. Based on the results of this study, the original Theory of Reasoned Action constitutes a more efficacious model for explaining drinking on game day than the more complex TPB, because TRA does not include the PBC construct. Nevert heless, additional research utilizing the PBC construct is needed before definitive conclusions can be made concerning the TPBs applicability in this area. 118

PAGE 119

Limitations This study contains several limitations. First, self-reported data may include recall bias (Portney & Watkins, 2000). Participants do not always report their behavi ors or respond to questions accurately. Some participants may be inclined to over-or underreport their behaviors, especially when answering sensitive questions. For example, respondents may be reluctant to admit driving under the influence or people may exaggerate how much alcohol they drank. Further, underage participants may be reluctant to divulge any information which they fear could lead to legal repercussions with the police or university, and may intentionally misrepresent their age or fail to acknowledge their drinking behaviors. In othe r situations par ticipants simply cannot recall their drinking behavi or correctly. Intoxicated individuals may not be able to remember certain segments of their drinking e xperience, let alone remember precisely how many drinks they consumed. Nevertheless, self-repo rted data constitute a valid and common method for collecting health informati on, especially when measures ar e taken to ensure anonymity (Brener et al., 2003; Cooper, Sobell MB Sobell LC, & Maisto, 1981; Midanik, 1988). The 36% response rate obtained for this survey merits serious consideration when interpreting the results fr om this study. While the sample matched the overal l student population with the exception of a slight overrepresentation of females a nd Caucasians, resource limitations precluded the ability to document whether non respondents would have answered the survey questions differently. Heavy drinkers may have been less likely to re spond to an alcohol questionnaire, because this group may be less res ponsible and compliant than others (Cottler, Robins, & Spitznagel, 1987; Knibbe, 1984 as c ited by Lemmens et al., 1988; Wild, Cunningham, & Adlaf, 2001). This would result in the under-re porting of ERAC and 5+4+ drinking rates. Heavy drinkers may respond differently to the various survey items, thus skewing the pseudo R2 119

PAGE 120

values (Cox & Snell, and Nagelkerke statistics sp ecific to multiple logistic regression), the odds ratio scores, and the path coefficent values. Conv ersely, nonsports fans an d abstainers may have chosen not to participant in the study because th ey are not concerned with college football and their experiences with alcohol are limited. Therefor e, they may not be interested in the survey nor appreciate the relevance of their responses (Lahaut, Jansen, Mheen & Garretsen, 2002). However, because the participants were randomly selected to participate in the survey, the response bias should be minimal (Alreck & Settle, 1995). The response rate may have been compromised by e-mail default mechanisms. For example, sometimes mass e-mails go directly into a recipients junk mail folder. With over 2000 e-mails sent out for this st udy, it is likely that so me participants did not check their junk mail folders and were unaware of their opportunity to participate in the study. Nonetheless, the response rate reported in this study is similar to that obtained in other al cohol related web-based surveys (Bormann & Stone, 2001, Glassman et al., 2007). The large sample size may have contributed to committing a Type II error, whereby some analyses appear to be statistically significant, but they may not have been if a smaller sample had been selected. For example, the goodness-of-fit Chi-square procedure (n=316) yielded a statistically significant result suggesting poor model fit. Ho wever, when conducting a path analysis with a large sample, it is not uncommo n for the Chi-square goodness-of-fit measure to be statistically significant, because the test is very sensitive to sample size (Kline, 1998). Further, the count data used for the endogenous variable (number drinks consumed on game day) resulted in a substantial number of abstainers and low-end drinkers, creating a positively skewed distribution, which may have contributed to the high Root M ean Square Error of Approximation value. The high RMSEA value wa rrants caution when interpreting the results 120

PAGE 121

from the path analysis. However, these goodness -of-fit measures may vary from acceptable to unacceptable depending on the index used (Hair, et al., 2006). For these reasons, as is standard practice when conducting a path analysis, multiple goodness-of-fit measures were utilized, which allow the reader to assess the streng ths and weaknesses of the analysis This sample was drawn from the University of Florida, a large public university in the southeastern United States. Because of regional differences, drinking patterns found in this study may not be representative of thos e at other universities. For exam ple, the rates of 5+4+ drinking tend to be the highest at college s and universities located in the Northeast (Presley et al., 2002). In addition, drinking rates change over time and are related to variables such as the football teams schedule (opponent), the time of day the games are played, the teams ranking, the schools football conference, the schools football history, and the weather. To improve external validity, researchers should conduct a need s assessment among randomly selected US universities with football programs, in order to determine the national ra te of 5+4+ and ERAC drinking on game day. The cross-sectional design employed for this st udy limits inferences concerning causality. It is not possible to assess whether TPB cons tructs (alcohol beliefs, attitudes, norms, and perceived control) lead to ERAC, or whethe r previous drinking experiences influence TPB constructs. Causal relationshi ps can only be established by using an experimental design (Cottrell & McKenzie, 2005). Another study limitation involves th e high rates of missing data in this investigation. For example, only about 50% of the data received from the 740 participants could be used for the multiple regression analyses due to the listwise ex clusion of cases (respondents). Therefore, any missing response to the relevant question(s) cau sed the entire case to be removed from the 121

PAGE 122

analysis. This issue was particularly problematic with the composite scales. Additionally, items with a nonapplicable response were also co ded as missing. The Normative Belief and Motivation to Comply scales included items concerning the approval or compliance with a significant others opinion on game day drinking. Because they were currently without a significant other, nearly a third of the sample pa rticipants indicated the not applicable response to these items. The use of this response option by such a large portion of the sample contributed to the high rates of missing data. No adjustments were made to account for the missing data. Consequently, it is impossible to determine the influence of the missing data on the logistic regression analyses. Missing data represents a common problem in multivariate analysis (Beale & Little, 1975). However, all of the logistic regression analys es in the present study included large numbers of participants, thereby increasing the reliability of the findings. Nevertheless, caution should be used when interpreting these results. Finally, the reliability measures, which consiste d of a test-retest and internal consistency analyses revealed, limitations with certain items. For example, the Cronbachs alpha scores for the Behavioral Beliefs and Evaluations of Behavioral Outcomes scales were 0.559 and 0.685, respectively, indicating only moderate reliabil ity. Table 3-16 illustrates the Cronbachs alpha value for several of the survey scales would im prove if certain items which had low Cronbachs alpha items were removed from the analysis. In addition, the test-retest results for Perceived Behavioral Control ranged from 0.85 to 0.54 with an average of 0.68, indicating modest stability over time. Thus, the findings from this investigat ion need to be interpreted with caution due to the moderate, but acceptable reliabilit y scores among certain items. 122

PAGE 123

Discussion Despite these limitations, the results of this study reveal important information that increases our understanding of alcohol use am ong college students, particularly within a ritualized context. Game day signifies a distinct social experience, in which tens of thousands of fans come to campus to watch football. For some, the event comprises a major opportunity to imbibe alcoholic beverages, which may include drinking before, during, and after the game. Indeed, roughly half of the sample population in this study indicated that they typically drank on game day, with the average time spent drinking equa ling nearly four hours. In addition, students who reported drinking on game day each consumed an average of approximately seven alcoholic drinks. In an effort to describe the unique drinking patterns which occur on game day, a customized term, Extreme Ritualistic Alcohol Consumption (ERAC) was created for this investigation. ERAC based on the measure White and colleagues created (2006), is defined as consumption of 10 or more drinks on game day for a male, and eight or more drinks for a female. This differs from the conventional 5+4+ drinki ng measure, meaning five or more drinks by a male, four or more drinks by a female in one s itting (Wechsler et al, 2002) While the latter is helpful in identifying unsafe drinking practices, it fails to provide a clear indication of how heavily people actually drink (White et al., 2006). Given the unique circumstances associated with game day drinking, the ERAC measure ma y better assist health advocates and school officials in assessing the extent of high-volume alcohol consumption on game day. Approximately 16% of the sample populati on reported engaging in ERAC on game day, with males participating in this behavior at more than twice the rate of fe males. Not surprisingly, the 5+4+ drinking rates follow a similar pattern. Nearly two out of every five students engaged in 123

PAGE 124

5+4+ drinking on game day, with females drinking at substantially lower rates than males. These drinking trends are consistent with th e literature (NIAAA, 2004b; UF Biennial Review, 2006; Wechsler, et al., 2002), which indicates that males typically consume more alcohol than females. The racial/ethnic drinking patterns among par ticipants in this study also affirm the literature on college students and alcohol consumption (Wechsler et al., 2002). Caucasians in the sample population drank at considerably highe r rates than other ra ces or ethnicities. Approximately two-fifths of Caucasians in th is study consumed 5+4+ drinks, while one-fifth engaged in ERAC. Hispanics re ported drinking at the second hi ghest rates followed by African Americans and Asians, respectively. In summar y, while no racial group or gender should be ignored, the need to target male Caucasians wi th alcohol interventions remains a public health priority on college campuses. The results of this study also justify creating specific initiatives for the Greek (social fraternities and soro rities) community. W ithin the sample of students in this study, approximately one-quarter of the Greek population (students in a social fraternity or sorority) engaged in ERAC on game day. Greeks were a lmost twice as likely to engage in ERAC on game day as non-Greeks. Further, nearly half of Greeks (46.9%) engaged in 5+4+ drinking, whereas approximately one-third (33.7%) of non-Gr eeks reported taking part in this behavior on game day. When analyzed by grade classification, the ra tes of 5+4+ and ERAC both increase with undergraduate grade classification. Additionally, graduate/professional students engaged in ERAC on game day at rates higher than those of freshmen, sophomores, and juniors, and at rates similar to those of seniors. A similar pattern existed with 5+4+ drinking on game day, which 124

