i

THE EFFECTS OF MESSAGE FRAMING, THEORY OF REACTANCE, AND EGO-

DEPELTION ON THE EFFICACY OF A

DRINKING REDUCTION CAMPAIGN

______

A Dissertation Presented to the Faculty of the Department of Psychology University of Houston

______

In Partial Fulfillment Of the Requirements for the Degree of Doctor of Philosophy

______

By

Chun-Han Chen

August, 2014

ii

THE EFFECTS OF MESSAGE FRAMING, THEORY OF REACTANCE, AND EGO-

DEPLETION ON THE EFFICACY OF A

DRINKING REDUCTION CAMPAIGN

______Chun-Han Chen

APPROVED:

______Clayton Neighbors, Ph.D. Committee Chair

______C. Raymond Knee, Ph.D.

______Suzanne Kieffer, Ph.D.

______F. Richard Olenchak, Ph.D. College of Education

______John W. Roberts, Ph.D. Dean, College of Liberal Arts and Social Sciences Department of English iii

THE EFFECTS OF MESSAGE FRAMING, THEORY OF REACTANCE, AND EGO-

DEPLETION ON THE EFFICACY OF A

DRINKING REDUCATION CAMPAIGN

______

An Abstract of a Dissertation Presented to the Faculty of the Department of Psychology University of Houston

______

In Partial Fulfillment Of the Requirements for the Degree of Doctor of Philosophy

______

By

Chun-Han Chen

August, 2014

iv

ABSTRACT

This dissertation focuses on message framing with the goal of designing effective advertisements that advocate healthy drinking styles. Rooted in persuasion models, message framing by gain or loss frame may directly convince people or indirectly prime subconsciousness. Two studies were conducted. Ego-depletion was considered with framed message in the first study, and the theory of reactance was utilized to predict the communication outcome in the second study, which also considered the roles of self-esteem, instability of self-esteem, and self-relevance. Reduction of future drinking intentions was the goal of message delivery.

Both studies recruited students who were regular drinkers, mostly females to complete online surveys, with random assignment to ego-depletion or waiting conditions (Study 1) and messages to evaluate (Study 1 & 2). Several regression models were tested with a unique measure of regular drinking in each equation. Ego-depletion among heavier drinkers was positively associated with ratings of health messages, which provided partial support for hypotheses of Study 1. Reactance to follow coercive health advice was supported indirectly by higher drinking intention. Higher self-esteem and unstable self-esteem were positively related to message acceptance, which was counter to expectations. Self-relevance with drinking was found to moderate the effects of message framing on drinking intention. This was interpreted as an indirect indication of reactance. Overall, findings were somewhat supportive of hypotheses and provide a unique contribution to understanding of message framing and alcohol communications. Limitations and future directions are summarized.

v

TABLE OF CONTENTS

Page

Introduction 1

Persuasion for Health Campaign 3

Health Communication: Processing and Perceived Message 4 Implication

Social Marketing as Message Processing 6

Message Framing, Mind Processing, and Other Exemplary Forms of 8 Health Communication

Framing: Gain Frame and Loss Frame 11

Gain Frame and Loss Frame in Miscellaneous Topics 13

Frame (Gain vs. Loss) on Health Issues 16

Known Moderators behind Gain vs. Loss Frame 18

Frame (Gain vs. Loss) and Emotion on Health Issues 21

Message Framing and Drinking 24

Bridging: Future Directions 25

Study 1: Message Framing and Ego-depletion on Drinking Reduction 26

Ego-depletion and its Implications 27

Ego-depletion affects Attitude Certainty and Information Processing 28

Moderating Drinking to be Controlled 29

Ego-depletion and Loss Frame on Health Campaign 30

vi

Study 1: Method 31

Participants 31

Experimental Materials: Drinking Scales 32

The Quantity-Frequency-Peak Alcohol Use Index 32

The Daily Drinking Questionnaire 33

The Rutgers Alcohol Problem Index 33

The Drinking Intention Scale 33

Experimental Materials: Imaginative Writing Task 33

Presented Health Messages 34

The Messages and Believability Scale 34

Experimental Procedures 34

Study 1: Results 35

Predictors by Experimental Design 35

Outcome 36

Drinking as the Statistical Covariates 36

Analyses 36

Descriptive Statistics 37

Regression 37

Study 1:Discussion 40

Study 2: Message Framing and Theory of Reactance on Drinking 44 Reduction Campaign

Psychological Reactance and Message Framing 45 vii

Uniqueness of Statistical Information plus Texts Messages 47

Moderators—Self-esteem and Self-esteem instability 49

Self-esteem 49

Instability of Self-esteem 50

Self-relevant Issue and Intellectual Reasoning 51

Self-Relevance and Health Decision 52

Study 2: Method 54

Design 54

Participants 55

Messages 55

Experimental Procedures: Recruiting and Survey 56

Measures 57

The Self-esteem Scale 57

Instability of Self-esteem 57

The Message Believability Scale 58

Self-relevance 58

Study 2: Results 59

Scores by Experimental Groups 59

Outcomes 59

Drinking as the Statistical Covariates 60

Analyses 60

Descriptive Statistics 61 viii

Regression for H3, H4 and H5 61

Study 2:Discussion 68

Conclusion 72

Research Limitation 75

Future Directions 79

References 85

Tables 104

Figures 156

Appendices 170

Appendix A: Tested Health Messages 170

Appendix B: Consent Forms 180

Appendix C: Measurement Tools 186

Appendix D: Experimental Manipulations for Study 1 200

ix

List of Tables

Table Page

1 Various Framing beside Gain vs. Loss 104

2 Framing (gain vs. loss) Researches in Miscellaneous Areas 105

3 Framing (gain vs. loss) Researches in Health Promotions. 107

4 Frame (Gain vs. Loss) and Emotion at Health Issues 109

5 Message Framing on Alcohol Problem 111

6 Correlation and Descriptive Statistics: Study 1:Ego-Depletion 113

7 Message rating as a function of ego-depletion and drinks per 114 week

8 Message rating as a function of ego-depletion and peak drinks 115

9 Message rating as a function of ego-depletion and typical drinks 116

10 Message Rating as a Function of Drinks per week, Ego-depletion 117 and Message Framing

11 Message Rating as a Function of Typical Drinks, Ego-depletion 118 and Message Framing

12 Weekly Drinking Intentions a Function of Drinks per week, Ego- 119 depletion and Message Framing

13 Typical Drinking Intentions a Function of Typical Drinks, Ego- 120 depletion and Message Framing

14 Correlation and Descriptive Statistics: Study 2:Theory of 121 Reactance

15 Study 2: Message Rating as a Function of Framing, Statistical 122 Information and Drinks per week. x

16 Study 2: Message Rating as a Function of Framing, Statistical 123 Information and Peak Drinks

17 Study 2: Message Rating as a Function of Framing, Statistical 124 Information and Typical Drinks

18 Study 2: Weekly Drinking Intention as a Function of Framing, 125 Statistical Information and Drinks per week

19 Weekly Drinking Intention as a Function of Framing, Statistical 126 Information and Peak Drinks

20 Study 2: Typical Drinking Intention as a Function of Framing, 127 Statistical Information and Typical Drinks

21 Message Rating as a Function of Statistical Information, 128 Message Framing, Self-esteem and Instability of Self-esteem, and Drinks per week

22 Message Rating as a Function of Message Framing, Self-esteem 129 and Instability of Self-esteem, with Statistical Information and Drinks per week controlled

23 Message Rating as a Function of Message Framing, Self-esteem 130 and Instability of Self-esteem, with Statistical Information and Drinks per week controlled (Continued)

24 Message Rating as a Function of Statistical Information, 131 Message Framing, Self-esteem and Instability of Self-esteem, with Peak Drinks controlled

25 Message Rating as a Function of Message Framing, Self-esteem 132 and Instability of Self-esteem, with Statistical Information and Peak Drinks controlled xi

26 Message Rating as a Function of Message Framing, Self-esteem 133 and Instability of Self-esteem, with Statistical Information and Peak Drinks controlled (Continued)

27 Message Rating as a Function of Statistical Information, 134 Message Framing, Self-esteem and Instability of Self-esteem, with Typical Drinks controlled

28 Message Rating as a Function of Message Framing, Self-esteem 135 and Instability of Self-esteem, with Statistical Information and Typical Drinks controlled

29 Message Rating as a Function of Message Framing, Self-esteem 136 and Instability of Self-esteem, with Statistical Information and Typical Drinks controlled (Continued)

30 Weekly Drinking Intention as a Function of Statistical 137 Information, Message Framing, Self-esteem and Instability of Self-esteem, with Drinks per week controlled

31 Weekly Drinking Intention as a Function of Message Framing, 138 Self-esteem and Instability of Self-esteem, with Statistical Information and Drinks per week controlled

32 Weekly Drinking Intention as a Function of Message Framing, 139 Self-esteem and Instability of Self-esteem, with Statistical Information and Drinks per week controlled (Continued)

33 Weekly Drinking Intention as a Function of Statistical 140 Information, Message Framing, Self-esteem and Instability of Self-esteem, with Peak Drinks controlled

34 Weekly Drinking Intention as a Function of Message Framing, 141 Self-esteem and Instability of Self-esteem, with Statistical Information and Peak Drinks controlled xii

35 Weekly Drinking Intention as a Function of Message Framing, 142 Self-esteem and Instability of Self-esteem, with Statistical Information and Peak Drinks week controlled (Continued)

36 Weekly Drinking Intention as a Function of Statistical 143 Information, Message Framing, Self-esteem and Instability of Self-esteem, with Typical Drinks controlled

37 Weekly Drinking Intention as a Function of Message Framing, 144 Self-esteem and Instability of Self-esteem, with Statistical Information and Typical Drinks controlled

38 Weekly Drinking Intention as a Function of Message Framing, 145 Self-esteem and Instability of Self-esteem, with Statistical Information and Typical Drinks week controlled (Continued)

39 Typical Drinking Intention as a Function of Statistical 146 Information, Message Framing, Self-esteem and Instability of Self-esteem, with Typical Drinks controlled

40 Typical Drinking Intention as a Function of Message Framing, 147 Self-esteem and Instability of Self-esteem, with Statistical Information and Typical Drinks controlled

41 Weekly Drinking Intention as a Function of Message Framing, 148 Self-esteem and Instability of Self-esteem, with Statistical Information and Typical Drinks week controlled (Continued)

42 Message Rating as a Function of Statistical Information, 149 Framing, Self-relevance and Drinks per week

43 Message Rating as a Function of Statistical Information, 150 Framing, Self-relevance and Peak Drinks

44 Message Rating as a Function of Statistical Information, 151 Framing, Self-relevance and Typical Drinks xiii

45 Weekly Drinking Intention as a Function of Statistical 152 Information, Message Framing, Self-relevance and Drinks per week

46 Weekly Drinking Intention as a Function of Statistical 153 Information, Framing, Self-relevance and Peak Drinks

47 Weekly Drinking Intention as a Function of Statistical 154 Information, Framing, Self-relevance and Typical Drinks

48 Typical Drinking as a Function of Statistical Information, 155 Framing, Self-relevance and Typical Drinks

xiv

List of Figures

Figure Page

1 Power Analysis Diagram 156

2 Two-way Interaction Predicting Message Rating by Both 157 Ego-depletion and Peak Drinking on Table 8

3 The Graph for the Two-way interaction between Drinks per 158 week and Frames on Table 10

4 The Graph for the Two-way Interactions between Drinks per 159 week and Ego-depletion on Table 12

5 The Graph Depicting the Three-way interaction on Table 13. 160 The Major Dimensions are Typical Drinks, Ego-depletion and Message Framing with Outcome of Typical Drinking Intention

6 The Graph Depicting the Three-way interaction on Table 161 18.The Major Dimensions are Stats, Frame and Drinks per week, with Outcome of Weekly Drinking Intention

7 The Graph Depicting the Three-way interaction on Table 162 19.The Major Dimensions are Stats, Frame and Peak Drinks, with Outcome of Weekly Drinking Intention

8 The Significant Pair of Two-way Interactions on Table 20 163 with the outcome of Typical Drinking Intention

9 The Two-way Interaction between Instability of Self-esteem 164 and Self-esteem with Outcome of Message Rating on Table 25

10 The Two-way Interaction between Self-esteem and Message 165 Framing with Outcome of Message Rating on Table 31 xv

11 The Two-way Interaction between Self-esteem and Message 166 Framing with Outcome of Message Rating on Table 32

12 The Three-way Interaction, Stats X Frame X Self-relevance 167 with Outcome of Weekly Drinking Intention on Table 45

13 The Three-way Interaction, Stats X Frame X Self-relevance 168 with Outcome of Weekly Drinking Intention on Table 46

14 The Three-way Interaction, Stats X Frame X Self-relevance 169 with Outcome Weekly Drinking Intention on Table 47

xvi

List of Appendices

Appendix Page

A Tested Health Messages 170

B Consent Forms 180

C Measurement Tools 186

D Experimental Manipulations for Study 1 200 MESSAGE FRAMING Framing 1

The Effects of Message Framing, Theory of Reactance, and Ego-depletion on

the Efficacy of a Drinking Reduction Campaign

Introduction

This dissertation has two main sections. The first section reviews current knowledge concerning message framing and relevant factors on persuasion. The second section describes two studies which were completed to further investigate possible moderating factors of framing effects: provision of statistical information, defensive reactance (i.e. theory of reactance), self-esteem, self-relevance, and ego-depletion. The application of persuasive messages to health persuasion represents the central focus of the proposed research. Of all health problems discussed, alcohol consumption serves as the foremost issue for application.

Excessive alcohol consumption represents a prevalent health problem in the United

States. According to the review by Hinson (2010), from 1999 to 2005 both collegiate and non-college young adults have had increasing rates of drinking. The average amount of college students’ drinking was higher than non-college students (Hinson, 2010). Heavy drinking is associated with a number of consequences including unintentional injury or deaths, such as traffic accidents and date rape (Hinson, 2010). Other problems which college students may experience include unwanted sexual contact, class absence, and low exam grades, etc. (Avci, & Fendrich, 2010).

Besides controls instituted by society and its laws, Hinson (2010) suggests that interventions to reduce college students’ drinking can be administered at the individual level or at the group level through normative re-education. Individual interventions refer to those in which individuals experience the intervention on their own, such as Motivational MESSAGE FRAMING Framing 2

Interviewing (Miller, & Rollnick, 2002). Normative re-education is conducted in groups or individually, and focuses on the instillation of accurate and appropriate norms regarding drinking behaviors (Hinson, 2010). Individual interventions incorporating normative re- education require certain forms of information-presentation in order to persuade individuals to change behavior. It is therefore important to understand how the presentation of messages affects the likelihood of acceptance and implementation of health promotion messages. This dissertation introduces current knowledge regarding the impact of message framing on the perception of these messages and examines current applications in health promotion.

Problem drinking can be reduced or intensified based on the influence of media or communication. According to a qualitative study conducted by Gordon, Moodie, Eadie, and

Hastings (2010), adolescents in the U.K. are aware of the alcohol manufacturers’ branding techniques in their alcohol advertisements. Branding is the linking of positive images or descriptions to a certain brand of alcohol (Gordon et al., 2010). Although the adolescents in the Gordon et al. (2010) focus group were able to sense the marketing intentions within the alcohol advertisements, alcohol marketing may still be problematic to the development of problem drinking. Thus, knowledge of persuasive strategies may not necessarily undermine their effectiveness.

Mass media can play a significant role in facilitating health behaviors. According to the review by Wakefield (2010), the effectiveness of mass media campaigns for the promotion of health behavior has been demonstrated by a number of published research studies. Broadcasting health promotion messages via television and radio, mass media can improve various health outcomes and health behaviors, such as better nutrition and higher likelihood of having Papanicolaou smears, etc. (Wakefield, 2010). Research studies about MESSAGE FRAMING Framing 3 the effects of health communication are also relevant to the design of interventions which may prevent or reduce problem drinking. This dissertation reviews existing research about framed message and health behaviors, and conducted two studies designed to extend our knowledge of health communication in order to more effectively address the problem of drinking.

Persuasion for Health Campaign. The following introduced components and routes in the process of persuasion can apply to the domain of health campaign. Persuasion has been widely studied by various disciplines, including psychology and communication.

Communication often exerts its persuasiveness through verbal logic within propositions, but can influence the perceived persuasiveness through other informal sources or channels.

Rothenthal’s (1972) publication on communication proposed the concept of paramessage, referring to the information to be considered besides the formal message per se. The categories of the paramessages are (1) messages associated with the primary message, such as cover letter and commentary; (2) information sources, such as the personality of the information provider; (3) contexts, such as the atmosphere of the physical environment where the communication occurs.

Two similar models have been widely used to describe the processes of verbal persuasion along with non-verbal persuasion. The first model was originally proposed in the research of Chaiken (1980). She suggested that people use two routes for evaluating the persuasiveness of a message: systematic processing and heuristic processing. Systematic processing refers to cognitive reflection about a conveyed message’s propositional implications, while heuristic processing is the consideration of the information source’s credibility or any other tangential cues which are readily observable (Chaiken, 1980). The MESSAGE FRAMING Framing 4 second model is from Cacioppo, Petty, and Stoltenberg (1985), the Elaboration Likelihood

Model (ELM). This model designates a central route and a peripheral route as two processing routes used in the consideration of persuasiveness of an argument. The central route refers to the consideration of a message or an argument in the context of existing previous knowledge, and is accompanied by significant effort for critical evaluations. The peripheral route describes a process where messages are considered quickly without diligent cognitive elaboration (Cacioppo et al., 1985).

Evans (2007) similarly referred to two processes, a conscious process which is an analytic one and a separate unconscious process which relies on fast intuition as heuristic processing. According to Evans (2007), it is possible to represent the models with both processes working concurrently or competing for the psychological resources at the same time with path diagrams and relevant probability.

Health Communication: Processing and Perceived Message Implication. Science has constructed relevant theories for health communication and education, and numerous empirical studies have evaluated these models. Many psychological processes are related to the processing of the presented health message. The review by De Wit (2007) suggests that

Chaiken’s (1980) dual process model (systematic and heuristic routes) is well suited to apply to the domain of health communication. Disease and treatment always involve potential risks to be perceived and estimated. Making decisions with potential risks is directly related to dual process models, and health communication is one domain where risk-related decisions are central (De Wit, 2007). Trumbo (2002) used cancer as one health issue and tested the process of persuasion as indicated by the heuristic-systematic model (HSM). The subject’s perceived self-intelligence and motivation to process information and the MESSAGE FRAMING Framing 5 message’s content sufficiency determined which processing was used (systematic or heuristic) before the potential cancer risks were estimated (Trumbo, 2002) According to

Trumbo (2002), motivation and content sufficiency were positively linked to adoption of systematic processing, while message sufficiency was positively linked to heuristic processing. Using the prostate cancer patients as subjects, Steginga and Occhipiati (2004) found that health locus of control, tolerance of ambiguity, and decision related uncertainty are moderating variables affecting whether systematic processing and/or heuristic processing is adopted in considering health information. Using sleep deprivation and spinal meningitis as the health topic, Averbeck (2001) found that students with prior knowledge about a health topic tended to use systematic processing more often, and students with little prior knowledge about a health topic tended to rely more on heuristic processing, such as appeal to the emotion of fear.

