Citation
Audience Influence on Health Information Avoidance

Material Information

Title:
Audience Influence on Health Information Avoidance
Creator:
Lipsey, Nikolette P
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (125 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Psychology
Committee Chair:
Shepperd,James A
Committee Co-Chair:
Ratliff,Kate
Committee Members:
Heesacker,Martin
Weigold,Michael Fredrick

Subjects

Subjects / Keywords:
avoidance -- health -- pathway
Psychology -- Dissertations, Academic -- UF
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Psychology thesis, Ph.D.

Notes

Abstract:
The decision to seek or avoid health information is an important, sometimes life-altering decision. Although many studies explore personal reasons for avoiding information, limited research exists exploring the influence of other people on peoples' decision to seek or avoid information. In five studies (total N = 1,709), I examine three proposed pathways by which audiences influence people's health information decisions: proactive impression management (Studies 1a and 1b), capacity to harm (Study 2), and social norms (Studies 3a and 3b). Results provide mixed support for the proactive impression management pathway, support for the capacity to harm pathway, and no support for the social norms pathway. Overall, results suggest that the health information context and manipulation itself may strongly influence whether audiences affect health information decisions. ( en )
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, 2019.
Local:
Adviser: Shepperd,James A.
Local:
Co-adviser: Ratliff,Kate.
Statement of Responsibility:
by Nikolette P Lipsey.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
LD1780 2019 ( lcc )

UFDC Membership

Aggregations:
University of Florida Theses & Dissertations

Downloads

This item has the following downloads:


Full Text

PAGE 1

AUDIENCE INFLUENCE ON HEALTH INFORMATION AVOIDANCE By NIKOLETTE P. LIPSEY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR O F PHILOSOPHY UNIVERSITY OF FLORIDA 2019

PAGE 2

2019 Nikolette Lipsey

PAGE 3

To my family, friends, and cats

PAGE 4

4 ACKNOWLEDGMENTS I extend my sincerest gratitude to my advisor James Shepperd, Ph.D, for his continued support, advice, and feedback. I also thank the other members of my dissertation committee Kate Ratliff, Ph.D, Micha el Weigold, Ph.D, and Martin Heesacker, Ph.D for their valuable contribution s. I am eternally grateful for my uncondit ional love and guidance I would not be where I am today without your solidarity and support. Finally, I would like to thank the University of Florida Psychology Department for the resources they provided to facilitate my research.

PAGE 5

5 TABLE OF CONTENTS p age ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST O F TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ 11 Information Avoidance ................................ ................................ ........................ 11 Interpersonal Influences ................................ ................................ ..................... 13 The P roactive Impression Management Pathway ................................ .............. 14 The Capacity to Harm Pathway ................................ ................................ .......... 15 The Social Norms Pathway ................................ ................................ ................ 16 Overview and Hypotheses ................................ ................................ .................. 18 2 STUDY 1A ................................ ................................ ................................ .......... 20 Method ................................ ................................ ................................ ............... 20 Parti cipants ................................ ................................ .............................. 20 Procedure ................................ ................................ ................................ 20 Measures ................................ ................................ ................................ 21 Results and Discussion ................................ ................................ ...................... 24 Zero order Correlations ................................ ................................ ............ 24 Manipulation Check ................................ ................................ ................. 24 Information Avoidance ................................ ................................ ............. 26 Exploring Other Predictors of Avoidance ................................ ................. 26 3 STUDY 1B ................................ ................................ ................................ .......... 30 Method ................................ ................................ ................................ ............... 31 Participants ................................ ................................ .............................. 31 Procedure ................................ ................................ ................................ 31 Measures ................................ ................................ ................................ 32 Results and D iscussion ................................ ................................ ...................... 34 Zero o rder Correlations ................................ ................................ ............ 34 Manipulation Check ................................ ................................ ................. 34 Information Avoid ance ................................ ................................ ............. 35 Exploring Other Predictors of Avoidance ................................ ................. 35 Exploratory Moderation Analyses ................................ ............................ 37 Summary of Findings ................................ ................................ ............... 38

PAGE 6

6 4 STUDY 2 ................................ ................................ ................................ ............ 40 Method ................................ ................................ ................................ ............... 41 Participants ................................ ................................ .............................. 41 Procedure ................................ ................................ ................................ 42 Results and Discus sion ................................ ................................ ...................... 45 Zero order Correlations ................................ ................................ ............ 45 Manipulation Check ................................ ................................ ................. 45 Information Avoidanc e ................................ ................................ ............. 46 Exploratory Mediation Analysis ................................ ................................ 47 Other Predictors of Avoidance ................................ ................................ 51 Summary of Findings ................................ ................................ ............... 53 5 STUDY 3A ................................ ................................ ................................ .......... 56 Method ................................ ................................ ................................ ............... 57 Participants ................................ ................................ .............................. 57 Procedure ................................ ................................ ................................ 57 Materials ................................ ................................ ................................ .. 57 Results and Discussion ................................ ................................ ...................... 59 Zero order C orre lations ................................ ................................ ............ 59 Manipulation C heck ................................ ................................ ................. 59 Information Avoidance ................................ ................................ ............. 60 Other Predictors of Avoidance ................................ ................................ 61 6 STUDY 3B ................................ ................................ ................................ .......... 64 Method ................................ ................................ ................................ ............... 64 Participants ................................ ................................ .............................. 64 Procedure ................................ ................................ ................................ 64 Materials ................................ ................................ ................................ .. 65 Results and Discussion ................................ ................................ ...................... 67 Zero order Correlations ................................ ................................ ............ 67 Man ipulation Check ................................ ................................ ................. 67 Information Avoidance ................................ ................................ ............. 68 Other Predictors of Avoidance ................................ ................................ 69 Ex plorat ory Condition Contrasts ................................ .............................. 71 Summary of Findings ................................ ................................ ............... 73 7 GENERAL DISCUSSION ................................ ................................ ................... 74 Proactive Impression Management ................................ ................................ .... 74 Capacity to Harm ................................ ................................ ................................ 76 Injunctive and Descriptive Norms ................................ ................................ ....... 78 Limitations ................................ ................................ ................................ .......... 80 Moderators of Interpersonal Information Avoidance ................................ ........... 82 Conclusion ................................ ................................ ................................ .......... 84

PAGE 7

7 APPENDIX A STUDY 1A MATERIALS ................................ ................................ ..................... 86 B STUDY 1B MATERIALS ................................ ................................ ..................... 98 C STUDY 2 MATERIALS ................................ ................................ ..................... 100 D STUDY 3A MATERIALS ................................ ................................ ................... 109 E STUDY 3B MATERIALS ................................ ................................ ................... 112 LIST OF REFERENC ES ................................ ................................ ............................. 121 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 125

PAGE 8

8 LIST OF TABLES Table page 2 1 Zero order correlations for Study 1a ................................ ................................ ... 24 2 2 Manipulation check results Study 1a ................................ ................................ .. 25 2 3 Predictors of Information Avoidance in Study 1a ................................ ................ 27 2 4 Predictors of Informati on Avoidance Proclivity in Study 1a ................................ 28 3 1 Zero order correlations for Study 1b ................................ ................................ ... 34 3 2 Predictors of Information Avoidance in Study 1b ................................ ................ 36 3 3 Predictors of Information Avoidance Proclivity in Study 1b ................................ 37 4 1 Zero order correlations for Study 2 ................................ ................................ ..... 45 4 2 Study 2 manipulation check means by condition ................................ ................ 46 4 3 Mediation results for dichotomous choice information decision. ......................... 49 4 4 Predictors of Information Avoidance in Study 2 ................................ .................. 52 4 5 Predictors of Information Avoidance Proclivity in Study 2 ................................ ... 53 5 1 Zero order correlations for Study 3a ................................ ................................ ... 59 5 2 Study 3a manipulation c heck means by condition ................................ .............. 60 5 3 Predictors of Information Avoidance in S tudy 3a ................................ ................ 62 5 4 Predictors of Information Avoidance Proclivity in Study 3a ................................ 62 6 1 Zero order correlations for Study 3b ................................ ................................ ... 67 6 2 Study 3b manipulation check means by condition ................................ .............. 68 6 3 Predictors of Information Avoidance in Study 3b ................................ ................ 70 6 4 Predictors of Information Avoidance Proclivity in Study 3b ................................ 70 6 5 Predictors of Information Avoidance in Study 3b ................................ ................ 72 6 6 Predictors of Information Avoidance Proclivity in Study 3b ................................ 72

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Audience influence information a voidance pathways. ................................ ........ 13 4 1 Exploratory mediation pathways. ................................ ................................ ........ 50

PAGE 10

10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfill ment of the Requirements for the Degree of Doctor of Philosophy AUDIENCE INFLUENCE ON HEALTH INFORMATION AVOIDANCE By Nikolette P. Lipsey August 2019 Chair: James Shepperd Major: Psychology The decision to seek or avoid health information is an importa nt, sometimes life altering decision. Although many studies explore personal reasons for avoiding decisio n to seek or avoid information. In five studies (total N = 1,7 09), I examine three proactive impression management (Studies 1a and 1b), capacity to harm (Study 2), and so cial norms (Studies 3a and 3b). Results provide mixed support for the proactive impression management pathway, support for the capacity to harm pathway, and no support for the social norms pathway. Overall, results suggest that the health information conte xt and manipulation itself may strongly influence whether audi ences affect health information decisions.

PAGE 11

11 CHAPTER 1 INTRODUCTION Why do people avoid information that could save their lives? Research across several disciplines finds that avoiding health information occurs regularly, even when the information could be vital to survival (Howell, Lipsey, & Shepperd, in press). Although, people often avoid health information for personal reasons, they sometimes avoid health information for interpersonal reasons. That is, concerns with perceptions and responses of audience s may lead people to avoid information they may otherwise have sough t. On the flipside, audiences may sometimes prompt people to seek information they may otherwise have avoided. The pr esent studies examine audience influences on health information seeking and avoidance. Information Avoidance Information avoidance refers to any behavior designed to delay or prevent the acquisition of available information (Sweeny et al., 2010). Information avoidance is prevalent people avoid medical test results (e.g. Howe ll & Shepperd, 2013; Melnyk & Shepperd, 2012), evaluations by others (Howell, Crosier, & Shepperd, 2014, Study 4; Sweeny & Miller, 2012, Study 2), and information about others (Afifi & Weiner, 2006; Yaniv, Benador, & Sagi, 2004). In a study surveying a nat ionally representative sample of US adults, 25% of participan ts agre Crosier, & Shepperd, 2014). However, measures of avoidance intentions and avoidance behavior do not always yield the same results. For example 9 7 % of participants in one study reported intentions to undergo genetic testing for hereditary non polyposis

PAGE 12

12 colorectal cancer but 51% of participants actually underwent genetic testing ( Hadley et al., 2003 ). Although information avoidance is not always harmful (Barbour, Rintamaki, Ramsey, & Brashers, 2012; Shipp et al., 2004), it sometimes has adverse consequences, especially in health cont exts (Cutler & Hodgson 2003; Knowles, Lowery, Chow, & Unzueta, 2014; Sweeny et al., 2010). For example, many types of cancer have high survival rates and better outcomes if detected and treated early (American Cancer Society, 2015). Avoiding cancer screen ing information can th us have life personally harmful and can risk infection of sexual partners. Yet, as much as 55% of people who tested for HIV failed to return for their results in one study (Hightow et al., 2003). Given the potential negative consequences, why do people avoid information? Theorists propose three broad motives that prompt information avoidance (Sweeny et al., 2010). The first motive is emotional regulation. People sometimes avoid infor mation that may lead to unpleasant emotions (Cutler & Hodgson, 2003), but also avoid information to maintain or boost positive emotions (Shipp et al., 2004). The second motive is obligation. People may avoid information that potentia lly obligates undesire d action (Sweeny et al., 2010). For example, in a series of studies, participants displayed greater avoidance when learning the information would obligate them to undergo an uncomfortable testing procedure or a time consuming pill reg imen (Howell & Shepper d, 2013). The third motive is to maintain valued beliefs. People may avoid information that

PAGE 13

13 threatens their beliefs about themselves, about others, or about their general worldviews) (Howell et al., 2013; Johnston, 1996; Knowles et a l., 2014; Sweeny et al .). Interpersonal Influences The three motives just described represent personal reasons for avoiding information (Sweeny et al., 2010). However, people may also avoid information for interpersonal reasons such as self presentation al concerns (reflectin g concerns with managing how audiences view the person) or audience influence (reflecting either concerns with how audiences will use information, or how audiences will view the information decision itself). Concerns about audiences m ay thus influence peop le to make different information decisions than decisions made in the absence of such concerns. I propose that audiences influence information decisions in via three pathways (Figure 1 1). Figure 1 1 Audience influence information avoidance pathways.

PAGE 14

14 T he first pathway concerns how people managing impressions once the information is known (proactive impression management). The second pathway concerns h ow the audience will use the information that follows the information decision (po tential to harm) t he information th at results from avoiding nor seeking could affect resources and/or image The third pathway concerns how the audience views the information decision (social norms). Both injunctive norms about the information deci sion itself (whether o ther people expect one should seek or avoid the information) and descriptive norms about avoiding/seeking behavior (whether other people tend to seek or avoid under similar circumstances) can influence information avoidance/seeking. I discuss each of these paths in turn. The Proactive Impression Management Pathway P articipants may be concerned with how audiences view them in light of new information. Information may negatively or positively affect how others view the information seeker and people may proac tively manage impressions by avoiding (or learning) the information. Pursuing information (such as undergoing a medical test) carries the risk that other people may gain access to the information. At the very least, knowing the inform ation may affect inter actions with others. For example, people who learn that they have a sexually transmitted infection likely feel obligated to tell their sexual partners. However, this news can create an unfavorable impression in the eyes of the partner s. Additionally, even though people may not tell their parents or friends, having the burden of this knowledge may lead the person to feel guilty, and the guilt may color interactions with others. Thus, the information may affect people public image and th eir interactions with others. Consequently, people may reason that it is best to proactively manage impressions by avoiding the information altogether.

PAGE 15

15 Although this pathway remains unexplored in the literature, research on information disclosure offers h ints regarding how the proactive impression management pathway may play out Information disclosure research demonstrates that people are concerned that audiences may learn certain information about them Thus, people regulate the information they disclose (Delaney, Serovich, & Lim, 2008) For example, some people withhold information about their HIV/AIDS status even to close others for two reasons. First, they are concerned that the information will become public and lead to stigma and ostracism. Second, t hey are concerned that the information may taint their friends and acquaintances (Delan e y, Serovich & Lim, 2008; Murphy, Roberts, & Hoffman, 2002) Notably, choosing not to disclose information is not information avoidance. The actor already knows the wit hheld information. Con ceivably, however, people may avoid learning certain information to avoid concerns of audience disclosure. Revealingly, in one genetic screening study participants indicated that receiving and sharing an undesirable diagnosis with cl ose others might cause others to perceive them as a burden or to desert them (Yaniv, Benador, & Sagi, 2004). These results suggest that people concerned that others may view them negatively or ostracize them may choose to avoid undesirable health informati on to mitigate these i nterpersonal concerns. The Capacity to Harm Pathway People also avoid health information when they are concerned that audiences may use their information to harm them (Lipsey & Shepperd, 2019a; 2019b). Seeking potentially threatening health information car ries the risk that powerful audiences with access to the information may cause people harm. testing suggests that people are concerned with the role powerful others play in health

PAGE 16

16 decisions. Of the part icipants who indicated testing, over 75% believed that getting tested might lead them to lose health insurance (Cutler & Hodgson, 2003). Additionally, although participants in one study expressed high willingness and desire to undergo a medical test that might yield unfavorable results, they expressed significantly less willingness and desire if they believed those unfavorable results might imperil their insurance coverage (Lipsey & Shepperd, 201 9b ). Perhaps more compellingly, recent experimental evidence suggests that powerful audiences play a significant role in actual information avoidance decisions. In two separate studies, participants who took a general health risk assessment were then led to believe that se eking their personal h ealth risk information would make that information available to either health insurers, employers, health researchers, or only the participant. Participants displayed dramatically greater information avoidance when they believed their information could be viewed by powerful audiences (insurers and employers) versus low power audiences (health researchers or only themselves) (Lipsey & Shepperd, 2019a). The Social Norms Pathway Finally, how people view the information decision itself in deci to seek or avoid under similar circumstances (descriptive norms) may influence information decisions. In support of this assertion, research on theory of planned behavior points to the importance of subjective norms (including injunctive and descriptive norms) in both intentions and behavior (Ajzen, 1991; Rivis & Sheeran, 2003). Injunctive norm s, also referred to as prescriptive norms, inform people whether

PAGE 17

17 certain behavior or decisions are approved or disapproved. Descriptive norms inform people whether certain behaviors or decisions are common or uncommon (Cialdini, 2003) Injunctive and descr iptive norms can produce powerful behavior change in areas such as alcohol and drug use, environmental behavior, and gambling (Cialdini; Neighbors, Larimer, & Lewis, 2004; Schultz, Nolan, Cialdini, Goldstein, & Griskevicius, 2007). Fo r example, among house hold residents who consumed above average energy, receiving information about how their energy consumption compared with the average resident (descriptive norm) and a message indicating social approval (injunctive norm) decreased thei r energy consumption ( Schultz et al., 2007). What remains less clear is the role of injunctive and descriptive norms in information decisions avoidance? Some preliminary evidence suggests they do. For example, one s tudy found that receiving injunctive norms about peer expectations predicted a lower likelihood of avoiding risk information (Calhoun, 2012), such that the more participants felt pressure from their friends to know about sexual aggres sion risk, the less li kely they were to avoid the information 1 Additionally, research on feedback seeking in the presence of audiences (Ashford & Northcraft, 1992) demonstrated that people match their feedback seeking to descriptive norms. If participants believed it was the n orm to seek feedback (versus having no norm information or having information that feedback seeking was not the norm), they were less likely to avoid feedback (Ashford & Northcraft, Study 1). This research hints that injunctive and de scriptive norms may 1 Although receiving descriptive norm information predicted less avoidance and predicted less heuristic (superficial) processing of information, it was not a significant predictor of information seeking (Calhoun, 2012).

PAGE 18

18 in fluence information decisions. Yet, it remains unclear whether these norms influence avoidance of health information more generally, and whether combining norms (as Schultz et al., 2007 recommend) elicits more potent effects on inform ation avoidance. Overv iew and Hypotheses The present research examined interpersonal influences on avoidance of health can have potentially life altering consequences (Sweeny et al., 2010). T hus, although it is important to examine the personal influences on health information avoidance (which have been examined relatively thoroughly, Howell et al., in press), it is perhaps equally important to examine the relatively unexplored interpersonal i nfluences on health information avoidance. Hypothesis 1 The proactive impression management pathway impression management concerns will correspond with higher likelihood of avoiding information. Concretely, people prompted to believe that seek ers or avoiders are viewed more favorably will be more likely to seek or avoid, respectively (Study 1a). Additionally, people will be more likely to avoid health information if they are prompted to consider the negative interpersonal consequences of receiv ing unfavorable hea lth information than if they are not (Study 1b). Hypothesis 2 The c apacity to harm pathway People will express greater information avoidance intention if they believe audiences with the capacity to harm them may have access to their i nformation than if they believe that audiences without a capacity to harm will have access to their information (Study 2). Hypothesis 3 The social norms pathway People will seek or avoid information in ways that match injunctive and descriptive norms. People will seek

PAGE 19

19 information more when descriptive and injunctive norms suggest information seeking than when de scriptive and injunctive norms are absent (Studies 3a, 3b). I tested these three hypotheses in five experimental studies. In Studies 1a and 1b, I experimentally manipulated impression management expectations for seeking or avoiding information and for rec eiving unfavorable health information. In Study 2, I experimentally manipulated whether participants believed their health information would be available to audiences with versus without a capacity to harm. In Studies 3a and 3b, I experimentally tested a p otential audience based intervention that encouraged participants to seek potentially threatening information they may have otherwise avoided.