PAGE 125

indicates an unusual finding. Previous research reveals that gra duate/professional students drink substantially less than undergraduates and that upperclassman tend to mature and dont drink as much as underclassman (Marlett, Baer, Ki vlahan, Dimeff, Larimer, Quigley, Somers, & Williams, 1998; Wechsler, et al., 2002). Similarly, these results reveal that older stude nts drink at higher rates on game day than do younger students. Nearly a quarter of the st udents ages 21 and over engaged in ERAC on game day compared to less than one-tenth of thos e under the legal drinking age. Students legally permitted to drink alcohol were almost three times more likely to drink on game day than students under the age of 21. Perh aps the minimum drinking age law is serving as a deterrent or a protective factor on game day among stude nts under the age of 21. Haun and colleagues (2007), attribute lower rates of 5+4+ drinking on game day among underage students to increased vigilance by police, bar/restaurant ow ners, school officials, and others. Conversely, underage participants may be unwilling to disclose any information which they fear would lead to legal consequences. As a result, these students may lie about their ages or not admit to consuming alcohol, leading to under-repor ted underage drinking statistics. The multivariate analyses conducted in this study provided additional insight. The odds ratios revealed that, while each of the TPB cons tructs was statistically significant at the 0.05 alpha level, only Behavioral In tention predicted ERAC at a level sufficient to influence intervention design. The modest odds ratios for engaging in ERAC among the constructs Attitude Toward the Behavior, Perceived Behavi oral Control, and Subjective Norm were 1.04, 1.03, and 1.02, respectively, whereas the odds ratio for Behavioral Intention was 1.40. A one unit change in the Behavioral Intention score was associated with a 40% in crease in the odds of engaging in ERAC, whereas the other TPB constr ucts each elicited less than a 5% increase, 125

PAGE 126

respectively. In summary, the odds ratio values for the composite measures were too small, in practical terms, to make programmatic decisions Additional research is needed to expand our understanding of how the TPB can be utilized to explain alcohol us e and abuse, especially as it relates to game day drinking. Another purpose of this invest igation was to assess gender differences when utilizing the TPB to predict ERAC. Separate multiple logi stic regression analyses were conducted with females and males to determine the extent to which gender influences ERAC. Behavioral Intention was the strongest predictor variable with both groups, while th e Subjective Norm was not statistically significant with either gender. Thus, based on the findings from this study, both males and females are likely to benefit from Beha vioral Intention interventions, whereas neither genders ERAC rates are apt to decrea se from norms-related initiatives. Consistent with the findings of Wall and co lleagues (1998), Perceived Behavioral Control was statistically significant for females, but no t for males. For females, the likelihood of excessive drinking increases as th e perceived control decreases. Males, conversely, appear not to be influenced by this psychosocial factor. The au thors concluded that fema les appraisal of the amount of control they have over their alcohol co nsumption is more accurate than that of their male counterparts (Wall et al., 1998). Females ma y be more aware or likely to acknowledge the limits of their control than men especially as it re lates to alcohol use. Males may perceive that it is not masculine to acknowledge their limits or lack of control. Attitude Toward the Behavior was found to be statistically significant among males, but not among females. This finding conflicts with the study by Johnson and colleagues (2006) which found that college-aged females who held fa vorable attitudes toward alcohol were more likely to drink excessively. However, their st udy only included undergraduate female students 126

PAGE 127

from Australia. Thus, these findings may not be applicable in the Unite d States, especially relating to game day drinking. Nevertheless, the findings from this study confirms previous research indicating the efficacy of challenging student alcohol expectancies (Wall et al., 1998; NIAAA, 2002b), and warrants the development of intervention strategies designed to reduce alcohol consumption among male college students on game day. While the multiple logistic regression results showed minimal differences based on grade classification, an additional analysis on the lega l drinking age variable revealed that TPB was slightly more effective in predic ting ERAC among underage participan ts than those of legal age. Behavioral Intention was predictive of ERAC for both underage students and those of legal drinking age, but the Subjective Norm constr uct was statistically significant only among underage participants. None of the other TPB constructs was significant with either group. These findings indicate that ERAC among college students who are underage is motivated, in part, by the approval or disapproval of key referents. Thus, desi gning a social norm intervention targeting underage students has merit. The results of the path analysis indicate that the original Theo ry of Reasoned Action (TRA) model may provide a better model than it s extension, the Theory of Planned Behavior (TPB), for explaining alcohol consumption on ga me day. The Perceived Behavioral Control construct was not predictive of Behavioral Intention or self-reported drinking behavior on game day, as the TPB model suggests. Nevertheless, Behavioral Intention to get drunk on game day predicted drinking behavior. Inte ntions, in turn, were predicted by Attitude Toward the Behavior and Subjective Norm constructs. These findings are consistent with the re search of OCallaghan and colleagues (1997), who found that Perceived Behavi oral Control (PBC) was not predictive of college students 127

PAGE 128

intentions to drink alcohol or of their drinking behavior. The authors speculated that because most of the participants in their study we re non-problem drinkers, they experienced high perceived control over their dri nking. Thus while a substantia l percentage of their sample intended to drink alcohol, they pe rceived their alcohol consumption to be within their control. This may explain why the PBC failed to elicit a si gnificant influence in their research or in the present study. The TPB may be le ss applicable in alcohol res earch with college students for these reasons, than other behavioral theories. The items used to assess the PBC construct may provide another potential explanation for its lack of statistical significance in the current study. A composite measure of the indirect constructs Control Beliefs and Perceived Power was used to create the PBC scale. Control Beliefs are designed to assess the presence or ab sence of facilitators and barriers to behavioral performance weighted by the Perceived Power or imp act of each factor to fa cilitate or inhibit the behavior (Glanz et al., 2002 pg. 75). While the items were constructed on the basis of the literature and were reviewed by experts, these su b-scales require refinement. The Control Belief items utilized in the current st udy assessed how often respondents participated in the behavior, but they did not measure the participants pe rceptions of the like lihood of performing the behavior. For example, one of the Control Belie f items asked how often participants Are given free alcoholic drinks on game da y? with always to never response options. The item could improve from a theoretical standpoint by assessing the perceived likelihood of the facilitating or constraining condition. An alternat ive version to this item is, How likely is it that you will be offered free alcoholic drinks on game day? with likely to unlikely response options. The Perceived Power items assessed the likeli hood of performing a be havior rather than the perceived ease or difficulty of engaging in the behavior. For example, one of the Perceived 128

PAGE 129

Power items on the Game Day Survey asked if People offering me free alcoholic drinks influence my decision to get drunk on game day with extremely likely, and extremely unlikely as response options. This item would be more consistent with the TPB if it were reworded as follows: Receiving free alcoholic drinks makes getting drunk on game day: with response options varying from easy to difficult. Poor theoretical fidelity with the design of the Control Belief and Perceived Power items may help to explain why the PBC construct failed to elicit a statistically significant value. Accurately assessing the PBC construct remain s a challenge for researchers. A metaanalysis revealed that only a limited number of studies demonstrated a significant relationship between PBC and Behavioral Intention (Ajzen, 1991). In the current study the PBC construct was statistically significant in the multiple logi stic regression analyses when all of the TPB variables, including Behavioral Intention, were treated equally as independent variables. However, PBC was not statistically significant in the path analys is when Behavioral Intention served as a mediating variable between the dependent variable alcohol consumption and the TPB composite measures. Thus, as Reinecke, Sc hmidt, and Ajzen (1996) found, controlling for Behavioral Intention weakens the statistical supp ort for the PBC construct. Consequently, the results of the logistic regression analyses should be interpreted with caution. Interpreting the findings from this investiga tion requires theoretical considerations as well. A weakness of the TPB is the underlying as sumption that behavior follows a linear course of action. However, behavior optimizes a dynamic, extremely complex phenomenon (Glanz et al., 2002). Certain predic tive factors are not included. For ex ample, TPB does not address past behavior, which serves as a strong predictor of alcohol use among college students (OCallaghan et al., 1997). Ones attitude toward drinking may be based mo re on past drinking experiences 129

PAGE 130

than on future expectations concerning alcohol us e. Perhaps it is more likely that ones past alcohol use creates the foundation for future alcohol expectancies A similar pattern emerges with the construct Perceived Behavioral Control. Ones previous performance of a behavior likely influences perception of personal cont rol over the behavior, unlike the notion that Perceived Behavioral Control merely influences behavioral intention and behavior (Kashima, Gallois, & McCamish, 1993). Behavioral Intentions may change or evolve as students drink on game day. For instance, a student may intend to drink in moderation on ga me day, but as the individual consumes more and more alcohol, situational cues for behavior become more salient. The pleasure of the drug, the excitement of the game, and the immedi ate influence of ones peers may shift the individuals Behavioral Intentions and, as a result, the behavi or changes as well. A respondents attitude, perceived acceptance of drinking, control of alcohol use, and intentions may fluctuate throughout the drinking encounter. In short, beliefs and attitude s are likely to shift before, during, and after the drinking experience. The present study, based on a simple linear model, measures the causal sequence after the entire episode, and thus fails to address potential fluctuations in attitudes, beliefs, and behavi or. Obtaining such information would shed additional insight into the complex and dynamic factors which influence behavior. Finally, while the TPB allows researchers to measure systematically the role of factors perceived by respondents to be salient in their in tentions and behaviors, it does not necessarily mean those perceptions are accurate. Cognitive bi ases among participants may result in an underor overestimation of what truly motivates their behavior (Alreck & Settle, 1995). For example, several key external factors may influence alcohol use on game day. The presence of law enforcement, the cost of alcohol, and the opportun ity to tailgate may all impact behavior much 130