Another important dimension affecting processing of health messages is psychological resistance, i.e., reactance, or defensiveness. According to Whitehead and

Russell (2004), psychological resistance is manifested by irritation to the conveyed health message or unwillingness to attend or process the information. Furthermore, psychological reactance is experienced as deprived psychological freedom which may be implied by a health message. For example, telling alcohol drinkers that excessive drinking may lead to death may provoke psychological resistance because of the perception of thwarted freedom.

People often subjectively estimate their potential health risks in a biased fashion and fail to estimate their risks with rational thought (Whitehead & Russell, 2004). Because of potential threat and perceived deprivation of freedom, which the health message can arouse, a demanding request of behavioral change indicated by a health message may result in higher MESSAGE FRAMING Framing 6 psychological reactance (Rains, & Turner, 2007). The second study of this dissertation took

Reactance Theory as the major research topic.

The other factors affecting the impact of health communications and health education include relevance and authenticity. Kruglanski, Pierro, Mannetti, Erb, and Spiegel

(2006) suggested that the perceived relevance of the health issue (i.e. cancer) is the most crucial factor related to the health decision. Authenticity of a message refers to how a message relates to the practical aspects of their conveyed communication. For example, if certain people care about how drinking alcohol affects their social life, then messages addressing this issue will be more relevant and practical to them. Petraglia (2009) proposed that professionals collect information, especially authenticity, from the message audience and their surroundings, identifying practical issues and managing to conduct dialogues

(mutual discussion) with them. Overall, the psychological processing of health messages is related to the process of persuasion (such as dual-route processing), psychological responses

(such as resistance), and implication of message (such as message relevance and authenticity).

Social Marketing as Message Processing. The models of persuasion involving how mind processing messages are assured to be suitable on convincing people to conduct environment-protective behaviors (Geller, 1989). Geller (1989) and Grier, and Bryant

(2005) describe social marketing with very similar concepts as those discussed in dual process models. Social marketing is “typically defined as a program planning process that applies commercial marketing concepts and techniques to promote public behavior change

(Grier, & Bryant, 2005, p321)”. In marketing, 4P’s can be used when analyzing the conditions under which an intervention program is being established. These 4P’s are: MESSAGE FRAMING Framing 7 product (the potential benefit which the recipients would make if a behavior change occurs according to the advice), price (the cost which the recipients will pay in order to make a behavioral change), place (the location which is necessary to let the behavioral change happen), and promotion (the persuasive communication to the recipients in order to let behavioral change occur), according to Geller (1989) and Grier (2005). Geller (1989) reviewed how social marketing was applied to the spread of environmental protection, while

Grier and Bryant (2005) reviewed how social marketing was applied to issues of public health promotion.

Researchers have conducted empirical studies to test the effectiveness of social marketing on health behaviors. These have shown that social marketing can be defined as improving the interpersonal relationship of patron and service provider (Al-Alak, 2010).

This 2010 study found that social marketing is effective in facilitating satisfaction toward the fitness centers in Jordan. In this context of fitness centers, the quality of interpersonal relationship was determined by how the employees provide the service. High profile service was associated with patron satisfaction, trust, and purchase of future service. Furthermore, patrons spread positive messages about this fitness service when they are more satisfied (Al-

Alk, 2010). Thus, Al-Alak (2010) found that interpersonal relationships represent a crucial variable which needs to be considered when designing a social marketing campaign to encourage physical activity. Social marketing was inspired by traditional marketing theory, but with emphasis on the non-profit sector (Andreason, 2003). Research has examined the potential effects of emotions on social marketing. Treating anti-tobacco advertisements as one kind of social marketing, Wolburg’s (2006) qualitative analysis found that anger, denial, and other defensive emotions tend to accompany smokers’ responses to anti-tobacco MESSAGE FRAMING Framing 8 messages, but not non-smokers. These occurrences of emotions fit the model for risky decision making because smoking is a choice which accompanies health risk. The reason that participants experienced anger and denial in the risk decision making process is because of psychological reactance (Wolburg, 2006).With the establishment of social marketing theory and relevant empirical studies, it has been established that health behavior can be facilitated through appropriate social influences and/or social communications.

There are a variety of categorizations for message presentation styles. Message framing is a very common category to interpret how a message is presented. The following reviews the literature of concepts which analyze message presentation styles and emphasizes message framing.

Message Framing, Mind Processing, and Other Exemplary Forms of Health

Communication. Several dimensions have been used to discriminate the presentation styles in which messages are communicated. Message framing is only one dimension which has been used to categorize types of communications. The following presents research from the literature regarding how to categorize styles of message presentation.

According to the review by Niederdeppe, Bu, Borah, Kindig, and Robert (2008), messages delivered to individuals can be analyzed by the following three categories: (1) narrative, (2) visual image, and (3) message framing. Narrative refers to the distinction between the message form of a personal story and other forms of description. Secondly, visual images refer to pictures or visual stimuli which accompany the presented message

(Niederdeppe et. al, 2008). MESSAGE FRAMING Framing 9

Narrative formats vary and have different effects. Kreuter et al (2007) asserted that presented information about cancer prevention and treatment represents a form of narrative.

A narrative about cancer benefits because it can reduce people’s distorted impressions about health and replace them with correct knowledge (Kreuter et al, 2007). According to Kreuter

(2007), cancer narratives differ in their sequence (how the message is arranged in terms of what is to be described early or late), characteristics (how the message catches the audience’s attention, identification, or comprehension), structure (how the message points out the issue which is currently unresolved), space-time bound ( the extent to which the audience perceives the message as understandable and likable, and the extent to which the audience is motivated to accept the message idea), and message production (whether the message presentation is accompanied by color and sound, etc.).

Particularly in regards to framing, Niederdeppe et al, (2008) suggest that messages implicating different attributions would have important influence on persuasiveness; therefore they made the distinction between internal framing and external framing. Internal framing refers to messages portraying personality qualities or characteristics that have caused a social problem, such as health disparity. External framing concerns messages which are presented to suggest that some environmental factors have caused a social problem (Niederdeppe et. al, 2008). They conclude that the most effective message framing to communicate the importance of health disparities is one that mixes both internal attributions and external attributions into the message to be presented. From their perspective, this will be the best frame for a message about a social problem (Niederdeppe et. al, 2008). Moreover, messages can be framed differently according to the time frame.

One study found that framing the question into “when you were drinking” or “in the past 12 MESSAGE FRAMING Framing 10 months” did not make a difference in drinking, whether applied to the general population or to a clinical sample undergoing drinking intervention because the different wording does not urge people to answer based on different memory (Weisner, Kaskutas, Hilton, &

Barlie,1999). Along with these presentation styles mentioned above (i.e. attribution, visual image, narrative, and time), there exists another major framing device which has been under intensive investigation, the framing in terms of gain vs. loss. This is discussed in the later section and serves higher theoretical interest. For these examples of definitions of message framing, besides Gain vs. Loss Frame, Table 1 presents details about various kinds of message framing.

Some research states that for different clients in different stages of readiness to change health behavior, different types of message framing exert maximal influence in terms of persuasion. There exists no one message frame that works best for all the stages. For example, for men who have sex with men (MSM), Gilbert, Peterson, and Scanlon (2010) found that different subjects in different readiness stages (pre-contemplation, contemplation, preparation, action and maintenance) perceived different types of framed messages as most persuasive for injection of hepatitis A and B vaccinations. According to the theory by

Pelletier and Sharp (2008), different phases of behavior change (detection phase, decision phase, and implementation phase) also demand different kinds of framed messages. In the detection phase, fear appeal in the message is useful; in the decision phase, gain-framed messages (to be discussed in the following paragraphs) are more persuasive, and action plan and implementation goals in communication may be received with less defensiveness

(Pelletier & Sharp, 2008). The individual difference on need for cognition plays a role as well. Putrevu (2010) generated an empirically supported hypothesis; people with low need MESSAGE FRAMING Framing 11 for cognition tend to rely more on message framing to determine a message’s persuasiveness. Also, the need for cognition as a personality trait is believed to play a role determining which category of the persuasion process occurs (Putrevu, 2010).

Compared to the ELM model and HSM model, persuasion caused by the message framing of positive vs. negative, (positive frame is also referred to as gain frame, and negative frame is also referred to as loss frame) is considered a more intuitive persuasion process which relies less on conscious evaluation (Putrevu, 2010). In sum, message framing is a very broad topic with many variations in operation definitions. However, the most frequently studied kind of framing has been gain frame vs. loss frame.

Framing: Gain Frame and Loss Frame. The earliest conceptualization of gain frame vs. loss frame can be traced to research in economics (i.e., prospect theory) by

Kahneman and Tversky (1979). To comprehend how people make decisions, Kahneman and

Tversky (1979) proposed prospect theory to generalize patterns of human choice. The prospect theory states that the perceived risk primarily drives choices. Specifically, human decision making is influenced by the presentation style of a message (Kahneman, &

Tversky, 1979). The prospect theory of Kahneman and Tversky (1979) was supported by

Newman (1980), whose theory suggested the value of including message framing into predictive equations for decision making. According to Newman (1980), the expected utility theory, which predicts choices based on the calculated expected gain, is still partially explanatory. Using prospect theory regarding a message presentation’s influence on decision making is much more practical in certain situations. Newman (1980) and Tversky &

Kahneman (1981) compared the prospect theory to the expected utility theory, and to the rational choice model, which was more accepted at that time. The rational choice model was MESSAGE FRAMING Framing 12 well-accepted with assumptions that people rationally choose according to their best interest.

However, biased by perceived messages which are technically framed or designed, people make choices which are inconsistent with the choices predicted by the traditional economic theory—rational choice theory (Tversky, & Kahneman, 1981). People’s choices are influenced by the framed message, and people genuinely believe in their choice, not because the choices are made hastily (Tversky, & Kahneman, 1981).

Later, in 1987, the term “gain vs. loss frame” was born, signifying when a message about possible alternative choices is presented. It may be displayed in a form about how much or what a person can acquire by pursuing a particular choice. This is called gain frame

(Meyerowitz & Chaiken, 1987). On the other hand, loss framed messages focus on possible consequences for people who do or do not make a certain choice (Meyerowitz & Chaiken,

1987; See Appendix A for example content from Meyerowitz & Chaiken). Meyerowitz and

Chaiken (1987) tested the effect of gain vs. loss frame about the issue of breast self- examination. Wilson, Purdon, and Wallston (1988) formally made a literature review and supported the meaning of applying message framing (gain, loss, or threat) to health communication. From prospect theory, the health belief model, and protection motivation theory, the perceived message can vary in relaying emotion (e.g. threat) and/or information

(e.g. survival rate), and these really matter for people’s decision and choices (Wilson,

Purdon, & Wallston, 1988). Over the years, the issue of positive vs. negative frame (i.e., gain vs. loss frame) has been under extensive study. For example, using pamphlets with health information Shamaskin, Mikels, and Reed (2010) found that age affects memory toward positive-framed messages compared to negative-framed messages, and the effect was shown from the higher recall and evaluation of the positive messages in the pamphlet. In MESSAGE FRAMING Framing 13 another example, Rucker, Petty, and Briñol (2008) presented either one-side (positive frame only) or two sides (both positive frame and negative frame) of framed message in their experiments. The information was about products to be sold. Results showed that two-sided messages brought more attitude certainty than one-side messages (Rucker et al., 2008).

Although theory about message framing is progressing through increasing relevant empirical research studies, there remain issues and criticisms of it. For example, Levy

(1997) who researched international relations and politics, suggested that the expected utility theory remains very useful in many cases. Levy (1997) suggested that prospect theory could work in situations which are or are not related to incentives. Thus, prospect theory is not a complete theory of decision making that can replace the expected utility theory. Also, much empirical evidence for prospect theory has been gathered over the years. Whether this evidence can apply to real situations, such as political decision making for international affairs, is still uncertain—especially since many experimental procedures were quite artificial and distant from real life (Levy, 1997). In summary, the validity of manipulating persuasiveness through message framing directly depends on each case or each category of persuasion.

Gain Frame and Loss Frame in Miscellaneous Topics. The effect of message framing has been tested within several research domains. For example, the process of negotiation was assumed to be affected by the framing of the message (Carnevale, 2007).

Using game simulation, with profits as the goal, the research conducted by Carnevale (2007) found that college students tended to make more concessions of virtual cash to imaginative others within the computer when the message about the potential result of a choice was expressed in a loss frame. Simultaneously, their positive moods were instigated by the MESSAGE FRAMING Framing 14 assignment of sweet food, such as candy. Also, positive affect was found to link with a consistent pattern regarding concessions to other people or self-profit (Carnevale, 2007).

This research by Carnevale (2007) is directly related to the study by Kahneman and Tversky

(1979), concentrating on the effect of message framing on the domain of economic behavior, or human decision making.

The frame of gain vs. loss has also been studied in the domain of political persuasiveness. Vraga, Carr, Nytes and Shah (2010) published one study with 519 students.

The outcome was the discrepancy between self-aversion and perceived aversion of other people. Four kinds of messages about a policy aiming to loosen the restrictions of health insurance to students were presented. Gain vs. loss frame was the first dimension for the categorizations of the message presentations. The other dimension was precision vs. realism-

-whether the message contained several credible facts (precision) or whether it focused on consistent logic to persuade readers (realism). Vraga et al. (2010) found that gain-framed messages increased the willingness of students to acknowledge potential limitations of the current situation of the Wisconsin state public healthcare, and thus supported a proposed policy that enlarged the students’ accessibility to health insurance. Written and video messages about health care and possible policies were manipulated by gain-loss and resonant facts-control categories. With the gain-framed messages included more credible facts to make the messages highly precise, the students’ awareness of the limitation of current Wisconsin healthcare policy was greater. They showed higher willingness to follow the proposed policy on enhancement of their health insurance access, when they compared between their own potential costs and imagined others’ potential cost to access health care.

(Vraga et al., 2010). MESSAGE FRAMING Framing 15

Negotiation and political persuasion have been another major area in which researchers have tested the effects of framing (gain vs. loss) on audiences. Chang and Lee

(2009) studied message framing with charity appeal advertisements. The messages appealing for charity contributions were framed either positively or negatively (Chang, &

Lee, 2009). The positively-framed message increased its persuasiveness with the addition of a positive image (i.e. pleasant picture). On the other hand, the negative-framed message increased its persuasiveness with the supplement of negative image (unpleasant picture)

(Chang, & Lee, 2009). Temporal framing depicting the risk of population in need (next hour vs. next year) had an effect on the persuasion as well. Short-term temporal frame increased the persuasiveness of a negative-framed message, and long-term temporal frame increased the persuasiveness of positive-framed message (Chang, & Lee, 2009).

McKenzie-Mohr (1994) tested the issues of message framing on the persuasiveness of advocated pro-environment behaviors. Similarly, the presented campaign to persuade people to conserve energy, (i.e. electricity) can be categorized into gain frame or loss frame

(McKenzie-Mohr, 1994). Reminding homeowners of the potential cost for excessive electricity usage is more effective for persuasion than reminding homeowners of the potential savings of cost from cutting excessive electricity usage (McKenzie-Mohr, 1994;

See Table 2 for a summary for these examples of framing studies for various topics).

While message foci have been different, many research studies have related framing

(gain vs. loss) to the persuasiveness of a message. Relevant variables, such as decision for self versus others, affect, and credible facts within the message, etc. have been explored and tested in experiments. As would be expected, framing has also been studied within the domain of health promotion, which aims to change people’s unhealthy behaviors through the MESSAGE FRAMING Framing 16 process of persuasion and message presentation. The next section reviews the current literature on framing studies related to health promotion.

Frame (Gain vs. Loss) on Health Issues. Regarding message framing of gain vs. loss, there have been many health messages in the public media for health promotion; many health messages use this message framing. With qualitative coding by content analysis,

Latimer et al (2010) examined the frames of printed health messages from the media named as New York State Smoker’s Quitline (NYSSQ) and 12 other messages by the service calls from Quitline. The percentage of gain-framed, loss-framed and non-framed messages was reported for both printed messages and telephone messages (Latimer et al, 2010). The most representative finding was that 21.6% of these printed messages were gain-framed, and

10.2% of these telephone messages were loss framed (Latimer, et al, 2010). In theory, this result by Latimer et al., (2010) confirms the existence of gain and loss message frame(s) within those health messages to which everyone had access. Using a qualitative method,

Gallopel-Morvan, Gabriel, Le Gall-Ely, Rieunier, and Urien (2009) found that social marketing of anti-tobacco messages in France similarly depicted both positively and negatively framed messages in the form of text and visual images. An anti-tobacco message with implications of self-efficacy and visual warning was found to be the most effective in persuasion (Gallopel-Morran et al., 2009). This finding can be understood in the sense that positive frame and negative frame combined together can result in the greatest persuasion in anti-tobacco social marketing.

At the end of the 20th century and the beginning of the 21st, various researches have questioned how message framing affects the result of health promotion. Banks, Salovey,

Greener, Rothman, and Moyer (1995) presented two kinds of video productions to two MESSAGE FRAMING Framing 17 groups of women. These videos provided information on mammography for the detection of breast cancer. “The two videos were identical in the factual information conveyed; they differed only in the way persuasive statements were framed” (Banks, et al., 1995, p180).

From the follow-up data 12 months after the experiment, women watching a loss-framed video about mammography and breast cancer had a higher rate of mammography check-ups

(Banks et al., 1995).

To test the effect of message framing on the promotion of breast self-examination

(BSE), Williams, Clarke and Borland (2001) recruited 539 women to complete a telephone interview and receive experimental material via mail. Half of the participants in the study received a version of a brochure titled “Breast Cancer and Family History” (Williams et al.,

2001, p932), and the other half of the participants received a different version of the brochure. The former group’s brochure contained a positive gain-frame message; the latter group received the version with a loss-framed message. The gain-framed message presented the estimated percentage of women who have avoided the onset of breast cancer by participating in breast self-examination, and the loss-framed message presented the estimated percentage of women who had gotten breast cancer (Williams et al., 2001).

Regardless of framing, all participating younger women, tended to perceive greater risk of positive diagnosis of breast cancer then their older counterparts, and, thus, increased intentions to perform breast self-examination after reading the brochures (Williams et al,

2001). For all the participants, the loss framed messages were related to greater perceived risk of positive diagnosis of breast cancer and higher intention to perform breast self- examination (Williams et al., 2001). MESSAGE FRAMING Framing 18

Framing (gain vs. loss) affects how decisions are made and perceived risks

(Armstrong, Schwarz, Fizgerald, Putt, & Ubel, 2002). In an experiment by Armstrong et al.,

(2002), the colon cancer survival curve, which indicates the possibility of being alive once colon cancer is diagnosed, whether under treatment or not, was represented in a gain frame message. The mortality curve, which indicates the possibility of dying once colon cancer is diagnosed, whether under treatment or not, was represented in a loss frame message

(Armstrong et al., 2002). The loss-framed message was associated with less understanding of the health message and lower intention to undergo surgery for colon cancer (Armstrong, et al., 2002).