PAGE 20

20 CHAPTER 2 STUDY 1A Study 1 a examined proactive impression management in the context of Sexually Transmitted Inf ection (STI) risk. Participants either learned that seeking STI risk information was favorable or unfavorable, or they were not provided any add itional information. I predicted that information avoidance would be highest in the seeking unfavorable conditio n and lowest in the seeking favorable condition. Method Participants Participants were 453 sexually active college students ( M age = 19.26, age r ange = 18 64; 334 White, 57 Asian, 45 Black, 6 Native American, 5 other (participants could select more than one racial identifier); 331 non Hispanic, 94 Hispanic; 322 female, 109 male, 2 non binary/genderqueer ) recruited from the University of Florida par ticipant pool for research credit and from several psychology classes for extra credit To confirm that particip ants were sexually active, after consenting to participate participants answered the following question: Are you (or have you ever been) sexual ly active? If participants Of the 529 participants who consented to partici pate, 76 identified as non sexually active, and were thus unable to complete the study. A post hoc power analyses for chi square test revealed s ufficient power to detect a moderate effect ( w = .21), power = .99, p = .05. Procedure Participants completed an online Qualtrics study (Appendix A ) After participants consented to participate and confirmed they were sexually active, they answered the Se xual Risk Scale (SRS) followed by demographic items. Participants next received

PAGE 21

21 condition specific in formation, answered manipulation check items, made the decision whether to receive their STI risk information, and completed the Information Avoidance Scal e. Participants then read an online debriefing, selected whether they wished to have their data inclu ded in analysis, and concluded the study. Measures Sexual Risk Scale (SRS). I used a modified version of the SRS (DeHart & Birkimer, 1997) to assess partic item scale questions about attitudes, norms expectation s, in tention s, susceptibility, and substance use. Because the original scale used binary response options (yes /no) that could provide participants with an indication of their results without actually seeking or avoiding the results, I changed the response options to a 7 point agreement scale, where 1 = disagree strongly ; 4 = neither agree nor disagree ; and 7 = agr ee strongly The original scale also asked questions about HIV/AIDs specifically and did not ask abou t other STIs. I chose to make the scale more broadly applicable by changing all questions that referred to HIV/AIDs to refer to STIs. Sample questions incl ude: The idea of using a condom does not appeal to me and I may have had sex with someone who was at risk for a Sexually Transmitted Infection (STI). The reliability of the modified scale was = .92. Perceived STI risk. After completing the SRS, part icipants answered the your The item used a 1 7 scale, where 1 = very low risk, 4 = moderate risk, 7 = very high risk. Demographics. Participants next answered additional demographic questions assessing whether they had ever been tested for an STI (and if so, when they were last

PAGE 22

22 tested), age, gender, sexual orientation, ethnicity, race, education, marital status, employment, health insurance, finan cial security, income, risk orientation ( Maestas & Pollock, 201 0 ) and political orientation. Manipulation. I randomly assigned p articipants to one of three conditions: Seeking Favorable, Seeking Unfavora ble, or Control condition. Participants in all three conditions first read the following text: In a moment, you will have the choice of whether to receive your personal sexually transmitted infection (STI) risk results. In our research we find that half o f people opt to receive risk results and half do not This text provided a descriptive norm about the level of seeking/avoiding, but did not provide clear information about what the participant should do to conform. Participants in the control condition d id not receive any additional information. Participa nts in the Seeking Favorable condition read the following additional information: We also find that people who opt to receive their risk results are viewed favorably by others they are viewed as responsi ble and proactive about their health. Deciding wheth er to receive your risk results is an important decision. In one survey, 89% of participants agreed that they would respect a person who chose to seek personal health risk information. Participants in the Seeking Unfavorable condition read the following ad ditional information : We also find that people who opt not to receive their risk results are viewed favorably by others they are viewed as thoughtful and prudent about their health. Deciding whether to r eceive your risk results is an important decision. I n one survey, 89% of participants agreed that they would respect a person who chose not to seek pe rsonal health risk information. Manipulation check. Participants answered five questions assessing the eff ectiveness of the manipulation. The items assessed t he extent to which participants felt others would favorably view the decision to seek or avoid STI risk results (e.g ., If I

PAGE 23

23 choose not to receive my STI risk results, other people will view me unfavorabl y ; People will respect me if I choose not to recei ve my risk results ). All five items used a 7 point scale (1 = not at all true, 7 = very true ). Information decision. All participants t hen chose whether they wished to receive their risk results Partici pants could either select : Yes, please tell me my STI risk results, or No, I am not interested in learning my STI risk results. After participants made their choice, participants saw an open ended comment box that prompted them to explain why they decided to seek or avoid their information. If participant s opted to receive their STI risk results, I provided them with feedback based on their responses to the SRS Information avoidance scale. Participants also completed an eight item information avoidance sc ale (IA Scale; Howell & Shepperd, 2016) that measu red the proclivity to avoid health risk feedback (e.g ., I would rather not know my risk for sexually transmitted infections [STIs] ). Th e scale allows researchers to modify item wording to meet their resea rch objectives. I modified the scale to assess STI risk information. All eight items used a 7 point Likert scale (1 = strongly disagree, 7 = strongly agree). I averaged responses to the items to create a single index of information = 0.88, M = 2.00, SD = 1.07). In Howell and Sheppe rd (2016), the scale exhibited good test retest validity over two weeks ( r = .68), and good convergent and discriminant validit y The means and standard deviations in the present study was comparable to tho se documented in the original scale paper (Howell & Shepperd, 2016).

PAGE 24

24 Results and Discussion Zero order C orrelations I initially computed the correlations among the predictors (Table 2 1). Information decision did not correlate significantly with any varia bles except the IA scale. However, sex correlated negatively with information avoidance proclivity such that males expressed greater proclivity to avoid STI information than did females. Both actual and perceived STI risk correlated positively with procliv ity to avoid STI information. Table 2 1. Zero orde r correlations for Study 1a Variable 1. 2. 3 4 5 6 7 1 Actual STI risk 2 Perceived STI risk .38** 3. Previous STI testing (1 = yes; 2 = no) .09 .19* 4 Age .12* .18** 12 5 Sex (1 = male, 2 = female) .27** .00 .11 .06 6 Information avoidance scale .26** .14** .07 .04 .23** 7 Avoidance decision (1 = not avoid; 2 = avoid) .09 .00 .04 .09 .05 .48** p < .05, ** p < .01 Manipulation C heck Du items, I conducted five linear regressions assessing manipulation check items by condition (using Helmert contrast codes H1: Seeking Unfavorable 2, Seeking Favorable 1, Co ntrol 1; H2: Seeking Unfavorable 0, Seeking Favorable 1, Control 1) Results indicate that the manipulation was likely ineffective: none of the five manipulation check items varied significantly by condition (Table 2 2).

PAGE 25

25 Table 2 2. Manipulation check r esults Study 1a Manipulation Check Item M SD p 2 If I choose not to receive my STI risk results, other people will view me unfavorably. 3.20 1.93 0.34 (.715) .002 Seeking Unfavorable 3.31 1.94 Seeking Favorable 3.13 2.04 Control 3.17 1.82 I think most people given the choice w ould choose to learn their STI risk results. 5.45 1.55 0.86 (.425) .004 Seeking Unfavorable 5.45 1.55 Seeking Favorable 5.33 1.63 Control 5.57 1.47 People will respect me if I choose not to receive my risk results. 3.98 1.81 0.37 (.691) .002 See king Unfavorable 3.93 1.80 Seeking Favorable 4.09 1.80 Control 3.93 1.85 Most people will view me as thoughtful and prudent if I choose not to receive my risk results. 2.54 1.50 2.02 (.133) .009 Seeking Unfavorable 2.38 1.32 Seeking Favorable 2.75 1.62 Control 2.51 1.52 Most people will view me as responsible and proactive if I choose to receive my risk results. 5.88 1.61 1.84 (.161) .008 Seeking Unfavorable 5.80 1.71 Seeking Favorable 5.74 1.63 Control 6.08 1.46

PAGE 26

26 Information A v oidance Participants did not differ by condition in their STI risk avoidance (Seeking favorable: 4.4% avoidance, Avoiding favorable: 3.7% avoidance, Control: 3.2% 2 (2) = 0.88, p = .64, = .05. A one way ANOVA of the information avoidance scale also revealed no significant differences in avoidance by condition (Seeking favorable: M = 2.05, SD = 1.12, Avoiding favorable: M = 2.03, SD = 1.1 1, Control: M = 1.92, SD = 0.99), F (2, 428) = 0.58, p = .56, p 2 = .003 Results suggest that the manipulation was ineffective in eliciting differences between conditions in STI risk avoidance or STI risk avoidance tendencies. Exploring Other Predictors o f Avoidance To assess whether other demographic variables might predict avoidance, I conducted an exploratory hierarchical logistic regression, with the effects of condition entered as two Helmert contrast codes in step one (H1: Seeking Unfavorable 2, Seek ing Favorable 1, Control 1; H2: Seeking Unfavorable 0, Seeking Favorable 1, Control 1), and demographic variables (perceived risk, actual risk, previous STI testing, age, sex) entered simultaneously with the contrast codes in step two (Table 2 3). Resul t revealed that both actual STI risk and whether participants had undergone previous STI testing were significant predictors of information avoidance. Participants who scored higher on the SRS (higher scores indicate greater risk for STIs) and participants who had not undergone STI testing previous were significantly more likely to avoid their STI risk information. Effects of condition did not emerge as significant in either step.

PAGE 27

27 Table 2 3. Predictors of Information Avoidance in Study 1a Model or Variabl e B Wald Odds ratio [95% CI] STEP ONE H1 H2 STEP TWO H1 H2 Perceived STI risk Actual STI risk Previous STI testing (1 = have been tested, 2 = have not been tested) Age Sex (1 = male, 2 = female) Results for two step hierarchical logistic regression, with information avoidance choice ( Information decision: 0 = did not avoid, 1 = avoided) as the DV. Actual STI risk calculated using raw total score for SRS. p < .05, ** p <.01, ***p < 001 I also conducted a parallel hierarchical linear regression with the same predictors predicting information avoidance proclivity (IA Scale Table 2 4 ). Similar to the information decision results, effects of condition were non significant in the first and second step. Also in line with previous results, both actual STI risk and testing for a STI in the past predicted greater proclivity to avoid STI risk information. Additionally, greater perceived STI risk predicted greater proclivity to avoid, whereas older participants and female participants expressed less proclivity to avoid than did younger and male participants.

PAGE 28

28 Table 2 4. Predictors of Information Avoidance Proclivity in Study 1a Model or Variable B t r p [CI 95% ] STEP ONE H1 0.01 0.21 .835 H2 0.06 0.98 .328 STEP TWO H1 0.01 0.35 .729 H2 0.07 1.11 .269 Perceived STI risk 0.10 2.25 .025 Actual STI risk 0.24 4.12 .000 *** Previous STI testing (1 = have been tested, 2 = have not been tested) 0.44 4.17 .000 *** Age (1 = male, 2 = female) 0.04 2.03 .043 Sex 0.39 3.34 .001 ** Actual STI risk calculat ed using raw total score for SRS. p < .05, ** p <.01, ***p < .001 In summary the results from Study 1a suggest that my manipulation of how people view people who seek or avoid STI risk information was an ineffective manipulation of proactive impres sion management. Additionally, overall avoidance in Study 1a was relatively low (1 1.3% of participants opted not to receive their STI risk information), which may have limited the ability to find reliable effects. However, results from the exploratory hi erarchical linear regression suggest that in the absence of an effective manipulati on of proactive impression management, both actual STI risk and previous STI testing are consistently significant predictors of STI risk information decisions and informatio n avoidance proclivity. Participants who had previously been tested for STIs were m ore likely to avoid STI risk information (and

PAGE 29

29 expressed greater proclivity to avoid). It is possible these participants were more likely not to seek STI risk information bec ause they were already well aware of their status (i.e., the information had little utility, Lipsey & Shepperd, 2019b). Participants at higher risk for an STI were more likely to avoid their STI risk feedback (and expressed greater proclivity to avoid) tha n did people at lower risk, suggesting a potential motivation to avoid information that would change their self beliefs or how others viewed them (Sweeny et al., 2010). Yet, it is noteworthy that while perceived STI risk did predict information avoidance p roclivity, it did not predict information decision, suggesting that the decision to seek or avoid STI risk information may not be motivated by strongly deliberate cognitive processes.

PAGE 30

30 CHAPTER 3 STUDY 1B I designed Study 1b to address two major limitatio ns of Study 1a. First, avoidance in Study 1a was low, limiting the ability to test for differences in levels of avoidance. Study timing may be one potential reason for the relatively low levels of avoidance I conducted Study 1a in the fall semester, and t he vast majority of participants were relatively young (40.7% 18 year olds, 33.1% 19 year olds). It is possible that participants this young at the beginning of the school year may have had limited sexual e xperience at the time of participation, and theref ore may have little reason to avoid STI risk information. To address this issue, I conducted Study 1b in the spring semester. Second, the manipulation in Study 1a was ineffective. The manipulation check re vealed no differences by condition, and condition did not significantly predict information avoidance or tendencies to avoid STI risk information. One reason Study 1a manipulated impressions of seekers/av oiders, rather than impressions of people at high risk. People may be less concerned with how others will view the decision itself to seek or avoid information, but more concerned with how others will view them if they receive unfavorable results. Another three conditions introduced a descriptive norm of seeking/avoiding by stating that half of participants avoid, and half of participants seek. It is possible that introducing this descriptive norm may have shifted overall avoidance levels. To address thes e issues, I revised the manipulation to vary how unfavorably participants would be viewed if they received high risk STI results. Study 1b also

PAGE 31

31 omitted descriptive norm information from the manipulation. I predicted greater likelihood of avoiding STI risk information when participants believed others would view them negatively if at high risk, than if participants believed others would view them no differently if at high risk. Method Participants Participant s were 109 sexually active college students ( M age = 20.17, age range = 18 46; 81 White, 7 Asian, 19 Black, 1 Native American, 1 other (participants could select more than one racial identifier); 83 non Hispanic, 23 Hispanic; 77 female, 31 male ) recruited f rom the University of Florida participant pool for research credit and from several psychology classes for extra credit To confirm that participants were sexually active, after consenting to participate participants answered the following question: Are y ou (or have you ever been) sexually active? If pa rticipants selected Of the 117 participants who consented to participate, 6 identified as non sexually active, and were thus unable to complete the study. A post hoc power analysis for chi square using G power, with a medium effec t size w .05, revealed sufficient power to detect effects, power = .83. Procedure Participants completed an online survey through Qualtrics ( Appendix B ). The Study 1b procedure was identical to t he Study 1a procedure, with the addition of a mani pulation reinforcement after participants received condition specific information.

PAGE 32

32 Measures Sexual Risk Scale (SRS). I used the same modified version of the SRS (DeHart & Birkimer y of the modified scale was = .93. Perceived STI risk. Participants answered the same item used in Study 1a to assess their perceived STI risk based on the SRS. Demographics. Par ticipants answered the same demographic items as in Study 1a, with one addition. Participants in this St udy 1b also entered how many sexual partners they had in the past year. Manipulation. Participants were randomly assigned to one of two conditions: The Unfavorable Image condition or Control condition All participants first read the following text: In a moment, you will have the choice of whether to receive your personal sexually transmitted infection (STI) risk results. We believe that it is important to inform you about research on how people view individuals who are at high risk. Participants in the Unfavorable Image condition then read the following: Sadly, research finds that the vast majority of people view high risk individuals less favorably t han low risk individuals. We tell you this information so that you can make an informed decision. Parti cipants in the Control condition read: Fortunately, research finds that the vast majority of people do not view high risk individuals any less favorably than low risk individuals. We tell you this information so that you can make an informed decision. Man ipulation reinforcement. To reinforce their condition information, participants next saw an item with three choices asking what r esearch shows a bout how people view those at high risk for an STI: ( view people at high risk more favorably; view

PAGE 33

33 people at hig h risk less favorably; view people at high risk the same as people at low risk ). Limitations in the survey software prevented us from determinin g whether Participants who answered incorrectly (based on the information they were provided in their randomly assigned condition) received the correct answer plus instruc tions to select that answer to continue. I included this manipulation reinforcement to assure that all participa nts fully understood the manipulation they received. Manipulation check. Participants answered two items assessing the effectiveness of the man ipulation. I tems assessed the extent to which participants believed others would view the participant unfavorabl y if the participant was at high risk for an STI ( If I receive high risk feedback, other people will view me unfavorably ; If I receive high risk feedback, other people will not view me negatively [R] ). Participants answered using a 1 7 scale, 1 = not a t all true, 7 = very true. The reliability of the two item manipulation check scale was = .59. Information decision. A s in Study 1a, par ticipants t hen chose whether they wished to receive their risk results Participants could either select : Yes, please tell me my STI risk results, or No, I am not interested in learning my STI risk results. If participants opted to receive their STI risk r esults, I provided them with feedback based on their responses to the SRS Information avoidance scale. Participants also completed the same eight item information avoidance scale (Howell & Shepperd, 2016) used in Study 1b to measure procliv ity to avoid health risk feedback I averaged responses to the items

PAGE 34

34 to create a single index of information = 0.8 5 M = 2. 30 SD = 1. 16 ) Results and Discussion Zero order C orrelations I initially computed the correlations among t he predictors (Table 3 1 ). Information decision correlated significantly with condition and with information avoidance proclivity ( IA Scale). Condition correlated significantly with information avoidance proclivity, and actual STI risk correlated positivel y with information avoidance proclivity, such that higher risk for an STI corresponded with greater likelihood of avoiding STI risk feedback. Notably, condition correlated significantly with actual STI risk, which likely is an anomaly. Participants were ra ndomly assigned to condition after taking the SRS. Table 3 1 Zero order correlations for Study 1b Variable 1. 2. 3. 4. 5. 6. 7. 8 1. Condition 2. Actual STI risk .27** 3. Perceived STI risk .07 .42** 4. Previous STI testing .13 .07 .28* 5. Number of Sex Partners .14 .22* .28** .16 6. Age .05 .01 .06 .15 .13 7. Sex (1 = male, 2 = f emale) .14 .12 .04 .11 .46** .05 8. Information avoid. scale .24* .37** .06 .18 .13 .09 .08 9. Avoidance dec. (0 = did not avoid, 1 = avoided). .28** .07 .08 .12 .06 .13 .05 .51** Condition coded as 1 = Unfavorable Image, 2 = control co ndition. Previous STI testing coded as 1 = have undergone testing, 2 = have not undergone testing. p < .05, ** p < .01 Manipulation C heck Analysis of the manipulation check suggests that the manipulation was effective. Participants were more likely to repo rt that high risk STI feedback would elicit a negative

PAGE 35

35 evaluation from others in the U nfavorable I mage condition ( M = 4.96, SD = 1.58) than in the C ontrol condition ( M = 4.25, SD = 1.42), t (99) = 2.38, p = .019 Cohen's d = 0.19 Information A voidance Ove rall avoidance levels were higher in Study 1b (33% avoided) than in Study 1a (11.3% avoided). A chi square analysis revealed a significant difference between condition s in STI risk information avoidance. Contrary to predictions, more participants opted not to receive their results in the control condition (46.3%) than in the unfavorable image condition 2 (1) = 8.52, p = .004, = .28. A one way ANOVA of the information avoidance scale revealed a significant difference in avoidance by condition with the same pattern, F (1, 104) = 6.42, p = .013, p 2 = .058 such that participants expressed great er STI risk avoidance proclivity in the control condition ( M = 2.58, SD = 1. 20) than in the unfavorable image condition ( M = 2.02, SD = 1. 06). Exploring Other P redictors of A voidance To assess whether other demographic variables might predict avoidance, I conducted a hierarchical logistic regression, with condition entered in step one. In step two I included the same demographic variables used in Study 1a (perceived risk, actual risk, previous STI testing, age, sex), with one additional predictor: number of sexual partners in the past year (Table 3 2 ). Contrary to Study 1a, the only significant predictor of STI risk avoidance was condition (in bot h steps), with participants in the unfavorable image condition avoiding less than participants in the control c ondition.