PAGE 131

more than respondents realize or are willing to admit. Creating in terventions designed to change behavior based on respondents perceptions may not be optimally effective. While the Game Day Survey included items which measured part icipants perceptions, future studies should include an objective environmental assessment that more accurately measures the external influences mentioned above. Implications The Game Day Survey was conducted to describe Extreme Ritualistic Alcohol Consumption patterns among college students usi ng the Theory of Planned Behavior as a guiding framework. The findings pr ovide important implications a ddressing the scope of alcohol use and abuse on game day among college st udents at the Univer sity of Florida. Alcohol consumption is a common occurrence on game day, with approximately half of the sample population reporting that they typically drink on game day. Further, 36% and 16% of those surveyed engaged in 5+4+ drinking and ER AC, respectively. Given the large number of students who attend college footba ll games at the University of Florida, game day drinking represents a significant public health concern. The alcohol related consequences which accompany game day drinking include DUI, fights, encounters with the police, hangovers, vomiting, black outs, sexual a ssaults, and injury. While the relationships among the variable s in the TRA/TPB vary depending on the sample and behavior of interest (Trafimow, 1996; OCallaghan et al., 1997; Johnson & White, 2004), the model effectively isolates variables likely to influence a persons motivation to drink (Budd et al., 1984). As far as game day drin king is concerned, the findings from this investigation further validate th e Theory of Reasoned Action. The applicability of its extension the Theory of Planned Behavior, the model employed in this study, remains in question. 131

PAGE 132

Nonetheless, the results from this study provide a useful model for explaining alcohol consumption among college students on game day. The model demonstrates a considerable degree of success in predicting alcohol consumpti on, and in predicting Beha vioral Intention in particular. Thus prevention e fforts should (a) focus on persuadi ng students to set limits for themselves concerning their alcohol consumption on game day, (b) create attitudes and beliefs which are inconsistent with excessive alcohol co nsumption, and (c) highlight peer disapproval of inappropriate drinking behavior. Recommendations Future Research Recommendations for future research include the following: Administer the Game Day Survey using re vised Perceived Behavioral Control (PBC) items to better assess if the insignificant PBC results were a function of theoretical underpinnings or psychometric error. In a ddition, determine whet her the TPB composite measures are more effective in predicting alcohol use among college students on game day than the original direct measures. Replicate this study utilizing a time series or longitudinal design. The present study was cross-sectional; thus, inferences about causa lity cannot be determined. In addition, a number of variables may influence game day drinking, such as the time the game is held, the opponent, the win-loss records of the hom e team and the opponent, rivalry between teams, etc. Studying these patterns may provide insight into the scope of the problem as well as future prevention efforts. A long itudinal study design requires a substantial incentive for the participants. A time-seri es method utilizing different samples may represent a realistic research de sign to address this issue. Conduct qualitative research w ith college students regardi ng their drinking behavior on game day. Information concerning what diffe rentiates game day drinking from normal drinking activities is lacki ng. In addition, collecting information from students on motivating factors which may mode rate their drinking on game da y is an essential step in designing prevention messages. In the present study approximately half of the sample reported abstaining from alcohol use on a typical game day. This led to a zero-inflated positively skewed distribution for the responses to this item. In addition, abst ainers, social drinkers, 5+4+ drinkers, and those who engage in ERAC may have unique motivations regarding alcohol use. Conducting a discriminant analysis categorizi ng participants based on their alcohol use 132

PAGE 133

on game day would address distributional conc erns while providing insights on how to tailor intervention design to each of these groups. Replicate this study and examine the rates of alcohol related cons equences of 5+4+ drinkers compared to those who engage in ERAC. The more people dr ink the more likely they are to experience alcohol related consequences. The results from this research may provide further justification fo r utilizing the ERAC measure if the rate of consequences differs considerably between the two groups. Administer the Game Day Survey utilizi ng the Expectancy Theory as a guiding framework. The males in this investigation appeared to be most influenced by the TPB construct Attitude Toward the Behavior. This construct is very similar to the Expectancy Theory found in the Social Cognitive Th eory (Bandura, 1977). The NIAAAs (2002d) Call to Action: Changing the Culture of Drinking at U.S. Colleges lists Challenging Alcohol Expectancies as an effective st rategy in reducing alc ohol use among college students. Modifying the Expectancy Th eory scale to match game day drinking expectancies may shed unique insights into the developmen t of prevention messages and the design of other relevant interventions. Assessing how males and females respond to this construct may benefit practitioners as well. Conduct research on game day utilizing br eathalyzers to collect Blood Alcohol Concentrations (BAC). Due to the inhe rent limitations of self-reported responses, acquisition of more objective data represents a research priori ty. Further, collecting BAC measures during tailgating activities on campus before, durin g, and after the game may provide school officials with the justification needed to im plement policy changes such as restricting the time and loca tion parameters associated with tailgating. Examine the drinking behaviors of alumni, vi sitors, and other foot ball fans who are not college students. The UFs football stad ium seats approximately 85,000 fans, many of whom are not current UF college students. Special interventions n eed to be designed, implemented, and evaluated for these groups pa rticularly alumni, whose actions may be emulated by impressionable undergraduates. Practice Recommendations for practice include the following: Implement a social marketing campaign that focuses on specific TRA constructs while segmenting the audience based on ge nder and legal drinking status. o To address the Behavioral Intention cons truct all demographic groups may benefit from messages which encourage fans to th ink in advance about, and to set limits on, how much alcohol they will drink on game day. o Utilize the Attitude Toward the Behavior construct to design messages targeting males. Challenging alcohol rela ted expectancies, such as needing to drink in order 133

PAGE 134

to have fun on game day, be more social and enjoy the game, is fundamental to changing attitudes towards alcohol. o The Subjective Norm construct should be used to develop messages targeting underage drinkers. Creating a message hi ghlighting the fact that approximately half of the student population does not drink on game day merits consideration. Another possibility is to produce a message stating that ones current or future partner would disapprove of his or he r significant other overindulging on game day. o While more research is needed, target ing females with messages which address the Perceived Behavioral C ontrol construct shows some promise. Addressing the facilitating or constraining issues related to alcohol, such as the financial costs associated with alcohol, the negative physical effects (h angover), the presence of police on game day, etc., may help motiv ate females to reduce their alcohol consumption. Students who engage in ERAC may meet the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition criteria for alcohol abuse or alcohol dependency. Screening students for this behavior provides mental health professionals with data to determine the types of treatment that will best serve these individuals. Students who are transported to the emergency room for the alcohol associated complications or who vi olate state, local, or campus alcohol related policies constitu te obvious intervention priorities. There is a variety of environmental strategi es which, if implemented, may reduce ERAC. One option is to limit the number of tailgating hours permitted before and after the game, thereby reducing access to alcohol on game day. Another possibil ity is to designate certain areas on campus where alcohol would be permitted. Over time these locations could become fewer in number and smaller in area. Conversely, ce rtain areas on campus should provide tailgating alternatives that do not permit alcohol use. Ideally, the University could designate sites on campus where students and other fans could tailgate (watch pre-game shows) with in an alcohol-free venue. Enforcement of existing alcohol laws constitutes a fundament al step in reducing alcohol consumption on game day. Currently, it is illega l in the City of Gainesville to have an open container in any public area, including the UF campus on game day. However, perceived and real pressure from alumni, students, and other fans makes it difficult to address this issue. Furthermore, inadequate resources, including the limited number of law enforcement officers available on game day, add to the enforcement problem. Nevertheless, the findings from this study on pu blic health and safety issues related to ERAC clearly warrant the need for more effective enforcement on game day. Conclusion This study contributes to the limited body of research concerning college student drinking patterns on game day. ERAC represents a unique concept, based on the measure White 134

PAGE 135

and Colleagues (2006) created to classify th e excessive drinking behaviors which occur on ritualized occasions, such as f ootball game day. Assessing dri nking rates by specific levels of consumption provides a fundamental step in monitoring the drinking behaviors of college students. Historically, prev ention efforts divide those w ho engage in 5+4+ drinking (consumption of five or more drinks in one sitting, four or more for a female) from those who do not. However, the prevention needs concerning ab stainers, social drinkers, 5+4+ drinkers, and those who engage in ERAC undoubtedly vary. The complexities associated with alcohol research cannot be overstated. Time, setting, a nd context merit special consideration when assessing alcohol usage. For example, student s may drink more at a night game than a day game; certain stadiums may be more renowned for their tailgating rituals than others; and certain events such as bowl or championship games may re sult in increased drin king. Regardless of the circumstances, students who engage in ERAC are at high risk for adverse alcohol-related consequences. Customized preven tion, screening, and treatment plans need to be designed for this group. Further research is needed to expand our understandi ng of this unique public health challenge. To date, this is the first st udy to utilize a health behavior theory to examine game day drinking. The findings from this investigation furt her validate the original TRAs usefulness as a health behavior theory, while le aving in question the applicability of its extension, the TPB. With the exception of Perceived Behavioral Cont rol, each of the constructs served as a significant predictor of either Behavioral Intentio n or Behavior. Additional research with more effective PBC measures is needed before definitive statements can be made concerning the TPBs applicability in predicting college student alcohol consumption on game day. The TRAs demonstrated parsimony and utility warrant its consideration when developing game day 135

PAGE 136

interventions. The results suggest that the college population should be segmented based on the TRAs theoretical underpinnings, and that each of the constructs be utilized to meet the unique prevention needs of a particular group. The results from this study provide the necessary justification for creating interventions to address alcohol use on game day. Decr easing 5+4+ drinking among college students represents a prominent health goal in Healthy Campus 2010 The specific objective is to reduce the 5+4+ drinking rate to 20% or lower by the en d of the decade. Reducing alcohol consumption on game day may help universities lower their over all 5+4+ drinking rates, especially during the fall semester. Suggestions for reducing 5+4+ drinking and ERAC on game day include alcoholfree game day alternatives, increased enforcemen t of underage drinking, designation of tailgating areas which clearly delineate wher e alcohol consumption is legal, limitation of the number of tailgating hours, and implementation of social marketing messages to discourage excessive drinking. Universities need to implement and ev aluate these and other game day interventions, and additional studies need to be conducted to de termine their relative effectiveness. 136