Known Moderators behind Gain vs. Loss Frame. The framed message based on gain vs. loss may influence the outcome persuasion depending on some moderating variables. Here some potential moderators are reviewed. They are message color, perceived risk, individual difference, consideration of future consequences, and self-efficacy. Color may also moderate the effect of message framing on health communication, according to

Gerend and Sias (2009) who recruited male undergraduate students to read pamphlets regarding the injection of HPV (Human Papillomavirus) vaccine. When the pamphlet was printed with the threatening color-red, the message in loss frame resulted in greater persuasion to the participants, compared to any other combinations of color (grey or red) and frame (gain vs. loss) (Gerend, & Sias , 2009).

Gain-framed messages may assure people and make them confident to pick the recommended choice, when this behavioral choice is presented as easy and with lower failure risk. Bartels, Kelly, and Rothman (2010) found that when the preventive health behavior (i.e., injecting vaccine against West Nile virus) was presented with a higher MESSAGE FRAMING Framing 19 success rate, a gain-framed message was found to be more persuasive. The same was true with lower success rate paired with loss-framed message. In another study when the detection of a fictitious toxin was presented with higher risk to the body, a loss-framed message was more persuasive; a gain-framed message was more persuasive when the toxin was paired with lower risk (Bartels et al., 2010).

Some individual differences or trait variables have been found to affect message framing and persuasion for health behavior. While many individual differences have been tested, moderating effects have been found in some but not others. Cohen’s (2008) conference paper tested the effect of attitude ambivalence (i.e. attitudinal inconsistency for the benefit or risk to become an organ donor). Lower ambivalence was predictive of higher intention to become an organ donor, but no significant statistical relation to message framing was found (Cohen, 2008). Age has also been found to play a role in the process of framing persuasion. Shamaskin et al (2010) distributed pamphlets about “influenza, cholesterol, skin cancer, sexually transmitted disease” to the participants; older participants, relative to younger participants, found gain-framed messages were more persuasive. Both age and frame are the main effects explaining for the higher evaluation of these pamphlets.

Consideration of future consequences (CFC) has also been studied as relevant to the perception of framed messages (O’Connor, Warttig, Conner, & Lawton, 2009). The evaluations of both immediate danger, and long term development in life were found to relate to the effect of message framing. From O’Connor et al. (2009), high CFC is a mindset that highlights the importance of long term consequences, while low CFC is a mindset that manages to gain benefit in the short term and neglects the long term gain-loss more. One hundred and seventy participants recruited via e-mail went through the study by O’Connor MESSAGE FRAMING Framing 20 et al. (2009). They read hypertension messages framed according to prevention or detection and gain or loss. The frame of prevention or detection of high-blood pressure did not affect the participants’ time spent reading the message. However, the loss-framed message about hypertension, no matter whether detection or prevention oriented, caused high CFC participants to spend a longer time reading the message compared to low CFC participants.

On the other hand, low CFC participants spent a longer time reading gain frame message, rather than loss frame message. The above are representative studies exploring how people consider the potential risk and benefit conveyed by health messages under gain or loss frame.

Message framing is also related to self-efficacy when the normal health behavior exercise plays a role (Latimer, Rench, Rivers, Katulak, Materese, Cadmus, Hick,

Hodorowski, & Salway, 2008). Latimer et al. (2008) conducted a longitudinal study of 517 sedentary people with mailing of brochures and telephone follow-ups. Compared to a loss frame message, a gain-framed message about physical exercise was linked to higher self- efficacy at Week 2 and higher participation in exercise at week 9 (Latimer et al,2008). The sample findings are shown in Table 3.

The effectiveness of gain vs. loss frame has been mixed and complicated. Some researchers have found non-significant effects from message framing. For example, with

103 Latino Americans as participants reading messages about the potential risk of kidney disease, Maguire et al. (2010) said that behavioral intention to understand kidney disease and ask for help from doctors was not predicted by categories of presented message frames

(gain or loss), controlling for individual susceptibility to the disease, severity and other covariates. Consistently, the message frames could not predict other variables, such as MESSAGE FRAMING Framing 21 perceptions of severity and susceptibility, response efficacy and self-efficacy, controlling for remaining dependent variables as covariates (Maguire et al, 2010). Findings appear to depend on each research setting, such as the culture of the sample group, etc. Not every research study has supported an effect of message framing (gain vs. loss) in health promotion messages.

Frame (Gain vs. Loss) and Emotion on Health Issues. Emotion (or affect) has been one popular variable under frequent investigation. How emotion plays a role in the process of social influence has been introduced in the previous section. This paragraph specifically addresses these studies that include both message framing (gain vs. loss) and emotion to explain how health campaign works. To construct a coherent theory about the relationship between emotions and frame (gain vs. loss) to outcome of health behavior or intention can be complicated because some researchers support definite links between mood and frame, but some others have contrary conclusions. Table 4 has summary findings regarding this paragraph.

Cho and Choi (2010) used 246 elementary, middle and high school students as participants and let them watch ads from the Office of National Drug Control Policy. These ads were for anti-drug promotion. The mood induced by observing these ads was hypothesized to affect persuasion by Cho and Choi (2010). In their experiment, positive emotion words, such as happy and contented, and negative emotion words, such as afraid and scared, were presented as alternative responses to the general question, “How much did this ad make you feel?” (Cho & Choi, 2010, p306). Taking the attitude toward the these anti- drug messages as the dependent variable, positive mood was more predictive of higher support toward anti-drug messages when presented in gain frames (Cho & Choi ,2010). On MESSAGE FRAMING Framing 22 the other hand, Cho and Choi (2010) failed to find significant association between negative mood and loss frame.

Keller, Lipkus and Rimus (2003) tested women’s response to information about breast cancer and mammography. State mood was manipulated in their first experiment; participants were told to read the Life Event Inventory and answer the questions. Positive mood was associated with more persuasive loss-framed messages; negative mood was associated with more persuasive gain-framed messages (Keller et al., 2003). In their second study, the CES-D scale measuring depression was used as indication of current mood. In general, positive mood plus a loss framed message predicted higher perception of risk about breast cancer and lower perceived cost of the mammography; negative mood plus a gain framed message similarly predicted higher perception of risk and lower perceived cost of the mammography as well. Keller et al. (2003) inferred that positive mood leads to higher motivation to manage future calamities and that positive mood is associated with higher persuasion in loss-framed messages.

Another study about mood and framing was conducted by Yan, Dillard and Shen

(2010). In their first study mood state was manipulated. Happy participants were more likely to adopt attitudes consistent with behavioral approach decisions, whereas sad participants were more likely to adopt attitudes consistent with behavioral inhibition decisions (note. behavioral approach means to encourage oneself to do a behavior, and behavioral inhibition means to constrain from doing a certain behavior.) In the Yan et al. (2010) study, health messages about moderation of junk food (i.e. appropriate intake of junk food) and participation in eye exams were presented to different subjects. Yan et al. (2010) treated moderation of junk food and participation in eye exams from different behavioral systems: MESSAGE FRAMING Framing 23 the former-behavioral inhibition system (i.e. prescriptive message) and the latter-behavioral approach system (i.e. proscriptive message). The procedures included mood manipulation

(i.e. reading Live Event Inventory and writing about a life event) and presented health information in either gain frame or loss frame. Yan et al. (2010) found that not only positive mood plus gain-framed messages can result in greater persuasion of proscriptive messages, but also that persuasiveness of a prescriptive message can be greater when negative mood and the loss-framed message were combined. (Yan et al., 2010).

Anxiety has also been examined as a moderating variable in this domain of research.

In Millar and Millar (1998) and Millar (2006), anxiety played a role in predicting cognitive processing or persuaded intention. Message framing (health detection vs. health promotion) is one key factor to the persuaded intention to behave healthily. State anxiety (Millar, &

Millar, 1998) and trait anxiety (Millar, 2006) serve as statistical moderators between message framing and behavioral intention to act according to the message.

However, some researchers have derived the opposite conclusions. For example,

Park, Simmer, Prevost, and Griffin (2010) published their study which rebutted this theory.

In their study, 116 participants, ages 40-69, joined the experiment after being selected based on the risk of having type 2 diabetes. Those participants read the invitations for screening for type 2 diabetes, which were written in gain, loss or neutral frame. However, no effect of message framing was found for the behavioral intention for screening for type 2 diabetes, nor did message framing significantly affect the state anxiety (Park et al., 2010). Overall, research examining emotion in the context of framing and behavioral intentions has produced mixed findings. Not all research has supported the utility of these variables, but recent studies seek to verify this. MESSAGE FRAMING Framing 24

Message Framing and Drinking. Some research has begun to test the effects of message framing (gain vs. loss) on messages related to alcohol consumption. One hundred college student drinkers participated in the study by Brown and Locker (2009) testing how drinkers’ defensiveness can be reduced by varying the presentation of a message. Two kinds of messages about drinking were randomly assigned and presented. One was loss-framed message, which portrayed the potential risk of drinking with autobiographical photos and texts (Brown & Locker 2009, named it as emotive condition). The control group read a pamphlet with less autobiographical materials, which was loss-framed to a lower degree

(Brown and Locker named this condition as non-emotive condition).To measure psychological defensiveness toward the message, the estimated health risk of drinking after reading the pamphlet was measured. Brown and Locker (2009) found that for people with higher indices of both denial and vulnerability, the loss-framed message (i.e., emotive condition with autobiographical images) with significant emotional tone decreased the participants’ estimated risk. That is, highly defensive and vulnerable participants showed the highest psychological defensiveness toward the loss-framed message compared to participants with low denial and low vulnerability, and to participants high in only one of these dimensions (Brown & Locker, 2009).

On the other hand, Quick and Bates (2010) tested the effect of both gain and loss- framed messages dissuading drinking with a sample of students from communication classes. The gain-framed message presented by Quick and Bates (2010) was, “By not abusing alcohol, you have better coordination and judgment, are likely to drive safely, and can make better decisions before engaging in sex. You will also be physically healthier”

(Quick & Bates, 2010, p609). The loss-framed message used by Quick and Bates (2010) MESSAGE FRAMING Framing 25 was, “If you abuse alcohol, you will suffer from impaired coordination and judgment, and increase your risk of drunk driving and unsafe sex. You may also suffer from stroke, damage to the liver and brain, and even death” (Quick and Bates, 2010, p609). However, the predicted effect of message framing in the study by Quick and Bates (2010) was not significant. There were no differences in terms of psychological defensiveness in relation to health promotion message. No relation between perceived threat and message frame (gain or loss) was found.

There has been some research that has found effects of message framing intentions related to fetal alcohol spectrum disorder (FASD). Yu, Ahen, Connolly-Ahern, and Shen

(2010) studied 213 female undergraduate students in their experiment; two dimensions of message framing (gain vs. loss & statistics vs. exemplar) were manipulated. The participants in the gain frame conditions showed higher intention to prevent FASD, and loss frame plus exemplar frame led to the highest perceived severity of FASD and highest perceived efficacy to prevent for FASD.

Research has not specifically targeted message framing and problems related to alcohol through any in-depth research studies. It is possible that message framing affects persuasiveness of messages encouraging drinking reduction. Whether this is the case awaits the results of future studies. Table 5 has a simplified summary.

Bridging: Future Research Directions. There has been a large amount of research published about the persuasive effect of message framing on health promotion. Research about abstinence or moderation of drinking represents one important topic in health promotion research. Comparing the uniqueness of anti-alcohol-campaigns to other health MESSAGE FRAMING Framing 26 promotions is very difficult because the nature of health behaviors vary considerably and the relevant variables affecting the relation between message framing and persuasion are abundant. For example, differences between individuals are known to affect the persuasiveness of health promotion. The proposed research is designed to extend research on framing and health promotion communications as a means of reducing problem drinking.

The identification of important factors associated with responses to health communications is important in the construction of health-promotion theory and the application of theory to drinking reduction. Moreover, the current issues and findings about drinking per se represent another important basis for the formulation of future research plans.

The empirical quests of this dissertation have been designed to extend the knowledge of health communication with manipulation of gain vs. loss message frame in the domain of problem drinking and the other two major topics: Ego-depletion and psychological reactance. Ego-depletion and psychological are thought to impede social influence conditionally.

Study 1. Message Framing and Ego-depletion on Drinking Reduction

Introduction

The two theories, ego-depletion and theory of reactance may interact with framed message regarding how campaign works. Study 1 focused on theories which assume human self-control as a limited resource at a given time, i.e. ego-depletion is limited. Study 2 investigated the connection between reactance theory and message framing. Accordingly, this first study evaluated whether ego-depletion leads to the different patterns of message processing when dealing with messages under different message frames. MESSAGE FRAMING Framing 27

Decision making always consumes cognitive resources when considering the alternative choices regarding temptation and/or decisions implicative of ethical consequences (Baumeister, Bratslavsky, Muraven, & Tice, 1998). Baumeister et al. (1998) refer to the difficult process to make choice as “ego depletion”, while some other terms describe the same basic process. For example, Muraven, and Slessavera (2003) designate it as self-control failure; Fischer, Greitemeyer, and Prey (2008) call it depleted self-regulation resource. Ego-depletion represents the term used here.

Ego-depletion and its Implications. Ego-depletion has been shown to affect many aspects of behaviors and volitional functions. In one example, ego-depletion was related to the behavioral constraint from tempting food. In the study by Baumeister et al. (1998), participants who were refraining from eating chocolate showed significant inefficient performance in an intellectual puzzle task. This means that refraining from eating tempting food (e.g., chocolate) results in consumption of mental resources and is likely to lead to ego- depletion (Baumeister et al., 1998). Ego-depletion can also be initiated from the practice of attention-directing tasks or emotion-suppressing tasks (Schmeichel, Vohs, & Baumeister,

2003). Ego-depletion functions with certain other factors. Motivation has been found to compensate cost of cognitive resources used in depletion tasks; when the subjects were told that an intellectual task would improve their general intelligence if sufficiently practiced, the participants could perform significantly better regardless of ego-depletion (Muraven, &

Slessavera, 2003). In contrast, Clarkson, Hirt, Jia, and Alexander (2010) argued that it is perceived depletion rather than actual depletion that relates to the weakening of intellectual performance. MESSAGE FRAMING Framing 28

Ego-depletion has also been tested for its implications with other variables, especially emotional variables. Two kinds of emotion have been shown to be associated with ego-depletion. First, rumination increases aggression via weakening of self-regulation

(i.e. resulting in ego-depletion), according to Denson, Pederson, Friese, Hahm, and Roberts

(2011). Secondly, cognitive regulation, the process adjacent to ego-depletion, is probably directed by envy because the activation of envy is linked to a different pattern of attention and memory recall (Hill, Del Priore, & Vaughan, 2011). In sum, ego-depletion has been widely implicated in various psychological processes.

Ego-depletion affects Attitude Certainty and Information Processing. The current research aims to understand how audiences accept certain persuasive messages. The attitude to reject a message or not is the key here. Ego-depletion models affect attitude in two dimensions: strength and attribution. First, attitude could be stronger (i.e. the initial pro or con becomes more intense) after experience of ego-depletion. Wan, Rucker, Tormala, and

Clarkson (2010) examined undergraduate students as potential consumers and tested how ego-depletion relates to their loyalty to product brand. After completing the task for ego- depletion in their Study 1, participants showed greater intention to buy an advertised food which they had previously supported (i.e. showed higher attitude certainty). Moreover, the experience of ego-depletion made the consumers perceive themselves as having detailed elaborations when they make decisions with attitude certainty (their Study 2). Especially, for those participants who recognized that the depleting task retarded their processing of new material (i.e. identifying the depletion task as the cause of weak mind state), attitude certainty was much higher (their Study 3, Wan et al., 2010). Furthermore, ego-depletion may affect the pattern of information seeking. Fischer et al. (2008) examined confirmatory MESSAGE FRAMING Framing 29 information seeking as focusing attention on the information behind which the perspective is consistent with a person’s initial attitude or ideology. Participants who completed the task of ego-depletion were found to seek more confirmatory information seeking. Therefore, once ego-depletion occurs, people tend to hold on to their original attitude more firmly, either by expressing the relevant intention (Wan, et al, 2010) or directing attention to more agreeable information (Fischer et al., 2008).This suggests that ego-depletion may cause drinkers to insist on their initial attitude which supports regular drinking. Hence, the first hypothesis of this dissertation is:

H1: Heavier drinkers will tend to show less acceptance of the health message after

experiencing the process of ego-depletion.

Three variables have to set equal on each comparison. These three are statistically controlled variables. They are self-esteem, instability of self-esteem and weekly drinking.

Moderating Drinking to be Controlled. Drinking may moderate the research data, so it was set as the controlled predictor. Drinking identity has been regarded as indirectly related to actual alcohol take (Foster, Yeung, and Neighbors, 2014). Since drinking identity is always closely related to one’s self and that implicates how incoming messages are understood and accepted, this research considered that higher drinking may indirectly accompany with a qualitatively different drinking identity, making the research product from attempted health persuasion deviate from participants having lower drinking regularly.

In other words, drinking is thought to make responses different; high drinking intake may make people more resistant to adjust drinking as a function of addiction. The control variable was drinking. MESSAGE FRAMING Framing 30

Ego-depletion and Loss Frame on Health Campaign. Message framing may play a role in promotional outcomes of health messages. More deductions about message framing are provided in the second study. The reactance theory postulates that psychological reactance occurs under perceived threat. Therefore, audiences tend to make opposite choices if a message is presented as threatening. This psychological threat is unpleasant to the psychological self or personal reputation. This research also hypothesizes that heavier drinkers tend to hold stronger attitudes for their drinking behavior at least temporarily after experiencing ego-depletion. When the two assumptions are combined, people might react more when the message sounds unpleasant and when people are ego-depleted.

To illustrate, this research claims again the logic from Wan et al. (2010) and Fischer et al. (2008) that ego-depletion shortly precedes affirmation of one’s original attitude. This means that people are more likely to insist on their original attitude in the mind state of ego- depletion. Combining with the reactance theory from Brehm and Brehm (1981), people in the state of ego-depletion would disobey the loss-framed message which looks threatening to their ego to a greater extent on the ground that ego-depletion leads to more attitude certainty and confirmatory information seeking. More explanation and further study about reactance theory and message acceptance are conducted and offered in the Study 2 section.

Also, peak, typical or regular drinking may moderate this hypothesized reactance.