PAGE 36

36 Table 3 2 Predictors of Information Avoidance in Study 1b Model or Variable B Wald Odds ratio [95% CI] STEP ONE Condition 1.33 9.01 .003 ** STEP TWO Condition 1.35 8.24 .004 ** Perceived STI risk Actual STI risk Pr evious STI testing (1 = have been tested, 2 = have not been tested) 0.04 0.01 Number of sexual partners Age Sex (1 = male, 2 = female) Results for two step hierarchical logistic regression, with information avoidance choice ( Information decision: 0 = did not avoid, 1 = avoided) as the DV. Condition coded as 1 = Unfavorable Image, 2 = control condition. Actual STI risk calculated us ing raw total score for SRS. p < .05, ** p <.01, ***p < .001 I conducted a parallel hierarchical linear regression with the same predictors predicting information avoidance proclivity (IA Scale Table 3 3 ). The effect of condition was significant in th e first step, indic ating that participants exhibited a higher proclivity to avoid STI risk information in the control condition than in the Unfavorable Image condition. However, when I entered demographic predictors, the effect of condition on information avoidance proclivit y became non significant In Step 2, actual STI risk was the only significant predictor of information avoidance proclivity, with higher risk participants exhibiting greater information avoidance proclivity. It is likely that the effect of condition became non significant in Step 2 because actual STI risk correlated

PAGE 37

37 positively with information avoidance proclivity ( r = .37) and with condition ( r = .27) and thus absorbed the variance explained by information condition observed in Step 1. Table 3 3 Predictors of Information Avoidance Proclivity in Study 1b Variable B t r p [CI 95% ] STEP ONE Condition 0.56 2.53 .013 STEP TWO Condition 0.34 1.56 .122 Perceived STI risk 0.12 1.37 .174 Actual STI risk 0.01 3.60 .001 ** Previous STI testing (1 = have been tested, 2 = have not been tested) 0.32 1.48 .142 Number of sexual partners 0.02 0.32 .750 Age 0.02 0.95 .342 Sex (1 = male, 2 = female) 0.01 0.02 .986 Condition coded as 1 = Unfavorable Image, 2 = control condition. Actual STI risk calculated using raw total score for SRS. p < .05, ** p <.01, ***p < .001 E xploratory M oderation A nalyses To better understand the results, I conducted exploratory moderation analyses testing whether perceived or act ual risk moderated the effect of condition on information avoidance. It is possible that participants in the Unfavo rable Image condition who were at low risk for an STI (perceived or actual) would be more likely to seek information because doing so could s erve to boost their image. Conversely, participants at high risk for an STI would be less likely to seek in the Unf avorable Image condition because doing so could tarnish their image. To test this possibility, I conducted two logistic

PAGE 38

38 regression analyses e xamining the main effects of condition, risk (actual and perceived), and their interaction. The interaction was non significant for both perceived risk ( B = 0.42, SE = 0.34, Wald = 1.54, p = .215, Odds Ratio = 0.66 [0.34, 1.27] and actual risk ( B = 0.01, SE = 0.01, Wald = 0.28, p = .599, Odds Ratio = 0.99 [0.97, 1.02]. Summary of F indings In sum, Study 1b revealed a significant effect of condition, such that participants were more likely to seek STI risk information in the Unfavorable Image condition than in the control condition. The finding that more participants avoided in the control condition than the unfavorable image cond ition may reflect psychological reactance (e.g., Brehm, 1966) Participants in the unfavorable image condition may have perceived the manipulation message as a threat to their freedom to make an independent decision and may have felt angry or ge nerated cou nterarguments that led them to seek their risk feedback (Rains, 2013). In other words, participants in the unfavorable image condition may have perceived the message as an attempt to control their information decision and reasserted their contro l by choosi ng to oppose this unwelcomed control, resulting in less avoidance. Participants in the control condition may also have displayed reactance, perceiving the manipulation as an attempt to control their information decision, prompting them to reasse rt control by avoiding their STI risk information. Participants in the present studies were likely already aware of the potentially negative interpersonal consequences of having or being at high risk for an STI. In support of this assertion, a major barri er college students face when testing for an STI is the perception of negative consequences and stigma (Barth, Cook, Downs, Switzer, & Fischhoff, 2002). As such, participants may have viewed the unfavorable image

PAGE 39

39 manipulation as obvious and the control man ipulation a s unrealistic, leading them to engage in reactance. Another possibility is that the manipulation may have led participants to perceive STIs as more serious or important in the unfavorable image condition than in the control condition. Specifical ly, partici pants in the unfavorable image condition learned that being at high risk for an STI has interpersonal consequences people perceive people at high risk for an STI less favorably. Knowing that their STI risk information could lead to an importan t shift in how others view them, these participants may have viewed the STI risk information as important both for their health and for their social status. Likewise, participants in the control condition learned that being at high risk for STIs carries no interperso nal consequences and may have thus perceived the information as less important. In support of this possibility, recent research indicates that people report that they would be less likely to avoid getting tested for a serious medical condition ( as opposed to a minor medical condition, Lipsey & Shepperd, 2019b). These results indicate that if the stakes are high, people may be less likely to avoid the information. Accordingly, participants in Study 1b may have equated interpersonal costs of being at high ri sk with the seriousness of the STI risk feedback, and therefore were less likely to avoid their risk feedback. This explanation suggests that perhaps participants were still engaging in proactive impression management but rather than avoiding unfavorable information to prevent being viewed poorly, sought unfavorable information to mitigate future impression dilemmas. Indeed, the importance of the information may have motivated participants to proactively impression manage by getting information that would assure their partners if they were at low risk for an STI.

PAGE 40

40 CHAPTER 4 STUDY 2 I designed Study 2 to test the hypothesis that people are more likely to avoid information if they believe others could use that information to harm them. Research d emonstrates that powerful audiences such as insurers and employers elicit greater levels of avoidance than non powerful audiences such as researchers or no audience (Lipsey & Shepperd, 2019a). Although anticipated likelihood of harm partially mediated the relationshi p between powerful audience conditions and information avoidance (Lipsey & Shepperd), researchers have yet to directly test the hypothesis that capacity to harm is the mechanism through which powerful audiences elicit avoidance. Study 2 was an extension o f Lipsey and Shepperd (2019a) and tested whether leading participants to believe their information would be made available to insurers with a capacity to harm (Insurer Harm), insurers without a capacity to harm (Insurer No Harm), or researchers without a c apacity to harm (Researcher No Harm). I predicted that insurers with a capacity to harm would elicit the greatest avoidance and researchers without a capacity to harm would elicit the least avoidance. I had competing hypotheses for the Insurer No Harm con dition. One possibility is that capacity to harm is the predominant mechanism by which powerful audiences elicit avoidance. In other words, people display greater avoidance if an audience has a capacity to harm (the harm condition) than if audie nce does no t (the two no harm conditions). A second possibility is that certain audiences (such as insurers) elicit higher avoidance regardless of their capacity to cause harm. There are many ways insurers can cause harm, such as raising premiums, denying coverage, or limiting coverage. If

PAGE 41

41 people h ave experienced insurer harm previously or know others who have experienced harm, they may not trust or believe that insurers can have no capacity to harm. In other words, the association of insurers with potenti al harm may be quite strong. If the audience itself is the predominant mechanism, then participants will avoid more in the Insurer conditions than in the Researcher condition, but participants in the two Insurer conditions will not differ in avoidance. A final possibility is that both the audience itself and the capacity to harm elicit avoidance. If true, participants in the Harm condition should display higher avoidance than participants in the two No Harm conditions, and participants in the Insurer condi tions should display higher avoidance than p articipants in the Researcher condition. Method Participants Participants were 423 ResearchMatch volunteers ( M age = 51.96, age range = 18 87; 285 White, 8 Asian, 25 Black, 4 Native American, 1 other (participants could select more than one racial identifie r); 305 non Hispanic, 6 Hispanic; 235 female, 76 male, 1 transgender, 1 genderqueer female). From 423 volunteers who consented to participate, 284 provided sufficient data for analysis. A post hoc power analysis for Chi square using G power, with a medium effect size w = .21, = .05 revealed sufficient power to detect effects, power = .90. ResearchMatch is a national health registry that was created by several academic institutions and supported by the U.S. National Institutes of Health as part of the Clinical Translational Science Award (CTSA) program ResearchMatch volunteers are individuals who reside in the United States and have consented to receive contact

PAGE 42

42 from researchers about studies for which they may be eligible. ResearchMatch accepts volunte ers of all ages and bac kgrounds. As of May, 2019, over 140,000 volunteers are registered on ResearchMatch (Race: White 75.8%, Black 11.3%, Multi Racial 5.1%, Asian 3.7%, American Indian or Alaskan Native: 0.7%, Native Hawaiian or other Pacific Isla nder 0.2%, Other 3. 1%; Ethnicity: Non Hispanic 91.6%, Hispanic 8.4%; Gender: Female 70.6%, Male 28.9%, Transgender 0.4%; Medication Use: Takes at least one medication 56.2%, No medications 48.2%; Medical conditions: At least one condition 64.3%, No conditions 35.7%). Procedure The Qualtrics survey randomly assigned the volunteers who consent ed to participate to one of three conditions: Insurer Harm, Insurer No Harm, or Researcher No Harm. After consenting, all participants received ins tructions to provide th eir full name and address, which I used to create credible audience manipulation. Participants then received a condition specific message informing them of who would have access to their information. After answering manipulation chec k items, participants l earned that they would take a general health risk questionnaire and had the choice to receive or not receive their risk results. Participants then completed the health questionnaire and demographic information. Audience harm manipul ation. Participants in the three conditions received different instructions. Participants in the Insurer Harm condition received the following information: As part of your participation in this study, if you elect to receive your risk results, your result s will automatically be sent to a registry that is accessible by health insurance companies However, per contract with University of Florida, please be advised that we cannot control how insurance companies use this information in the future. Although cur rent

PAGE 43

43 laws forbid insura nce companies from raising premiums or denying coverage because of preexisting conditions, the law has not always been successful in protecting consumers. Moreover, it is possible that the law will be changed or repealed in ways that are unfavorable to peo ple with preexisting conditions. Participants in the Insurer No Harm condition read the following: As part of your participation in this study, if you elect to receive your risk results, your results will automatically be sent to a registry that is acces sible by health insurance companies Thus, insurance companies can learn if you are at risk for the health conditions assessed. However, per contract with University of Florida, health insurance companies CANNOT use this information in any capacity to incr ease premiums, increase the deductible, deny coverage, or limit access to providers. Participants in the Researcher No Harm condition read the following: If you elect to receive your risk results, your results will automatically be sent to a registry that is accessible by health scientists and researchers However, as per the Institutional Review Board requirements, researchers will maintain absolute confidentiality and your results will not be used in any capacity beyond research. Notably, participants i n all three conditions learned that their information would be immediately visible to or would be forwarded automatically to the specified audience requiring no additional efforts from participants Manipulation Check. After receiving condition specific information, all participants answered three items assessing perceptions of insurer harm. Specifically, parti cipants answered how likely insurers would be to harm them ( How likely is it that insurance will raise your premiums or deny you coverage if they learn that you are at high risk for the health conditions assessed in this study? ), how worried they were about insurer harm ( How worried are you that insurance companies might have access to your ri sk results if you choose to receive them? ), and how fe arful they were about being harmed by insurers ( How fearful are you that you will experience higher insurance rates or denial of coverage if you receive results that you are at high risk for

PAGE 44

44 the healt h conditions assessed in this study? ). All items used a 1 7 scale, with item specific anchors (e.g., 1 = not at all likely, 7 = very likely; 1 = not at all worried, 7 = very worried). I averaged the three items together to create a perception of insurer h arm manipulation check 92, M = 3.63, SD = 2.01). Information decision. Participants read that later in the study they would complete a risk assessment and could choose now whether they wanted the risk assessment results (Yes, plea se tell me my risk resu lts) or did not want the results (No, I am not interested in learning my risk results). Because the health questionnaire participants took was not diagnostic, I did not provide p articipants with their risk feedback. Information avoi dance scale. Participan ts completed the same eight item information avoidance scale (Howell & Shepperd, 2016) used in Studies 1a and 1b. However, the scale wording was modified to assess general health risk information (e.g ., I would rather not know my ri sk for the health condi tions assessed in this study ). I averaged responses to the items to create a single index of information avoidance M = 2.13, SD = 1.36). Lifestyle Health Questionnaire. diet, habits, and medic al risk factors (e.g., family history, smoking, high blood pressure, etc.). Previous studies have used versions of this questionnaire (Howell & Shepperd, 2012) to convince participants that the computer could calculate their health r isks. Demographics. Par ticipants answered the same demographic items as in Studies 1a and 1b, except for the sexual behavior items (as those items were relevant specifically to the paradigm used in Studies 1a and 1b).

PAGE 45

45 Results and Discussion Zero order C orrelations I initially co mputed the correlations among the predictors (Table 4 1 ). Information decision correlated significantly with condition, anticipated insurer harm, and information avoidance proclivity (IA Scale). Information decision al so correlated significantly with sex, such that females were more likely to avoid than males. Table 4 1 Zero order correlations for Study 2 Variable 1. 2. 3. 4. 5. 6. 7. 1. Anticipated insurer harm 2. Education .08 3. Income .01 .24** 4. Financial Security .05 15** .13* 5. Age .06 .07 .08 .07 6. Sex (1 = male, 2 = female) .03 .06 .10 .05 .24** 7. Information avoidance scale .16** .04 .12* .15** .07 .17** 8. Avoidance decision (0 = did not avoid, 1 = avoided) .16** .03 .08 .11 .0 2 .21** .65** p < .05, ** p < .01. Manipulation C heck To test whether condition affected insurer harm perceptions (see Table 4 2 for mean insurer harm perceptions by condition), I conducted two linear regressions using two sets of Helmert contrast codes a s predictors, and information avoidance as the outcome. In the first linear regression, I compared the researcher condition to th e two insurer conditions (H1: Researcher No Harm 2, Insurer No Harm 1, Insurer Harm 1) and the insurer no harm to the insurer harm condition (H2: Researcher No Harm 0, Insurer No Harm 1, Insurer Harm 1). Participants expressed greater harm perceptions in the insurer conditions than in the researcher condition, ( B = 0.26, p = .001, r p = 0 19

PAGE 46

46 [0.08, 0.29] ), but avoidance did not differ within the two insurer conditions ( B = 0.17, p = .202, r p = 0 07 [ 0.04, 0.17] ). In the second linear regression, I compared harm to no harm (H3: Insurer Harm 2, Researcher No Harm 1, Insurer No Harm 1) and the researcher no harm condition to th e insurer no harm condition (H4: Insurer Harm 0, Researcher No Harm 1, Insurer No Harm 1). Participants expres sed greater harm perceptions in the harm condition than in the no harm conditions, ( B = .21 p = .006, r p = 0 15 [0.05, 0.25] ), and also express ed greater harm perceptions in the Insurer No Harm condition than in the Researcher No Harm condition ( B = .31, p = .018, r p = 0 13 [0.02, 0.23] ). These results indicate that participants seem to view insurers and researchers as two distinct audiences and provide a conceptual replication of prior research (Lipsey & Shepperd. 2019a). These results also suggest that capacity to harm plays a role in harm perceptions. Collectively, the manipulation check analyses provide initial support for the audience itself and the capacity to harm both playing a role in perceptions of harm. Table 4 2 Study 2 manipulation check means by condition Condition M SD Insurer Harm 4.09 1.97 Insurer No Harm 3.75 2.09 Researcher No Harm 3.15 1.87 Information A voidance A chi s quare analysis revealed a significant difference between condition s in health risk information avoidance 2 (1) = 6.65, p participants avoiding health risk information in the Insurer Harm condition, 16.7% in the

PAGE 47

47 Insurer No Harm condition, and 9.7% in the Researcher No Harm condition. A one wa y ANOVA using the IA scale as an outcome measure of avoidance revealed no significant difference in avoidance by condition, F (2, 313) = 1.18, p = .310, p 2 = .007. To better understand differences in avoidance between conditions, I conducted two logist ic regressions using the same two sets of Helmert contrast codes used in the manipulation check analysis as predictors, and information avoidance as the outcome. In the first logistic regression, I compared effects of the audience itself (H1: Researcher No Harm 2, Insurer No Harm 1, Insu rer Harm 1; H2: Researcher No Harm 0, Insurer No Harm 1, Insurer Harm 1). Similar to the manipulation check, participants were significantly less likely to avoid their risk information in the researcher condition than in t he two insurer conditions ( B = .2 7, Wald = 5.25, p = .022, Odds ratio [95% CI] = 1.31 [1.04, 1.64]). However, participants in the two insurer conditions did not differ ( B = .18, Wald = 1.05, p = .306, Odds ratio [95% CI] = 1.20 [0.85, 1.68]). In the second logistic regression, I compared the capacity to harm (H3: Insurer Harm 2, Researcher No Harm 1, Insurer No Harm 1; H4: Insurer Harm 0, Researcher No Harm 1, Insurer No Harm 1). Participants were significantly more likely to avoid their risk infor mation in the harm condition than in th e two no harm conditions ( B = .22, Wald = 4.60, p = .032, Odds ratio [95% CI] = 1.25 [1.02, 1.53]). However, participants in the two no harm conditions did not differ in avoidance ( B = .31, Wald = 2.51, p = .113, Odds ratio [95% CI] = 1.37 [0.93, 2.01]). E xploratory Mediation Analysis To explore whether expectations of harm mediated the relationship between condition and information decision, I conducted two mediation analyses. The first examined the effect of harm (ha rm versus no harm) on information avoid ance, via the

PAGE 48

48 harm expectations mediator. To run this analysis, I used M odel 4 in the PROCESS dialogue for SPSS ( Hayes, 2013 ) with bootstrapped standard errors ( 5 ,000 draws) (Zhao et al., 2010). The predictor variable was harm versus no harm (Harm conditio n = insurers ( manipulation check measure ), and the outcome was the information decision (1 = avoid, 0 = seek). I observed a statist ically significant direct effect of har m condition. As evident in Table 4 3 and Figure 4 1 fewer participants avoided in the no harm conditions than in the harm condition. I also observed a statistically significant indirect effect (harm versus no harm expectation of insurer harm informati on avoidance) as indicated by a confidence interval for the indirect effect that did not include zero. Consistent with full mediation, the direct effect of harm versus no harm was no longer significant after controlli ng for the effect of expectation of ins urer harm on information avoidance. These results suggest that the greater expectations of insurer harm prompt participants in the harm condition avoidance the health information.