PAGE 137

APPENDIX A UF GAME DAY SURVEY The University of Florida is interested in the social and personal activities in which our football fans engage on Game Day. Pleas e take about 15 minutes to comp lete this brief (37 questions) and anonymous survey. Your participation and honest answers are grea tly appreciated. Go Gators! 1. Which football games did you attend in 2006? UF vs. Southern Miss on Saturday, September 2, 2006 UF vs. Central Florida on Saturday, September 9, 2006 UF vs. Kentucky on Saturday, September 23, 2006 UF vs. Alabama on Saturday, September 30, 2006 UF vs. Louisiana State on Satu rday, October 7, 2006 (Homecoming) UF vs. South Carolina on Saturday, November 11, 2006 UF vs. Western Carolina on Saturday, November 18, 2006 Didnt attend any home football games this season 2. What is your sex? Male Female 3. What is your classification? Freshman Sophomore Junior Senior Graduate Student/Professional Other (please specify) 4. How do you describe yourself? American Indian/Alaskan Native Asian or Pacific Islander Black (non Hispanic) Hispanic or Latino White (non Hispanic) Other (please specify) 137

PAGE 138

5. Are you currently a member of a Greek fr aternity or sorority (IFC, NPHC, PC, MGC)? Yes No 6. How old are you? 7. How much do you weigh (in pounds)? Game Day Public Relations Please select the most appropriate response for the following questions. 8. Since the beginning of the fall 2006 school year, please indicate if you have seen the following: Yes No Ad in the Alligator with Coach Meyers picture encouraging Gators to be responsible fans? GatorLight tip sheet about drinking in moderation? The Gator Fans Code of Conduct? GatorHealth Guide? The public service announcement Gators Set the Standard Respect the Swamp? 9. The public service announcement Gators Set th e Standard Respect the Swamp concerns which of the following: Reducing litter Improving traffic flow Courteous fan behavior Reducing fights Preventing excessive drinking Not Sure 138

PAGE 139

Game Day Behaviors Game Day is defined as a typical home foot ball game, including activities before, during, and after the game (i.e., tailgating either on or off campus). 10. Do you typically drink alcohol on Game Day? Yes No 11. Where do you spend the majority of your time drinking alcoholic beve rages on game day? Dont Home Friends Restaurant Bar Tailgate Other Drink Home Area Before the Game? During the Game? After the Game? Game Day Drinking One drink is defined as 12 oz of beer, 12 oz of wine cooler, 5 oz of wine, 1.25 oz of liquor either straight or in a mixed drink. Note: please answer the following questions regardless of your location on Game Day. 12. What is the total number of alcoholic drinks that you typically cons ume before, during, and after a Gator home football game? 13. How many alcoholic drinks do you typically consume during the two hours before a Gator home football game? 14. How many alcoholic drinks do you typically consume during a Gator home football game? 15. How many alcoholic drinks do you typically consume during the two hours after a Gator home football game? 139

PAGE 140

16. What is the total number of alcoholic drinks that you consumed before, during, and after the Gator home football game ag ainst Western Carolina? 17. What is the total number of hours that you typically spend drinking on Game Day? 18. Not including Game Day how many alcoholic drinks di d you have the last time you partied/socialized? 140

PAGE 141

Game Day Health You are now halfway through the Game Day Survey. Please finish taking the survey to obta in information regarding the incentives. At the end of the survey you will be able to view the aggregate group responses to each of the questions. 19. During the 2006 football season, how often did you experience the following due to drinking alcohol on Game Day? Never Rarely Sometimes Often Always Had a hangover Vomited Drove after drinking alcohol Drove after having 5 or more drinks Had a memory loss (blackout) Was hurt or injured Got into a fight or an argument Got reprimanded by the police Arrested/ticketed by the police Took advantage of someone sexually Had been taken advantage of sexually Perceptions 20. What percentage of people in the following categories do you believe typically get drunk on Game Day? Your Friends Current UF Students Gator Fans (who are not current students) 141

PAGE 142

Drunk is defined as having ones mental and physical abilities impa ired by alcohol. In the state of Florida a blood alcohol level th at is equal to or exceeds .08 is considered legally intoxicated or drunk. 21. Please check the circle that shows how lik ely or unlikely it is that you would do the following. Extremely Likely Extremely Unlikely I would have more fun if I got drunk on game day: I would be more soci al if I got drunk on game day: My chances of hooking up with someone (having sex) would increase if I got drunk on game day: I would have a hangover if I got drunk on game day: I would enjoy watching the game less if I got drunk on game day: I would embarrass myself if I got drunk on game day: 22. Please check the circle that indicates your level of agreement or disagreement with each of the following statements. Strongly Agree Strongly Disagree Having fun on game day is important to me: Being social on game day is important to me: Meeting someone and hooking up (having sex) with them on game day is important to me: Having a hangover on game day is a concern of mine: Watching the game is important to me: Embarrassing myself due to my drinking on game day is a concern of mine: 142

PAGE 143

Please select the response that most accura tely reflects your beliefs for each of the following statements. 23. For me to get drunk on game day is: Good Bad 24. For me to get drunk on game day is: Beneficial Harmful 25. For me to get drunk on game day is: Enjoyable Unenjoyable 26. For me to get drunk on game day is: Healthy Unhealthy 27. For me to get drunk on game day is: Favorable Unfavorable 28. For me to get drunk on game day is: Wise Foolish 143

PAGE 144

29. Please check the circle that indicates your level of agreement or disagreement with each of the following statements. Strongly Agree Strongly Disagree N/A My best friend would approve of me getting drunk on game day: My close friends would approve of me getting drunk on game day: My mother (legal guardian) would approve of me getting drunk on game day: My father (legal guardian) would approve of me getting drunk on game day: My current partner would approve of me getting drunk on game day: My ideal future partner would approve of me getting drunk on game day: 30. When it comes to drinking alcohol, how mo tivated are you to meet the expectations of your: Very Motivated Not motivated at all N/A Best friend? Close friends? Mother (legal guardian)? Father (legal guardian)? Current partner? Ideal future partner? 144

PAGE 145

31. Please check the circle that indicates your level of agreement or disagreement with each of the following statements. Strongly Agree Strongly Disagree The people in my life whom I value get drunk on game day: The people in my life whom I value would approve of me getting drunk on game day: Most people I hang out with get drunk on game day: The people in my life whom I value encourage me to get drunk on game day: 32. Please indicate how often you: Always Never Use a designated driver or safe transportation on game day. Attend pre-game tailgating activities on game day. Are given free alcoholic drinks on game day. Notice the police on game day. Consider the financial costs associated with consuming alcoholic beverages on game day. Feel hungover from drinking alcohol on the day after game day. 145

PAGE 146

33. Please check the circle that indicates how likely or unlikely you would be influenced by each of the following scenarios. Extremely Likely Extremely Unlikely Having a designated driver or safe transportation influences my decision to get drunk on game day. Attending pre-game tailgating opportunities influences my decision to get drunk on game day. People offering me free alcoholic drinks influence my decision to get drunk on game day. The presence of police deters me from getting drunk on game day. The financial costs associated with alcoholic beverages deter me from getting drunk on game day. Having a hangover the day after a game, deters me from getting drunk on game day. 34. Please check the circle that indicates your level of agreement or disagreement with each of the following statements. Strongly Agree Strongly Disagree I am confident that I can limit my alcohol consumption on game day: I can resist pressure from friends to consume alcohol on game day: Its difficult for me to drink moderately on game day: As I get drunk, I start to lose control over the number of drinks I consume: Its difficult for me to refuse free alcoholic drinks on game day: 146

PAGE 147

147 35. Please check the circle that indicates your level of agreement or disagreement with each of the following statements. Strongly Agree Strongly Disagree N/A I intend to get drunk at the next Gator home football game I attend: I intend to drink in moderation at the next Gator home football game I attend: I intend to not drink any alcoholic beverages at the next Gator home football game I attend: I intend to get drunk at every Gator home football game I attend: 36. Approximately how many alcoholic drinks does it take for you to become drunk? 37. Is there anything else that you think affects your leve l of alcohol consumption on game day (Gator home football game)? Thank You! You have now completed this survey. To be eligible for the $50 gift card please e-mail GatorWell at gatorwell@ufl.edu and provide your name, e-mail, and phone number. In the subject line please write Game Day Survey Incentive. Your contact information will not be linked to your survey responses. The first three, middle three, and last three participants to complete the survey will be awarded the incentive. Survey results are available on the subsequent webpage. We appreciate your time and opinions.