Higher amounts and frequency of drinking is thought to boost reactance potential because this behavioral pattern implies the person could be a heavier drinker who supports drinking more firmly. The idea that a higher drinking profile may lead to stronger support for drinking and higher reactance to health messages challenging heavy drinking is based on the MESSAGE FRAMING Framing 31 theory from Bem (1967), which indicates that one tends to construct one’s attitude by perceiving his/her own behavior. Therefore, the following is the second Hypothesis:

H2: Participants who finish a cognitive task leading to ego-depletion are more likely

to report lower message ratings and greater drinking intentions when they read a loss

framed message; and this will be particularly true among heavier drinkers.

The above are the two hypotheses for Study 1. Ego-depletion and message framing are the grounding variables to the addition of the psychological reactance and message framing model. The researcher expects that ego-depletion will change the flexibility of drinker attitude alone, but also will increase drinker’s psychological reactance to health messages when it functions along with the presentation of loss-framed messages.

Study 1. Method

Participants. Study 1, the ego depletion study, included a 2X2 online experiment. A link to the survey was sent to the participants who enrolled via the psychology experiment

SONA web-site. Participants included students who were taking at least one psychology course and identified themselves as regular drinkers. The major experimental dimensions were ego-depletion vs. control and Gain-framed message vs. Loss-framed message.

Participants earned their extra course credits as the incentive.

This method section first describes the measures used and then describes the entire experimental procedures. As indicated by the power analysis graph in Figure 1, collecting a sample of 200 participants was estimated to provide statistical power up to .9 for medium effect sizes (f = .25) or greater. MESSAGE FRAMING Framing 32

In total, 207 participants provided valid data and did not leave portions of incomplete data on the internet survey. The first condition was the ego-depletion manipulation plus the gain-framed message group, which included 42 participants. The second condition was the ego-depletion manipulation plus the loss-framed message group, which included 61 participants. The third condition was the control group with the gain-framed message, which included 45 participants. Finally, the fourth group which was control with the loss-framed message and had 59 participants. In total, 168 females and 39 males composed this experimental sample.

Experimental Materials: Drinking Scales. The following three drinking scales, plus one intention scale for Study 1 and Study 2 were: the Quantity-Frequency-Peak Alcohol

Use Index, (Baer, 1993; Marlatt et al, 1995), The Daily Drinking Questionnaire (Collins et al., 1985), the Rutgers Alcohol Problem Index (White & Labouvie, 1989), and the Drinking

Intention Scale modified from the original work of Dimeff, Baer, Kirlahan, and Marlatt

(1999).

The Quantity-Frequency-Peak Alcohol Use Index. This five-item scale was created by Baer (1993) and reapplied by Marlatt, Baer, & Larimer (1995). This scale measures alcohol consumption over the past month. Two items are about the participants’ peak drinking, for example, “Think of the occasion you drank the most this past month.

How many hours did you spend drinking on that occasion?” Two items are about general drinking, e.g., “On a given weekend evening during the past month, how many hours did you spend drinking? Estimate for the past month.” Also, one item is for the participants to summarize their overall drinking frequency. MESSAGE FRAMING Framing 33

The Daily Drinking Questionnaire (DDQ; Collins, Parks, & Marlatt, 1985). The

DDQ is a widely used measure which asks participants to list the average number of standard drinks consumed on each day of a typical week over the previous three months.

Scores represent the average number of drinks per week over the past three months.

The Rutgers Alcohol Problem Index (White & Labouvie, 1989). This 25-item scale lists possible alcohol-related problems which drinkers may encounter and asks the participants to report the frequencies of problem occurrence in the past 3 months. For example, “noticed a change in your personality” and “neglected your responsibilities” are two example items. The measure is taken by summing all items. Thus the possible range is from 0 to 100. Item responses include 0 (never), 1(1 to 2 times), 2(3 to 5 times), 3(6 to 10 times), and 4 (more than 10 times).

The Drinking Intention Scale. This scale is modified from the scale by Dimeff,

Baer, Kirlahan, and Marlatt (1999). The research used 4 of the 5 items in this scale. They are: (1) The intended amount of drinking each day of a week during the next month;(2)

Intended Drinking hours on each day of a general week during the next month; (3) Arranged frequency of drinking during the next month; (4) Intended amount of average drinking on each weekday during the next month. See the Appendix C for the content.

Experimental Materials: Imaginative Writing Task. To experimentally reproduce the state of ego-depletion, the participants in the experimental group were assigned an imaginative task. The experimental group was told that this task was to test their imagination. The instruction of ego-depletion for the experimental group was modified from

Fischer et al. (2008). The intense self-control requirement that is thought to lead to ego- MESSAGE FRAMING Framing 34 depletion was facilitated by instructing the participants to write about a real or fictional zoo trip. The constraints requiring self-control which added to ego-depletion was their not being able to use the words: “I”, “like”, “and”, or “or.” Appendix D has the content of this experimental manipulation. For the participants in the control group, they received the instruction telling them to wait for three minutes.

Presented Health Messages. The health messages were displayed after the participants completed their ego-depletion task or waiting. In Study 1, the participants either read the gain-framed message with statistical information or the loss-framed message with statistical information. As listed in Appendix A, both messages discussed the potential health implications (i.e. benefit of avoiding drinking or cost of drinking), such as health issues and birth defects. The messages were compiled with several information sources on the Internet.

The Messages and Believability Scale. To test how participants perceive the persuasiveness of the presented health information, the Realism/Believability Scale designed by Flanagin and Metzeger (2000) was deployed. This scale has five items: “Believable”,

“Accurate”, “Trustworthy”, “Biased”, and “Complete”. Each item asked how participants evaluated the presented message using the qualities above. The scale had the range between one and seven. The higher the score, the higher the message rating was. The current data indicated that the Cronbach’s α was .73.

Experimental Procedures. The online SONA computer system served as the platform for the recruitment of participants. The recruitment information invited students who identified themselves as regular drinkers with the minimum of one alcohol drink per MESSAGE FRAMING Framing 35 week. On SONA, potential participants were presented with a link to the study, which was programmed using the Datstat Illume platform. The first step for participation was to check the online informed consent. The informed consent was written according to guidelines of the University of Houston Committee for the Protection of Human Subjects (See Appendix

B). After the informed consent was obtained, the participants entered their experimental surveys depending on their self-reported birth months. Then, they filled in the questionnaire for demographic information. Next, the major research questionnaires, including drinking behaviors, were presented for the participants to complete. Then, the ego-depletion or control condition was administered. Ego-depletion involved writing about the imaginative zoo trip. The control condition remained waiting. When the task of depletion was accomplished, the health messages and the Realism/Believability scale displayed for the participants to rate the health message they saw (Texts shown in Appendix C). To relieve the participants’ burden of memory, when the participants evaluated the qualities of the health message, the computer program allowed them to review it as carefully as they wished before they clicked on the icon saying they had finished reading.

Study 1 Results

Predictors by Experimental Design. For analyses, the two independent variables, ego-depletion and message framing were dummy coded. For each participant who experienced ego-depletion, the numerical value was represented by 1; for each participant who was in the control group without depletion, the numerical value was represented by 0.

On Message framing, the gain-framed message group was coded as 1, and the loss-framed message group was coded 0. The message rating revealed how the health campaign worked psychologically; if the participants accepted the health advice. MESSAGE FRAMING Framing 36

Outcome. The average scores of the entire items on the Message and Believability

Scale by Flanagin and Metzeger (2000) was the dependent variable for the first hypothesis test. Further, drinking intention, as well as message rating, were the outcomes in the equations of Hypothesis 2. Drinking intention was represented by weekly drinking intention from the first item of the Drinking Intention Scale by Dimeff, Baer, Kirlahan, and Marlatt

(1999) (i.e. the sum of all the blanks for Monday through Sunday drinking amounts); and also the other, typical drinking intention which represents the total intended drinks reported with the fourth item (Dimeff, Baer, Kirlahan, & Marlatt, 1999).

Drinking as the Statistical Covariates. Drinking served as a covariate and potential moderator, given the assumption that effects might not be present for light drinkers. This first study separately constructed drinking by the following three variables (1) Drinks per week: the sum of the first item from the Daily Drinking Questionnaire by Collins, Parks and

Marlatt (1985). That item measured Monday to Sunday, asking: “Consider a typical week during the last three months. How much alcohol, on average (measured in number of drinks), do you drink on each day of a typical week? ” (2) Peaks Drinks: the first item in the

Quantity-Frequency-Peak Alcohol Use Index (Baer, 1993), “Think of the occasion you drank most this past month. How much did you drink?” (3)Typical Drinks: the third item in the Quantity-Frequency-Peak Alcohol Use Index (Baer, 1993),” On a typical occasion during the last month, how much alcohol did you typically drink? Estimate for the past month”

Analyses. The procedures of ANCOVA were basic to this report section. The SPSS software downloaded the collected data from the DatStat Illume program, and SAS was used to calculate the basic descriptive statistics and regression models with covariates MESSAGE FRAMING Framing 37

(ANCOVA). The primary outcomes in the statistical regressions were message rating, which indicates how the participants accepted the advice from the health messages, and drinking intention, which was the major outcome in this health promotion research. As introduced above, drinking was separately represented by drinks per week, peak drinks and typical drinks. Each was included in separate regression analysis. The outcomes were message rating, weekly drinking intention and typical drinking intention; each was put in the calculation for the hypothesis involved.

Descriptive Statistics. Means and correlations are provided in Table 6. Participants reported drinking on average 6.16 drinks per week. Following experimental procedures, participants reported intending to drink on average 4.47 drinks per week over the next month. Correlational analyses revealed significant positive correlations between drinks per week and each measure of drinking intentions. Each intention measure correlated with one another significantly and positively. Also drinking items correlated with all the intention items.

Regression. The first hypothesis (H1) was that ego-depletion would result in lower message ratings. Three regression equations were examined with different drinking measures included as covariates in step 1 and potential moderators in step 2. Thus, the first hypothesis regressed message rating as the outcome on drinking and ego-depletion. The regression including both ego-depletion and drinks per week as predictors appears in Table

7. There were no main effects and no interaction. The next equation regressing message rating on ego-depletion and peak drinks presents in Table 8. Again, there were no main effects, but there was a significant interaction (b = .10, β=.22, t=2.44, p<.05). Figure 2 presented the interaction where high and low drinking values were specified as one standard MESSAGE FRAMING Framing 38 deviation above and below the mean of peak drinks (Cohen, Cohen, Aiken, & West, 2003; each interaction and simple slope analysis used the same procedure in this dissertation).

Results suggested a positive association between ego-depletion and message rating for participants who were higher in peak drinks and a negative association for participants who were lower in peak drinks. Tests of simple slopes suggested both of these associations were marginally different from zero. Again, the third table (Table 9) for the first hypothesis put ego-depletion and typical drinking as regression predictors and message rating as the outcome. No predictors appeared as significant.

As a whole, only the second regression for H1 showed a significant pattern, and only as an interaction between drinking and ego-depletion. However, the direction was opposite of the original anticipation. With high peak drinks and experience of ego-depletion, message rating as acceptance of persuasion on drinking reduction turns out to be the highest. The first hypothesis was not supported, but a finding worth further consideration and research was found.

The second hypothesis extended the scope of H1 and incorporated message framing in addition to ego-depletion. Moreover, two dependent variables were examined, message framing and drinking intention. Thus, the second hypothesis took turns examining the two dependent variables, message rating and drinking intentions, as a function of ego-depletion, message framing and drinking. Drinking intentions were operationalized with two variables

(drinks per week and typical drinks per occasion). In each case the drinking measure used as a covariate and potential moderator was specified to match the dependent measure for drinking intentions. Thus, four models were tested for H2; two outcomes X two drinking MESSAGE FRAMING Framing 39 measures. The predictors of drinking were centered. All models evaluated main effects, two- way interactions, and three-way interaction (ego-depletion X framing X drinking).

In the first model of H2, message rating was examined as a function of ego- depletion, message framing and drinks per week. Results are presented in Table 10. Results revealed no main effects. There was a significant two-way interaction between (b =-.08, β =

-.23, t =-2.09, p<.05) drinks per week and message framing. No other effects were significant. The graph for this two-way interaction between drinks per week and message framing is in Figure 3. The pattern of predicted values of Drinks X Frame interaction suggested that gain frame message was associated with higher message acceptance relative to loss-framed messages, when participants reported consuming more drinks per week.

The second model of H2 examined message rating as a function of peak drinks, ego- depletion, and message framing. The results are presented in Table 11. No effects were significant.

The third model of H2 examined weekly drinking intentions as a function of drinks per week, ego-depletion and message framing. The results appear in Table 12. Drinks per week predicted weekly drinking intentions (b = .75, β = .90, t = 11.82, p<.0001). The only other significant effect was a two-way interaction between drinks per week and ego- depletion. As shown by Figure 4, the association between drinks per week and weekly drinking intentions was slightly weaker for those participants in the ego-depletion condition.

The simple slopes analyses suggested that both slopes were significant. Both associations indicate that the higher the weekly drinking, the higher the drinking intention, regardless of ego-depletion. The three-way interaction term only approached significance. MESSAGE FRAMING Framing 40

The fourth model of H2 examined typical drinking intentions as the outcome and included typical drinks, ego-depletion and message framing as independent variables. The results are listed in Table 13. Typical drinks were positively associated with typical drinking intentions, suggested by the main effect (b =.60, β = 1.12, t =7.7, p<.0001).There were two significant two-way interactions. The interaction between typical drinks and ego-depletion suggested that the association between typical drinks and typical drinking intentions was weaker for those in the ego-depletion condition (b =-.51, β = -.76, t =-5.54, p<.0001). The other two-way interaction was between typical drinks and message framing and suggested the association between typical drinks and typical drinking intention was weaker among participants presented with the gain-framed message (b =-.46, β = -.47, t =-3.93, p<.0001).

Both of these two-way interactions were qualified by the three-way interaction among typical drinks, ego-depletion, and message framing. (b =.46, β = .34, t =3.04, p<.01). Figure

5 has the graph for the three-way interaction. The pattern of the results suggests, relative to all others, that heavier drinkers had particularly high typical drinking intentions when in the loss-framed and when not in the ego-depletion group. Thus, the pattern of results was opposite to the original hypothesis on the axis of ego-depletion. The original H2 postulated that for heavy drinkers, ego-depletion plus loss message frame would result in greatest drinking intention. However, the data shows that for heavy drinkers, the loss-framed message plus a very low ego-depletion state can make drinking intention very high.

Study 1. Discussion

In Study 1, ego-depletion was hypothesized to impede both the direct and indirect persuasiveness of framed messages. The acceptance of health messages as the direct response to new health information served as the evaluated outcome in the first hypothesis MESSAGE FRAMING Framing 41

(H1) and the first half of the second hypothesis (H2). H1 suggested that heavy drinkers would have disregarded the health messages in the moment of ego-depletion; the first part of

H2 suggested that the heavy drinkers would have shown more concern and rejection to the loss-framed messages in the moment of ego-depletion. Results did not confirm these expectations. The significant evidences suggest that ego-depletion in heavier drinkers created greater message acceptance. This stands opposite from the original expectation.

Suggested by the data, theory guides new research to the perspective that ego- depletion may boost acceptance of health advices, instead of rejection. The possible explanation for this recognizes that ego-depletion represents a state which leads to different decision makings, but this depends on which kind of choice are to be made. The research by

Wan, Rucker, Tormala and Clarkson (2010) affirms that ego-depletion potentially leads to attitude certainty; but they studied consumer choice among daily groceries. Choosing which products to purchase may be different from thinking about reducing drinking. Drinking can have both positive and negative implications for personal reputation, and this may make drinkers feel less justified as their drinking attitudes are considered, which is especially true for heavier drinkers. This may explain why in some given contexts, ego-depletion links to determination to follow health instructions. Ego-depletion may enhance evaluation of a behavior having both positive and negative aspects.

Three more explanations as to how ego-depletion leads to compliance with health advice about alcohol reduction are notable. Ego-depletion has been known as a mind state with lower mental resources to consider over behavioral alternatives. Philosophically, heavier drinkers may deceive themselves with respect to negative impacts drinking has on them (Walker, 2010). Therefore, it is a logical interpretation that psychological defense MESSAGE FRAMING Framing 42 against health information requires mental capacity. Then, the mind becomes suggestible to the health campaign while the ego-depletion state is temporally onset. Consequently, heavy

(peak) drinkers who see drinking as more important, tend to become more influenced by health campaign messages and thus show willingness to follow the health advice.

A second explanation concerns the focus on short-tem vs. long term benefit. Ein-

Gar, Goldberg and Sagiv (2012) asserted that consumers with inferior ability to self-control prefer purchasing health products emphasizing short-term benefit. Ego-depletion leads to acceptance of health advice by heavy drinkers, which may mean that during the process of low self-control, heavy drinkers tend to sense certain short-term benefits from health messages about drinking reduction. Whether this study indeed presented short-term benefit is uncertain, how to examine this hypothesis is discussed in the Conclusion.

Thirdly, drinking may count as a risky behavior. Unger and Stahlberg (2011) state that ego-depletion is a state which makes people avoid risk choices. Reading health messages under an ego-depletion state may make heavy drinkers aware of possible risks from alcohol and then produce higher ratings on the health messages under evaluation. The intention to avoid risk, which is more accessible when the mind is in a state of depletion, may lead heavy drinkers to rate the messages as more credible and positive.

Ego-depletion may have incited the constructive and experimental questions because how to induce this psychological state within available labs is a topic to address. In terms of the construct validity of ego-depletion, Study 1 modified the procedure (Fischer et al., 2008) by asking the participant in the experimental group to write an essay about a zoo trip.

However, the restriction of not using “I”, “like”, “and”, “or”. The restriction of not using “I” MESSAGE FRAMING Framing 43 might prime the participants’ self-concepts. As suggested by MacDonald, Sandry, and Rice

(2012), certain priming of self enhances efficiency of certain retrieval from short-term memory (i.e. STM). Ego-depletion was associated with higher message acceptance, and priming of self may account for this since it may make messages in short-term memory more accessible. Besides, the manipulation of ego-depletion may be imperfect in its control group. The control groups of ego-depletion in the previous studies (Fischer et al., 2008;

Muraven, Tice, & Baumeister, 1998) did not include waiting. Study 1 allowed participants in the control group to wait and proceed once they found the three-minute waiting time is over. The participants in the control group were told that cell phones were free to use. This control group may experience a certain disturbance when they volunteered to wait, which was not monitored. Thus, the results may not fully indicate if ego-depletion lowers of elevate message rating because the manipulation may be imperfect. Nevertheless, the procedure seems to be a logical extension of previous manipulations and experimental designs related to ego-depletion.

The first part of the second hypothesis (H2) regressed message rating on message framing and ego-depletion as the predictors and checked how they related to health message rating and drinking intention respectively. Low drinkers (i.e. drinkers drinking less) showed favoring evaluation on the message of gain frame. This may suggest that lighter drinkers do not closely identify with alcohol, so they prefer to follow the health information which looks appealing. For these people, gain frame had better persuasiveness on low drinkers. Although gain frame leading to higher message acceptance as the result is consistent with the meaning of the original hypothesis that loss frame leads to less acceptance, this pattern was found on only the regularly low drinkers. The question of how gain frame differentiates its status from MESSAGE FRAMING Framing 44 loss frame in the domain of healthy drinking campaign was pursued further in the next study.