PAGE 49

49 Table 4 3 Mediation results for dichotomous choice information decision. N = 337 Harm versus No Harm Insurer versus Researcher Pathway b t or Wald r p or OR 95% CI b t or Wald r p or OR 95% CI I nformation Avoidance Condition (c) 0.30 4.37 .037 0.74 [0.56, 0.98] 0.38 5.22 0.02 0.68 [0.49, 0.95] Harm Expect. Condition (a) 0.15 2.81 .005 0.15 [0.04, 0.25] 0.18 3.46 .001 0 .18 [0.08, 0.28] Information Avoidance Harm Expect. (b) 0.43 7.52 .006 1.54 [1.13, 2.10] 0.41 6.87 .009 1.51 [1.11, 2.06] 0.26 3.06 .080 0.77 [0.58, 1.03] 0.32 3.56 .059 0.73 [0.52, 1.01]

PAGE 50

50 Figure 4 1 Exploratory mediation pat hways Regression coefficients are standardized. Numbers in brackets reflect 95% confidence intervals. Numbers in parentheses show the mediated effect of powerful audience conditions on inform ation avoidance with bootstrapped 95% confidence intervals. = p < .05; ** = p < .01. In the second mediation analysis, the predictor variable was insurer versus researcher (Insurer conditions = 1; Researcher condition = 2), the meditator was s expectation of harm from insurers ( manipulation check measure ) and the outcome was the information decision (1 = avoid, 0 = seek). Similar to the harm analysis, I observed a statistically significant effect of audience condition, such that fewer partici pants avoided in the researcher condition than in the insurer co nditions Harm versus No Harm [ 0 23 0 02 ] [ 0 12 0 74 ] ** [ 0.58 0 02 ] [ 0. 54 0.03 ] ) Insurer versus Researcher [ 0 26 0 06 ] [ 0 10 0 72 ] ** 0.38 [ 0. 71 0.05 ] [ 0.65, 0 01 ] )

PAGE 51

51 (Table 10, Figure 2). I also observed a statistically significant indirect effect (insurer versus researcher expectation of insurer harm information avoidance) as indicated by a c onfidence interval for the indirect effect that did not include zero. Consistent with full mediation, the direct effect of insurer versus researcher was no longer significant after controlling for the effect of expectation of insurer harm on information av oidance. These results suggest that participants in the insurer conditions had greater expectations of insurer harm, which led to a greater rate of information avoidance. These mediation results should be interpreted with caution. The analyses were exploratory and harm expectations were measured rather than manipulated (although I did manipulate capacity to harm and audience). Testing for mediation without manipulating the mediator cannot clearly inform whether a mediation effect is valid (Bullock, 2010; Pirlott & Mackinnon, 2016). Other P redictors of A voidance To asses s whether other demographic variables might predict avoidance, I conducted a hierarchical logistic regressi on in which I entered condition in step one (Helmert contrast codes) and demographic variables (education, income, financial security, age, sex) that predicted in other research in step two (see Howell, Lipsey, & Shepperd, 2019 for a review) (Table 4 4 ). R esults revealed a significant effect of condition, and the effect remained significant after entering demographic predictors in Step 2 (p = .04) 1 In terestingly, gender was the strongest predictor of information avoidance, with males avoiding significantly more than females, overall (p < .001). 1 Entering the alternate set of Helme rt contrast codes (H3 & H4) did not change the overall results of the logistic regression. Harm remained a significant predictor of information avoidance ( p = .024) even after entering the demographic predictors in Step 2.

PAGE 52

52 Because is an inconsistent and somewhat weak predictor of information avoidance in other studies (Howell et a l., 2019), and because I made no a priori predictions about gender as a significant predictor, I urge cauti o n in interpreting this finding. Table 4 4 Predictors of Information Avoidance in Study 2 Model B Wald Odds ratio [95% CI] STEP ONE H1 0.27 4.03 .045 1.31 [1.01, 1.70] H2 0.30 2.11 .147 1.34 [0.90, 2.00] STEP TWO H1 H2 Education 0.03 0.06 Income Financial Security Age Sex Results for two step hierarchical logistic regression, with information avoi dance choice ( Information decision: 0 = did not avoid, 1 = avoided) as the DV. Sex coded as 1 = male, 2 = female. p < .05, ** p <.01, ***p < .001 I also conducted a parallel hierarchical linear regression with the same predictors predicting information a voidance procl ivity (IA Scale Table 4 5 ). The results differed considerable from the results for information avoidance decision. The effects of condition were non significant in both steps. In Step 2, income and financial security negatively predicted inf ormation avoid ance proclivity, such that participants with higher income and a greater sense of financial security expressed lower information avoidance proclivity. Age and sex were also significant predictors of information avoidance

PAGE 53

53 proclivity, with olde r participants and female participants exhibiting lower information avoidance proclivity. Table 4 5 Predictors of Information Avoidance Proclivity in Study 2 Variable B t r p [CI 95% ] STEP ONE H1 0.07 1.41 .160 0.08, [ 0.03, 0.19] H2 0.04 0.41 684 0.02, [ 0.09, 0.14] STEP TWO H1 0.06 1.16 .249 0.07, [ 0.05, 0.18] H2 0.03 0.28 .781 0.02, [ 0.10, 0.13] Education 0.01 0.25 .807 0.01, [ 0.13, 0.10] Income 0.17 2.50 .013 0.15, [ 0.26, 0.03] Financial Security 0.93 2.84 .005 ** 0.1 7, [ 0.27, 0.05] Age 0.01 2.18 .030 0.13, [ 0.24, 0.01] Sex 0.55 3.16 .002 ** 0.18, [ 0.29, 0.07] Sex coded as 1 = male, 2 = female. Summary of F indings Collectively, these suggest that both the type of audience (i.e. insurers versus research information) predict health information avoidance. However, the capacity to harm may not be the sole determinant of why insurers elicit higher levels of health information avoidance th an do researchers (Lipsey & Shepperd, 2019a). It is possible that other factors, such as trust in the audience and privacy concerns, may play a significant role in wha t makes an audience powerful. I discuss this possibility in the General Discussion. I ob served less avoidance in the present Study than in two prior, similar studies where Insurer avoidance ranged from 51.6% to 55.8%, and Researcher avoidance

PAGE 54

54 ranged from 14.3% to 17.6% (Lipsey & Shepperd, 2019a). Although all three studies used the same sampl e (ResearchMatch volunteers), the present study differed from the prior studies in the order of events. In the prior studies, participants complete d the health risk as sessment receive d condition specific info rmation and then made their information decisi on. In the present study participants received condition specific info rmation made the ir information decision and then completed the health risk assessment. I changed the order to decrease how much time participants committed to getting their health inf ormation (i.e., taking the survey) prior to making an information decision. I assumed that this change would elicit higher avoidance levels than I o bserved in the Lipsey & Shepperd studies because participants had not committed a significant amount of tim e to getting their results (Burger, 1999) In hindsight, it is possible that once participants took the health risk assessment, they formed a more concrete sense of the personal relevance of the health risk results and whether they were worried about the results reaching a potentially harmful audience. Lacking additional information about importance or potential harm from the health risk assessment, participants may have felt less motivated to avoid their risk information. It is also possible that partic ipants displayed less avoidance because the present study lacked a manipulation reinforcement. In Lipsey and Shepperd (2019a), after reading conditi on specific information, participants answered a question about who would have access to their information. This question aimed to ensure that participants understood the manipulation. In the present study, I did not use a manipulation

PAGE 55

55 reinforcement. A man ipulation reinforcement itself may be essential for the manipulation work and thus essential to elicit high avoidance. One potential limitation in the present study is the wording of the manipulation check items. Rather than ask about general harm expecta tions, (e.g ., How likely are you to experience harm if the results reveal that you are at high risk for the health conditions assessed in this study? ) the manipulation check items focus solely on harm from insurers. As such, it is difficult to determine whether the manipulation would affect general harm expectations to the same extent that the manipulation aff ected insurer harm expectations. Another potential limitation in the present study is the harm framing of the manipulation. In all three conditions, participants expressly learned that the audience had or did not have the capacity to harm them. These expli cit instructions may have inadvertently activated thoughts about harm even in the no harm conditions. It is possible that participants in the no har m conditions would have responded differently had they received no mention of harm whatsoever.

PAGE 56

56 CHAPTER 5 STUDY 3A Study 3a aimed to reduce health information avoidance by tested whether using normative information condoning seeking information would eli cit lower levels of information avoidance. Notably, avoidance levels in Studies 1a, 1b, and 2 were relativel y low, raising questions about whether information avoidance requires intervention. However, levels of avoidance elicited in the present studies did not match levels of avoidance elicited in prior studies (e.g., Lipsey & Shepperd, 2019a). Several factors m ay affect avoidance levels including when participants are sampled and the type of information to seek versus avoid. For example, if participants ar e sampled during a time when healthcare policy is prominent in the news, they may respond differently than i f healthcare issues are less salient. Participants may also be more or less likely to avoid health information depending on the nature of the health condition, for example, whether the condition is treatable, serious, expensive to treat, or stigmatizing (L ipsey & Shepperd, 2019b). It is likely that the low avoidance in the present studies were an artifact of nature of the health information. Study 3a used the same health risk paradigm used in Study 2, with the expectation that baseline levels of avoidance would be around 15% (Lipsey & Shepperd, 2019a). Specifically, Study 3a aimed to harness the power of audience injunctive and descriptive norms to re duce potentially harmful avoidance of health risk information. I predicted that the use of either descriptiv e or injunctive norm endorsing seeking would result in lower levels of avoidance than the use of no norm. I also predicted the use of both an injunc tive and descriptive norm together would elicit the lowest level of seeking compared to all other conditions

PAGE 57

57 Method Participants Participants were 386 ResearchMatch volunteers ( M age = 5 0 7 6, age range = 18 87; 338 White, 9 Asian, 12 Black, 7 Native American, 5 other (participants could select more than one racial identifier); 3 53 non Hispanic, 9 Hispanic; 2 75 f emale, 8 6 male, 1 transgender, 2 non binary/ genderqueer 1 neutrois ). From 386 volunteers who consented to participate, 311 provided s ufficient data for analysis. A post hoc power analysis for Chi square using G power, with a medium effect size w = .21, = .05 revealed sufficient power to detect effects, power = .92. Procedure The Qualtrics survey (Appendix D) randomly assigned the volunteers who consent ed to participate to one of four conditions: Injunctive Norm, Descriptive Norm, Both Norms, or No Norm Similar to Study 2, participants received a condition specific message. After answering manipulation check items, participants learned that they would take a general health risk questionnaire and had the choice to receive or not receive their risk result s. Participants then completed the health questionnaire and demographic information. M at erials Audience manipulation. Participants in the three norm conditions first read the following text: Before you make your decision and take the risk assessment, we b elieve it is important to provide you with additional information Participants in the In junctive Norm condition then received the following information: Learning your risk allows you to make important life choices, to monitor for signs before problems ari se, and to plan for the future. In many ways, it is the right thing to do. Participants i n the Descriptive Norm

PAGE 58

58 condition read the following: Most people who take the Lifestyle Health Questionnaire Risk Calculator choose to receive their risk results. Part icipants in the Both Norms condition received both the injunctive and descriptive norm te xt together. Participants in the No Norm condition received no additional information. Manipulation Check. After receiving condition specific information, all particip ants answered two items assessing endorsement of injunctive norm to seek information ( Ch oosing to see my risk results is the right thing to do ) and descriptive norm to seek information ( If given a choice, most people will opt to receive their risk resul ts ). Both items used a 1 7 scale, where 1 = strongly disagree, 7 = strongly agree. Information decision. As in Study 2, participants read that later in the study they would complete a risk assessment and could choose now whether they wanted the risk asse ssment results (Yes, please tell me my risk results) or did not want the results (No, I am not interested in learning my risk results). As with Study 2, because the health questionnaire participants took was not diagnostic, I did not provide p articipants w ith their risk feedback. Information avoidance scale. Participants completed the sam e eight item information avoidance scale (Howell & Shepperd, 2016) used in Study 2. I averaged responses to the items to create a single index of information avoidance (Cr = 0.8 1 M = 2. 03 SD = 1. 01 ). Lifestyle Health Questionnaire. Participant s answered the same 16 item questionnaire used in Study 2. Demographics. Participants answered the same demographic items used in Study 2.

PAGE 59

59 Results and Discussion Zero orde r C orrelations I initially computed the correlations among the predictors and outcome s (Table 5 1 ). Information decision correlated significantly with endorsing an injunctive norm, such that greater endorsement of an injunctive norm to seek corresponded to lower likelihood of information avoidance. This same relationship was true for infor mation avoidance proclivity. Additionally, greater endorsement of a descriptive norm to seek also corresponded to lower information avoidance proclivity. Table 5 1 Zero o rder correlations for Study 3 a Variable 1. 2. 3. 4. 5. 6. 7. 8. 1. Endorsing injunctive norm 2. Endorsing descriptive norm .31** 3. Education .05 .06 4. Income .10 .07 .32** 5. Financial Security .01 .03 .11* .10 6. Age .11* .09 .13* .13* .01 7. Sex (1 = male, 2 = female) .06 .00 .02 .14* .05 .28** 8. Information avoidance scale .59** .20** .03 .07 .01 .07 .01 9. Information decision (0 = did not avoid, 1 = avoided) .14** .07 .06 .01 .08 .01 .05 .23** p < .05, ** p < .01. Manipulation C heck I conducted two 2x2 ANOVAs to test the effects of the descriptive norm and injunctive norm on each of the two manipulation check items (endorsement of a descriptive norm for seeking; endorsement of an injunctive norm for seeking; see Tab le 5 2 for means by condition). Results suggest that the descriptive norm manipulation was effective. Analysis revealed a main effect of descriptive norm on endorsement of descriptive norm ( F [1, 376] = 13.18, p < .001, p 2 = .034 but no effect of injunctive norm

PAGE 60

60 ( F [1, 376] = 0.12, p = .728, p 2 = .000) and no interaction ( F [1, 376] = 3.19, p = .075, p 2 = .008). Participants were more likely to endorse the descriptive norm in the descriptive norm condition However results suggest that the injunctive norm manipulation was ineffective. Analysis revealed no main effect of injunctive norm ( F [1, 376] = 2.48, p = .116, p 2 = .007) or descriptive norm ( F [1, 376] = 0.18, p = .672, p 2 = .000), and no interaction ( F [1, 376 ] = 0.26, p = .608, p 2 = .001). Table 5 2 Study 3a manipulation check means by condition Endorsement of Descriptive Norm Endorsement of Injunctive Norm Condition M SD M SD No Norms 4.79 1.33 6.14 1.31 Injunctive Norm 5.07 1.26 6.38 0.91 Descriptive Norm 5.51 1.31 6.24 1.17 Both Norms 5.32 1.24 6.37 1.11 Information A voidance A chi square analysis revealed no significant difference between condition s in health risk information avoidance 2 (3) = 2.11, p = .551 08 Notably overall avoidance in this study was extremely low: only 2.6% of participants avoided their health risk information (No Norms: 3.3%, Injunctive: 1%, Descriptive: 4.3%, Both Norms: 2.1%). Attempting to predict avoidance when average level of avoidan ce is so low is problematic. To better understand whether condition affected information avoidance, I also conducted a one way ANOVA using the IA scale as an outcome measure of avoidance. However, results revealed no difference in avoidance between conditi ons, F (3, 362) = 0.80, p = .493, p 2 = .007.

PAGE 61

61 I also conducted a logistic regression using a set of three Helmert contrast codes as predictors, and information avoidance as the outcome. I compared no norms to norms (H1: No Norm 3, Both Norms 1, Injunct ive 1, De scriptive 1), both norms to individual norms (H2: No Norm 0, Both Norms 2, Injunctive 1, Descriptive 1), and injunctive to descriptive (H3: No Norm 0, Both Norms 0, Injunctive 1, Descriptive 1). Results revealed no significant differences in an y conditi on comparisons ( p > .05 for all three comparisons). I conducted a parallel linear regression using the Helmert contrast codes as predictors and the IA scale as the outcome measure. Similar to the logistic regression results, linear regression resu lts revea led no significant differences for any of the condition comparisons ( p > .05 for all three comparisons). Other P redictors of A voidance To assess whether other demographic variables predicted avoidance, I conducted a hierarchical logistic regressio n, with e ffects of condition entered in step one (Helmert contrast codes). In step two I included the same demographic variables used in Study 2 (education, income, financial security, age, sex) (Table 5 3 ). The results revealed no significant predictors o f informa tion avoidance. Similarly, results from a parallel hierarchical linear regression using health risk information avoidance proclivity (IA Scale) as the outcome also revealed no significant predictors of information avoidance (Table 5 4).

PAGE 62

62 Table 5 3 Pre dictors of Information Avoidance in Study 3a Model B Wald Odds ratio [95% CI] STEP ONE H1 0.03 0.02 .892 0.97 [0.64, 1.47] H2 0.07 0.05 .820 1.07 [0.59, 1.96] H3 0.57 0.95 .330 1.76 [0.56, 5.52] STEP TWO H1 0.04 0.04 .850 0.96 [0.63, 1.47] H2 0.05 0.03 .862 1.06. [0.57, 1.95] H3 0.59 0.99 .321 1.80 [0.57, 5.72] Education 0.34 1.34 .247 1.40 [0.79, 2.48] Income 0.14 0.15 .696 0.87 [0.43, 1.76] Financial Security 2.78 1.91 .167 0.06 [0.00, 3.19] Age 0.00 0.03 .874 1.00 [0.95, 1.04] Sex (1 = male, 2 = female) 0.82 1.08 300 0.44 [0.10, 2.07] Results for two step hierarchical logistic regression, with information avoidance choice ( Information decision: 0 = did not avoid, 1 = avoided) as the DV. Table 5 4 Predictors of Information Avoidance Proclivity in Study 3a Variab le B t p r p [CI 95% ] STEP ONE H1 0.00 0.09 .930 0.01, [ 0.11, 0.10] H2 0.02 0.02 .688 0.02, [ 0.08, 0.13] H3 0.10 0.07 .178 0.07, [ 0.03, 0.18] STEP TWO H1 0.00 0.05 .962 0.00, [ 0.10, 0.11] H2 0.01 0.25 .807 0.01, [ 0.09, 0.12] H3 0.11 1 .48 .140 0.08, [ 0.03, 0.19] Education 0.02 0.56 .575 0.03, [ 0.08, 0.14] Income 0.06 1.17 .242 0.06, [ 0.04, 0.17] Financial Security 0.11 0.45 .652 0.03, [ 0.08, 0.13] Age 0.01 1.48 .139 0.08, [ 0.19, 0.03] Sex (1 = male, 2 = female) 0.02 0.15 .879 0.01, [ 0.11, 0.10]

PAGE 63

63 Because levels of avoidance in this study were quite low, I could not reliably determine whether the norms reduce information avoidance. Moreover, my manipulation of the injunctive norm was ineffective in this paradigm. It is a lso likely that avoidance in this paradigm was low because, similar to Study 2, participants made their information decision prior to taking the health questionnaire, and thus had little information to determine whether the information was important or wor th avoiding.

PAGE 64

64 CHAPTER 6 ST UDY 3B Study 3b was designed to address a major limitation in Study 3a the low baseline level of information avoidance in the health risk assessment paradigm. To address this limitation, Study 3b used a new paradigm, previousl y demonstrated to elicit high level s of avoidance (Newell & Shepperd, unpublished manuscript). Participants received information about a fictitious but serious medical condition, TAA deficiency, and learned that they could receive their risk information fo r TAA deficiency with an uncomforta ble test that entailed an anal swab. Using this paradigm, I again tested whether participants displayed less avoidance when seeking was normative than when it was not, and that providing descriptive and injunctive norm in formation together would elicit the lowest levels of avoidance overall. Method Participants Participants were 338 University of Florida students ( M age = 19.09, age range = 18 30; 237 White, 57 Asian, 44 Black, 1 Native American, 9 other (participants coul d select more than one racial ident ifier); 283 non Hispanic, 49 Hispanic; 237 female, 101 male ) who received research credit for their participation A post hoc power analysis for Chi square using G power, with a medium effect size w = .21, = .05 reveal ed sufficient power to detect effec ts, power = .92. Procedure Research Assistants (RAs) wearing scrubs greeted participants and led them to a lab room with several computers. Participants sat at a computer with the Qualtrics survey on the screen and the R A briefed them about the study ( Appendix E).