PAGE 148

APPENDIX B GAME DAY SURVEY IRB 1. TITLE OF PROTOCOL: Game Day Survey 2. PRINCIPAL INVESTIGATOR(s): (Name, degree, title, dept., address, phone #, e-mail & fax) Virginia Dodd, PhD, MPH, Assistant Professor, College of Health & Human Performance, Department of Health Edu cation & Behavior, PO Box 118210, Gainesville, FL 32611-8210, 352-392-0583 ext.1359, vdodd@hhp.ufl.edu. Tavis Glassman, MSEd, MPH, CHES, Coordinator Alcohol of and Other Drug Prevention, Student Health Care Center, PO Box 117500, 392-1161, ext. 4281, tavis@ufl.edu 846-2628. 3. SUPERVISOR (IF PI IS STUDENT): (Name, campus address, phone #, e-mail & fax) 4. DATES OF PROP OSED PROTOCOL: From: November 1, 2006 To: November 1, 2007 5. SOURCE OF FUNDING FOR THE PROTOCOL : (A copy of your grant proposal must be incl uded with this protocol if DHHS funding is involved.) The Student Health Care Cent er is collaborating with the Department of Health Education and Behavior, College of Health and Human Performance, to implement a grant awarded by the U.S. Department of Educations Grant Competition to Prevent High-Risk Drinking Among Colle ge Students. The grant entitled, Using Social Marketing Principles to Change Soci al Norms, award number Q184H060086, will be used to collect data regarding the consump tion of alcohol and be havior of students on home football games. This data will be used to help develop specific game day messages to be promoted throughout campus. 6. SCIENTIFIC PURPOSE OF THE INVESTIGATION: Determine the proportion of students who engage in hi gh-risk drinking on home footba ll games, as well as assess their level of support for game day prevention messages. 7. DESCRIBE THE RESE ARCH METHODOLOGY IN NONTECHNICAL LANGUAGE. The UFIRB needs to know what will be done wi th or to the research participant(s). The online survey is voluntary and anonymous No person under the age of 18 will be selected to participate. Over the course of the month, students will receive two additional reminders to log on and complete the surve y. There is no compensation for taking this survey. 148

PAGE 149

Note: This survey has been previously implemented by Gator Well Health Promotion Services (UF Student H ealth Care Center). 8. POTENTIAL BENEFITS AND ANTICIPATED RISK. (If risk of physical, psychological or economic harm may be involved describe the steps taken to protect the participant.) There are no anticipated risks. Potential benefits include: Capture UF student data to determin e the Game Day drinking behavior of students. Assess what other groups besides students engage in risky behavior on Game Day. Determine if football fans noticed any of the health promoting messages the University of Florida produced concerning game day behavior. Provide baseline data to help create prevention messages. Students will receive a cover letter, via e-ma il, and will be instructed to click on the Game Day Survey link. Thus, e-mail addresse s will not be linked to the survey. 9. DESCRIBE HOW PARTICIPANT(S) WILL BE RECRUITED, THE NUMBER AND AGE OF THE PARTICIPANTS, A ND PROPOSED COMPENSATION (if any): Two thousand students will be randomly selected (by the Registrar) to complete an online game day survey. All participants will be no tified via e-mail, to visit the online survey. Students will be required to read an inform ed consent message prior to taking the survey, which includes instructions. 10. DESCRIBE THE INFORMED CONSENT PROCESS. INCLUDE A COPY OF THE INFORMED CONSENT DOCUM ENT (if applicable). Students will be required to read an inform ed consent message prior to taking the survey, which includes instructions. S ee attached survey for details. Please use attachments sparingly. __________________________ Principal Investigator's Signature _________________________ Supervisor's Signature I approve this protocol for submission to the UFIRB: ____________________________ Dept. Chair/Center Director Date 149

PAGE 150

APPENDIX C GAME DAY SURVEY E-MAIL INSTRUCTIONS Dear Gator Fan, You are among 2,000 Gator fans that have been randomly selected to participate in an anonymous online Game Day survey. Your partic ipation and honest answers are crucial for assessing Game Day public relations efforts, alcohol related issues, and campus safety. Please participate in this survey even if you do not attend home football games. The first three, middle three, and last three survey participants, who complete the survey (and incentive protocol), will receive a $50 gift card to the University of Florida Bookstore. Instructions: Do not take this survey if you are under the age of 18. This short survey will take approximately ten minutes to complete. Please log onto 2006 UF Game Day Survey to begin the survey. This survey is completely anonymous You may choose not to participate or not respond to any questions you do not wish to answer. Your participation is voluntary. You do not have to answer any question you do not wish to answer. You will not be compensated for participating in this survey. Potential benefits for study participation include: Aiding in collecting data to dete rmine Game Day drinking behavior Aiding in the development of health promoting messages concerning Game Day drinking behaviors produced by the UF Provide data to help create future Game Day prevention messages. There are no anticipat ed risks related to study participation. If you have any questions about your rights as a research participant, contact UFs Institutional Review Board, at 352-392-0433 or irb2@ufl.edu If you have concerns about this survey please contact Tavis Glassman at the Student Health Care Ce nter at 352-392-1161 or tavis@ufl.edu Thank you for taking the time and thought to complete this survey. Go Gators! 150

PAGE 151

LIST OF REFERENCES Abbey, A. (2002). Alcohol-related sexual assault: A common problem among college students. Journal of Studies on Alcohol, 63 (Suppl. 14), 118-128. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Process, 50, 179-211. Ajzen, I. (2002). Constructing a TPB questionnaire: Conceptual and methodological considerations Retrieved August 7, 2006, from http://www.people.umass.edu/aizen/pdf/tpb.measurement.pdf Ajzen, I., & Fishbein, M. (1980). Understanding attitudes an d predicting social behavior. Englewood Cliffs, NJ: Prentice Hall. Albarracin, D., Johnson, B. T., Fishbein, M., & Muellerleile, P.A. (2001). Theories of reasoned action and planned behavior as m odels of condom use: A meta-analysis. Psychological Bulletin, 127 142-161. Alreck, P. L., & Settle, R. B. (1995). The survey research handbook (2nd ed.). Boston, MA: Irwin/McGraw-Hill. American College Health Association. H ealthy Campus 2010 Manual. Baltimore, MD: September 2002. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders DSM-IV (4th ed.). Washington, DC: Author. Arndorfer, B. (2005, November 2). Alcohol, sports can be deadly mix. Gainesville Sun, pp. 1A, 4A. Arnett, J. J. (2005). The developmental context of subs tance use in emerging adulthood. Journal of Drug Issues, 35, 235. Baer, J. S. (2002). Student factors: Unders tanding individual va riation in college drinking. Journal of Studies on Alcohol, 63(Suppl. 14), 40. Babor, T. F., Higgins-Biddle, J.C., Sa unders, J.B., & Monteiro, M.G. (2001). AUDIT: The Alcohol Use Disorders Identification Test: Guidelines for use in primary care. World Health Organization. Retrieved September 22, 2006, from http://whqlibdoc.who.int/ hq/2001/WHO_MSD_MSB_01.6a.pdf Bandura, A. (1977). Social learning theory. Upper Saddle River, NJ: Prentice Hall. Barrios, L., Everett, S., Simon, T., & Br ener, N. (2000). Suicide ideation among US college students. Journal of American College Health, 48 229. 151

PAGE 152

Beale, E. M. L., & Little, R. J. A. (1975) Missing values in multivariate analysis. Journal of the Royal Statistical Society, 37, 129-145. Bobko, P. (2001). Correlation and regression: Applicati ons for industrial organizational psychology and management (2nd ed.). Thousand Oaks, CA: Sage. Bormann, C. A., & Stone, M. H. (2001). The e ffects of eliminating alcohol in a college stadium: The Folsom field beer ban. Journal of American College Health, 50, 81-89. Bosompra, K. (2001). Determinants of condom use intentions of university students in Ghana: An application of the theory of reasoned action. Social Science & Medicine, 52, 1057-1069. Brener, N. D., Billy, J. O. G., & Grandy, W. R. (2003.) Assessment of factors affecting the validity of self-reported health-ris k behavior among adolescents: Evidence from the scientific literature. Journal of Adolescent Health, 33, 436-457. Budd, R. J., & Spencer, C. P. (1984). Predic ting undergraduates inte ntions to drink. Journal of Studies on Alcohol, 45, 179-183. Cameron, A. C., & Travedi, P. K. (1998). Regression Analysis of Count Data. New York: Cambridge University Press. Campo, S., Brossard, D., Frazer, M. S., Marche ll, T., Lewis, D., & Talbot, J. (2003). Are social norms campaigns really magic bu llets? Assessing the effects of students misperceptions on drinking behavior. Health Communication, 15, 481-497. Carson, E. (1995). Purging binging. Reason, 27, 61-63. Centers for Disease Control and Preven tion. (2006). Surveillance Summaries. Morbidity and Mortality Weekly Report, 55 (SS05); 1-108. Centers for Disease Control and Preven tion. (2002). Alcohol consumption among pregnant and childbearing-ag e women United States, 1991-1999. Morbidity and Mortality Weekly Report 51(13), 273-6. Centers for Disease Contro l and Prevention. (2006). Communication at CDC. Retrieved October 8, 2006, from http://www.cdc.gov/communication/ practice/social marketing.htm Center for Science in the Public Interest. (2007). Campaign for Alcohol-Free Sports TV. Retrieved June 5, 2007, from http://cspinet.org/booze/CAFST/index.htm 152

PAGE 153

Chaloupka, F. J., & Wechsler, H. (1996). Binge drinking in college: The impact of price, availability, and alc ohol control policies. Contemporary Economic Policy, 14, 112-124. Chen, M. J., Grube, J. W., Bersamin, M., Wa iters, E., & Keefe, D. B. (2005). Alcohol advertising: What make s it attractive to youth? Journal of Health Communication, 10, 553-565. Clarke, M. (1995). Women and alcohol, a gender difference? Retrieved July 1, 2004, from http://www.oznet.ksu.edu/humannutrition/_timely/ALCOHOL.htm Conner, M., & Armitage, C. J. (1998). Exte nding the theory of planned behavior: A review and avenues for further research. Journal of Applied Social Psychology, 28, 1429-1464. Coons, C. J., Howard-Hamilton, M., & Waryold, D. (1995). College sports and fan aggression: Implications fo r residence hall discipline. Journal of College Student Development, 36, 587-593. Cooper, A. M., Sobell, M. B., Sobell, L. C., & Maisto, S. A. (1981). Validity of alcoholics self-repo rts: Duration data. International Journal of the Addictions, 16, 401-406. Core Institute at Southern Illinois University. (1994). Alcohol and drug survey long form. Retrieved July 25, 2006, from http://www.siu.edu/~coreinst/ Core Institute at Southern Illinois University. (2005). Core alcohol and drug survey results: Consequences. Retrieved September 5, 2006, from http://www.siu.edu/~coreinst/ Cottrell, R. R., & McKenzie, J. F. (2005). Health promotion and education research methods: Using the five-chapter thesis/dissertation model Sudbury, MA: Jones and Bartlett. Couper, M. (2000). Web surveys: A review of issues and approaches. Public Opinion Quarterly, 64, 464-494. Dawson, D. A. (1998). Editorial-measuring alcohol consumption: Limitations and prospects for improvement. Addiction, 93, 965-968. Dawson, D. A. (2003). Methodological issues in measuring alcohol use. Alcohol Research and Health, 27, 18-29. Dejong, W., Vince-Whitman, C., Colthurst, T., Cretella, M., Gilbreath, M., Rosati, M., & Zweig, K. (1998). Environmental management a co mprehensive strategy for reducing alcohol and other drug use on college campuses. Newton, MA: Higher Education Center for Alcohol and Other Drug Prevention. 153