The second half of H2 appointed drinking intention as the regression outcome.

Heavier drinkers had greater drinking intentions, which was not surprising. Ego-depletion significantly reduced the positive association between drinks and drinking intention, although the magnitude reduced was not large. A postulation is that ego-depletion represents not only a mind state exhibiting exhaustion; it may also promote confusion or attitude ambivalence. Therefore, when people experience ego-depletion, it would be more complicated to predict their future drinking intention from their past drinking habits. When the drinking variables become typical drinking intention and typical drinks, high drinkers who directly saw the loss-framed message and experienced very low ego-depletion reported the greatest typical drinking intention. It might mean that message framing exerts its greater influence by instigating psychological reactance, which makes typically heavy drinkers express stronger future drinking intention.

The next study considered psychological reactance as a framework for examining message framing effects. The outcomes were again message evaluation and drinking intention.

Study 2. Message Framing and Theory of Reactance on Drinking Reduction Campaign

Introduction

The first study examined how ego-depletion might have affected the evaluation of health communications. From study 1, ego-depletion had some effect on decision making MESSAGE FRAMING Framing 45 about drinking. The first study largely focused on ego-depletion and health communication.

The second study focuses more on reactance theory.

The second study aimed to understand psychological reactance and persuasion by focusing on the potential reactance that may proceed from loss-framed messages. Under the circumstances of thwarted freedom to choose, people experience psychological reactance and thus responses may be affected (Brehm, & Brehm, 1981). Reactance was also presumed to be associated with self-esteem and instability of self-esteem.

Overall, the models of persuasiveness and promotion on health are complicated and contradictory. Even the effects of message framing have been largely inconsistent in research studies. Some research indicates that gain-framed messages generate more persuasiveness, whereas other research suggests that loss-framed messages appear to be more effective. This difference is often found when at least one controlling variable or moderator is considered. Study 2 posited that Reactance Theory (Brehm & Brehm, 1981) may account for some of the mixed findings, especially for loss-framed messages. In brief, the loss-framed message may instigate psychological reactance in the process of health communication because it is presented with an unpleasant tone. The next section introduces a possible experimental context which may activate reactance via presentation of loss- framed messages.

Psychological Reactance and Message Framing. Loss-framed communication is expected to create more defensiveness and reactance based on Reactance Theory (Brehm &

Brehm, 1981). This serves as one of the fundamental expectations underlying this dissertation. The theory of reactance was initially composed by Brehm and Brehm (1981) MESSAGE FRAMING Framing 46 with detailed demonstrations. Brehm and Brehm (1981) and the subsequent work by Brehm

(1989) suggest that when a communication instigates a sense of threat or loss of freedom, and when the feeling of control is thwarted, people are more likely to disobey the transferred implication from certain proposed messages. Then, Lee and Lee (2009) proceeded from this paradigm and found some empirical support. For example, students who sampled a web service with higher perceived threat on future behavioral choices displayed lower intentions to continue the usage of this service by internet recommendation (Lee, & Lee, 2009).

According to Silvia (2006), the threatening sentences (i.e. a command forcing students to follow a student policy) either indirectly increased reactance through mistrust and counter arguing, or directly increased reactance of the college students.

However, whether loss or gain-framed messages convince people better has been a topic under intense debate. Loss-framed messages are usually considered as relatively threatening because of the adopted language and phrasing. Quick and Bates (2010) attempted to test the theory of reactance on the effectiveness of message framing on alcohol drinkers. Quick and Bates (2010) found that a loss-framed message could be persuasive when the audience was composed of heavy drinkers, even though this is contradictory to the assumption of Reactance Theory. Nevertheless, Quick, Scott, and Ledbetter (2011) extended this work and found confirming empirical evidence of Reactance Theory by targeting the intention to donate organs as the main outcome. Quick et al. (2011) found that a freedom- threatening message increased the state reactance by indirectly increasing the perceived threat. The research by Quick et al. (2011) is discussed at greater length in the subsequent sections. MESSAGE FRAMING Framing 47

Uniqueness of Statistical Information plus Texts Messages. The current dissertation suggests that it is possible to generate greater reactance when the health message is presented in a more precise form with more sophistication. One possible distinction of message presentations is based on text or statistical information. Text messages, consisting of only sentences is one message format. Another form of message presentation includes the addition of statistical information plus texts messages. When both sentences and numerical data are presented, the outcome of convincing an audience may be different. In the health promotion research by Greene and Brinn (2003), intention to support reduction of tanning bed use was the targeted healthy means to prevent skin cancer. Statistical messages were defined as messages with statistical information; narrative messages were defined as case stories (i.e. text messages). Although either narrative or statistical message could persuade the participants to decrease the frequency of using the tanning bed, the statistical message was regarded as having higher information value than the narrative message, which was treated as more real (Greene, & Brinn, 2003). Could this categorization between text message and statistical data message combine with the theoretical dimension of gain message frame vs. loss? To focus on drinking, the research studies about Reactance Theory and alcohol (Quick, & Bates, 2010; Quick, et al., 2011) are worth further consideration.

Why did Quick, Scott, and Ledbetter (2011) find more reactance from the loss-framed message, whereas Quick and Bates (2010) did not?

The difference may lie within the factor of the adopted experimental stimuli. The experiment by Quick and Bates (2010) did not include statistical information, but another by

Quick et al. (2011) included both statistical message and narrative message; also the perceived threat was manipulated with the message (Quick et al, 2011). Because Quick et al. MESSAGE FRAMING Framing 48

(2011) found more reactance, but Quick and Bates (2010) did not, the former experiment might have caused a stronger effect by impressing the participants more, with the more persuasive presenting style of salient messages including both text and statistical information. Based on these findings, in this research, reactance to a loss- framed message is predicted to strongly occur when the message combines both statistical information and texts.

Besides the salience of text plus statistical information, the second reason that adding statistical information to the loss-framed message along with the textual message may result in greater resistance is that the statistical information may contradict the drinker’s current perceptions about his or her own behavior or health status. For example, drinkers tend to have incorrect estimations about how other people drink (i.e. incorrect descriptive social norm; Neighbors, Dillard, Lewis, Bergstrom, & Neil, 2006). Providing relevant statistical information about drinking behavior may adjust drinkers’ extant knowledge. This process may make some drinkers feel ambivalent about their current drinking. Moreover, Jung, Shim and Mantaro (2010) proposed that exposure to descriptive social norms with sufficient statistical numbers may instigate drinker reactance.

Therefore, the current research suggests that any statistical information about drinking (does not have to be about the drinking norm) may create reactance in heavy drinkers, especially when the message is written with a loss frame. For example, if a person was told that drinking would decrease health, this would be less likely to provoke defensiveness than if told that heavy drinking was likely to decrease life-span by 12 years.

Therefore, the first hypothesis in Study 2 (the third hypothesis for the entire study) is: MESSAGE FRAMING Framing 49

H3: The addition of statistical information plus the basic texts message will create

more reactance, which will be manifested by lower message rating and higher

drinking intention of the health message when the message is presented with loss

frame, compared to gain frame.

Moderators—Self-esteem and Self-esteem instability. Some of the predictions made in this dissertation are related to self. High self-esteem, low stability (instability) of self-esteem, as well as loss framed messages are thought to provoke more defensiveness.

The sections below illustrate the importance of self-esteem and instability of self-esteem.

Self-esteem. The effect of self-esteem has been demonstrated in the domain of health behavior. However, whether self-esteem is related to healthy or risky lifestyle has been a very complicated issue. Baumeister, Campbell, Kruger, and Vohs (2003) reviewed the extant studies and asserted that adolescents with high self-esteem tend to take risks in experimenting with alcohol and other drugs. After engaging in these risk behaviors, these adolescents tend to downplay their own vulnerability from these risky behaviors. There have also been conclusions in the opposite direction. For example, Stein, Dixon, and Nyamathi

(2008) found that for homeless adults, self-esteem affects drug use and alcohol use through emotional distress and negative coping (i.e. “denial and behavioral disengagement,” p.412).

The lower the self-esteem, the more likely problems related to substance abuse occurs for homeless adults (Stein et al., 2008). Some moderators may be intermediate variables between self-esteem and health behaviors. For example, gender is a potential moderating variable, according to Neumann, Leffingwell, Wagner, Mignogna, and Mignogna, (2009).

From their research, both men and women with low self-esteem showed higher intention to reduce alcohol intake after exposure to health messages about the risk of drinking. MESSAGE FRAMING Framing 50

Especially, females with lower self-esteem rated the health message about drinking risk as high in scientific accuracy. This is consistent with the assumption that high self-esteem will be associated with more defensiveness in response to loss-framed messages (Neumann et al.,

2009). There also have been findings which devalue the role of self-esteem. For example,

Larsen, Overbeck, Vermulst, Granic, and Engels (2010) failed to find a significant moderating self-esteem effect between adolescents’ friends drinking and their own drinking.

Similarly, using adolescents as the participants, Mullan and NicGabhainn (2002) failed to discover any relation between self-esteem and risky behaviors (i.e. smoking, drinking, drunkenness and drug use).

This research dissertation predicted that high self-esteem will be associated with more defensiveness regardless of gender, and this is supported by Neumann et al. (2009).

Additional evidence is from Di Paula, and Campbell, (2002). Lower self-esteem was linked to lower tolerance to frustrating messages. For example, people with lower self-esteem tended to ruminate more when acknowledging an embarrassing low-performance feedback

(Di Paula, & Campbell, 2002). Therefore, this current research hypothesizes that low self- esteem leads to higher psychological defensiveness to the loss-framed message because loss- framed message is one kind of threatening message to one’s self-status. However, self- esteem is more complicated that just being high or low.

Instability of Self-esteem. If self-esteem increases or decreases sharply and quickly, it means that self-esteem is unstable. From the study by Kuppens and Van

Mecheden (2007), it is the instability of self-esteem that contributes to the fluctuation of self-esteem and perception of self, not self-esteem per se. According to Kuppens and Van

Mecheden (2007), with lower stability of self-esteem, one is more likely to sense any threat MESSAGE FRAMING Framing 51 to the self. The two way interaction between level of self-esteem and instability of self- esteem contributes to perceived threat in information processing. Specifically, it is unstable and high self-esteem that relates to higher sensed threat to messages about performance failure (Seery, Weisbuch, & Vick, 2004). Therefore, the current research regards unstable plus high self-esteem as one critical characteristic that may contribute to greater reactance

(i.e. psychological defensiveness) to loss-framed health messages, which are threatening in their tone. This is Hypothesis 2 (the fourth hypothesis to the entire research) under Study 2:

H4: High self-esteem, matched with unstable self-esteem (i.e. low stability of self-

esteem), will make the corresponding participants display lower message rating (or

higher drinking intention) after reading the loss-framed message. Again, there will

be health messages with statistical information and texts or only texts.

Another moderator is a construct of self, named as self-relevance. The following discusses why self- relevance was considered as one of the potential moderators.

Self-relevant Issue and Intellectual Reasoning. When people make judgments about self-relevant issues they often behave irrationally, according to Kross and Grossman

(2012). To define and measure responses to self-relevant issues, Kross and Grossman (2012) recorded essay responses to economic climate and ideals for current society regarding cooperation and openness to diverse views as the outcomes and used content analysis to measure the results. Those participants who were instructed to think in a distant perspective before answering significantly displayed more intellectual reasons than those who were instructed to think in an immersed perspective (Kross, & Grossman, 2012). Inferring from MESSAGE FRAMING Framing 52

Kross and Grossman (2012), people rely less heavily on intellectual reasons when they consider issues which are more relevant to their self-concepts.

Self-Relevance and Health Decision. Self-relevance is expected to moderate the effect of message frame on defensiveness. This dissertation predicts that loss-framed messages will provoke more defensiveness when the behavior described in the message has greater relevance to the self. Self-relevance can be defined as how a behavior or a habit is tied to one’s self concept (Chaiken, & Liberman, 1992). To operationally define self- relevance, if a person drinks coffee very often, and answers that coffee is very important to him or her, this means higher self-relevance. This dissertation suggests self-relevance as one potentially important moderator between psychological acceptance of messages and message framing (gain/loss). Research has found that self-relevance affects consumers’ perceptions about healthy products. For example, when breakfast cereals are tagged with claims about Type-II Diabetes, those consumers who perceived themselves to be at risk for diabetes tended to perceive the cereal as healthier, more beneficial, and they reported greater intention to buy these products with information claims about diabetes (Dean et al., 2012).

The messages in Dean’s et al. (2012) research were defined as benefit claim, risk-reduction claim, and no claim. The benefit claim was, “Promotes regulation of blood sugar balance”

(p.131), and the risk-reduction claim was, “Contains cereal-based compounds which balance the blood glucose levels and therefore lower the risk of diabetes” (p.131). These two messages were framed in a relatively positive manner. Dean et al. (2012) did not investigate how negative language or a loss-framed message influences perception and attitude to messages. In this dissertation self-relevance has been hypothesized to facilitate focused MESSAGE FRAMING Framing 53 attention to health messages, but the audience may not necessarily accept the health recommendations.

The current project hypothesizes that self-relevance to a health issue is related to more concentrated attention to the health messages; but when being presented with a loss- framed message, the audience is more likely to display psychological defensiveness or reactance due to the possible phenomenon as predicted by Reactance Theory. According to

Liberman and Chaiken (1992), self-relevance may increase reactance when participants are reading a loss-framed message. Using the relation between coffee and fibrocystic disease as the topic, Liberman and Chaiken (1992) found that female participants showed more acknowledgment to the link between coffee consumption and fibrocystic disease after reading a high threat-message compared to a low threat message. However, consistent with the aim to incorporate the theory of reactance, this dissertation assumes that the message with loss frame would excite more psychological reactance than the message with gain frame, and this will be especially true when accompanied by statistical information.

Moreover, Liberman and Chaiken (1992) found that higher self-relevance facilitated more psychological reactance: that is, participants with higher self-relevance showed relatively less intention to follow the instructions of the high threatening messages.

The current research offers a novel contribution: testing both gain and loss framed messages as one experimental dimension and examining the validity of Reactance Theory.

The third hypothesis in Study 2 is grounded both on the assumed validity of Reactance

Theory and possible role of self-relevance. The inclusion of statistical information will be tested in the equation without a specific assumption regarding its effect. The following is

H5: MESSAGE FRAMING Framing 54

H5: High self-relevance will lead to more resistance to the health message (or higher

intention to drink) when the health message is presented from the loss frame.

In sum, study 2 serves to investigate the Reactance Theory and message framing on health promotion, and test the relevant moderators. The following is the method section.

Study 2. Method

Design. Study 2 is an experimental study with two dimensions of message presentation manipulated: Framing (gain or loss) and Statistical Information (included or no). Study 1 only included two types of messages (gain or loss-framed), while Study 2 had the 2 X 2 (presence versus absence of statistical information and gain versus loss frame), so each participant only saw one of the four types of messages. Appendix A has the actual content of these four types of messages. The first dimension of message contents to vary was gain vs. loss frame. Consistent with the definition explained in the introduction section and the first study, the gain-framed messages concentrated on health benefit from drinking abstinence or moderate drinking as it displayed the information. The loss-framed message talked about the harm of heavy drinking with a negative writing style. The second dimension

(statistical information vs. control) to be manipulated was the statistical content of messages

(as shown in Appendix A). The experimental group read the messages suggesting the benefit or harm regarding alcohol consumption with supporting statistical numbers (e.g. 50% of severe trauma). The control group read the messages with descriptions targeting identical issues without any numbers to corroborate. The following paragraph has more detailed description of presented messages on drinking and health. MESSAGE FRAMING Framing 55

Participants. The undergraduate students at the University of Houston who were taking at least one psychology course and identified themselves as regular drinkers composed the participating sample. Participants were again recruited from the psychology subject pool using the SONA system. Course extra credits served as the incentive for experimental participation. The total number of participation was 206 subjects, as the estimated power fits the plan (again, shown in Figure 1).

Participants included 206 (157 females and 49 males) students who completed the internet survey. Some participants (n=42) had also attended Study 1 and returned. The potential limitation is addressed in the Conclusion. The first condition delivered the text- only message presentation plus the grain-framed message group (n = 50). The second condition presented the text plus statistics message with gain message framing (n = 56). The third condition was text-only message group with the loss message framing (n = 57).

Finally, the fourth group was delivered with the text plus statistics message which was by loss frame (n = 44).

Messages. For each experimental condition, a unique message was presented to the participants. The message frame (gain vs. loss) and inclusion of statistical information (yes or no) were the two dimensions which differentiated these four experimental conditions with unique message contents. Therefore, four messages were created: (1) loss X text only, (2) loss X text + statistics, (3) gain x text only, and (4) gain X text + statistics. Each message with statistical information addressed possible alcohol benefit or harm with specific numbers, while the text only messages introduced the harm or benefit with plain sentences without numbers. The two loss-framed messages included information about injury as a result of drinking. The gain-framed messages did not include information about injury, but MESSAGE FRAMING Framing 56 this is considered acceptable protocol because loss-framed messages often do not directly compare to gain-framed messages. Beside this, all the messages had corresponding content.

Therefore, a sufficient experimental control was expected because of the writing and arrangement of the description of these four messages. See the relevant Appendix provided.

Experimental Procedures: Recruiting and Survey. The participants were recruited via SONA system. Only the undergraduate students who were then taking at least one psychology course and identified themselves as regular drinks were encouraged to sign up.

The participants enrolled in this study through the SONA system. When they actually logged into the computer link in the DataStat system, the experimental procedures began.

All measures were administered by web survey.

First, the participants provided informed consent by clicking the agree button. They then responded to the demographic questionnaire. The informed consent was written according to guidelines of the University of Houston Committee for the Protection of

Human Subjects (See Appendix B). Then, a battery of research questionnaires, including drinking, self-esteem, instability of self-esteem, and self-relevance (i.e. drinker identity) were presented for the participants to complete. The third step was the presentation of health messages and the evaluation of the corresponding health message depending on which group they were in. The assignment of participants to the experimental group depended on which month they were born, as they disclosed this once after they electronically signed the informed consent.

After they finished reading the health message, the participants automatically proceeded to the program and answered the Message Realism/Believability Scale (text MESSAGE FRAMING Framing 57 shown in Appendix C) which evaluated participants’ assessment of the quality of the message, and served as an indicator of level of defensiveness. To relieve the participants’ burden of memory, when the participants evaluated the qualities of the health message, the computer program allowed them to return to the presented message and review as carefully as they wished. The second outcome variable in addition to message evaluation was future drinking intention (texts shown in Appendix C). Future drinking intention was assumed to approximate how the participants perceived the health message, accept or reject. More details are provided in the following scale descriptions.