PAGE 65

65 Participants learn ed they would listen to a short podcast about a (fictitious) medical condition known as Thioamine Acetylase Deficiency (TAA Deficiency). After participants consented to participate, participant s listened to the podcast, which led participants to believe TA A Deficiency was a real medical condition ( see Appendix E for script ). Next, p articipants read the following information: Testing for TAA Deficiency involves a simple anal swab which takes ab out 5 minutes to process. We can provide a free screening for T AA Deficiency. This process should only take a few minutes at the conclusion of your study session. Next, I randomly assigned participants to view condition specific information, similar to St udy 3a. After reading condition specific information and answer ing manipulation check items, participants made their information decision and completed the IA scale and brief demographic information. The computer screen then prompted participants to tell t he experimenter they were ready for debriefing. The RA probed f or suspicion and thoroughly debriefed participants. After debriefing, participants selected whether they wished to include their data in analysis. Materials Audience manipulation. Participants in the three norm conditions first read the following text: B efore you make your decision about whether to receive the free anal swab screening for TAA Deficiency, we believe it is important to provide you with additional information. Participants in th e Injunctive Norm condition then received the following informa tion: Learning your risk for TAA Deficiency allows you to make important life choices, to monitor for signs before problems arise, and to plan for the future. In many ways, it is the right thi ng to do. Participants in the Descriptive Norm condition read the following: Most people who are given the option for free testing

PAGE 66

66 choose to get screened for TAA Deficiency. Participants in the Both Norms condition received both the injunctive and descr iptive norm text together. Participant s in the No Norm conditio n received no additional information. Manipulation c heck. Similar to Study 3a, a fter receiving condition specific information, all participants answered two items assessing endorsement of injun ctive norm to seek information ( Choosing to get screened for T AA deficiency is the right thing to do ) and descriptive norm to seek information ( If given a choice, most people will opt to get screened for TAA deficiency ). Both items used a 1 7 scale, wh ere 1 = strongly disagree, 7 = strongly agree. Information dec ision. Participants then chose whether they wanted to receive a TAA Deficiency screening (Yes, I would like to receive a screening for TAA Deficiency) or did not want the screening (No, I am no t interested in a screening for developing TAA Deficiency). In formation avoidance scale. Participants completed the same eight item information avoidance scale (Howell & Shepperd, 2016) used in all previous studies. I adjusted the items to be specific to TAA Deficiency and averaged responses to the items to create a single index of information avoidance ( M = 3.50, SD = 1.22 ). Demographics. Because this study was conducted in lab and I wanted to ensure that the study did not last too long, participants answered a subset of the demographic items used in Study 3a (age, gender, ethnicity, race).

PAGE 67

67 Results and Discussion Zero order C orrelations I initially computed the correlations among the predictors and outcomes (Table 6 1 ). As in Study 3a, information decision correlated significa ntly with endorsing an injunctive norm, such that greater e ndorsement of an injunctive norm to seek corresponded to lower likelihood of information avoidance. This same relationship was true for information avoidance proclivity. Additionally, greater endor sement of a descriptive norm to seek also corresponded to l ower information avoidance and information avoidance proclivity. Table 6 1 Zero order correlations for Study 3b Variable 1. 2. 3. 4. 5. 6. 1. Endorsing injunctive norm 2. Endorsing descriptive norm .43** 3. Age .01 .09 4. Sex (1 = male, 2 = female) .04 .06 .12* 5. Information avoidance scale .49** .39** .11* .05 6. Information decision (0 = did not avoid, 1 = avoided) .38** .36** .10 .07 .46** p < .05, ** p < .01. Manipulation C heck As in Study 3a, I conducted t wo 2x2 ANOVAs to test the main effects of the descriptive norm and injunctive norm, and their interaction on each of the two manipulation check items (endorsement of a descriptive norm for seeking; endorsement of an injunctive norm for seeking; see Table 6 2 for means by condition). Replicating Study 3a r esults suggest that the descriptive norm manipulation was effective. There was a main effect of descriptive norm on endorsement of descriptive norm ( F [1, 3 37 ] =

PAGE 68

68 1 7 57 p < .001 p 2 = .023, but no effect of injunctive norm ( F [1, 3 37 ] = 0. 22 p = 642, p 2 = .001 ) and no interaction ( F [1, 3 37 ] = 0.24 p = 628, p 2 = .001 ). However, as in Study 3a, results s uggest that the injunctive norm manipulation was ineffective. There was no main effect of injunctive norm ( F [1, 3 37 ] = 0.32 p = 574, p 2 = .001 ) or descriptive norm ( F [1, 3 37 ] = 1.80 p = 180, p 2 = .005 ), and no interaction ( F [1, 3 37 ] = 0. 78 p = 379, p 2 = .002 ). Table 6 2 Study 3b manipulation check means by condition Endorsement of Descriptive Norm Endorsement of Injunctive Norm Condition M SD M SD No Norms 3.89 1.54 4.83 1.47 Injunctive Norm 3.74 1.33 5.05 1.32 Descriptive Norm 4.27 1.48 4.76 1.37 Both Norms 4.27 1.62 4.72 1.30 Information A void ance A chi square analysis revealed no significant difference between conditions in 2 (3) = 4 47 p = 215 12 O verall avoidance in this study was much higher than in Study 3a : 76 .6% of participants avoided their health risk information (No Norms: 75.0 %, Injunctive: 75.0 %, Descriptive: 71.8 %, Both Norms: 84.7 %). I also conducted a one way ANOVA using the IA scale as an outcome measure of avoidance. However, results revealed no diff erence in avoidance between conditions, F (3, 3 32 ) = 1.76 p = 155 p 2 = .016 I also conducted a logistic regression using a set of three Helmert contrast codes as predictors, and information avoidance as the outcome. I compared no norms to

PAGE 69

69 norms (H1 : No Norm 3, Both Norms 1, Injunctive 1, Descriptive 1), both norms to individual norms (H2: No Norm 0, Both Norms 2, Injunctive 1, Descriptive 1), and injunctive to descriptive (H3: No Norm 0, Both Norms 0, Injunctive 1, Descriptive 1). As in Study 3a r esults revealed no significant differences for H1 and H3 ( p > .05 ), but unlike Study 3a, results revealed a significant difference in avoidance when comparing Both Norms to the two single norm (Injunctive and Descriptive) conditions (H2): B = 0.23, SE = .12, Wald = 4.00, p = .046, Odds Ratio = 1.26 (1.01, 1.58) P artici pants avoided more in the Both Norms condition compared to the two single norm conditions I conducted a parallel linear regression using the Helmert contrast codes as predictors and the IA scale as the outcome measure. Similar to the logistic regression results, linear regression results revealed no significant differences for H1 and H3 ( p > .05) but did reveal a significant difference for H2: B = .11, SE = .05, t = 2.07, p = .040, again indicating a higher proclivity to avoid TAA risk information in the Both Norms condition compared to the two single norm conditions. Other P redictors of A voidance To assess whether other demographic variables predicted avoidance, I conducted a hierarchical logistic regression, with effects of condition entered in step one (Helmert contrast codes). In step 2, I included the demographic variables (age, sex) (Table 6 3 ). Although the contrast between Both Norms versus the two single norm conditi ons was significant in Step 1, this effect became non significant after adding the demographic predictors in Step 2.

PAGE 70

70 Table 6 3 Predictors of Information Avoidance in Study 3 b Model B Wald Odds ratio [95% CI] STEP ONE H1 0.04 0.26 .613 1.04 [0.90, 1.20] H2 0.23 4.00 .046 1.26 [1.01, 1.58] H3 0.08 0.23 .634 0.92 [0.65, 1.30] STEP TWO H1 0.03 0.16 .690 1.03 [0.89, 1.19] H2 0.22 3.68 .055 1.25 [1.00, 1.57] H3 0.09 0.28 .595 0.91 [0.65, 1.29] Age 0.16 2.66 .103 0.85 [0.70, 1.03] Sex (1 = male, 2 = female) 0.25 0.80 .372 1.28 [0.74, 2.21] Results for two step hierarchical logistic regression, with information avoidance choice ( Information decision: 0 = did not avoid, 1 = avoided) as the DV. p < .05, ** p <.01, ***p < .001 Table 6 4 Predictors of Information Avoidance Proclivity in Study 3b Variable B t p r p [CI 95% ] STEP ONE H1 0.02 0.46 .647 0.03, [ 0.13, 0.08] H2 0.11 2.07 .040 0.11, [0.01, 0.22] H3 0.08 0.88 .381 0.05, [ 0.15, 0.06] STEP TWO H1 0.02 0.54 .590 0.03, [ 0.14, 0.08] H2 0.11 2.00 .046 0.11, [0.00, 0.21] H3 0.09 0.91 .361 0.05, [ 0.16, 0.06] Age 0.11 2.01 .045 0.11, [ 0.21, 0.00] Sex (1 = male, 2 = female) 0.07 0.49 .623 0.03, [ 0.08, 0.13] Similarly, results from a parallel hierarchical linear regression using health risk information avoidance proclivi ty (IA Scale) as the outcome also revealed a significant contrast between Both Norms versus the two single norm conditions. This effect remain ed significant after adding the demographic predictors in Step 2. Additionally, age

PAGE 71

71 was a significant predictor, s uch that older participants expressed less information avoidance proclivity than did younger participants (Table 6 4). Exploratory C ondition C ontrasts I originally predicted that avoidance would be highest in the No Norm condition and lowest in the Both No rms condition. The Helmert contrast codes used in the previous analyses reflect these hypotheses. However, the results from Study 3b suggest t esting a different set of contrast codes. That is, avoidance levels in the No Norm, Injunctive Norm, and Descripti ve Norms conditions were relatively equal, and avoidance was highest in the Both Norms condition, and a more appropriate comparison would be t o compare the Both Norms condition to the three remaining conditions. As such, I conducted an exploratory logisti c regression analysis to determine whether the avoidance was significantly higher in the Both Norms condition compared to the three remaining conditions ( H 4 : Both Norms 3, No Norm 1, Injunctive 1, Descriptive 1; H 5 : Both Norms 0, No Norm 2, Injunctive 1, Descriptive 1; H 6 : Both Norms 0, No Norm 0, Injunctive 1, Descriptive 1). Results revealed that the contrast between Both Norms versus th e three remaining conditions was significant in Step 1, but this effect became non significant after adding the de mographic predictors in Step 2 (Table 6 5 ).

PAGE 72

72 Table 6 5 Predictors of Information Avoidance in Study 3b Model B Wald Odds ratio [95% CI] STEP ONE H4 0.17 4.01 .045 1.18 [1.00, 1.39] H5 0.03 0.07 .787 1.03 [0.84, 1.26] H6 0.08 0.23 .634 1.09 [0.77, 1.53] STEP TWO H4 0.16 3.60 .058 1.17 [1.00, 1.38] H5 0.03 0.07 .787 1.04 [0.85, 1.27] H6 0.08 0.23 .634 1.10 [0.78, 1.55] Age 0.16 2.66 .103 0.85 [0.70, 1.03] Sex (1 = male, 2 = female) 0.25 0.80 .372 1.28 [0.74, 2.21] Results for two step hierarchical logistic regression, with information avoidance choice ( Information decision: 0 = did not avoid, 1 = avoided) as the DV. p < .05, ** p <.01, ***p < .001 Table 6 6 Predictors of Information Avoidance Proclivity in Study 3 b Variable B t p r p [CI 95% ] STEP ONE H4 0.07 1.79 .074 0.10, [ 0.01, 0.20] H5 0.06 1.12 .262 0.06, [ 0.05, 0.17] H6 0.08 0.88 .381 0.05, [ 0.06, 0.15] STEP TWO H4 0.07 1.71 .089 0.09, [ 0.01, 0.20] H5 0.06 1.18 .238 0.07, [ 0.04, 0.17] H6 0.09 0.91 .361 0.05, [ 0.06, 0.15] Age 0.11 2.01 .045 0.11, [ 0.21, 0.00] Sex (1 = male, 2 = female) 0.07 0.49 .623 0.03, [ 0.08, 0.13] p < .05, ** p <.01, ***p < .001. However results from a parallel hierarchical linear regression using health risk informa tion avoidance proclivity (IA Scale) as the outcome did not reveal a significant

PAGE 73

73 contrast between Both Norms versus the remaining three conditions. Age was the only significa nt predictor; such that older participants expressed less information avoidance pr oclivity than younger participants (Table 6 6 ). Summary of Findings Overall, the results suggest that manipulating normative information did not reduce health information avoidance. Indeed, contrary to hypotheses, participants displayed equal levels of avo idance whether presented with either an injunctive norm or a descrip tive norm or no norm. Also contrary to hypotheses, in most analyses, participants displayed higher levels of avoidance in the both norms condition than the remaining three conditions. Stu dy 3b results may indicate similar concerns raised in Study 1b. That is, results seem to suggest that heavy handed manipulations may result in psychological reactance (Brehm, 1966; Rains, 2013). Indeed, perhaps participants in the Both Norms condition disp layed higher levels of avoidance because they perceived the descript ive norm and injunctive norm information together to be an excessive and blatant attempt to encourage less avoidance. Future studies would benefit to include measures of psychological reac tance to test this possibility.

PAGE 74

74 CHAPTER 7 GENERAL DISCUSSION The p resent research aimed to answer the overarching question do audiences influence health information avoidance? I had three aims: (1) To test whether proactive impression management influe nces health information avoidance, (2) To test whether capacity to harm influences health information avoidance, and (3) To test whether including normative information endorsing seeking information would elicit less health information avoida nce. Proactive Impression Management Results from Studies 1a and 1b provide mixed evidence for proactive impression management. Study 1a tested whether participants proactively managed impressions of being a seeker or avoider by seeking or avoiding inform ation. Results from Study 1a suggest that the manipulation was ineff ective, and overall avoidance of STI risk information was low. Study 1b attempted to address the limitations in Study 1a by using a different manipulation and sampling participants at a di fferent time of year (to recruit participants with potentially more sexual experience). Notably, the participants in Study 1b did report a higher (albeit, not statistically significant) average score on the SRS ( M = 3.10, SD = 1.09) than did participants i n Study 1a ( M = 2.96, SD = 0.96). However, the SRS is not a measure of sexual experience specifically, so it is difficult to determine whether time of year affects sexual experience. In addition, Study 1b tested whether participants proactively managed imp ression of being viewed unfavorably (or no differently) if at high r isk for an STI by seeking or avoiding STI risk information. The manipulation produced differences in avoidance by condition, but not in the hypothesized direction. Indeed, participants wer e less likely to avoid if they learned that

PAGE 75

75 people at high risk for an STI are viewed unfavorably than if they learned that people at high risk for an STI are viewed no differently than people at low risk. As discussed previously, these findings in Study 1b may be the result of psychological reactance participants may hav e perceived the unfavorable image manipulation as a heavy handed nudge towards avoiding risk information, and subsequently exerted their freedom by opting to seek instead of avoid. Alterna tively, participants may have viewed the manipulations as indicating the seriousness of the risk feedback. As such, participants in the Unfavorable Image condition may have perceived serious interpersonal costs to being at high risk for an STI, prompting t hem to learn their status and potentially mitigate interpersonal cos ts. By contrast, participants in the Control condition may have perceived minimal interpersonal costs to being at high risk for an STI, prompting them to view the STI risk information as l ess useful and to avoid their risk feedback. Notably, exploratory an alyses revealed that neither perceived risk nor actual risk moderated the effect of condition on information avoidance. Taken together, it remains unclear whether people engage in proacti ve impression management via information seeking or avoidance. The l ack of clarity suggests two directions for future research. First, researchers may benefit from using a health paradigm where participants do not have preconceptions about how people at hi gh risk are viewed. Although Studies 1a and 1b aimed for external va lidity by employing an important real perceptions of people at high risk for an STI may be too difficult to change with a brief text manipulation. Perhaps using a fictious disease (e.g., TAA Defic iency, used in Study 3b) would elicit different results. Additionally, participants may have responded

PAGE 76

76 differently had we offered diagnostic testing for an STI rather than feedback regardi ng participants risk for an STI. Second, researchers may also benefi t from developing manipulations that are more subtle in their attempts to influence participants, therefore minimizing potential reactance. Capacity to Harm Results from Stud y 2 provide d mixed evidence that capacity to harm drives audience related avoidan ce of health risk information. The results from Study 2 indicated that both the type of audience (insurers versus researchers) and the capacity to harm influence health risk information av oidance. When comparing insurers (both with the capacity to harm and without the capacity to harm) to researchers (with no capacity to harm), participants in the insurer conditions displayed higher levels of information avoidance than participants in the r esearcher condition. Yet, when comparing the capacity to harm to the incapacity to harm, participants displayed higher levels of avoidance in the Insurer Harm condition than the two no harm conditions (Insurer No Harm and Researcher No Harm). These results are consistent with the finding that perceived likelihood of harm o nly partially mediates avoidance of health risk information accessible to powerful audiences (Lipsey & Shepperd, 2019b). It seems that itself are likely independent predictors of health information avoid ance. Results from Study 2 raise important questions: What makes an audience powerful beyond capacity to harm? What other aspects of powerful audiences lead people to avoid information? T here are several possibilities. One possibility is that audience tru st is an important factor in determining whether people allow other audiences to access their information. People may acknowledge that an audience has

PAGE 77

77 the capacity to harm, but if they tru st the audience not to use that capacity, they may be unconcerned th at the audience will have access to the information. Similarly, an audience may have limited capacity to harm, but if people do not trust the audience, they may be unwilling to allow the a udience access to that information. Accordingly, it is possible part icipants in the Insurer No Harm condition did not fully believe that insurers would not cause any harm. Lending support to this possibility, research examining predictors of trust in syste ms that share health information for research and health care practi ce find that health care, altruism, and generalized trust positively predicted system trust. Convers ely, privacy concerns and knowledge about health information sharing negatively predicted system trust (Platt & Kardia, 2015). Although these results do not examine whether system trust predicts information avoidance or information withholding, it seems fe asible that having lower system trust predicts greater information a voidance/withholding. Experiences with healthcare providers also predict system trust. Specifically, regular contact with a healthcare provider (i.e. visiting the provider once or more a year) predicte d trust in the health system (Platt & Kardia, 2015). M ore tellingly, research finds that the quality of care people receive from healthcare providers predicted the probability of withholding information from the provider because of concerns a bout privacy and security of their health information (Patel, Beckjo rd, Moser, Hughes, & Hesse, 2015). Taken together, these results suggest that people have variable levels of trust in their healthcare providers, and that having less positive experiences with

PAGE 78

78 healthcare providers contributes to less willingness to allow h ealthcare providers access to health information. These results suggest that trust may be an important mediator explaining why people seek or avoid health information on learning that insu rers or employers would be privy to that information. Another relate d possibility is that participants are primarily concerned with privacy and unauthorized audiences accessing their information. Participants may have been fine with powerful audiences acce ssing their information so long as the information remained private and secure (is not shared with unauthorized audiences). In support of this hypothesis, research on perceptions of privacy and security of electronic health records (EHRs) indicates that a sizeable minority of American adults lacked confidence in the securi ty and privacy of their EHRs (Patel et al., 2015). Additionally, nearly 60% of participants indicated at least moderate concern about unauthorized individuals viewing their health informat ion when sent between health care providers (Patel et al.). These results suggest that people may be more concerned with information privacy and less concerned with health care providers (or insurers) using their information to cause harm (via added costs or difficulties obtaining coverage). Results from Study 2 suggest avenues for future research. Researchers could test whether trust in health insurers, healthcare providers, and other powerful audiences affects levels of health information avoidance. Resea rch ers could also test whether manipulating or measuring privacy concerns (rather than harm concerns) predict health information avoidance. Injunctive and Descriptive Norms Results from Stud ies 3a and 3b suggest that employing social norms interventions to re duce health information avoidance may not be effective. In both Study