PAGE 154

Dillman, D. A. (2000). Mail and internet surveys: The tailored design method (2nd ed.). New York: John Wiley & Sons. Dimeff, L. A., Baer, J. S., Kivlah an, D. R., & Marlatt, G .A. (1999). Brief Alcohol Screening and Intervention for Colleg e Students (BASICS): A harm reduction approach. New York, NY: Guilford. Dodd, V. J., & Glassman, T. (2006). [Focus group comments by UF students on Game day drinking versus typical drinki ng behaviors]. Unpublished raw data. Dooley, D. (2001). Social research methods (4th ed.). Upper Saddle River, NJ: Prentice-Hall. Edwards, P., Roberts, I., Clarke, M., DiGuiseppi, C., Pratap, S., Wentz, R., & Kwan, I. (2002). Increasing response rates to pos tal questionnaires: Sy stematic review. British Medical Journal, 324, 1183. Eigan, L. (1991). Alcohol practices, policies and pot entials of American colleges and universities. Washington, DC: United States De partment of Health and Human Services. Fazio, R. H., & Zanna, M. P. (1981). Di rect experience and attitude-behavior consistency. Advances in Experiment al Social Psychology, 14, 161-202. Fisher, B. S., Cullen, F. T., & Turner, M.G. (2000). The sexual victimization of college women. Washington, DC: Bureau of Justice Statistics, Department of Justice. Frauenfelder, M. (2001, December 9). Social-norms marketing. New York Times Magazine, p. 100. Friis, R. H., & Sellers, T. A. (1996). Epidemiology for public health practice. Gaithersburg, MD: Aspen. GatorWell Health Promotion Services. Alcohol tobacco and ot her drug statistics. Retrieved August 7, 2006, from http://www.shcc.ufl.edu/ gatorwell/atod/pdf/long05.pdf Geisner, I. M., Larimer, M. E., & Neighbor s, C. (2004). The rela tionship among alcohol use, related problems, and symptoms of psychological distress: Gender as a moderator in a college sample. Addictive Behaviors 29, 843-849. Glanz, K., Rimer, B., & Lewis, F. (Eds.). (2002). Health behavior and health education: Theory, research, and practice. San Francisco, CA: Jossey-Bass. Glassman, T., Werch, C., Jobli, E., & Bian, H. (2007). Alcohol-related fan behavior on college football game day Journal of American College Health, 56, 255-260. 154

PAGE 155

Goksen, F. (2002). Normative vs. attitudinal considerations in breastfeeding behavior: Multifaceted social influences in a developing country context. Social Science & Medicine, 54, 1743-1753. Graham, J. W., Marks, G., & Hansen, W. B. (1991). Social influence processes affecting adolescent substance use. Journal of Applied Psychology, 76, 291-298. Gruenewald, P. J., Millar, A.B., & Roeper, P. (1996). Access to alcohol: Geography and prevention for local communities. Alcohol Health & Research World, 20 244-251. Gruenewald, P. J., Ponicki, W. R., & Hold er, H. D. (1993). The relationship of outlet densities to alcohol consumption: A ti me series cross-sectional analysis. Alcoholism: Clinical and Experimental Research, 17 38-47. Gruenewald, P., Ponicki, W., Holder, H., & Romelsj, A. (2006). Alcohol prices, beverage quality, and the demand for alc ohol: Quality substitutions and price elasticities. Alcoholism: Clinical and Experimental Research, 30 96-105. Haines, M. (1998). Social norms: A wellne ss model for health promotion in higher education. Wellness Management, 14, 1-10. Haines, M., & Spear, S. F. (1996). Changing the perception of the norm: A strategy to decrease binge drinking among college students. Journal of American College Health, 45, 134-140. Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2006). Multivariate Data Analysis. (6th ed.). Upper Saddle River, NJ: Pearson Prentice Hall. Hansen, W. B., & Graham, J. W. (1991). Prev enting alcohol, marijuana, and cigarette use among adolescents: Peer pressure resi stance training versus establishing conservative norms. Preventive Medicine, 20, 414-430. Haun, J., Glassman, T., Dodd, V., & Young, G. (2007). Game-day survey results: Looking at football fan alcohol-related behaviors. American Journal of Health Education, 38, 91-96. Higher Education Center for Alcohol and Other Drug Abuse and Violence Prevention. (1997). Be vocal, be visible, be visionary: Recommendations for college and university presidents on alcohol and other drug prevention. Retrieved July 17, 2006, from http://www.higheredcenter .org/pubs/plgvisionary.pdf Higher Education Center for Alcohol and Other Drug Abuse and Violence Prevention. (2000). Note to the field: On binge dr inking. Retrieved January 18, 2008, from http://www.higheredcenter.or g/press-releases/001020.html 155

PAGE 156

Hingson, R. W., Heeren, T., Winter, M., & Wech sler, H. (2005). Magnitude of alcoholrelated mortality and morbidity among U.S. college students ages 18-24: Changes from 1998 to 2001. Annual Review of Public Health, 26, 259-279. Hingson, R., Heeren, T., Zakocs, R., Kopstein, A., & Wechsler, H. (2002). Magnitude of alcohol-related mortality and morbidity among U.S. college students ages 18. Journal of Studies on Alcohol, 63, 136. Holder, H. D., Gruenewald, P. J., Ponicki, W. R., Treno, A. J., Grube, J. W., Saltz, R. F., Voas, R. B., Reynolds, R. I., Davis, J., Sanchez, L., Gaumont, G., & Roeper, P. (2000). Effect of community-based in terventions on highrisk drinking and alcohol-related injuries. Journal of the American Medical Association 284 23412347. Jackson, K. M., Sher, K. J., Gotham, H. J., & Wood, P. K. (2001). Transitioning into and out of large-effect drinking in young adulthood. Journal of Abnormal Psychology, 110, 378. Jackson, K. M., Sher, K. J., & Park, A. (2005). Drinking amo ng college students: Consumption and consequences. In M. Galanter (Ed.), Recent developments in alcoholism, Vol. 17: Alcohol pr oblems in adolescents and young adults (pp. 85-117). New York: Springer. Jennings-Dozier, K. (1999). Predicting inte ntions to obtain a pap smear among African American and Latino women: Testing the theory of planned behavior. Nursing Research, 48, 198-205. Johnson, R. B., & Onwuegbuzie, A .J. (2004) Mixed methods research: A research paradigm whose time has come. Educational Researcher, 33, 14-26. Johnston, K. L., & White, K .M. (2004). Beliefs underlying bi nge-drinking in young female undergraduate students: A theory of planned behavior perspective. Youth Studies Australia, 23, 22-30. Johnston, L. D., OMalley, P.M., & Bachman, J.G. (2005). Monitoring the Future national survey results on drug use, 1975. Volume I: Secondary school students (NIH Publication No. 01, 492 pp). Bethesda, MD: National Institute on Drug Abuse. Kashima, Y., Gallois, C., & McCamish, M. (1993). The theory of reasoned action and cooperative behavior: It takes two to use a condom. British Journal of Social Psychology, 32, 227-239. Katz, J., & Kilbourne, J. (2004). Spin the bottle: S ex, lies, & alcohol [Educational DVD]. (Available from the Media Educat ion Foundation, 60 Masonic Street, Northampton, MA 01060). 156

PAGE 157

Kline, R. B. (1998). Principles and practices of structural equation modeling New York: Guilford. Komro, K. A., Perry, C.L., Munson, K. A., Stigler, M. H. & Farbakhsh, K. (2004). Reliability and validity of self-report meas ures to evaluate drug and violence prevention programs. Journal of Child & Adolescent Substance Abuse, 1 3, 17-51. Korcuska, J. S., & Thombs, D. L. (2003) Gende r role conflict and sex-specific drinking norms: Relationship to alcohol us e in undergraduate women and men. Journal of College Student Development 44, 204-216. Kotler, P., Roberto, N., & Lee, N. (2002). Social marketing improvi ng the quality of life (Rev. ed.). Thousand Oaks, CA: Sage. Kotler, P., & Zaltman, G. (1971). Social mark eting: An approach to planned social change. Journal of Marketing, 35, 3-12. Krejcie, R. V., & Morgan, D.W. (1970). Dete rmining sample size for research activities. Educational and Psychological Measurement, 30, 607-610. Kuo, M., Dowdall, G., Koss, M., & Wechsler, H. (2004). Correlates of rape while intoxicated in a national sample of college women. Journal of Studies on Alcohol, 65, 37-45. Kuo, M., Wechsler, W., Greenbe rg, P., & Lee, H. (2003). The marketing of alcohol to college students: The role of lo w prices and special promotions. American Journal of Preventive Medicine, 25, 204-211. Kuther, T. L. (2002). Rational decision pe rspectives on alcohol consumption by youth: Revising the theory of planned behavior. Addictive Behaviors, 27 35-47. Kypri, K., & Gallagher, S. J. (2003). Incentives to increase participation in an internet survey of alcohol use: A controlled experiment. Alcohol & Alcoholism, 38, 437441. Lahaut, V. M. H. C. J., Janse, H. A. M., Mheen, D. V. D., & Garretsen, H .F. L. (2002). Non-response bias in a sample survey on alcohol consumption. Alcohol & Alcoholism, 37, 256-260. Lange, J. E., & Voas, R. B. (2001). Defini ng binge drinking quantities through resulting blood alcohol concentrations. Psychology of Addictive Behaviors, 15, 310-316. 157