Measures. Daily Drinking Questionnaire (Collins, Parks, & Marlatt, 1985),

Quantity-Frequency-Peak Alcohol Use Index (Baer, 1993; Marlatt, Baer, & Larimer, 1995) and Rutgers Alcohol Problem Index (White & Labouvie, 1989) represent the three measures to depict the participants’ current drinking. These measures are the same as those described for Study 1 for the measure of drinking. Again, drinking intention was measured by the scale modified from the work by Dimeff, Baer, Kirlahan, and Marlatt (1999) Measures are provided in the Appendix C; measures about self and message rating were also included.

The Self-esteem Scale (Rosenberg, 1965). The self-esteem scale by Rosenberg

(1965) is one of the historical scales measuring self-esteem. This scale has 10 items, and each item has five response options (from Strongly Agree to Strongly Disagree). The questions indirectly test the participants’ self-esteem by phrasing the questions in an abstract and philosophical writing style. Scores are calculated by taking the average of the ten items.

Instability of Self-esteem (Charbol, Rorsseau, & Callahan, 2006). Charbol,

Rorsseau, and Callahan (2006) created a four item scale testing the instability of self-esteem. MESSAGE FRAMING Framing 58

According to the test by these researchers, this four item scale had good reliability (i.e. internal consistency and retest reliability) and validity. Of these four items in this short scale, “sometimes” and “at other times” are consistently applied to implicitly ask how stable one’s self-esteem is.

The Message Believability Scale (Flanagin & Metzeger, 2000). The communication study for how people evaluate message quality on the web-pages by

Flanagin and Metzeger (2000) adopted this seven point Likert scale. This scale lists five major dimensions and participants utilize these dimensions to evaluate the quality (i.e. credibility) of the presented message. These five dimensions describe whether this message is believable, accurate, trustworthy, biased, and complete. The acceptance of the health message would be represented by a high score from this scale, and a high score means the participants are more persuaded by the message. In contrast, a low score means the participant shows more resistance to the health message. In terms of reliability, the current data indicated that the Cronbach’s α was .73.

Self-relevance. A modified version of the Drinking Identity Scale (Lindgren,

Neighbors, Teachman, Wiers, Westgate, & Greenwald, 2012) was adopted to assess self- relevance. This five-item scale tests how fundamental drinking is to one’s self-image or self- cognition. One such item states, “Others view drinking as part of my personality.” The response alternatives consist of a 7-point continuum, from 1 (strongly disagree) to 7

(strongly agree). The score was the mean of the five items. The content of the Drinking

Identity Scale (Lindgren et al., 2012) is in Appendix C.

MESSAGE FRAMING Framing 59

Study 2. Results

Scores by Experimental Groups. Each participant was assigned one message of totally four types of health messages, based on the genres of message framing and included statistical information. The dummy coding processed the data and quantified the experimental assignment of each participant. As said, the experimental design has been 2 X

2; Stats X Framing. The grouping was based on stats and framing. For each participant who saw the messages including both statistics and texts, the numerical value for the dimension of stats was represented by 1; for each participant who was in the control group without statistical information, the numerical value was represented by 0 on the dimension of stats.

On the dimension of message framing, the gain-framed message group was coded as 1 as the label on the axis of framing, and the loss-framed message group was coded 0.

Outcomes. Consistent with the previous study, message rating and drinking intention served as the two outcomes. The message rating revealed how the health campaign worked; if the participants accepted the health advice. The average scores of the entire items on the Message and Believability Scale by Flanagin and Metzeger (2000) was the dependent variable for the first hypothesis test. Drinking intention as well as message rating were the outcomes in the equations of Hypothesis 2. Drinking intention was represented by weekly drinking intention from the first item of the Drinking Intention Scale by Dimeff, Baer,

Kirlahan, and Marlatt (1999) (i.e. the sum of all the blanks for Monday through Sunday drinking amounts); and also the other, typical drinking intention which is the total intended drinks reported with the fourth item (Dimeff, Baer, Kirlahan, & Marlatt, 1999). MESSAGE FRAMING Framing 60

Drinking as the Statistical Covariates. Drinking served as a covariate and potential moderator, given the assumption that effects might not be present for light drinkers. This first study separately constructed drinking by the following three variables. First variable was, (1) Drinks per week: the sum of the first item from the Daily Drinking Questionnaire by Collins, Parks and Marlatt (1985). That item was from Monday to Sunday, “Consider a typical week during the last three months. How much alcohol, on average (measured in number of drinks), do you drink on each day of a typical drink? ”. The second variable was

Peak Drinks: the first item in the Quantity-Frequency-Peak Alcohol Use Index (Baer, 1993);

“Think of the occasion you drank most this past month. How much did you drink?” The third variable was, Typical Drinks: the third item in the Quantity-Frequency-Peak Alcohol

Use Index (Baer, 1993),” On a typical occasion during the last month, how much alcohol did you typically drink? Estimate for the past month“.

Analyses. The primary outcomes in the statistical regressions were message ratings, which indicated how the participants accepted the advice from the health messages, and drinking intentions. As introduced above, drinking was separately represented by drinks per week, peak drinks and typical drinks. Each was included as a predictor in separate regression analyses. For all regressions, the drinking measure was included first as a single predictor (i.e. covariate to be controlled); then a predictor along with other interaction terms merely consisting of these independent variables besides drinking; finally served as a moderator and thus generated some interaction terms, two-way and three-way, in the equation. For all the regressions in Study 2, the predictors of drinking were centered. The outcomes were message rating, weekly drinking intention and typical drinking intention. MESSAGE FRAMING Framing 61

Descriptive Statistics. Means and correlations are provided in Table 14. Participants reported drinking on average 6.42 drinks per week. Following experimental procedures, participants reported intending to drink on average 3.83 drinks per week over the next month. Participants reported and average of 4.65 drinks at once on their peak drinking occasion in the previous month. The numbers are different from Study 1. Correlational analyses revealed significant positive correlations between drinks per week and each measure of drinking intentions. The measure of weekly drinking intention correlated with typical drinking intention significantly and positively (r =.70, p <.0001). Also drinking items correlated with all the intention items. Self-esteem was moderately and negatively correlated with instability of self-esteem. (r =-.50, p<.0001). This may have compromised predictive validity (Renger, 1993) for the H4 regressions to some degree.

Regressions for H3, H4 and H5. For each hypothesis, three analyses were conducted, one for each measure of drinking (drinks per week, peak drinks, and typical drinks). Main effects were entered at Step 1 (Model A). Model B added the two-way product between statistical information and message framing. Model C included main effects, all two-way product terms, and the three-way interaction term. As in study 1, there were uniform main effects of drinking on drinking intentions.

The first part of H3 aimed at understanding how message framing, on dimensions of gain vs. loss (named as message frame) and inclusion of statistical information or not, affect the consequential message rating and future drinking intention, with past drinking being controlled. Controlling for drinks per week , message framing and statistical information did not alone or jointly predict the outcome of message rating (Table 15), nor did they when the control was peak drinks (Table 16) or typical drinks (Table 17). MESSAGE FRAMING Framing 62

Next, weekly drinking intention was examined as a function of message frame, statistical information, and drinking. First, Table 18 displays the results where the measure of drinking was drinks per week. Results revealed a significant three-way interaction, Frame

X Stats X Drinks per week, which was generally supportive of expectations (b=-.49, β=-.29, t=-3.14, p<.001 ). For heavy drinkers, the predicted pattern emerged such that statistical information under loss frame was associated with higher drinking intentions. In contrast, for heavier drinkers in the gain frame, statistical information appeared to reduce drinking intentions (See Figure 6). The same pattern (b=-.72, β=-.24, t=-2.13, p<.05) emerged when peak drinks was the drinking variable (See Table 19 and Figure 7).

The remaining part of H3 ran the analyses by replacing the moderator with typical drinks and the measured outcome with typical drinking intention. A pair of two way- interaction terms manifested their significant effects (See Table 20 and Figure 8). Gain- framed message plus statistical information reduced drinking intention (b=-1.29, β=-.27, t=-

2.37, p<.05); for high typical drinkers, statistical information significantly reduced drinking intention (b=-.20, β=-.33, t=-2.57, p<.05). However, the corresponding three way interaction

(Frame X Stats X Typical Drinks) was not significant for predicting typical drinking intentions (See Table 20). When typical drinking intention served as the outcome, it was gain-framed messages plus statistical information that reduced drinking intention, so this has been consistent with the hypothesis that loss frame increases psychological reactance and thus drinking intention is higher.

The third hypothesis suggested that psychological reactance would be greater when messages were presented under loss-frame and when accompanied by statistical information; also drinking intentions would be higher with the above conditions implicating MESSAGE FRAMING Framing 63 reactance. The statistical output supported the hypothesis that loss frame along with statistical information would boost drinking intention, especially for heavier drinkers. But that had no direct relation with message rating. Perhaps psychological reactance indirectly appears via drinking intention, and message rating was not the indicator which directly fluctuated with psychological reactance.

The fourth hypothesis regarded self-esteem and instability of self-esteem, each would have predicted the outcomes of lower message rating and higher drinking intention; and especially, when messages are under loss frame, the message rating would be low and drinking intention would be high. Although some participants read the messages with statistical information, while some did not, statistical information served as a statistical covariate in the analytic process. Each regression took three different forms of statistical examination: (1) regression with all the main effects with statistical information and drinking controlled, (2) all main effects, with all possible two-way interactions by message framing, self-esteem and instability of self-esteem, with statistical information and drinking controlled, (3) all possible main effects, and all possible interaction terms (up to three-way) by message framing, self-esteem, instability of self-esteem with statistical information and drinking controlled.

Message rating was tested as the outcome. Three models were tested, each corresponding to one of the three different drinking measures: drinks per week, peak drinks, and typical drinks. As shown in Table 21, Table 22 and Table 23 with drinks per week serving the drinking measure, self-esteem had a positive main effect (b=.26, β = .19, t= 2.39, p<.05, Table 21) with the outcome of message rating, as did instability of self-esteem

(b=.21, β =.17, t= 2.15, p<.05, Table 21). This means that the higher self-esteem, the higher MESSAGE FRAMING Framing 64 message rating was given. The higher instability of self-esteem, the higher message rating was given. This was opposite of the expectation.

Next, the drinking covariate was replaced with peak drinks. Again, self-esteem and instability of self-esteem each positively correlated with the outcome of message rating. For the magnitudes of these main effects, please refer to Table 24, 25 and 26. A significant two- way interaction, Self-esteem X Instability of self-esteem emerged (b=-.23, β =-.14, t= -2.02, p<.05, Table 25). Figure 9 depicts the specific trends of this interaction. Message rating was significantly reduced when low self-esteem matched with low instability (i.e. stable) of self- esteem (Figure 9). The three-way interaction by message framing, self-esteem and instability of self-esteem was not significant (Table 26).

Next, typical drinks represented the drinking variable (see Table 27 to Table 29).

Compared with drinks per week and peak drinks, typical drinks did not negatively correlate with message rating. Once again, self-esteem and instability of self-esteem positively and individually had a positive relation with message rating (Table 27 28, & 29). The two-way interaction between self-esteem and instability of self-esteem on Table 28 was marginally significant.

Overall, the three sets derived some consistent information. Results revealed that self-esteem and instability of self-esteem were positive predictors of message rating. The higher the self-esteem, the higher the message rating was. Also, the more unstable of self- esteem, the higher the message rating was. This is aside from the fourth hypothesis. No major relation with message framing was found. Psychological reactance did not associate MESSAGE FRAMING Framing 65 with the loss-framed message, but it could increase when self-esteem is low and self-esteem is very stable.

The other important dependent variable of H4 was drinking intention. Weekly drinking intention was representative of this construct, but typical drinking intention was tested once as well. Drinks per week was the first drinking covariate to be controlled for.

The order of statistical examination was testing all the possible main effects, regression with all the two-way interactions by message framing, self-esteem and instability of self-esteem, and then regression with all the possible interactions by message framing, self-esteem and instability of self-esteem, up to three ways. There were not any significant main effects to predict drinking intention, controlling for statistical information and drinks per week (Table

30, 31 & 32). The two-way interaction between message framing and self-esteem was significant (Table 31 & Table 32). The graphs on Figure 10 and 11 visualized the slopes.

Drinkers with higher self-esteem tended to show higher drinking intentions when reading the gain-framed messages. Gain-framed messages did not act according to the original hypothesis reducing psychological defense.

Next, the regressions with peak drinks derived another similar set of output, with no significant variables predicting drinking intention (Table 33, Table 34 & Table 35). With typical drinks serving as the drinking measure, still no significant predictor correlated well with the outcome of drinking intention (Figure 36, 37 & 38).

Finally for H4, typical drinking intention as the outcome was regressed and typical drinks were operationalized as the drinking variable. Self-esteem was marginally associated with typical drinking in intention (Table 39). The interaction between message framing and MESSAGE FRAMING Framing 66 self-esteem was also marginally significant (Table 40 & 41). Thus, some evidence indicated that drinkers with high self-esteem showed higher drinking intention when reading the gain- framed message in spite of marginal significance in statistics. It is contradictory that gain frame indirectly aroused drinking intention. The relation between self-esteem and drinking intention was consistent with the hypothesis. However, message framing in a positive tone may instigate self-assured drinkers to drink more by reducing their psychological reactance.

This differs from the indication of the original hypothesis as well.

Broadly observing, low self-esteem or stable self-esteem makes participants resist the presented health advice more. This has been significant, but unexpected. In terms of drinking intention, high self-esteem could indirectly induce reactance which is higher drinking intention, and this result has been consistent with the hypothesis. However, it has been the messages under gain frame that facilitated drinkers with high self-esteem to show reactance which manifested by higher drinking intention.

The fifth hypothesis predicted high self-relevance to drinking and loss frame would enhance psychological reactance against health messages. Predictors included statistical information, message framing and self-relevance, and drinking at step 1. Then, all possible interaction terms involving statistical information, message framing and self-relevance were entered. Each regression predicted the outcome (i.e. message rating, drinking intention and typical drinking intention) according to message framing, statistical information self- relevance, and drinks. Drinking variables were drinks per week, peak drinks and typical drinks. Message rating was the first dependent variable addressed. The first set of regressions used drinks per week to represent drinking. As illustrated by Table 42, there was not any variable, except for drinks per week, which predicted message rating as the MESSAGE FRAMING Framing 67 outcome. Next, the regressions repeated with drinking replaced with peak drinks (Table 43).

There was no significant pattern to predict message rating from these variables. As typical drinks were inserted, no new principle showed up statistically (Table 44).

Then, drinking intention was the dependent variable examined, with the three different kinds of drinking measures examined in turn. One three-way interaction, Stats X

Frame X Self-relevance, was evident on Table 45, when all the interaction terms entered, including drinks per week as the only covariate (b=-2.22, β = -.25, t= -2.33, p<.05). The simple slope graph suggests that drinking intention was higher when the highly self-relevant drinkers read messages with loss frame and statistical information (Figure 12). The following regression tested drinking intention as the outcome again, with drinking measured by peak drinks. The significant three-way interaction Stats X Frame X Self-relevance (b=-

3.25, β = -.37, t= -2.87, p<.05, Table 46) had a similar pattern suggesting that high self- relevant drinkers showed high drinking intentions when reading the loss-framed messages with statistical information (Figure 13 ).

Next, typical drinks was included as the drinking variable in the set of regressions

(Table 47). Regressing drinking intention as the outcome by the variables and interaction terms taking typical drinks as the only covariate, a three-way significant interaction, Stats X

Frame X Self-relevance, was found (b=-.3.83, β = -.44, t= -2.98, p<.05). The graph (Figure

14) suggested that loss frame plus statistical information led to high drinking intention, especially when the drinkers reported high self-relevance regarding drinking. Finally, typical drinking intention was examined as another kind of drinking intention and specified as the outcome, with only typical drinks as the drinking measure (Table 48). The major variables, statistical information, message framing, and self-relevance did not predict typical drinking MESSAGE FRAMING Framing 68 intention. Overall, high self-relevance indirectly increased drinking intention by functioning simultaneously with loss-framed message providing statistical information. This successfully corroborated the fifth hypothesis, with the additional requirement, existence of statistical information.

The fifth hypothesis received mixed support. Attitudes toward health advice have been found having no relation with the proposed predictors. Drinking intention as the outcome provided results that were generally consistent with Reactance Theory.

Accordingly, loss message framing generated psychological reactance, especially when messages were strengthened by presenting statistical information, and this indirectly appeared through higher reported drinking intention.

The results of H5 resembled that of H4. Message rating was not the outcome which was predicted by some significant predictors. Again, the H5 part taking drinking intention as the outcome revealed some cases in which a loss-framed message, when paired with statistical information raised reactance from high self- relevant drinkers by indirectly expressing this on higher drinking intentions.

Study 2. Discussion

Study 2 represents the Reactance Theory study on drinking as a health behavior. The main issues were to discover when psychological reactance would be instigated. It was hypothesized that loss frame as a major form of message framing, when combined with statistical health information, would activate higher reactance and thus lower message ratings and higher drinking intentions could have been reported. Also, other variables regarding psychological self were studied. Self-esteem, instability of self-esteem and self- MESSAGE FRAMING Framing 69 relevance were expected to positively boost psychological reactance as message ratings would be lower and drinking intention would be higher. Only some of these hypotheses were supported. Some findings removed from these predictions have been found with different revelations.

When message rating was the outcome, Reactance Theory did not work. As the indirect outcome, drinking intention has been relatively reflective of psychological reactance. Controlling for drinking, when loss frame is combined with statistical information, drinking intentions tend to be higher. Since health messages in this experiment were presented in a professional academic context, perhaps an academic atmosphere was conveyed, especially with the offer to students of course credits. Perhaps the students who joined as participants were psychologically open to presented health advice and treated them as potentially useful messages by identifying the messages as professional and scientific, as these students took a serious attitude toward their academic progress. Besides, reactance to the offered health advice was indirectly expressed through higher future drinking intentions, instead of lower message rating per see. The smoking research by Erceg-Hurn and Steed

(2011) proposed that texts without graphs are never impressive enough to instigate reactance. Perhaps this also explains the findings here. It is also worth noting that the internet survey included a packet of questionnaires which participants could finish at their own pace. This free-style message reading and processing may have blocked psychological reactance by allowing more freedom to respond.

High self-esteem and high instability of self-esteem increased message rating. This was opposite of the expected direction. Indeed the research by (Seery, Weisbuch, & Vick,

2004) suggests that unstable high self-esteem leads to sensed psychological threat when MESSAGE FRAMING Framing 70 receiving negative performance feedback. However, the sensed threat represented in that research was related to cardiovascular index. This context (Seery, Weisbuch, & Vick, 2004) may imply that the so-called sensed threat may have occurred at the subconscious level.

Given that this is true, the subconscious irritation may not always lead to psychological reactance, which is more conscious. Also, the context of Study 2 did not include any performance test or feedback. Therefore, high and unstable self-esteem may logically not arouse reactance but instead facilitate message acceptance.