PAGE 79

79 3a and 3b, social norm manipulations elicited little to no difference in avoidance by condition. The manipulation only seemed to affect avoidance in the Both Norms condition, but cont rar y to predictions, participants avoided more in the Both Norms condition than participants in the remaining three conditions. The context of the intervention (i.e., what behavior researchers are attempting to change) could explain differences in the e ffi cacy of social norm interventions. Indeed, perhaps people are more receptive to consider normative information in certain contexts. Why social norms interventions are more effective in certain contexts (e.g., environmental behavior [Cialdini, 2003; Schu ltz et al., 2007] drinking behavior [Neighbors et al., 2004 ) than other contexts (fruit consumption [Stok, De Ridder, De Vet, & De Wit, 2014], information avoidance) remains unclear. Another important factor to consider is the manipulation itself. Studie s v ary considerably in how elaborate or personalized the normative manipulations are. For example, Schul t z et al. (2007) effectively manipulate descriptive norms using a con sumption. Additionally, the researchers effectively manipulate injunctive norms by using a smile emoticon if energy consumption was below average, and a sad emoticon if energy consumption was above average. Thus, Schul t z et al. provide a more personaliz ed compared to other people. By contrast, Stok et al. (2014) use single sentence manipulations for injunctive and descriptive norms. Their results indicate the descriptive norm was so mew hat effective, whereas the injunctive norm used in their study seemed to elicit psychological reactance. Perhaps more personalized/elaborate interventions are

PAGE 80

80 more successful, to the extent that they do not appear to be obvious manipulations of behavior F inally, it is possible that the reference point for the social norms influenced how participants responded. For example, the injunctive norm manipulation in the present studies may have elicited lower levels of avoidance if participants had learned that th eir parents or their close family members believed they should get tested for TAA deficiency. Likewise, the descriptive norm manipulation may have been more effective if participants had learned that most people their same age/gender or most students at th e University of Florida chose to get tested for TAA deficiency. Limitations The samples in my five studies were volunteers and students who were mostly White, female, affluent and educated The sample characteristics limit the generalizability of my r esu lts. Notably, several of the sample characteristics predominant in the present studies are sources of sociopolitical power. For example, research indicates that both healthcare and employment systems disproportionately benefit White people (Malat, Mayor ga Gallet, & Williams, 2018). Research also indicates that race is a predictor of trust in physicians and insurers, such that non Hispanic Black people report less trust in their physicians and more trust in their health insurance plans compared with non H isp anic White people (Boulware, Cooper, Ratner, LaVeist, & Powe, 2003). Tellingly, Non Hispanic B lack participants (compared with non Hispanic White people) report greater concern about personal privacy and potential for harmful experimentation in hospita ls (Boulware et al., 2003). Although knowledge of health

PAGE 81

81 Tuskegee Syphilis Study) does not directly predict willingness to participate in biomedical research (Katz et al. 2008), the history of mistreatment combined with a healthcare system that disproportionately benefits White people (Malat et al., 2018) are important factors to consider in health research. As such, it seems plausible that studies conducted with a lar ger proportion of non White participants might elicit greater health information for audience reasons. Another possible limitation in these studies was the lack of attention check items or items assessing believability. Participants in the present studies, especially the studies conducted online (Studies 1a, 1b, 2, and 3a), may not have dedicated full atten tion to the experimental manipulations or study items, or may have doubted the manipulation information. Although the manipulation check items help deter mine, to some degree, whether participants attended to and believed the manipulation, these items canno t diagnose lack of attention or believability specifically. Indeed, for studies where the manipulation appeared to fail, it is difficult to determine whe ther the manipulation was ineffective due to manipulation wording, manipulation check wording, inattent ion to the manipulation, inattention to the manipulation check, lack of believability, or something else. Finally, although I aimed to create an experie nce that mimics real world health decision making, the present studies (especially the online studies: Studies 1a, 1b, 2, and 3a) likely failed to induce the level of concern with privacy or harm that occurs when people are in a healthcare setting facing r eal medical decisions. It remains unknown whether the results found in the present studies would replic ate in actual healthcare settings.

PAGE 82

82 Moderators of Interpersonal Information Avoidance I used three different paradigms to study health information avoidan ce: STI risk (Studies 1a and 1b), General health risk (Studies 2 and 3a), and a fictious medical condit ion tested using an anal swab (Study 3b). Across the five studies, level of avoidance varied considerably, suggesting that specific health information or context are important moderators of interpersonal information avoidance. Several factors may moderate the extent to which interpersonal concerns influence health information decisions. Perhaps most obvious is the type of audience. Concerns about how other s may view the information decision, use the information, and view the decision maker may vary consider ably across audiences. Audiences differ in several important ways (e.g., their capacity to harm, their ability to provide social support, their familiari ty, the level of trust they elicit, etc.), and these differences likely influence how motivated a perso n is to seek or avoid information. Put simply, different audiences may have different effects on each of the proposed interpersonal pathways. Concerning the potential to harm pathway, certain audiences have greater pacity to harm those resources, or elicit greater mistrust than do other audiences. For example, in a health information decision context, people general ly perceive the greatest likelihood of harm from health insurers and express less concern about other a udiences that have a lower capacity to cause harm (Lipsey & Shepperd, 2019a). The present research indicates that this relationship between powerful audi ence and health information avoidance is more complicated than originally hypothesized. Capacity to har m is one potential moderator of audience influence, but there are likely other important factors to consider (e.g., trust or privacy concerns).

PAGE 83

83 Concernin g social norms, one study found that perceiv ed social pressures from peers predicted less information a voidance. However, perceived social pressure from parents did not significant ly alter information decisions (Calhoun, 2012) Recent research also demonst rated that both descriptive norms (referencing other study participants) and injunctive norms (referenc ing friends and separately referencing family) predicted intentions to receive genomic sequencing results, indicating that, at least in the context of ge nomic sequencing, several different normative reference groups (study participants, friends, and family ) are all associated with information seeking intentions (Reid et al., 2018). Although the present research and previous research together indicate that several audience characteristics may be important predictors of health information seeking or avoidance, I know of no research that systematically examines how various audience characteristics influence information decisions. Future research would benefit from exploring various audience characteristics together and their independent influences on h ealth information decisions. Characteristics of the health information may also moderate health information decisions. People may consider treatability, cost of treatment, seriousness, and condition stigma and embarrassment when deciding whether they want to learn their risk or prognosis for that health condition. Thus, skin damage may differ from HIV and tability, and cost to treat. As a result, people may display differing levels of avoidance of a UV photo indicating skin damage (Dwyer, Shepperd, & Stock, 2015), versus an HIV test (Sullivan et al., on, 2003). In one study,

PAGE 84

84 participants indicated that they would be less likely to avoid a serio us medical condition (as opposed to a minor medical condition), more likely to avoid an untreatable medical condition than a treatable medical condition, more li kely to avoid an expensive to treat than an inexpensive to treat medical condition, and more li kely to avoid an embarrassing/stigmatizing medical condition than a non stigmatizing/non embarrassing medical condition (Lipsey & Shepperd, 2019b). These results suggest that variations in the underlying psychological meaning of different medical condition s can elicit different levels of information avoidance. Additionally, just as personal reasons for avoiding information do not exist in a vacuum, neither do inte rpersonal reasons for avoiding information. Personal and interpersonal reasons for avoiding inf ormation may interact and affect information avoidance. For example, participants in Study 1b learned that others would view them negatively if they were at high risk for an STI. This interpersonal information may not affect avoidance directly, but rather may affect avoidance through its interaction with risk information would change how they think about themselves, feel about themselves, or obligate them to get diagnostic STI testing (Sweeny et al., 2010). As such, futu re studies would benefit to measure both personal and interpersonal reasons for avoidance, and test how these variables interact to predict information avoidance Conclusion Many important factors influence the decision to seek or avoid health information. Although many studies have examined personal reasons for avoiding information (e.g., Sweeny et al., 2010), little research has examined the influence of other p eople on

PAGE 85

85 management, capacity to harm, and social norms. My studies provide unclear evidence supporting or refuting the concept of proactive impression management and social norm interventions. However, the results provided preliminary evidence that the audience and tion avoidance. Overall, results from five studies provide some initial insight into the vario us ways audiences may influence (or not influence) health information avoidance. Although audiences may have a limited influence on health information avoidance in some contexts, results suggest that manipulating audience influence can affect health inform ation decisions. As such, it remains important for health researchers and on decisions.

PAGE 86

86 APPENDIX A STUDY 1A MATERIALS Study 1a Overview & Informed Consent This survey will ask you questions about your sexual health risk information and decisions. The survey should take no more than 15 minutes to complete. Informed Consent Pl ease read this consent form carefully before you decide to participate in this study. Purpose of the research study: This study examines sexual risk health decisions. What you will be asked to do in the study: You will complete a questionnaire about your sexual health risk and answer questions about your desire for sexual health risk infor mation. Time required: 7 15 minutes Compensation: (For Participant Pool participants): You will receive 1 credit for your participation. (For extra credit participa nts): You will receive the extra credit designated by your instructor, not to exceed 1% of your final grade points for the course. Confidentiality: Your responses to all items will be confidential and anonymous. There is no connection between any identify ing information and your responses. Your name and identify will not be disclos ed in any published reports. The online host for this survey is Qualtrics. There is a minimal risk that security of any online data may be breached, but because no identifying i nformation will be collected, and because the online host (Qualtrics) uses sev eral forms of encryption, it is unlikely that a security breach of online data will result in any adverse consequence for you. Voluntary participation & right to withdraw: Your participation is completely voluntary. There is no penalty for not participa ting. You have the right not to respond to any questionnaire item that you feel uncomfortable answering. You have the right to withdraw from the study at any time without consequ ence. Whom to contact if I have questions about the study: Investigators: N ikolette P. Lipsey nlipsey@ufl.edu & Dr. James A. Shepperd Shepperd@ufl.edu Dept of Psychology, Un iv of Florida Whom to contact about my rights as a research participant in th e study: UFIRB office, Box 112250, University of Florida, Gainesville, FL 32611 2250; ph 392 0433. I acknowledge that I have read the consent form and agree to participate in the study.

PAGE 87

87 I DO NOT wish to participate in this study.

PAGE 88

88 Before proceeding to the survey, we need to ask you one question to assess your eligibility for this study. If you a re not eligible for the study, you will be sent to the end of the survey wh ere we will document your participation. You will still receive the extra credit designated by your instructor for your participation. Extra credit compensation will not exceed 1% o f your final grade points for the course. Are you (or have you ever been) sexually active? Yes No ( if no is selected participants are sent to the end of the survey)

PAGE 89

89 Sexual Risk Scale (DeHart & Birkimer, 1997) option for all items 1 7 Likert scale, 1 = disagree strongly, 4 = neither agree nor disagree, 7 = agree strongly 1. If my partner wanted me to have unprotected sex, I would probably give in. 2. The proper use of a condom could enhance sexual pleasure. 3. I ma y have had sex with someone who was at risk for a sexually transmitted infection ( STI). 4. If I were going to have sex, I would take precautions to reduce my risk of a sexually transmitted infection (STI). 5. Condoms ruin the natural sex act. 6. When I think that one of my friends might have sex on a date, I ask him/her if he/she has a cond om. 7. I am at risk for a sexually transmitted infection (STI). 8. I try to use condoms when I have sex. 9. Condoms interfere with romance. 10. My friends talk a lot about saf er sex. 11. If my partner wanted me to participate in risky sex and I said that we needed t o be safer, we would still probably end up having risky sex. 12. Generally, I am in favor of using condoms. 13. I would avoid using condoms if at all possible. 14. If a friend knew that I might have sex on a date, he/she would ask me whether I was carrying a condom 15. There is a possibility that I have a sexually transmitted infection (STI). 16. If I had a date, I would probably not drink alcohol or use drugs. 17. Safer sex reduces the physical pleasure of sex. 18. If I thought that one of my friends had sex on a date, I would ask him/her if he/she used a condom. 19. 20. Safer sex is a habit for me. 21. If a friend knew that I had sex on a date, he/s used a condom or not. 22. If my partner wanted me to participa te in risky sex and I suggested a lower risk alternative, we would have the safer sex instead. 23. The sensory aspects (smell, touch, etc.) of condoms make them unpleasant 24. 25. really give yourself over to your partner. 26. I am determined to practice safer sex. 27. If my partner wanted me to have unprotected sex and I made some excuse to use a c ondom, we would still end up having unprotected sex. 28. If I had sex and I told my friends t hat I did not use condoms, they would be angry or disappointed. 29. I think safer sex would get boring fast. 30. My sexual experiences do not put me at risk for a sexually t ransmitted infection (STI).

PAGE 90

90 31. Condoms are irritating. 32. My friends and I encourage each ot her to practice safer sex. 33. When I socialize, I usually drink alcohol or use drugs. 34. If I were going to have sex in the next year, I would use condoms. 35. If a using condoms. 36. People can get t he same pleasure from safer sex as from unprotected sex. 37. Using condoms interrupts sex play. 38. It is a hassle to use condoms. Additional Questions 14. Based on your responses to the previous questions, how at risk for an STI do you believe you are? 1 7 sca le, 1 = very low risk, 4 = moderate risk, 7 = very high risk 15. Have you undergone testing for an STI? o If yes: When was the last time you were tested for an ST I? Within the last month Within the last 6 months Within the last year Within the last 3 years More than 3 years ago

PAGE 91

91 Demographics 1. What is your age? ______ 2. What is your gender? Male Female Transgender Non binary/Genderqueer Prefer to self describe: _____________ Choose not to respond 3. Which of the following best describes your sexual orienta tion? Heterosexual (straight) Homosexual (gay/lesbian) Bisexual Queer Prefer to self describe: _____________ Choose not to respond 4. P lease specify your ethnicity. Hispanic or Latino Not Hispanic or Latino Choose not to respond 5. Please specify your r ace. White Asian / Pacific Islander Indian (from India) Black or African American Native American or American Indian Other (please specify below) Choose not to respond 6. What is your highest level of education? No high school (less than grade 9) High school (grades 9 12, no degree) High school graduate (or equivalent) Some co llege (1 4 years, no degree) Associates degree College/university degree (for example, BA or BS) Some graduate school

PAGE 92

92 Graduate degree Choose not to respond 7. Wha t is your marital status? Single, never married Divorced Married or domestic partnership Separated Widowed Choose not to respond 8. What is your current employment status? Employed for wages Student Self employed Military Unemployed Retired Homem aker Choose not to respond 9. Do you currently have health insurance? Yes/No/C hoose not to respond 10. Which of these statements best describes your present financial status? I manage to get by I have enough to manage pl us some extra Money is not a problem; I can buy about whatever I want Choose not to respond 11. If you were faced with an unexpected $500 medical bill that was not covered by insurance, how would you best describe your situation? Not able to pay the bill Able to pay but with difficulty Able to pay comfortably Choose not to respond 12. What is your annual household income from all sources? $25,000 or less $25,001 $50,000 $50,001 to $75,000

PAGE 93

93 $75,001 or more 13. In general, people often face risks when making financial, career, or other life decisions. Overall, do you feel comfortable, uncomfortable, or neither comfortable nor uncomfortable taking risks? Extremely comfortable taking risks Moderately comfortable taking risks Slightly comfortable taking risks Neither comfortable nor uncomfortable taking risks Slightly uncomfortable taking risks Moderately uncomfortable taking risks Extremely uncomfortable taking risks 14. Indicate your political orientatio n: Very Liberal ( 3) to Very Conservative (+3)

PAGE 94

94 Audi ence views manipulation and manipulation check Seeking Favorable: In a moment, you will have the choice of whether to receive your personal sexually transmitted infection (STI) risk results. In our research we find that half of people opt to receive risk results and half do not. We also find that people who opt to receive their risk results are viewed favorably by others they are viewed as responsible and proactive about their health. Deciding whethe r to receive your risk results is an important decision. In one survey, 89% of participants agreed that they would respect a person who chose to seek personal health risk information. Avoiding Favorable: In a moment, you will have the choice of whether t o receive your personal sexually transmitted infection ( STI) risk results. In our research we find that half of people opt to receive risk results and half do not. We also find that people who opt not to receive their risk results are viewed favorably by o thers they are viewed as thoughtful and prudent about th eir health. Deciding whether to receive your risk results is an important decision. In one survey, 89% of participants agreed that they would respect a person who chose not to seek personal health ris k information. Control: In a moment, you will have the choice of whether to receive your personal sexually transmitted infection (STI) risk results. In our research we find that half of people opt to receive risk results and half do not. Manipulation Ch eck Not at all True Very True 1. If I choose not to rece ive my STI risk results, other people will view me unfavorably. 1 2 3 4 5 6 7 2. I think that most people given the choice would choose to learn their STI risk results. 1 2 3 4 5 6 7 3. People will respect me if I choose not to receive my risk results. 1 2 3 4 5 6 7 4. Most people will view me as thoughtful and prudent if I choose not to receive my risk results. 1 2 3 4 5 6 7 5. Most people will view me as responsible and proactive if I choose to receive my risk results. 1 2 3 4 5 6 7

PAGE 95

95 Information Avo idance Scale and Feedback Decision Questionnaire Feedback Decision Questionnaire Please select one of the options below: Option 1 No, I am not interested in learning my STI risk results. Option 2 Yes, please tell me my STI risk results. We are inter ested in your thoughts regarding this decision. Please write down any thoughts you had while making this decision, and provide your thoughts for the following question. What thoughts led you to choose the option you chose? ______________________ ________________________________________________ ______________________________________________________________________ ________________________________________________ ______________________ Information Avoidance Scale IA Scale Strongly Disagree Stron gly Agree 1. I would rather not know my risk for sexually transmitted infections (STIs). 1 2 3 4 5 6 7 2. I would avoid learning my risk for sexually transmitted infections (STIs). 1 2 3 4 5 6 7 3. Even if it will upset me, I want to know my risk for sexual ly transmitted infections (STIs). 1 2 3 4 5 6 7 4. When it comes to my risk for sexually transmitted infections (STIs), sometimes ignorance is bliss. 1 2 3 4 5 6 7 5. I want to know my risk for sexually transmitted infections (STIs). 1 2 3 4 5 6 7 6. I can think of situations in which I would rather not know my risk for sexually transmitted infections (STIs). 1 2 3 4 5 6 7 7. It is important to know my risk for sexually transmitted infections (STIs). 1 2 3 4 5 6 7 8. I would want to know my risk for sexuall y transmitted infections (STIs) immediately. 1 2 3 4 5 6 7

PAGE 96

96 Feedback (for those who opt to receive it) High Risk: Results from the Sexual Risk Survey indicate you are at HIGH RISK for having or contracting a sexually transmitted infection (STI). It is important to note that these risk results are only an estimate based on your responses. We encourage everyone who is sexually active to get tested regularly for STIs. Low Risk: Results from the Sexual Risk Survey indicate you are at LOW RISK for having or contracting a sexually transmitted infection (STI). It is important to note that these risk results are only an est imate based on your responses. We encourage everyone who is sexually active to get tested regularly for STIs.

PAGE 97

97 Debriefing and Data Relea se Statement. Thank you for participating in this study. Your responses will be helpful in our research. The purpose of this study: 1. What we are examining in this experiment are factors that influence the decision past studies that people sometimes opt to avoid scary health information such as the result of a medical or genetic t est. Such information can be unsettling and sometimes people would just rather not know. 2. In this study, we wanted to examine whether peop le might be more likely to seek/avoid sexual risk feedback if we advised people that others would view their seeking/a voidance decision favorably. 3. We are not interested in your individual response but rather in the response of all participants in the study Thus your responses will be combined with the responses of other participants to create a group or average response You may contact Dr. James Shepperd, Box 112250, phone number 352 273 2165, email, shepperd@ufl. edu or Nikolette Lipsey, email, nlipsey@ufl.edu if you have any feedback or concerns. If you have any questions about your rights as a research participant, you can contact the IRB office at UFIRB Office, Box 11225 0, University of Florida, Gainesville, FL 32611 2250, ph 352 392 0433, IRB2@ufl.edu If you requested to receive your sexual risk feedback, we provided you with results. If your results indicated you were at high risk, we recommend you get tested for STIs if you have not recently done so. You have the op portunity to choose whether you want your data included in our study or whether you would like to withdraw your data at this point. I would like to point out again, as the consent form stated, that no data are analyzed individually. We combine all of the d ata from all of our participants and look at the group averages. Data Release Statement Now that I have received a full disclosure of the purpose of experiment protoco l 2017 02410, I give permission for the researchers to use my responses in their data an alyses. I give the researchers in this study permission to analyze my responses in this study. I do NOT give permission to analyze my responses in this study.