PAGE 158

Langford, L. (2006). Preventing violence and promoting safety in higher education settings: Overview of a comprehensive approach Newton, MA: The Higher Education Center for Alcohol and Ot her Drug Prevention. Funded by the U.S. Department of Education. Lemmens, P. H. H. M., Tan, E. S. & Knibbe, R. A. (1988). Bias due to non-response in a Dutch survey on alcohol consumption, British Journal of Addiction, 83, 10691077. Linkenbach, J. W. (1998). Building a Bridge: Applying th e social norms approach to sexual health. The BACCHUS & GAMMA Se xual Responsibility Manual. (Available from the Bacchus & Gamma Peer Education Network, Minneapolis, MN). Listiak, A. (1974). Legitimate deviance and so cial class: Bar behavior during Grey Cup week. Sociological Focus, 7, 13-43. Maibach, E. W., Rothschild, M.L., & Novelli, W. D. (2002). Social Marketing. In K. Glanz, B. Rimer, & F. Lewis (Eds.), Health behavior and health education: Theory, research, and practice (pp. 437-461). San Francisco, CA: Jossey-Bass. Marlett, G. A., Baer, J. S., Kivlahan, D. R ., Dimeff, L. A., Larimer, M. E., Quigley, L. A., Somers, J. M., & Williams, E. (1998). Screening and brief intervention for high-risk college student dr inkers: Results from a 2-year follow-up assessment. Journal of Consulting and Clinical Psychology, 66, 604-615. McCabe, S. E., Boyd, C. J., Couper, M. P., Crawford, S., & DArcy, H. (2002). Mode effects for collecting alcohol and other drug use data: Web and U.S. mail. Journal of Studies on Alcohol, 63, 755-761. Mertler, C., & Vannatta, R. (2002). Advanced and multivariate statistical methods (2nd ed.). Los Angeles, CA: Pyrczak Publishing. Michels, T. C., & Kugler, J. P. (1998). Pred icting exercise in olde r Americans: Using the theory of planned behavior. Military Medicine, 163, 524-529. Midanik L. Validity of self report alcohol use: A literature review and assessment. (1988). British Journal of Addiction, 83, 1019-1030. Miller, T. R., Levy, D. T., Spicer, R. S., & Taylor, D. M. (2006). Societal costs of underage drinking. Journal of Studies on Alcohol, 67, 519-527. Montano, D. E., & Kasprzyk, D. (2002). The theo ry of reasoned acti on and the Theory of Planned Behavior. In K. Glanz, B. Rimer, & F. Lewis (Eds.), Health behavior and health education: Theory, research, and practice (pp. 67-98). San Francisco, CA: Jossey-Bass. 158

PAGE 159

Montano, D. E., Kasprzyk, D., & Taplin, S. H. (1997). The theory of reasoned action and the Theory of Planned Behavior. In K. Gl anz, F. M. Lewis, & B.K. Rimer (Eds.), Health Behavior and health educatio n: Theory, research, and practice (pp. 85112). San Francisco, CA: Jossey-Bass. Mosher, J. F. (1999). Alcohol policy and the young adult: Establishing priorities, building partnerships, overcoming barriers. Addiction, 94, 358-369. Naimi, T., Brewer, R. D., Mokdad, A., Denny, C., Serdula, M. K., & Marks, J. S. (2003). Binge drinking among US adults. Journal of the American Medical Association, 289, 70-74. Naimi, T. S., Lipscomb, L. E., Brewer, R. D ., & Gilbert, B. C. (2003). Binge drinking in the preconception period and the risk of unintended pregnancy: Implications for women and their children. Pediatrics 111, 1136-1141. National Center on Addiction and Substance Abuse at Columbia University. (1999). Dangerous liaisons: Substance abuse and sex. Funded by the Henry J. Kaiser Family Foundation and The Carnegie Cor poration of New York. Retrieved July 14, 2006, from http://www.casacolumbia.org/Absolutenm/articlefiles/Dangerous_Liasons_12_7_ 99.pdf National Center on Addiction and Substance Abuse. (2006). Women under the influence. Chapter one: Pathways to substance abuse among girls and women Baltimore, MD: Johns Hopkins University Press. National Collegiate Athle tic Association. (2003). Report on the sportsmanship and fan behavior summit Retrieved June 1, 2004, from http://www.ncaa.org/sportsmanship/s portsmanshipFanBehavior/report.pdf National Institute on Alcohol Abuse and Alcoholism. (1995). Alcohol Alert Number 29: Message from Secretary of Health and Human Services. Retrieved on July 13, 2006 from http://www.niaaa.nih.gov/ publications/aa29.htm National Institute on Alcohol Abuse and Alcoholism. (1997). Alcohol alert number 38: Alcohol, violence, and aggression. Retrieved on September 27, 2006, from http://pubs.niaaa.nih.gov/publications/aa38.htm National Institute on Alcoholism and Alcohol Abuse. (2002a). College factors that influence drinking Retrieved October 7, 2006, from http://www.collegedrinkingprevention.gov/SupportingResearch/Journal/presley.as px. 159

PAGE 160

National Institute on Alcoholism and Alcohol Abuse. (2002b). A call to action: Changing the culture of drinking at U.S. colleges Retrieved July 27, 2006, from http://www.collegedrinkingprevention.gov/ Reports/TaskForce/TaskForce_TOC.a spx. National Institute on Alcohol Abuse and Alcoholism. (2002c). Alcohol Alert Number 57: Alcohol and HIV/AIDS. Retrieved September 6, 2006, from http://pubs.niaaa.nih.gov/publications/aa57.htm National Institute on Alcohol Abuse and Alcoholism. (2002d). Alcohol Alert Number 58: Changing the Culture of Campus Drinking. Retrieved August 7, 2006 from http://pubs.niaaa.nih.gov/publications/aa58.htm National Institute on Alcohol Abuse and Alcoholism. (2004a, Winter). Copy Editor, 3, 14. Retrieved on September 7, 2006 from http://pubs.niaaa.nih.gov/publications/N ewsletter/winter2004/Newsletter_Number 3.pdf National Institute on Alcohol Abuse and Alcoholism. (2004b). Alcohol Alert Number 62: Alcohol-An Important Womens Health Issue. Retrieved August 7, 2006 from http://pubs.niaaa.nih.gov/publications/aa62/aa62.htm National Institute on Alcohol Abuse and Alcoholism. (2006). Alcohol Alert Number 68: Young Adult Drinking. Retrieved August 7, 2006 from http://pubs.niaaa.nih.gov/publications/aa68/aa68.htm Neighbors C., Oster-Aaland L., & Bergstrom, R. L. (2006). Event-and context-specific normative misperceptions and high-risk drinking: 21st birthday celebrations and football tailgating. Journal of Studies on Alcohol, 67, 282-289. Neal, J. N., & Fromme, K. (2007). Hookem horns and heavy drinki ng: Alcohol use and collegiate sports. Addictive Behaviors, 32, 2681-2693. Nelson, T. F., & Wechsler, H. (2002). School sp irits: Alcohol and collegiate sports fans. Addictive Behaviors, 28, 1-11. Norman, P., Bennett, P., & Lewis, H. ( 1998). Understanding binge drinking among young people: An application of the Theory of Planned Behavior. Health Education Research, 13, 163-169. OCallaghan, F., Chant, D., Callan, V., & Bagl ioni, A. (1997). Models of alcohol use by young adults: An examination of vari ous attitude-beh avior theories. Journal of Studies on Alcohol, 58, 502-507. 160

PAGE 161

OMalley, P. M., & Johnston, L. D. (2002). Epidemiology of alcohol and other drug use among American college students. Journal of Studies on Alcohol, 63(Suppl. 14), 23-39. Perkins, H. W. (2002). Surveying the damage: A review of research on consequences of alcohol misuse in college populations. Journal of Studies on Alcohol 63(Suppl. 14), 91-100. Perkins, H. W. (2003). The social norms approach to prevention school and college age substance abuse: A handbook for educat ors, counselors, and clinicians. San Francisco, CA: Jossey-Bass. Perkins, H. W., & Berkowitz, A.D. (1986). Perceiving the community norms of alcohol use among students: Some research impli cations for campus alcohol education programming. The International Journal of Addictions 21, 961-976. Portney, L. G., & Watkins, M.P. (2000). Foundations of clinical research applications to practice (2nd ed.). Upper Saddle River, NJ: Prentice Hall Health. Prentice, D. A., & Miller, D. T. (1993). Pl uralistic ignorance and alcohol use on campus: Some consequences of misperceiving the social norm. Journal of Personality and Social Psychology, 64, 243-256. Presley, C. A., Leichliter, J. S., & Meilman, P. W. (1999). Alcohol and drugs on American campuses: Findings from 1995, 1996, and 1997 A report to college presidents. Carbondale, IL: Core Institute Southern Illinois University. Presley, C. A., Meilman, P. W., Cashi n, J. R., & Leichliter, J. S. (1997). Alcohol and drugs on American college campuses: Issues of violence and harassment. Carbondale, IL: Core Institute, S outhern Illinois University. Presley, C. A., Meilman, P. W., & Leichliter, J. S. (2002). College fa ctors that influence drinking. Journal of Studies on Alcohol, 63 (Suppl. 14), 82. Reinecke, J., Schmidt, P., & Ajzen, I. (1996) Application of the theory of reasoned action to adolescents c ondom use: A panel study. Journal of Applied Social Psychology, 26, 749-772. Schafter, A. (2002). Girls and alcohol: Closing the gender gap? The National Clearinghouse for Alcohol and Drug Info rmation, Retrieved June 12, 2004, from http://www.health.org /newsroom/rep/188.htm Schlegel, R. P., D'Avernas, J. R., Zanna, M. P., Decourville, N. H., & Manske, S. R. (1992). Problem drinking: A problem fo r the theory of reasoned action? Journal of Applied Social Psychology, 22, 358-385. 161