In Study 2, high self-esteem and unstable self-esteem, each individually led the participants at the University of Houston to accept the offered health advices. Perhaps for the participants, most of whom were undergraduate students, high self-esteem implies a state of willingness to listen to new health advice and information. High self-esteem means stronger resilience to self-hurting messages telling drinkers to reduce alcohol intake.

Unstable self-esteem for young adults may be indicative of an open mind to change their daily habits. In terms of drinking intention, high self-esteem plus gain frame leads to higher drinking intention. Gain frame may induce a certain release of psychological defense and high self-esteem related to the will to express oneself. That may be the reason why gain frame plus high self-esteem leads to higher drinking intention. In sum, regarding the fourth hypothesis, Reactance Theory was not supported. However, the results indicate the idea that high self-esteem is accompanied by a willingness to be challenged by health advice.

Similarly, instability of self-esteem often features readiness to change one’s regular behaviors. Unexpectedly, self-esteem and instability of self-esteem are moderately and negatively correlated. The higher self-esteem, the more stable self-esteem is. This may have lowered the predictive validity (Renger, 1993). With this moderate correlation, a common MESSAGE FRAMING Framing 71 variable may exist behind both self-esteem and instability of self-esteem. More research needs to address this question.

Regarding hypothesis five, the research supports that reactance tends to rise when highly self-relevant drinkers read the message under loss frame, providing statistical information, but it emerges on the other measure, drinking intention rather than message rating. Self-relevance was measured by drinking identity. Drinking identity has been found to be a solid self-concept, which enhances psychological defense (i.e. psychological defensiveness) against health information. Especially, when a loss-framed message is paired with statistical information, the reactance becomes observable. Reactance may not directly relate to medical compliance (Seibel, & Dowd, 1999). The current finding is consistent with the perspective that reactance may not be diagnosable by some approximate measure.

Reactance may indirectly emerge through other signs or behaviors. This may be meaningful in understanding how mind indirectly expresses its uncivil desire or impulse. People do not express their psychological reactance by confronting scientific information. Instead, certain behavioral intentions (i.e. drinking intention) are increased.

One major anomaly to hypotheses deserves further consideration. Two regressions on Table 40 and Table 41 indicated marginally significant findings that high self-esteem drinkers showed higher typical drinking intention when reading the gain-framed messages.

This may reflect subconsciously higher psychological reactance to the presented health message. However, this contradictory pattern may be caused by practice effect. As noted, 42 participants took Study 1 and then finished Study 2. According to Ozubko and Fugelsang

(2011), repeated message communication makes the messages be perceived as more valid and familiar. Besides feeling familiar, the 42 participants might not feel so much uncertainty MESSAGE FRAMING Framing 72 toward the messages. Then, these participants may have greater capacity (i.e. tolerance), to ignore how threatening the loss-framed messages were, and thus accepted the advice by the loss-framed message more. Context might be important as well. The research was on students who took psychology classes. It is possible that university students tend to analyze the message content calmly without being bothered by threatening tone of loss frame.

Consequently, in some situations, the participants disliked the gain-framed messages more.

Conclusion

Message framing represents one category of how to communicate. Research has found many kinds of message framing and produced theories regarding how message framing works on persuasion. Some theories are empirically tested on health behavior, such as smoking reduction or health examination, as well as drinking reduction. Many studies separate messages by gain vs. loss, and expound the gain framing of message, how it works, compared with the loss framing of message. Most published studies theorize that a third variable or moderator is required to intermediate the relation between message framing and successful audience persuasion. The third intermediate variable could represent traits, individual difference or a situation. Determining which variable should match with gain frame or loss frame is very complicated, each research study has unique conclusions.

The current dissertation tested how message framing by gain or loss functions on simulated drinking reduction campaigns with other proposed variables and moderators. The research included two studies. The first study enquired how ego-depletion affected the evaluation of presented health advice and future drinking intention. The statistical analysis gave some affirmation to posit that ego-depletion facilitates acceptance of health messages if MESSAGE FRAMING Framing 73 drinkers regularly drink much during the peak days. Ego-depletion was hypothesized as leading to solid attitude and defense against advices, specifically to reduce future alcohol intake. Thus, results were opposite of what was anticipated.

The dissertation did not support the statement that ego-depletion always strengthens rigidity of regular attitude or choice. Certain choices or behaviors, such as drinking may be more flexible for behavioral change upon receiving relevant messages. Drinking intention is the indirect reflection of psychological reactance or acceptance of health campaign because drinking intention was collected after message reading. Ego-depletion reduced the positive connection between drinking and drinking intention. Heavy drinking may be associated with some degree of stigma. Thus, when the mind is depleted, the stigmatized behavior, drinking, becomes burdensome, then the capacity to defend one’s existing perspective decreases, the drinking intention tends to decrease even for regularly high drinkers.

To progress, the second study focused on message framing and Reactance Theory.

The experimental survey added statistical information and plain-text messages attempting to amplify social influence. In addition, self-esteem, instability of self-esteem and self- relevance with drinking were hypothesized to arouse reactance as shown by message rating and indirectly by drinking intention. Instead of message rating, the direct measure of psychological reactance to health messages, the hypothesis about Reactance Theory was supported by the statistical output which took drinking intention as the indirect indication of psychological reactance. Statistical information successfully boosted drinking intention by assisting the loss-framed message to prime psychological awareness of the potential health cost of alcohol drinking. Wilson Linsey, and Schooler (2000) suggest that humans have two attitudes, implicit attitude which is learned or developed previously and strongly, and MESSAGE FRAMING Framing 74 explicit attitude which is learned recently and is fragile. The present finding is harmonious with the model suggesting implicit attitude is deeply rooted and difficult to change.

Therefore, when participants in Study 2 evaluated the health messages, with their explicit attitude confronted, a relative openness existed for attitude change. Also, the participants are students, so they may have put more attention and contemplation to new scientific and professional messages. Since implicit attitude is hard to change and it functions indirectly, drinking intention as an implicit measure displayed the unexpressed reactance. The participants who read messages under loss-framed plus statistical information reported higher, reactance to unfriendly or threatening messages. This reactance was better represented by drinking intention rather than message rating.

The dissertation establishes a direct theory that self-esteem or instability of self- esteem alone enhances message acceptance of drinking reduction. When message framing is considered, the theory appears contradicting and paradoxical. High self-esteem is found to boost ratings of drinking reduction message. This may instead suggest that high self-esteem weakens potential psychological reactance of challenging health information. Self-esteem from this research is known to signify a healthy and open mind, not a protective decision making tendency. Instability of self-esteem signifies a trait of changing according to useful strategies, not an insecure mind avoiding change. Interacting with message framing, the finding becomes paradoxical. Gain frame message adds up with high self-esteem to increase drinking intention. This may imply high self-esteem leads to self-assurance if a health message uses a very pleasant tone to present. How self-esteem exactly functions with intention to drink and evaluation on health advices will be worth pursuing in future research projects. MESSAGE FRAMING Framing 75

Similarly, higher drinking intention suggests that reactance rises when self-relevant drinkers read the messages under loss frame plus statistical information. Self-relevance of drinking means the importance of drinking to self-identification. Consistent with the theory that self-relevance impedes the process of health advice and this reactance amplifies under unappealing presentation, the indirect measure, drinking intention captures this phenomenon. The scale about drinking identity represented self-relevance of drinking in this research. Drinking identity could be rooted firmly in the unconscious part of memory and resist behavioral reduction of alcohol, which may explain why only drinking intention as the indirect index of psychological acceptance of health information reflected the difference.

Perhaps for the participants who are students, drinking is one part of their developing identity, so psychological reactance did not appear from the direct measure, message rating.

Research Limitations. Some procedures and practices of this research have their limitations. First, the presented messages were planned to communicate differently according to theoretical designation on framing and statistical information. However, message rating as the outcome did not reflect many statistical patterns according to research hypotheses. The possible explanation was that messages were short and plain. Since the two kinds of messages in Study 1 and the four kinds of messages in Study 2 were short, the participants may not have been able to feel differently toward these experimental stimuli.

Moreover, the participants as students may have had some significant knowledge regarding the health consequences of drinking before they took this experimental survey. Therefore, the health messages in this research did not differentiate these participants according to presumed group-difference, such as simply gain frame vs. loss frame. For follow-up MESSAGE FRAMING Framing 76 research, elongating the message contents or adding some graphs may make persuasiveness of health message different to how persuasion worked in this research.

Future research might consider gender in the context of ego-depletion and reactance in relation to message framing. An obviously larger portion of the subjects for the two studies was females, which limited the ability to consider gender factor in the present work.

Gender difference may motivate drinking differently, as male drink out of traditional masculine ideologies featuring insensitivity and competiveness (Uy, Massoth, & Gottdiener,

2014). Men also tend to drink more. Since the majority of participants consist of males, the generalizability of the current research may be only suited to female drinkers.

Ego-depletion was manipulated. Since ego-depletion indirectly boosted acceptance, how ego-depletion was manipulated need to be reviewed. The ego-depletion in this research had two potential limitations: (1) Priming of self which may affect the results. (2) The control group may experience disturbance when waiting.

Furthermore, ego-depletion may manifest itself in a more severe form or in a more stressful situation. This dissertation used the zoo-trip composition with certain restriction to stand for the experience of ego-depletion. Logically this is accurate, but this task may generally incur moderate ego-depletion. It would be interesting to see what might happen under more severe examples of ego-depletion and whether attitudes toward advice would be receptive or reactive. The current research is not capable of addressing this question. For example, a more thorough ego-depletion may be car driving. However, this takes extensive research resources and might present ethical challenges. Finally, the context where ego- depletion occurs, such as in the lab or in front of a laptop anywhere, may make a difference MESSAGE FRAMING Framing 77 in how ego-depletion works on message processing. Waiting in the lab may lead to much less distraction if disturbance is prevented. The control group of ego-depletion may function with higher validity than online data-collection.

The definition of drinking may have been somewhat vague in this study. The second hypothesis alternately used drinks per week and peak drinks to represent for drinking. All other models tested drinks per week, peak drinks and typical drinks. How the three measures differ in meaning of construct is not entirely clear. Is it possible that drinkers who regularly drink more each week have a unique psychological pattern compared with drinkers who tend to drink much every time during a peak mood? Alcohol may be a specific substance which has a special psychosocial effect. The theory to differentiate the psychological implication of each drinking index would require a statement which describes how this drinking functions on mind precisely.

Presently, the researcher found a negative connection between self-esteem and instability of self-esteem. It was not expected in the original hypothesis. The corresponding predictive validity involving how message framing, self-esteem, instability of self-esteem functioned on the outcomes are lowered because of this correlation. How the U.S. culture educates people with high self-esteem to stabilize their self-esteem can be investigated further. Here the reason is not clear. In some cultures the two variables may be correlated differently than in other cultures. When future research delivers the idea that self-esteem and instability of self-esteem are connected differently in another cultural region, treating the two variables as two predictors in regression will require additional hypothesis. In theory, gain-framed message and loss-framed messages should be strongly different from each other. This experimental research did not include a group reading no message at all. Perhaps MESSAGE FRAMING Framing 78 the effect of reading a health message could be further located by comparing with participants having no message to read.

Study 2 of this dissertation failed to avoid familiarity for 20% of participants because they took Study 1 already. A major limitation is ignorance of perceived familiarity to the participants, especially to those who already took Study 1. Corresponding future research can explore the issue of practice effect on ego-depletion and reactance upon reading framed messages. The Future Direction section addresses how to progress new studies based on this limitation.

The last limitation of the current research concerns the freedom which the participants had. The participants were allowed to finish this experimental survey with a computer anywhere they felt convenient. They were under no compulsion to finish each question (the researcher deleted the responses with too many missed responses), and the participants could decide the pace and speed to complete this survey. Some participants may have carelessly gone through each question and rushed through this survey. Especially, given the fact that even using a Smartphone, one could possibly finish these surveys. The concentration of participants may be low. Lieberman (2008) validated internet research by suggesting that web-based research data is stable across rounds of data collection. However,

Lieberman (2008) took sample from recruiting web-sites on famous search engine pages, such as Google. It is possible that recruiting psychology students at the University of

Houston could lead to less validity since some students may not fully commit to responding when they take their online surveys. MESSAGE FRAMING Framing 79

Therefore, there is some leeway to consider that some of the responses may be less than authentic. Perhaps more meaningful significances could have been found by using the indirect attitude measure, drinking intention, because some participants were hurrying, and this may have interfered with their deliberate attention to the survey, especially the part about message evaluation.

Future Directions. Several possible extensions can be realized if more studies are planned based on the findings from this dissertation. First, persuasiveness of message framing was examined by the student participants who had no serious consequences choosing, applauding or rejecting the health advice. For example, if a follow-up research study practically offered the reward or punishment according to framed message, or served as another experimental dimension to be manipulated, acceptance of message may depend on the consequences involved. For example, in a scenario is which cash or a prize is to be gained, the meaning of accepting a message is associated with monetary acquisition. The vividness of this kind of research may amplify the effect of Reactance Theory or diminish it, depending on other moderating factors or the experimental situation. Also, the dissertation did not include a prime which activated the participants’ attention of message framing, gain or loss. Perhaps following research can make participants more sensitive as to how a message is framed by priming with explicitly irrelevant information. The information could be emotional and thus prime participants’ notice of message framing after they are primed.

Exactly which kind of prime still needs further consideration. Also, health advice about drinking reduction can be reached in any community via online web-sites or paper flyers.

The participants in this dissertation are thought to have certain understanding of drinking by the time they joined the experimental survey. This may influence or determine how MESSAGE FRAMING Framing 80 participants would treat messages in the first place. Then, even reactance might not have occurred. The dissertation captured the indirect reactance by measuring drinking intention. If a future experiment focuses on broadcasting scientific aspects of health behavior, such as the biological and pharmacological processes of addiction treatment, the participants may display different explicit attitudes toward these professional messages with knowledge, compared with current research presenting in more conventional terms. The messages in the dissertation included more conventional health information. The participants may use their general understanding to regard the contents. To apply the theory to practical health education or communication, a health advice or advertisement is not required to be so complicated. The current research may still have its importance. However, it is possible to reach out to the domain of conveying professional knowledge since the current research only delivered health advices with easy science knowledge. As a presentation includes professional information which needs thorough evaluation, those who would reject the offered message may differ from the obtained findings. Another direction is based on the theory that messages featuring short-term benefit may persuade different people under different states (ego-depletion or full self-control) compared with messages on long-term benefit (Ein-Gar, Goldberg, & Sagiv 2012). Future research would experiment on messages on short-term health benefit of alcohol reduction vs. long-term benefit. Ego-depletion could reduce attention to long-term benefit (Ein-Gar, Goldberg, & Sagiv 2012). This can go through replication or be combined with other variables in research.

Gender would potentially affect how message framing, ego-depletion and reactance function on health education or campaign. Adding gender as a dimension in the data of the current research is not meaningful since male participants were rare in both Study 1 and 2. MESSAGE FRAMING Framing 81

Some researches like that by Uy, Massoth, and Gottdiener(2014) support gender difference on drinking motivation. Gender also predicts structural brain difference (Belfi, Conrad,

Dowson, & Nopoulos, 2014). Perhaps ego-depletion and reactance work differently by gender. Just replicating this research by recruiting more male participants would be the simple, but appropriate first-step expansion for this research. The advanced studies can consider if brain difference by gender affects the implication of ego-depletion, and if culture assigns a gender, maybe female, more responsibility not to experience psychological reactance.

Study 1 tested potential effects of ego-depletion and loss frame. Perhaps future research can experiment on both ego-depletion and gain frame. As Cho and Choi (2010) found that gain frame leads to higher health education acceptance when mood is positive, maybe next study could induce a positive mood to the experimental group. If audiences are in positive mood, does gain frame cause efficient persuasion? It will be one study testing this issue. Ego-depletion can be another variable included. In general, being able to receive course credit should increase participant’s mood. However, the next study should practically ask this question to confirm. Thus, does ego-depletion plus positive mood, or ego-depletion plus negative mood cause mind open to health education? So this can be studied. Induction of negative mood may need care because of ethics which must be upheld.

This research found a moderate and negative correlation between self-esteem and instability of self-esteem. Follow-up studies can establish why and how they are associated, positively or negatively. Also, testing a third common variable behind both may be another plausible research option. The original theory of this dissertation treated the two as independent, but found a moderate and negative correlation between the two. This MESSAGE FRAMING Framing 82 association may still depend on each culture or region. Following studies can theorize why or why not self-esteem is linked with stability of self-esteem, positively or negatively.

Future studies can expand research on self-esteem and stability of self-esteem according to how Kernis, Lakey, and Heppener (2008) operated, and new research can add in message framing, ego-depletion and reactance. Self-esteem has other dimensions, such as contingency of self-esteem, which means whether self-esteem fluctuates much depending on life history, such as frustrations or victories depending on life experiences (Kernis, Lakey, and Heppener, 2008). Measuring implicit self-esteem instead of self-esteem (Kernis, Lakey, and Heppener, 2008) contributes to possible research extension about self-esteem as well.

Could other dimensions and forms of self-esteem function differently on ego-depletion and reactance? This will be a meaningful research task in terms of instability of self-esteem.

Future study can measure self-esteem for several times on the same sample, and examine the fluctuation across time (Kernis, Lakey, and Heppener, 2008). This longitudinal study takes more time and will take greater effort recruiting participants, but the external validity will increase.

Self and the relevant constructs may not be the only moderating variables to determine message acceptance. Intelligence and knowledge may drive how a person evaluates messages and their decisions to follow the advice or not. The domain of cognitive psychology may provide more sources for future construction of theory regarding how intelligence, educated expertise, and knowledge lead to attitude persistence in front of incoming messages with suggestions. Being more learned does not necessarily mean higher acceptance of health information. It requires further consideration before a hypothesis can be set. Future studies can engage in longitudinal tracking of drinking intentions after message MESSAGE FRAMING Framing 83 reading. Attitude changes as time goes by. With other moderators added, investigating how drinking intention varies as time goes by would be meaningful as the social influence from presented health message may exert its persuasive influence psychologically with a lag of time.

Messages can exert social influence through other forms. Garcia-retamero, and

Cokely (2014) argues that adding messages with graphs helps young people with higher graph literacy accept more of messages advocating sexual disease prevention. Then, could messages with graphs look extremely attractive at the moment of ego-depletion? Who rejects messages with graphs more? These questions can be addressed during future studies adopting some messages with graphs.

In terms of sampling, if the dissertation would be replicated across other universities or colleges, since the data would include samples from various schools, any inattention caused by online response from a certain school may not make the research results as biased.

Alternatively, lab studies which replicate this dissertation can serve as a valid comparison to check if major research results of this research were subverted because of inattention.