PAGE 98

98 APPENDIX B STUDY 1B MATERIALS Study 1b Overview & Informed Consent Same as Stud y 1a Sexual Risk S urvey Same as Study 1a Additional questions Same as Study 1a with one new question added: 1. How many sexual partners have you had in the past year? ______ (Choose not to respond was an option) Demographics Same as Study 1a Audience views manipulation manipulation reinforcement, and manipulation check Avoidance condition: We believe that it is important to inform you about research on how people view individuals who are at high risk. Sadly, research finds that the vast majority of people view high r isk individuals less favorably than low risk individuals. We tell you this i nformation so that you can make an informed decision. Control condition: We believe that it is important to inform you about research on how people view individuals who are at h igh risk. Fortunately, research finds that the vast majority of people do no t view high risk individuals any less favorably than low risk individuals. We tell you this information so that you can make an informed decision. Manipulation reinforcement: Pa Participants in the wrong (based on their randomly assigned condition), we will tell them the correct answer and proceed. We want to make sure everyone understands what research shows regarding how people v iew high STI risk individuals. Research shows that people generally: a. View people at high risk more favorably b. View people at high risk less favorably c. View people at high risk the sa me as people at low risk. Manipulation Check Not at all true Very true 1. If I receive high risk feedback, other people will view me unfavorably. 1 2 3 4 5 6 7 2. If I receive high risk feedback, people will not view me negatively. 1 2 3 4 5 6 7 Inf ormation avoidance decision and scale Same as Study 1a Feedback (for those who opt to receive it) Same as Study 1a

PAGE 99

99 Debriefing Thank you for participating in this study. Your responses will be helpful in our research. The purpose of this study: 1. What we are examining in this experiment are factors that influence the decision opt to avoid scary health information such as the result of a medical or genetic test. Such informati on can be unsettling and sometimes people would just rather not know. 2. In this study, we wanted to examine whether people might be more likely to seek/avoid sexual risk feedback if we advised people that others would view high risk individuals unfavorably. 3. Some participants were told that research shows that high risk individuals are viewed less favorably by others. Some participants were told that high risk individuals and low risk individuals are not viewed differently. Actual research on this topic is no t conclusive, though generally high risk individuals are viewed less favorably overall. 4. We are not interested in your individual response but rather in the response of all participants in the study. Thus your responses will be combined with the responses of other participants to create a group or average re sponse. You may contact Dr. James Shepperd, Box 112250, phone number 352 273 2165, email, shepperd@ufl.edu or Nikolette Lipsey, email, nlipsey@ufl.edu if you have any feedbac k or concerns. If you have any questions about your rights as a research participant, you can contact the IRB office at UFIRB Office, Box 112250, University of Florida, Gainesville, FL 32611 2250, ph 352 392 0433, IRB2@u fl.edu If you requested to receive your sexual risk feedback, we provided you with results. If your results indicated you were at high risk, we recommend you get tested for STIs if you have not recentl y done so. You have the opportunity to choose wheth er you want your data included in our study or whether you would like to withdraw your data at this point. I would like to point out again, as the consent form stated, that no data are analyzed individua lly. We combine all of the data from all of our parti cipants and look at the group averages. Data Release Statement Now that I have received a full disclosure of the purpose of experiment protocol 2017 02410, I give permission for the researchers to use my responses in their data analyses. I give the researchers in this study permission to analyze my responses in this study. I do NOT give permission to analyze my responses in this study.

PAGE 100

100 APPENDIX C STUDY 2 MATERIALS Study 2 Study Overview & Informed Consent This survey will ask you ques tions abo ut your health risk and decisions. The survey should take no more than 15 minutes to complete. Informed Consent Please read this consent form carefully before you decide to participate in this study. Purpose of the research study: This study ex amines health risk decisions. What you will be asked to do in the study: You will complete questionnaires about your health and your desire for health risk information. Time required: 7 15 minutes Compensation: You will receive no compensation for you r participation. Confidentiality: There is a minimal risk that security of any online data may be breached, but our survey host (QUALTRICS) uses strong encryption and other data security methods to protect your inf ormation. Your data will be analyzed by group averages and not by individual responses. Complete confidentiality cannot be guaranteed. Voluntary participation & right to withdraw: Your participation is completely voluntary. There is no penalty for not pa rticipating. You have the right not to r espond to any questionnaire item that you feel uncomfortable answering. You have the right to withdraw from the study at any time without consequence. Whom to contact if I have questions about the study: Investiga tors: Nikolette P. Lipsey nlipsey@ufl.edu & Dr. James A. Shepperd Shepperd@ufl.edu Dept of Psychology, Univ of Florida Whom to contact about my rights as a research participan t in the study: UFIRB office, Box 112250 University of Florida, Gainesville, FL 32611 2250; ph 352 392 0433. I acknowledge that I have read the consent form and agree to participate in the study. I DO NOT wish to participate in this study.

PAGE 101

101 Personal Information First Name? Last Nam e? What is your current home address? Audience views manipulation and manipulation check In this study, we will ask questions your risk for several medical conditio ns (heart disease, diabetes, and some forms of cancer). You can decide whether or not you wish to learn your health risk information. Before you make your decision and take the risk assessment, we believe that it is important to inform you about what wil l happen if you elect to receive your risk results. Insurer; Capacity to harm: As part of your participation in this study, if you elect to receive your risk results, your results will automatically be sent to a registry that is accessible by health insu rance companies. However, per contract with University of Florida, please be advised that we cannot control how insurance companies use this information in the future. Although current laws forbid insurance companies from raising premiums or denying covera ge because of preexisting conditions, the law has not always been successful in protectin g consumers. Moreover, it is possible that the law will be changed or repealed in ways that are unfavorable to people with preexisting conditions. Insurer; No capacity to harm: As part of your participation in this study, if you elect to receive your ris k results, your results will automatically be sent to a registry that is accessible by health insurance companies. Thus, insurance companies can lear n if you are at risk for the health conditions assessed. However, per contract with University of Florida, health insurance companies CANNOT use this information in any capacity to increase premiums, increase the deductible, deny coverage, or limit access to providers. Health Researcher; No capacity to harm: If you elect to receive your risk results, your res ults will automatically be sent to a registry that is accessible by health scientists and researchers. However, as per the Institutional Review Board requirements, researchers will maintain absolute confidentiality and your results will not be used in any capacity beyond research

PAGE 102

102 Manipulation Check Not at all Fearful Very Fearful 1. How fearful are you that you will experience higher insurance rates or denial of coverage if you receive results that you are at high risk for the health conditions assessed in this study? 1 2 3 4 5 6 7 Not at all Worried Very Worried 2. How worried are you that insurance companies might have access to your risk results if you choose to receive them? 1 2 3 4 5 6 7 Very Unlikely Very Likely 3. How likely is it that insu rance will raise your premiums or deny you coverage if they learn that you are at high risk for the health conditions assessed in this study? 1 2 3 4 5 6 7

PAGE 103

103 Information Decision and Information Avoidance Scale Later in this study you will complete a risk assessment. Some people want to see the results of the risk assessment whereas others do not. After you complete the study, you can choose to learn your risk for the conditions assessed in this study (heart disease, diabetes, and some forms of cancer ) based on the assessment. Please select one of the options below: Option 1 No, I am not interested in learning my risk results. Option 2 Yes, please tell me my risk results. We are interested in your thoughts regarding this decision. Please write down any thoughts you had while making this decision, and provide your thoughts for the following question. What thoughts led you to choose the option you chose? ______________________________________________________________________ _______________ _______________________________________________________ Information Avoidance Scal e IA Scale Strongly Disagree Strongly Agree 1. I would rather not know my risk for the conditions assessed in this study. 1 2 3 4 5 6 7 2. I would avoid learning my risk for the conditions assessed in this study. 1 2 3 4 5 6 7 3. Even if it will upset me, I want to know my risk for the conditions assessed in this study. 1 2 3 4 5 6 7 4. When it comes to my risk for the conditions assessed in this study, sometimes ignorance is bliss. 1 2 3 4 5 6 7 5. I want to know my risk for the conditions assessed in this study. 1 2 3 4 5 6 7 6. I can think of situations in which I would rather not know my risk for the conditions assessed in this study. 1 2 3 4 5 6 7 7. It is important to kn ow my risk for the conditions assessed in this study. 1 2 3 4 5 6 7 8. I would want to know my risk for the conditions assessed in this study immediately. 1 2 3 4 5 6 7

PAGE 104

104 Lifestyle Health Questionnaire ms 1. What is your age (in years)? ______ 2. In the last 12 months how often have you participated in some kind of exercise? Not at all 1 to 2 times per month 1 to 2 times per week 3 to 4 times per week More than 3 to 4 times per week 3. What is your height in inc hes? (12 inches = 1 foot) _________ inches 4. What is your weight in pounds? _________ pounds 5. Have you or someone in your family ever been told that you/they have high blood pressure (hypertension) or have you ever been given blood pressure medication? Yes No Not sure 6. Have you or someone in your family ever h ad a heart attack or been told that you/they have heart disease? Yes No Not sure 7. Have you or someone in your family ever been told that you/they have diabetes or a problem with high blood sugar? Yes No Not sure 8. Have you or someone in your family ever b een told that you/they have high cholesterol? Yes No Not sure 9. Do you smoke or chew tobacco? Yes No Not sure 10. (If yes) How many times a day do you smoke or chew tobacco? 1 2

PAGE 105

105 3 4 5 or more 11. Are you e xposed to smoke from other people's cigarettes or cigars? Regularly Occasionally Rarely Never 12. How many times a week do you eat 5 or more servings of fruit and vegetables in a day? A serving is one medium apple, banana or orange, 1 cup of raw leafy vegeta ble (like spinach or lettuce), cup of cooked beans or peas, cup of chopped, cooked or canned fruit/vegetable or cup of fruit/vegetable juice. Rarely/Never 1 to 2 times per week 3 to 4 times per week More than 3 to 4 times per week 13. How many times a we ek do you eat 3 or more servings of whole grains in a day (wheat bread, whole grain pasta, brown rice, oatmeal, whole grain breakfast cereal, bran or popcorn)? A serving is one slice of bread, 1 ounce of breakfast cereal or cup of cooked cereal, pasta o r rice. Rarely/Never 1 to 2 times pe r week 3 to 4 times per week More than 3 to 4 times per week 14. How many times a week do you usually eat 2 or more servings of butter, lard, red meat, cheese or whole milk 2 in a day? Rarely/Never 1 to 2 times per week 3 to 4 times per week More than 3 to 4 times per week 15. How many servings of alcohol do you have on a typical day? One serving is a can of beer, a glass of wine or a shot of hard liquor. 0 1 2 3 or more 16. What is your total cholesterol level? Low Less than 2 00mg/dL Borderline High 200 239mg /dL High 240 mg/dL 17. What is your Blood Pressure?

PAGE 106

106 Normal 139/89 or below Mild Hypertension 140/90 160/100 Moderate Hypertension 161/101 120/200 Severe Hypertension above 200/above 120 D emographics Same as Study 1a and Study 1b

PAGE 107

107 Debriefing and Data Release Statement Thank you for participating in this study. Your responses will be helpful in our research. If you requested to receive your risk results, unfortunately we are unable to provide them to you, the questionnaire we use is not diagnostic (see information below). If you would like to learn more about your risk for heart disease, please visit this li nk: https://www.mayoclinic.org/diseases conditions/heart disease/in depth/heart disease risk/itt 20084942 If you would like to learn more about your risk for diabetes, please visit this link: http://www.diabetes.org/are you at risk/diabetes risk test/ lways tell people about the true purpose of the experiment for several reasons 1. It might affect our results. For example, if we tell people the purpose of the experiment or how we predict people will act in the experiment, they may deliberately do whateve r it is they think we want them to do, just to help us out and give us the res ults that they think we want. If this happened, we would not have a very good indication of how they would act in situations in everyday life. 2. It is also possible that the oppos ite might occur. That is, if we tell people our predictions, they might delibe rately act in the opposite direction to show us that not have a good sample of how people act in ev eryday life. The purpose of this study: 1. What we are really examining in this experiment are factors that influence the sometimes opt to avoid scary health information such as the result of a medical or genetic test. Such information can be unsettling and sometimes people would just rather not know. 2. In this study we wanted to examine whether perceiving that a potentially threatening audience such as an insurance company or employe r might lead people to decline health feedback. To examine this question we ne eded some people to think that their information would be available to a particular audience. 3. We are not giving you feedback about your risk results. Indeed, the risk calculator you filled out cannot be used for this purpose. However, to examine our predi ctions, we needed you to believe that this questionnaire was accurate and that we could provide you with your risk results. As such, your fake test results will not be sent to an y database and will not be examined. Any personally identifying information yo u provided (name, home address) will be deleted from our database and will not be used in any capacity.

PAGE 108

108 4. We are not interested in your individual response but rather in the respon se of all participants in the study. Thus your responses will be combined wit h the responses of other participants to create a group or average response. 5. You may contact Dr. James Shepperd, Box 112250, phone number 352 273 2165, email, shepperd@ufl.edu or Nikolette Lipsey, email, nlipsey@ufl.edu if you have any feedback or concerns. If you have any questions about your rights as a research participant, you can contact the IRB office at UFIRB Office, Box 112250, University of Florida, Gainesville, FL 32 611 2250, ph 352 392 0433, IRB2@ufl.edu You have the opportunity to choose whether you want your data included in our study or whether you would like to withdraw your data at this point. I would like to point out again, as the consent form stated, that no data ar e analyzed individually and all personally identifying information will be deleted. We combine all of the data from all of our participants and look at the group averages. So, if you would like to include your da ta, simply sign and initial this data releas e form to say that we can include your data in our study. I understand that all personally identifying information collected in this study will be deleted and that my responses will not be analyzed individua lly and are not accessible to anyone but the research team. Data Release Statement Now that I have received a full disclosure of the purpose of experiment protocol 2017 02410, I give permission for the researchers to use my responses in their data analyse s. I give the researchers in this study permission to analyze my responses in thi s study. I do NOT give permission to analyze my responses in this study.

PAGE 109

109 APPENDIX D STUDY 3A MATERIALS Study 3 Study Overview & Informed Consent This survey will ask you qu estions about your health risk decisions. The survey should take no more than 15 minutes to complete. Informed Consent Please read this consent form carefully before you decide to participate in this study. Purpose of the research study: This study exam ines health risk decisions. What you will be asked to do in the study: You will complete questionnaires about your health and your desire for health risk information. Time required: 7 15 minutes Compensation: You will receive no compensation for your participation. Confidentiality: Your responses to all items will be confidential and anonymous. There is no connection between any identifying information and your responses. Your name and identify will not be di sclosed in any published reports. The onl ine host for this survey is Qualtrics. There is a minimal risk that security of any online data may be breached, but because no identifying information will be collected, and because the online host (Qualtrics) use s several forms of encryption, it is unlik ely that a security breach of online data will result in any adverse consequence for you. Voluntary participation & right to withdraw: Your participation is completely voluntary. There is no penalty for not part icipating. You have the right not to respo nd to any questionnaire item that you feel uncomfortable answering. You have the right to withdraw from the study at any time without consequence. Whom to contact if I have questions about the study: Investigato rs: Nikolette P. Lipsey nlipsey@ufl.edu & Dr. James A. Shepperd Shepperd@ufl.edu Dept of Psychology, Univ of Florida Whom to contact about my rights as a research participant in the study: UFIRB office, Box 112250, Un iversity of Florida, Gainesville, FL 32611 2250; ph 352 392 0433. I acknowledge that I have read the consent form and agree to participate in the study. I DO NOT wish to participate in this study.

PAGE 110

110 Audience norm manipulation and manipulation check In this study, we ask questions to assess your risk for several medical conditions (heart disease, diabetes, and some forms of cancer). You can decide whether or not you wish to learn your health risk information. No norms (No Injunctive + No Descriptive): No additional text. 3 Norm conditions: Before you make your decision and take the risk assessment, we believe it is important to provide you with additio nal information. Both norms (Injunctive + Descriptive): Learning your risk allows you to make important life choices, to monitor for signs before problems arise, and to plan for the future. In many ways, it is the right thing to do. Most people who take t he Lifestyle Health Questionnaire Risk Calculator choose to receive their risk results. Injunctive only (Injunctive + No Descriptive): Learning your risk allows you to make important life choices, to monitor for signs before problems arise, and to plan for the future. In many ways, it is the right thing to do. Descriptive only (No Injunctive + Descriptive): Most people who take the Life style Health Questionnaire Risk Calculator choose to receive their risk results. Manipulation Check Strongly Disagree Strongly Agree 1. Choosing to see my risk results is the right thing to do. 1 2 3 4 5 6 7 2. If given a choice, most people will opt to receive their risk results. 1 2 3 4 5 6 7 Information Decision and Information Avoidance Scale Same as Study 2 Lifestyle Health Questionnaire and Demographics Same as Study 2

PAGE 111

111 Debriefing and Data Release Statement. Thank you for participating in this study. Your responses will be helpful in our research. If you requested to receive your risk results, unfortunately we are unable to provide them to you, the questionnaire we used is not diagnostic. If you would like t o learn more about your risk for heart disease, please vi sit this link: https://www.mayoclinic.org/diseases conditions/heart disease/in de pth/heart disease risk/itt 20084942 If you would like to learn more about your risk for diabetes, please visit this link: http://www.diabetes.org/are you at risk/diabetes risk test / The purpose of this study: 1. What we are examining in thi s experiment are factors that influence the decision opt to avoid scary health information such as the result of a medica l or genetic test. Such information can be unsettling an d sometimes people would just rather not know. 2. In this study, we wanted to examine whether people might be more likely to seek health risk feedback if we advised people that they should seek informat ion and/or let people know that other people typically se ek that information. 3. We are not interested in your individual response but rather in the response of all participants in the study. Thus, your responses will be combined with the responses of other participants to create a group or average response. You may contact Dr. James Shepperd, Box 112250, phone number 352 273 2165, email, shepperd@ufl.edu or Nikolette Lipsey, email, nlipsey@ufl.edu if you have any feedback or concerns If you have any questions about your rights as a research participant, you can contact the IRB office at UFIRB Office, Box 112250, University of Florida, Gainesville, FL 32611 2250, ph 352 392 0433 IRB2@ufl.edu You have the opportunity to choose whether you want your data included in our study or whether you would like to withdraw your data at this point. I would like to point out again, as the consent form sta ted, that no data are analyzed individually. We combine a ll of the data from all of our participants and look at the group averages. Data Release Statement Now that I have received a full disclosure of the purpose of experiment protocol 2017 02410, I giv e permission for the researchers to use my responses in t heir data analyses. I give the researchers in this study permission to analyze my responses in this study. I do NOT give permission to analyze my responses in this study.

PAGE 112

112 APPENDIX E STUDY 3B MATERIALS Research Assistant Instructions/Script 1. Before th e participant arrives: a. Change into scrubs b. For one participant: i. Set up the computer in Room 288 Pull up the survey and make sure the headphones are plugged in and the volume is up. c. For two participa nts: i. Set up one computer in Room 288 AND one computer in R oom 292. 2. Greet the participant(s) and walk them over to 288 (and/or 292) and have them sit at the computer. Say the following: condition known as Thioamine Acetylase Deficiency, or TAA Deficiency. Please read the informed consent carefully. If you choose to participate, select the appropriate checkbox and proceed through the survey until you reach the STOP sign. When you reach the STOP sign, please notify me and I will give you furthe r instructions. I will be in the room next door, room 290. Please just come knock on the door when you are ready for further instructions. We ask that you put your phone on silent and do not use it during the study. If you need to silence your phone or p ut it away, ple ase do so now. When you are ready, please go ahead and put on the provided headphones so 3. When participant(s) reach the STOP sign and notify you, this means it is time for you to debrief them (see the s cript on the ne xt page). It is important that all participants are thoroughly debriefed. a. If you have two participants, keep them in separate rooms to probe for suspicion (the first paragraph of the debriefing). Once you do that, you can finish debriefing them together i f they finish around the same time.