PAGE 162

Schmidt, W. C. (1997). World-wide web surv ey research: Benefits, potential problems, and solutions. Behaviour Research Methods, Instruments and Computers, 29, 274-279. Sheeran, P., & Abraham, C. (2003). Mediator of moderators: Tem poral stability of intention and the inten tion-behavior relation. Personality and Social Psychology Bulletin, 29 205-215. Sher, K. J., Trull, T. J., Bartholow, B. D., & Vieth, A. (1999). Personality and alcoholism: Issues, methods, and etiologica l processes. In K.E. Leonard & H.T. Blane (Eds.), Psychological theories of drinking and alcoholism (pp. 54-105). New York: Guilford. Smith, G. S., Branas, C. C., & Miller, T. R. (1999). Fatal non-traffic injuries involving alcohol: A meta-analysis. Annals of Emergency Medicine, 33, 659-668. Sullivan, M., & Risler, E. (2002). Understa nding college alcohol abuse and academic performance: Selecting appropr iate intervention strategies. Journal of College Counseling, 5, 114-124. Swirko, C. (2000). Police are cracking dow n on the statistics. Retrieved on July 16, 2004, from http://gainesvillesunarchives.colony1.ne t/sunarchives/public _html/news/almanac_ 00/topics/crime.html Thombs, D., Dotterer, S., Olds, S., Sharp, K, & Raub C. G. (2005). A close look at why one social norms campaign di d not reduce student drinking. American Journal of College Health, 53, 61-67. Thombs, D. L., Olds, R. S., & Snyder, B. M. (2003). Field assessm ent of BAC data to study late-night college drinking. Journal of Studies on Alcohol, 64, 322-330. Timmreck, T. C. (1994). An introduction to epidemiology. Boston, MA: Jones and Bartlett. Toomey, T. L., & Wagenaar, A. C. (2002). Environmental policies to reduce college drinking: Options and research findings. Journal of Studies on Alcohol, 63 (Suppl. 14), 193. Trafimow, D. (1996). The importance of attitu des in the prediction of college students intentions to drink. Journal of Applied Social Psychology, 26, 2167-2188. Trafimow, D., & Finlay, K. (1996). The importa nce of subjective norm s for a minority of people. Personality and Social Ps ychology Bulletin, 22, 820-828. 162

PAGE 163

Treno, A. J., Gruenewald, P. J., & Johnson, F. W. (2001). Alcohol avai lability and injury: The role of local outlet densities. Alcoholism: Clinical and Experimental Research, 25, 1467-1471. University of Florida. (2004). Biennial review. Retrieved July 20, 2006, from http://www.ufsa.ufl.edu/ovp/ alcohol/resources/br04.pdf UF falls to No.18 party school. (2005, August 23). Gainesville Sun p. 4A. Vicary, J. R., & Karshin, C. M. (2002). College alcohol abuse: A revi ew of the problems, issues, and prevention approaches. The Journal of Primary Prevention, 22, 299-331. Voas, R. B., Furr-Holder, D., Lauer, E., Bri ght, K., Johnson, M. B., & Miller, B. (2006). Portal surveys of time-out drinking loca tions: A tool for studying binge drinking and AOD use. Evaluation Review, 30, 44-65. Wagenaar, A. C. (1983). Alcohol, young drivers, and traf fic accidents: Effects of minimum-age laws. Lexington, MA; Lexington. Wagenaar, A. C., Harwood, E.M., Toomey, R.L., Denk, C.E., & Zander K.M. (2000). Public opinion on alcohol policies in the United States: Results from a national survey. Journal of Public Health Policy, 21, 303-327. Wall, A. M., Hinson, R. E., & McKee, S. A. (1998). Alcohol outcome expectancies, attitudes toward drinking and the Theory of Planned Behavior. Journal of Studies on Alcohol, 59, 409-419. Walters, T. S., & Bennett, M. E. (2000). Addressing drinking among college students: A review of the empirical literature. Alcoholism Treatment Quarterly, 18, 61-77. Wechsler, H., Lee, J. E., Kuo, M., Seibring, M., Nelson, T. F., & Lee, H. (2002). Trends in college binge drinking during a period of increased prevention efforts: Findings from four Harvard School of Public Health College alcohol study surveys: 19932001. Journal of American College Health, 50, 203-217. Wechsler, H., & Nelson, T. F. (2001). Binge dr inking and the American college student: Whats five drinks? Psychology of Addictive Behaviors, 15, 287-291. Wechsler, H., Nelson, T. F., Lee, J. E., Se ibring, M., Lewis, C., & Keeling, R. (2003). Perception and Reality: A national eval uation of social norms marketing interventions to reduce st udents heavy alcohol use. Journal of Studies on Alcohol, 64, 484-494. 163

PAGE 164

Weitzman, E. R., Folkman, A., Folkman, K. L. & Wechsler, H. (2002). The relationship of alcohol outlet density to heavy and frequent drinking and drinking-related problems among college students at eight universities. Health and Place, 9, 1-6. Werch, C. E., Pappas, D. M., Carlson, J. M., DiClemente, C. C., Chally, P. S., & Sinder, J. A. (2000). Results of a social norm in tervention to prevent binge drinking among first-year residential college students. Journal of American College Health, 49, 85-92. White, A. M., Kraus, C. L., & Swartzwelder, H. S. (2006). Many co llege freshmen drink at levels far beyond the binge threshold. Alcoholism: Clinical and Experimental Research, 30, 1006-1010. White, H. R., Johnson, V., & Buyske, S. (2000). Parental modeling and parenting behavior effects on offspri ng alcohol and cigarette use: A growth curve analysis. Journal of Substance Abuse, 12, 287. Wild, T. C., Cunningham, J. & Adlaf, E. (2001). Nonresponse in a follow-up to a representative telephone survey of adult drinkers. Journal of Studies on Alcohol 62, 257-261. Witmer, D. F., Colman, R.W., & Katzman, S. L. (1999). From paper-and-pencil to screen-and-keyboard: Toward a methodology for survey research on the internet. In S. Jones (Ed.), Doing internet research: Crit ical issues and methods for examining the net (pp. 145-161). Thousand Oaks, CA: Sage. Word, R. (2005, May 29). UF chief fights binge drinking. Orlando Sentinel, pp. B5. Zimmerman, R. (1997). Social marketing strategies for campus prevention of alcohol and other drug problems. Newton, MA: Higher Education Center for Alcohol and Other Drug Prevention. Zucker, R. A., Fitzgerald, H. E., & Moses, H. D. (1995). Emergence of alcohol problems and the several alcoholisms: A developmen tal perspective on e tiologic theory and life course trajectory. In D. Cicchetti (Ed.), Developmental psychopathology: Vol. 2. Risk, disorder, and adaptation (pp. 677-711). New York: John Wiley & Sons. 164

PAGE 165

BIOGRAPHICAL SKETCH Tavis Glassman was born on July 17, 1971 in Greeley, Colorado. He grew up in Ohio and attended high school in the Toledo P ublic School system at Roy C Start. He began his undergraduate studi es at the University of Toledo, where he obtained a bachelors degree in education in health and physical educat ion. During that time period, he coordinated and conducted a variety of h ealth and safety courses while working at American Red Cross Greater Toledo Chapter. In 1995, Tavis last year at the Red Cross, he earned the Tiffany Award, which is awarded to the employee of the year. He later graduated magna cum laude and was awarded the Outstanding Undergraduate Student of the Year by the Health Education Department Tavis immediately started his graduate work at the University of Toledo pursuing a masters degree in science and education speci alizing in public health. During that time period he taught sex education in the Toledo Public School system. He also taught a variety of health and safety courses at the university including lifeguarding, personal conditioning, and swimming. Upon graduation, Tavis moved to Columbus where he earned a Master of Public Health specializing in health behavior/health promotion in the School of Public Health at The Ohio State University (OSU). While at tending graduate school, he worked at the Student Wellness Center as the Sexual Health Coordinator. He successfully obtained a grant from the Columbus Health Depart ment designed to reduce high-risk sexual behavior among fraternity and so rority members at OSU. Prior to graduation, he earned his status as a Certified Health Education Specialist. 165

PAGE 166

166 Shortly after graduating, in 2000, Tavis m oved to Gainesville, Florida to accept employment as the Coordinator of Alcohol & Other Drug Prevention at the University of Florida. During his tenure, he was awarded the Whos Who in Prevention in Prevention Leadership by the state of Florida and the Community Recognition Award for outstanding prevention service awarded by a lo cal community coalition. In addition, he co-authored a grant from the U.S. Department of Education Grant which was ranked number one out of a 100 applicants. In the summer of 2003, Tavis was accepte d into the doctoral program in the College of Health and Human Performance to pursue a Ph.D in health education and behavior. Throughout his graduate work, Ta vis was involved in a variety academic and service initiatives. In the 2004 school year, he worked as a doc toral fellow at the University of Florida Addictive & Health Be haviors Research Institute in Jacksonville Florida. He was involved in several research projects which culminated in presentations at national meetings and manuscripts. In 2006, he received the Outstanding Graduate Student Award by the Department of Health Education and Behavior. Currently, Tavis maintains his position as Coordinator of Alcohol & Other Drug Prevention at the University of Florida, wh ere he continues to c onduct research on game day drinking and othe r related areas.