PAGE 113

113 4. Once the participant(s) are debriefed, instruct them to return to their survey and opportunity to check a box saying they understa nd the purpose of the study and the opportunity to decide whether they would like their data included in analysis. 5. When the participant is done, they will see another STOP sign. Thank the participant(s) once more and once they leave the lab, click past th e second STOP s ign so that you can fill out your notes about the participant(s). 6. After you fill out your notes about the participant(s), log onto SONA ( https://ufl.sona systems.com/ ) and grant the participant( s) credit.

PAGE 114

114 Informed Consent Informed Consent Please read this consent form carefully before you decide to participate in this study. Purpose of the research study: This study examines health risk decisions. What you will be asked to do in the study: You will complete questionnaires about your health and your desire for health risk information. You will be given the opportunity to get tested for a medical condition. Time required: 30 minutes Compensation: You will receive one research credit for you r participation. Confidentiality: Your responses to all items will be confidential and anonymous. There is no connection between any identifying information and your responses. Your name and identify will not b e disclosed in any published reports. The o nline host for this survey is Qualtrics. There is a minimal risk that security of any online data may be breached, but because no identifying information will be collected, and because the online host (Qualtrics) uses several forms of encryption, it is unl ikely that a security breach of online data will result in any adverse consequence for you. Voluntary participation & right to withdraw: Your participation is completely voluntary. There is no penalty for not participating. You have the right not to res pond to any questionnaire item that you feel uncomfortable answering. You have the right to withdraw from the study at any time without consequence. Whom to contact if I have questions about the study: Investi gators: Nikolette P. Lipsey nlipsey@ufl.edu & Dr. James A. Shepperd Shepperd@ufl.edu Dept of Psychology, Univ of Florida Whom to contact about my rights as a research particip ant in the study: UFIRB office, Box 112250, University of Florida, Gainesville, FL 32611 2250; ph 352 392 0433. I acknowledge that I have read the consent form and agree to participate in the study. I DO NOT wish to participate in this study.

PAGE 115

115 rtbeat. Research teams at the University of Florida College of Medicine and The Feinberg School of Medicine at Northwestern University are working to have a previously understudied enzyme deficiency officially recognized as a disease by the American Medic al Associ ation. Known to medical researchers for decades, Thioamine Acetylase deficiency, or TAA deficiency, is finally getting the attention of the larger medical community. This announcement comes on the heels of recent evidence that TAA deficiency a ffects approximately one in five college age adults and that most adults living with TAA deficiency are completely unaware of their risk. Adults with TAA deficiency lack the Thi oamine Acetylase enzyme in their pancreas. Symptoms of TAA deficiency are lar gely indistinguishable from typical symptoms of exhaustion and stress. However, the physical consequences of TAA deficiency tend to worsen with age. Other symptoms include disrupt ed sleep, an inability to concentrate, general fatigue, achiness, and prolong ed cold like symptoms. Until recently, researchers had not linked the lack of the TAA enzyme to these outcomes. Now, researchers believe that TAA deficiency may be one of the lead ing reasons why some adults experience chronic fatigue and attentional defici ts. Although there is no known treatment, experimental trials are underway. The most promising treatment, however, has concerning side effects including 10 to 20% weight gain an d heightened risk for diabetes. The American Medical Association will discus s the status of TAA deficiency at their next annual meeting before officially recognizing it as a disease. Health in a Heartbeat is produced by University of Florida Health com mitted to advancing excellence in patient care research and education by WUFT FM.

PAGE 116

116 TAA Testing Information Testing for TAA Deficiency involves a simple anal swab which takes about 5 minutes to process. We can provide a free screening for TAA Deficien cy. This process should only take a few minutes at the conclusion of your stu dy session. Manipulation and Manipulation Check No norms (No Injunctive + No Descriptive): No additional text. 3 Norm conditions: Before you make your decision and take the risk assessment, we believe it is important to provide you with additional in formation. Both norms (Injunctive + Descriptive): Learning your risk for TAA Deficiency allows you to make important life choices, to monitor for signs before problems arise, and to plan for the future. In many ways, it is the right thing to do. Most peop le who are given the option for free testing choose to get screened for TAA Deficiency. Injunctive only (Injunctive + No Descriptive): Learning your risk for T AA Deficiency allows you to make important life choices, to monitor for signs before problems ar ise, and to plan for the future. In many ways, it is the right thing to do. Descriptive only (No Injunctive + Descriptive): Most people who are given the optio n for free testing choose to get screened for TAA Deficiency. Manipulation Check Strongly Disag ree Strongly Agree 1. Choosing to get screened for TAA Deficiency is the right thing to do. 1 2 3 4 5 6 7 2. If given a choice, most people will opt to get screened for TAA Deficiency. 1 2 3 4 5 6 7

PAGE 117

117 Information Decision and Information Avoidance Scale Would you like to receive a TAA Deficiency screening? Option 1 No, I am not interested in a screening for developing TAA Deficiency. Option 2 Yes, I would like to receive a screening for TAA Deficiency. We are interested in your thoughts regarding this decision. Please write down any thoughts you had while making this decision and provide your thoughts for the following question. What thoughts led you to choose the option you chose? _______________________________________________________ __________ _____ ______________________________________________________________________ ______________________________________________________________________ Information Avoidance Scale IA Scale Strongly Disagree Strongly Agree 1. I would rather not know my risk fo r TAA Deficiency. 1 2 3 4 5 6 7 2. I would avoid learning my risk for TAA Deficiency. 1 2 3 4 5 6 7 3. Even if it will upset me, I want to know my risk for TAA Deficiency. 1 2 3 4 5 6 7 4. When it comes to my risk for TAA Deficiency, sometimes ignorance is bliss. 1 2 3 4 5 6 7 5. I want to know my risk for TAA Deficiency. 1 2 3 4 5 6 7 6. I can think of situations in which I would rather not know my risk for TAA Deficiency. 1 2 3 4 5 6 7 7. It is important to know my risk for TAA Deficiency. 1 2 3 4 5 6 7 8. I would want to know my risk for TAA Deficiency immediately. 1 2 3 4 5 6 7

PAGE 118

118 Debriefing and Data Release First, I would like to thank you for participating in this study. Next, I would like to ask you what you think the study was about. What do you ( Encourage them to guess the purpose of the study; if they appear suspic ious/not deceived, remember to try to figure out the time of and reasons for their suspicion ). Sometimes in research it is necessary to use dec eption. We can't always reveal the experimental purpose because: It might affect our results If we tell peopl e the purpose or predictions of the experiment, they may deliberately do whatever it is they think we want them to do, just to help us out. It is also possible that if we tell people our predictions, they might deliberately act in the opposite direction to show us that we can't figure them out. In either situation, we would not have a good indication of how people normally act. Do you understan d why it might sometimes be necessary to conceal the real purpose of an experiment? We have not been frank abou t the real purpose of this experiment. Indeed, there is more to it than what we have told you: To study how people respond to information abou t their health it was important for us to use a disease without any attached feelings or thoughts. How could we do this? We decided the best method was to use a fictional disease. We decided to use the fictional disease, TAA Deficiency, you are now familia r with from the study. Using TAA Deficiency allowed us to avoid many of the problems that we have when using rea l medical conditions. We also needed to make sure our fabricated disease was believable. How could we do this? We decided to provide information that our study was part of collaboration with Shands Hospital. As you have guessed, our study is not associated with Shands Hospital and is solely the research of the Psychology Department at the University of Florida. Let me tell you a bit more abou t our study, we are interested in the factors that influence whether people want to learn about their risk for m edical conditions. We are particularly interested in finding methods of increasing the rate at which people seek information about risks to thei r health. condi tion, it very likely would have affected your response. Thus it was critical for us that you believe that TAA Deficiency was real and that you w ould at some point receive test results for TAA if you chose to get tested. We have gone to great lengths to make you believe that TAA Deficiency was real and that we could test you for it, and that you would receive feedback at some point. In fact

PAGE 119

119 the success of this experiment depends on people believing this. I hope we were successful. If not, please tell us a nd let me know what we can do to make the procedures seem more believable. Let me also say that we are not interested in your individual respon se but rather in the response of all participants in the study. Thus your responses to will be combined with th e responses of other participants to create a group or average response. Avoid disclosing: To draw conclusions, we will combine the data from y ou with data from other people. What this means is that we must ask you not to say anything about the study to anyone else. If you talk to other people about the purpose of the study and those people ar e later in the study, they may not respond naturally As a result, we wouldn't have valid data to draw conclusions about the average person. What this means is that the study, our time and your time would be wasted. We want everyone to get some educational value out of this experiment. Thus, I am telling y ou what our true hypothesis was. However, if you tell someone else what happened and they participate in this study, then they won't get the same experience from this experiment that you do. I hope you s ee why it is important not to tell anyone the purpos e of the experiment. You may wonder what difference it makes to tell a friend or roommate or partner because they will never be in the study. But they may tell someone else who will be in the study. I realize you may have an urge to tell people about wh at happened in this experiment. However, I am going to ask that you not mention anything about the experiment. Can I get a commitment from you not to say anything about the experiment? If anybody asks you about the experiment, just tell them that it was an experiment about people thoughts about a health condition. Don't make a big mystery about the study. Just say that you were in such and such experiment and that you are not at liberty to discuss the nature of the experiment. Do you have any question s? Comments? Suggestions? Now that you have been debriefed, we ask that you proceed past the STOP sign on the survey, so that you can indicate to us whether you wish to include your data in our analysi s. Once again, thank you for your participation in our research study.

PAGE 120

120 Data Release Statement Now that I have received a full disclosure of the purpose of experiment protocol 2017 02410, I give permission for the researchers to use my responses in thei r data analyses. I give the researchers in this study permission to analyze my responses in this study. I do NOT give permission to analyze my responses in this study.

PAGE 121

121 LIST OF REFERENCES Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes 50 (2), 1 79 211. Afifi W. A., & Weiner, J. L. (2006). Seeking information about sexual health: Applying the theory of motivated information management. Human Communication Research, 32, 35 57. American Cancer Society (2015). Breast Cancer. American Cancer Society Retrieved fr om: http://www.cancer.org/acs/groups/cid/documents/webcontent/003090 pdf.pdf Ashford, S. J., & Northcraft, G. B. (1992). Conveying more (or less) than we realize: The role of impression management in feedback seeking. Organizational Behavior and Human Dec ision Processes, 53, 310 334. Barbour, J. B., Rintamaki, L. S., Ramsey, J. A., & Brashers, D. E. (2012). Avoiding health information. Journal of Health Communication 17 (2), 212 229. Barth, K. R., Cook, R. L., Downs, J. S., Switzer, G. E., & F ischhoff, B. (2002). Social stigma and negative consequences: Factors that influence college students' decisions to seek testing for sexually transmitted infections. Journal of American College Health, 50 (4), 153 9. http://dx.doi.org/10.1080/07448480209596 021 Boulware, L. E., Cooper, L. A., Ratner, L. E., LaVeist, T. A., & Powe, N. R. (2016). Race and trust in the health care system. Public Health Reports 188 358 365. Brehm, J. W. (1966). A theory of psychological reactance New York: Academic Press. Bull ock, J. G., G expect an easy answer). Journal of Personality and Social Psychology 98 (4), 550 558. Burger, J. M. (1999). The foot in the door compliance procedure: A multiple process an alysis and re view. Personality and Social Psychology Review 3 (4), 303 325. Calhoun, G. J. (2012). Seeking Safety? Applying the Risk Information Seeking and Processing Model to Sexual Aggression on a College Campus. Marquette University. C ialdini, R. B (2003). Crafting normative messages to protect the environment. Current Directions in Psychological Science 12 (4), 105 109. Cutler, S. J., & Hodgson, L. G. (2003). To test or not to test: Interest in genetic testing among middle aged adults. American Journal of 9 20.

PAGE 122

122 DeHart, D. D., & Birkimer, J. C. (1997) Trying to Practice Safer Sex: Development of the Sexual Risks Scale. The Journal of Sex Research 34 11 25. Delaney, R O., Serovic h, S. M., & Lim., J. Y. (2008). Reasons for and against maternal HIV disclosure to children and perceived child reaction. AIDS Care, 20, 876 880. Dwyer, L. A., Shepperd, J. A., & Stock, M. L. (2015). Predicting Avoidance of Skin Damage Feedbac k Among College Students. Annals of Behavioral Medicine 49 (5), 685 695. Hadley, D. W., Jenkins, J., Dimond, E., Nakahara, K., Grogan, L., Liewehr, D. J., ... & Kirsch, I. (2003). Genetic counseling and testing in families with hereditary nonpolyposis colo rectal c ancer. Archives of Internal Medicine 163 (5), 573 582. Hayes, A. F. (2012). PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling [White paper]. Retrieved from http://www.afhayes.com/ public/process2012.pdf Hightow, L. B., Miller, W. C., Leone, P. A., Wohl, D., Smurzynski, M., & Kaplan, A. H. (2003). Failure to return for HIV posttest counseling in an STD clinic population. AIDS Education and Prevention 15 (3), 2 82 290. Howell, J. L., C ollisson, B., Crysel, L., Garrido, C. O., Newell, S. M., Cottrell, C. A., ... & Shepperd, J. A. (2013). Managing the threat of impending implicit attitude feedback. Social Psychological and Personality Science 4 (6), 714 720. Howell J. L., Crosier, B. S., & Shepperd, J. A. (2014). Does lacking threat management resources increase information avoidance? A multi sample, multi method investigation. Journal of Research in Personality, 50, 102 109. Howell, J.L., Lipsey, N. P., Shepperd, J.A. ( in press ). Health Information Avoidance. In K. Sweeny & M. Robbins (Eds.), The Wiley Encyclopedia of Health Psychology: The Social Bases of Health Behavior Wiley. Howell, J. L., & Shepperd, J. A. (2012). Reducing information avoidance through affirm ation. Psychological Sci ence, 23, 141 145. Howell, J. L., & Shepperd, J. A. (2013). Behavioral obligation and information avoidance. Annals of Behavioral Medicine, 45, 258 263. Johnston, L. (1996). Resisting change: Information seeking and stereotype chang e. European Journal of S ocial Psychology, 26, 799 825. Katz, R. V., Green, B. L., Kressin, N. R., Kegeles, S. S., Wang, M. Q., James, S. A., ... & McCallum, J. M. (2008). The legacy of the Tuskegee Syphilis Study: assessing its impact on willingness to par ticipate in biomedical s tudies. Journal of Health Care for the Poor and Underserved 19 (4), 1168 1180.

PAGE 123

123 Knowles, E. D., Lowery, B. S., Chow, R. M., & Unzueta, M. M. (2014). Deny, distance, or dismantle? How White Americans manage a privileged identity. Pers pectives on Psychologica l Science, 9, 591 609. Lipsey, N. P., & Shepperd, J. A. (2019a). Powerful audiences are linked to health information avoidance: Results from two surveys. Social Science and Medicine 225 51 59. https://doi.org/10.1016/j.socscimed.2 019.01.046 Lipsey, N. P. & Shepperd, J. A. (2019b). Powerful audiences prompt health information avoidance. Social Science and Medicine, 220, 430 439. https://doi.org/10.1016/j.socscimed.2018.11.037 Maestas, C. D., & Pollock, W. M. (2010). Measuring gener alized risk orientation with a single survey item. Available at SSRN 1599867 Malat, J., Mayorga Gallo, S., & Williams, D. R. (2018). The effects of whiteness on the health of whites in the USA. Social Science and Medicine 199 148 156. Melnyk D., & Shepperd, J. A. (2012). Avoi ding risk information about breast cancer. Annals of Behavioral Medicine, 44, 216 224. Murphy, D. A., Roberts, K. J., & Hoffman, M. A. (2002). Stigma and ostracism associated with HIV/AIDS: Children carrying the secret o serostatus. Jo urnal of Child and Family Studies, 11, 191 202. Neighbors, C., Larimer, M. E., & Lewis, M. A. (2004). Targeting misperceptions of descriptive drinking norms: Efficacy of a computer delivered personalized normative feedb ack intervention. Journal of Consult ing and Clinical Psychology 72 (3), 434 447. Patel, V., Beckjord, E., Moser, R. P., Hughes, P., & Hesse, B. W. (2015). The role of health care experience and consumer information efficacy in shaping privacy and security perceptions of medical records: nati onal consumer survey results. JMIR Medical Informatics 3 (2), e14. doi:10.2196/medinform.3238 Pirlott, A. G., & MacKinnon, D. P. (2016). Design approaches to experimental mediation. Journal of Experimental Social Psychol ogy 66 29 38. Platt, J., & Kardia, S. (2015). Public trust in health information sharing: Implications for biobanking and electronic health record systems. Journal of Personalized Medicine 5 (1), 3 21. Rains, S. A. (2013). The nature of psychological rea ctance revisited: A meta analytic re view. Human Communication Research 39 (1), 47 73.

PAGE 124

124 Reid, A. E., Taber, J. M., Ferrer, R. A., Biesecker, B. B., Lewis, K. L., Biesecker, L. G., & Klein, W. M. (2018). Associations of perceived norms with intentions to le arn genomic sequencing results: Role s for attitudes and ambivalence. Health Psychology 37 (6), 553. Rivis, A., & Sheeran, P. (2003). Descriptive norms as an additional predictor in the theory of planned behaviour: A meta analysis. Current Psychology 22 (3) 218 233. Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological Science 18 (5), 429 434. Shipp, T. D., Shipp, D. Z., Bromley, B., Sheahan, R., Cohen, A., Lieberman, E ., & the sex of their unborn child? Birth: Issues in Perinatal Care, 31, 272 279. Stok, F. M., De Ridder, D. T., De Vet, E., & De Wit, J. B. (2014). Don't tell me what I should do, but what others do: The influence of descriptive and injunctive peer norms on fruit consumption in adolescents. British Journal of Health Psychology 19 (1), 52 64. Sullivan, P. S., Lansky, A., & Drake, A. (2004). Failure to return for HIV test results among persons at high risk for HIV infection: Results from a multistate interview project. Journal of Acquired Immune Deficiency Syndromes, 35 (5), 511 518. doi:10.1097/00126334 200404150 00009 Sweeny, K., Melnyk D., Miller, W., & Shepperd, J. A. (2010). Information avoidance: Who, what, when, and why. Review of General Psychology, 14, 340 353. Sweeny, K., & Miller, W. (2012). Predictors of information avoidance: When does ignorance seem most blissful? Self and I dentity, 11 185 201. Yaniv, I., Benador, D. & Sagi, M. (2004). On not wanting to know and not wanting to inform others: Choices regarding predictive genetic testing. Risk, Decision & Policy, 9, 317 336. Zhao, X., Lynch Jr, J. G., & Chen, Q. (2010 ). Reconsidering Baron and Kenny: Myths and truths a bout mediation analysis. Journal of Consumer Research 37 (2), 197 206

PAGE 125

125 BIOGRAPHICAL SKETCH Nikolette received her Doctor of Philosophy in social psychology at the University of Florida in 2019, under the advisement of Dr. James Shepperd She graduated magn a cum laude with a Bachelor of Arts degree from Furman University in 2013, with a double major in psychology and Asian studies. She received her Master of Science degree in social psychology at the Univer sity of Florida in 2016. She examines how people mak e decisions about potentially threatening information, with an emphasis on interpersonal influences in health contexts.