One simple direction next time would be to replicate Study and restrict participation from the people who served as the subjects already. Or next time both studies can replicate by recruiting the participants who will definitely answer both surveys from Study 1 and

Study 2. Then, how familiarity affects response can be statistically controlled with techniques for repeated measures. Also next time a participant could evaluate a message across more than one session, and then whether the response changes out of individual character or out of situation could be analyzed. MESSAGE FRAMING Framing 84

Message framing, ego-depletion and Reactance Theory constitute the major part of this dissertation. How message framing enhances its persuasiveness to convince people is a challenging enigma. The current research found that Reactance Theory works in some instances involving indirect attitude expression to drinking reduction messages. Willingness to follow health communication to reduce drinking depends on ego-depletion, but other variables as well are critical. Ego-depletion may be a state which indirectly allows drinkers to cast aside their insistence on regular drinking and start to drink less. Other related variables are self-esteem, instability of self-esteem, and self-relevance. Self-esteem and instability of self-esteem help acceptance of health messages about drinking reduction. Self- relevance fosters more psychological reactance to drinking reduction campaign. MESSAGE FRAMING Framing 85

References

Al-Alak, B. A. M. (2010). Impact of marketing actions on relationship quality in the health

and fitness sector in Jordan and its implications for social marketing: A field

analytical study. Social Marketing Quarterly, 16(2),70-93. DOI:

10.1080/15245001003746733

Andreason, A.R. (2003). The life trajectory of social marketing: Some implications.

Marketing Theory, 3(3),293-303. DOI:10.1177/147059310333004.

Armstrong, K., Schwartz, S., Fitzgerald, G., Putt, M., & Ubel, P. A. (2002). Effects of

framing as gain versus loss on understanding and hypothetical treatment choices:

Survival and mortality curves. Medical Decision Making, 22(1), 76-83.

url:http://mdm.sagepub.com/content/22/1/76.abstract.

Avci, O., & Fendrich, M. (2010). Student drinking-related problems in an urban campus:

Implications for research and prevention. Journal of American College Health,

58(6), 545-554. DOI:10.1080/07448481003705933

Averback, J. M., Jones. A., & Robertson, K. (2011). Prior knowledge and health messages:

An examination of affect as heuristics and information as systematic processing for

fear appeals. Southern Communication Journal, 76(1), 35-54.

DOI:10.1080/10417940902951824.

Baer, J.S. (1993). Etiology and secondary prevention of alcohol problems with young adults.

In Baer, J.S., Marlatt, G.A., McMahon,R.J., editors p.111-p137. Addictive behaviors

across the life span. Sage Publication, Newbury Park, CA. ISBN:9780803950795 MESSAGE FRAMING Framing 86

Banks, S. M., Salovey, P., Grenner, S., Rothman, A. J., Moyer, A, Beauvais, J., & Epel, E.

(1995). The effects of message framing on mammography utilization. Health

Psychology, 14(2), 178-184. DOI:10.1037/0033-2909.131.6.856.

Bartels, R. D., Kelly. K. M., & Rothman, A.J. (2010). Moving beyond the function of the

health behavior: The effect of message frame on behavioral decision-making.

Psychology and Health, 821-838. DOI:10.1080/0887044090289.3708.

Baumeister, R.F., Bratslavsky, E., Muraven, M., & Tice, D.M. (1998). Ego depletion: Is the

active self a limited resource? Journal of Personality and Social Psychology,

74(5),1252-1265. DOI:10.1037/0022 -3514.74.5.1252.

Baumeister, R.F., Campbell, J.D., Krueger, J.I., & Vohs, K.D. (2003). Does high self-esteem

cause better performance, interpersonal success, happiness, or healthier lifestyles?

Psychological Science in the Public Interest, 4(1), 1-44. DOI:10.1111/1529-

1006.01431

Belfi, A.M., Conrad, A.L., Dawson, J., & Nopoulos, P. (2014). Masculinity/Femininity

predicts brain volumes in normal healthy children. Developmental

Neuropsychology, 39(1), 25-36. DOI:10.1080/87565641.2013.839681

Bem. D.J. (1967). Self-perception: An alternative interpretation of cognitive dissonance

phenomena. Psychological Review, 74(3),183-200. DOI:10.1037/h0024835

Bilewicz, M., & Kofta, M. (2011). Less biased under threat? Self-verificatory reactions

social identity threat among groups with negative self-stereotypes. Journal of MESSAGE FRAMING Framing 87

Applied Social Psychology, 41(9), 2249-2267. DOI/10.1111/j.1559-

1816.2011.00802.x.

Brehm, J.W. (1989). Psychological reactance: Theory and applications. Advances in

Consumer Research, 16, 72-75.

url: http://connection.ebscohost.com/c/articles/6487682/psychological-reactance-

theory- applications.

Brehm, S.S. & Brehm, J.W. (1981). Psychological reactance: A theory of freedom and

control. New York: Academic Press. ISBN:012129840X.

Brown, S. & Locker, E. (2009). Defensive responses to an emotive anti-alcohol message.

Psychology and Health, 24(5), 517-528. DOI:10.1080/0887044080191130.

Caccioppo, J.T., Petty, R.E., and Stoltenberg, C.D. (1985). Processes of social influence:

The elaboration likelihood model of persuasion. Advances in Cognitive- Behavioral

Research, 4, 215-273. Academic Press. CA. ISBN:0120106043.

Carnevale, P. J. (2007). Positive affect and decision frame in negotiation. Group Decision

Negotiation, 17, 51-63.DOI: 10.1007/s10726-007-9090-x.

Cassoti, M., Habib, M., Poirel, N., Aïte, A., Houdé , O., & Moutier, S. (2012). Positive

emotional context eliminates the framing effect in decision making. Emotion, 1-6.

DOI:10.1037/a0026788.

Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source

versus message cues in persuasion. New York: Plenum Press. ISBN:0-306-42022-8. MESSAGE FRAMING Framing 88

Chang, C., & Lee, Y. (2009). Framing charity advertising: Influences of message framing,

image valence, and temporal framing on a charitable appeal. Journal of Applied

Social Psychology, 39(12),2910-2935.DOI:10.1111/j.1559-1816.2009.00555.x.

Charbol, H., Rousseau, A., & Callahan, S. (2006). Preliminary results of scale assessing

instability of self-esteem. Canadian Journal of Behavioral Science, 38(2), 136-141.

DOI:10.1037/cjbs2006003

Cho, H, & Choi, J. (2010). Predictors and the role of attitude toward the message and

perceived message quality in gain-and loss- frame antidrug persuasion of

adolescents. Health Communication, 25,303-311.

DOI:10.1080/10410231003773326.

Clarkson, J.J., Hirt, E.R., Jia, L., & Alexander, M.B. (2010). When perception is more than

reality: The effects of perceived versus actual resource depletion on self-regulatory

behavior. Journal of Personality and Social Psychology,98(1), 29-46.

DOI:10.1037/a0017539.

Cohen, E. L., (2008). "My Loss is Your Gain": Examining the Role of Message Frame,

Perceived Risk, and Ambivalence in the Decision to Become an Organ Donor" Paper

presented at the annual meeting of the International Communication Association,

Montreal, Quebec, url:http://www.allacademic.com/meta/p232529_index.html.

Cohen, J., Cohen, P., West, S. G. & Aiken, L.S. (2003). Applied multiple

regression/correlation analysis for the behavioral sciences. ISBN:10-0805822232.

Collins, R. L., Parks, G. A., & Marlatt, G. A. (1985). Social determinants of alcohol

consumption: The effects of social interaction and model status on the self- MESSAGE FRAMING Framing 89

administration of alcohol. Journal of Consulting and Clinical Psychology, 53,189-

200. DOI:10.1037/0022-006X.53.2.189.

Dean, M., Lampila, P., Shepherd, R. Arvola, A., Saba, A., Vasallo, M. Claupein, E.,

Winkelmann, M., & Lähteenmäki, L. (2012). Perceived relevance and foods with

health- related claims. Food Quality and Preference, 24, 129-135.

DOI:10.1016/j.foodequal.2011.10.006.

De Wit, J. B.F., Das, E., & De Hoog, N. (2007). Self-regulation of health communications:

A motivated processing approach to risk perception and persuasion. In The scope of

social psychology; theory and applications; essays in honour of Wolfgang Stroebe,

209-226. Psychology Press, New York, NY. U.S.A. ISBN10:1-84169-645-5.

Denson, T.F., Pedersen, W.C., Friese, M., Hahm, A., & Roberts, L. (2011). Understanding

impulsive aggression: Angry rumination and reduced self-control capacity are

mechanisms underlying the provocation–aggression relationship, Personality and

Social Psychology Bulletin, 37(6),850-862.DOI:10.1177/0146167211401420.

Dimeff, L.A., Baer, J.S., Kivlahan, D.R., & Marlatt, G.A., (1999). Brief Alcohol Screening

and Intervention for College Students (BASICS): A Harm Reduction Approach.

Guilford Press, New York, NY.ISBN:1-57230-392-1

Di Paula, A. & Campbell, J.D. (2002). Self-esteem and persistence in the face of failure.

Journal of Personality and Social Psychology, 83(3), 711-724. DOI:10.1037//0022-

3514.83.3.711.

Durkin, S. J., Biener, L., & Wakefield, M.A. (2009). Effects of different types of

antismoking ads on reducing disparities in smoking cessation among socioeconomic MESSAGE FRAMING Framing 90

subgroups. American Journal of Public Health, 99(12), 2217-2223.

DOI:10.2105/AJPH.2009.161638.

Ein-Gar, D., Goldenberg, J., & Sagiv, L.(2012). The role of consumer self-control in the

consumption of virtue products. International Journal of Research in

Marketing,29,123-133. DOI:10.1016/j.ijresmar.2011.08.003

Erceg-Hurn, D.M., & Steed, L.G. (2011). Does exposure to cigarette health warnings elicit

psychological reactance in smokers? Journal of Applied Social Psychology,

41(1),219-237. DOI:10.1111/j.1559-1816.2010.00710.x

Evans, J.S.B.T. (2007). On the resolution of conflict in dual process theories of reasoning.

Thinking and Reasoning, 13(4), 321-339. DOI: 10.1080/13546780601008825.

Fischer, P., Greitemeyer, T., & Frey, D. (2008). Self-regulation and selective exposure: The

impact of depleted self-regulation resources on confirmatory information processing.

Journal of Personality and Social Psychology, 94(3), 382-395. DOI:10.1037/0022-

3514.94.3.382.

Flanagin, A.J., & Metzger, M.J. (2000). Perceptions of internet information credibility.

Journalism & Mass Communication Quarterly, 77(3),515-540.

DOI:10.1177/107769900007700304

Foster, D.W., Yeung, N., and Neighbors, C. (2014). I think I can’t: Drink refusal self-

efficacy as a mediator of the relationship between self-reported drinking identity and

alcohol use. Addictive Behaviors, (39),461-468. DOI:10.1016/j.addbeh.2013.10.009 MESSAGE FRAMING Framing 91

Garcia-retamero, R., & Cokely, E.T. (2014). The influence of skills, message frame, and

visual aids on prevention of sexually transmitted diseases. Journal of Behavioral

Decision Making, 27,179-189. DOI:10.1002/bdm.1797

Gallopel-Morvan, K., Gabriel, P., Le Gall-Ely, M., Rieunier, S, & Urien, B. (2009). The use

of visual warnings in social marketing: The case of tobacco. Journal of Business

Research,64, 7-11.DOI:10.1016/j.jbusres.2009.09.012 .

Gerend, M. A., & Sias, T. (2009). Message framing and color priming: How subtle threat

cues affect persuasion. Journal of Experimental Social Psychology, 45, 999-1002.

DOI:10.1016/j.jesp.2009.04.002.

Gilbert, K. L., Peterson, R. S., Scanlon, K. E. (2010). Promoting sexual health among MSM

online: A hepatitis campaign comparison study. Journal of Gay and Lesbian Social

Sciences, 22, 446-462. DOI:10/1080/10538720.2010.491750.

Geller, E.S. (1989). Applied behavior analysis and social marketing: An integration for

environmental preservation. Journal of Social Issues, 45(1), 17-36.

DOI:10.1111/j.1540- 4560.1989.tb01531.x.

Gordon, R., Moodie, C., Eadie, D., & Hastings, G. (2010). Critical social marketing – The

impact of alcohol marketing on youth drinking: Qualitative findings, International

Journal of Nonprofit and Voluntary Sector Marketing, 15, 265-275.

DOI:10.1002/nvsm.388.

Greene, K., & Brinn, L.S. (2003). Messages influencing college women’s tanning bed use:

Statistical versus narrative evidence format and a self-assessment to increase MESSAGE FRAMING Framing 92

perceived susceptibility. Journal of Health Communication, (8),443-461.

DOI:10.1080/1081073039233271.

Grier, S., and Bryant, C.A. (2005). Social marketing in public health. Annual Review Public

Health, 26, 319-339. DOI:10.1146/annurev.publhealth.26.021304.144610.

Hill, S.E., DelPriore, D.J., & Vaughan, P.W. (2011). The cognitive consequences of envy:

Attention, memory, and self-regulatory depletion. Attitudes and Social Cognition,

101(4), 653-666. DOI:10.1037/a0023904.

Hinson, R.W. (2010). Magnitude and prevention of college drinking and related problems.

Alcohol Research & Health, 33, 45-54.

url: http://www.niaaa.nih.gov.ezproxy.lib.uh.edu/Publications/AlcoholResearch/.

Hyde, M. K., & White, K. M. (2009). To be a donor or not to be? Applying an extended

theory of planned behavior to predict posthumous organ donation intentions,

Journal of Applied Social Psychology, 39(4), 880-900. DOI:10.1111/j.1559-

1816.2009.00464.x

Jung, T., Shim, W., Montaro, T. (2010). Psychological reactance and effects of social norms

messages among binge drinking college students. Journal of Alcohol and Drug

Education, 54 (3), 7-18.url: http://trove.nla.gov.au/work/10887758

Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk

Econometrica , 47(2), 263-291.url: http://www.blackwellpublishers.co.uk.ezproxy.

lib.uh.edu/journal.asp?ref=0012-9682&site=1 MESSAGE FRAMING Framing 93

Keller, P. A., Lipcus, I.M., & Rimer, B. K. (2003). Affect, framing, and persuasion. Journal

of Marketing Research, 40, 54-64. URL: http://www.jstor.org/stable/30038835

Kopfman, J.E., Smith, S.W., Ah Yun, J.K., & Hodges, A. (1998). Affective and cognitive

reactions to narrative versus statistical evidence organ donation messages. Journal of

Applied Communication Research, 26, 279-300. DOI:10.1080/00909889809365508

Kreuter, M., Holmes, K., Hinyard, L.J., Houston, T., Woolley, S., Green. M. C., Capella,

J.N.,Slater, M. D.,Wise, M. E., Strorey, D., Clark, E.M., O’Keefe, D.J., Erwin,

D.O.(2007).Narrative communication in cancer prevention and control: A framework

to guide research and application. Annals of Behavioral Medicine, 33(3),221-235.

DOI:10.1007/BF02879904

Kross, E., & Grossman, I. (2012). Boosting wisdom: Distance from the self enhances wise

reasoning, attitudes, and behavior. Journal of Experimental Psychology: General,

141(1),43-48. DOI:10.1037/a0024158

Kruglanski, A.W., & Chen, X., Pierro, A., Mannetti, L., Erb, H, & Spiegel, S. (2006).

Persuasion according to the unimodel: Implications for cancer communication.

Journal of Communication, 56, S105-S122. DOI:10.1111/j.1460-2466.2006.00285x

Kuppens, P., & Van Mechelen, I. (2007). Interactional appraisal models for the anger

appraisals of threatened self-esteem, other-blame, and frustration. Cognition and

Emotion, 21(1), 56-77. DOI:10.1080/02699930600562193

Larsen (2010). Initiation and continuation of best friends and adolescents’ alcohol

Consumption: Do self-esteem and self-control function as moderators? International MESSAGE FRAMING Framing 94

Journal of Behavioral Development, 34(5), 406-416.

DOI:10.1177/0165025409350363

Latimer, A.E., Green, K.E., Schmid, K., Tomasone, J., Abrams, S., Cummings, K.M.,

Celestino, P., Salovey, P., Seshadri, S., & Toll, B.A. (2010). The identification of

framed messages in the New York State Smoker’s Quitline materials. Health

Education Research, 25(1),54060. DOI:10.1093/her/cyp041

Latimer, A.E., Rench, T.A., Rivers, S.E., Katulak, N.A., Materese, S.A., Cadmus, L., Hicks,

A. Hodorowski, J.K, & Salovey, P. (2008). Promoting participation in physical

activity using framed messages: An application of prospect theory. British Journal

of Health Psychology, 13, 659-681.DOI:10.1348/135910707X246186

Lee, L.G., & Lee, W.J. (2009). Psychological reactance to online recommendation services.

Information and Management, 46,448-452. DOI:10.1016/j.im.2009.07.005

Levy, J. S. (1997). Prospect theory, rational choice, and international relations. International

Studies Quarterly,41,87-112.url:http://www.jstor.org/stable/2600908.

Lieberman, D.Z., (2008). Evaluation and stability of participants samples recruited over the

Internet. Cyberpsychology and Behavior, 11(6), 743-745.

DOI:10.1089/cpb.2007.0254

Liberman, A., & Chaiken, S. (1992). Defensive processing of personally relevant health

messages. Personality and Social Psychology Bulletin, 18(6), 669-679.

DOI:10.1177/0146167292186002 MESSAGE FRAMING Framing 95

Lindgren, K.P., Neighbors, C., Teachman, B.A., Wires, R.W., Westgate, E., & Greenwald,

A. (2012). I drink therefore I am: Validating alcohol-related implicit association

tests. Psychology of Addictive Behaviors,Mar.DOI:10.1037/a0027640

Maguire, K.C., Gardner, J., Sopory, P., Jian, G., Roach, M., Amschlinger, J., Moreno, M.,

Pettey, G., & Piccone, G. (2010). Formative research regarding kidney disease

health information in a Latino American frame, threat, efficacy, message

effectiveness, and behavioral intention. Communication Education, 59(3), 344-359.

DOI:10.1080/03634521003628271

Marlatt GA, Baer JS, & Larimer ME. (1995) Preventing alcohol abuse in college students: a

harm-reduction approach. In Boyd G, Howard J, Zucker R (eds). Alcohol Problems

Among Adolescents: Current Directions in Prevention Research, 147–172.

Northvale, NJ: Lawrence Earlbaum.ISBN:08058195150

MacDonald, J.A., Sandry, J., & Rice, S. (2012). Self-construal priming affects speed of

retrieval from short-term memory. PLOS ONE,7(11),e50007.

DOI:10.1371/journal.pone.005.0007

McKenzie-Mohr, D. (1994). Social marketing for sustainability: The case of residential

energy conservation Futures, 26(2), 224-233. DOI:10.1016/0016-3287(94)90111-2

Meyerowitz, B.E., & Chaiken, S. (1987). The effect of message framing on breast self-

examination attitudes, intentions, and behavior. Journal of Personality and Social

Psychology, 52(3), 500-510.

url:http://www.wiley.com.ezproxy.lib.uh.edu/WileyCDA/ MESSAGE FRAMING Framing 96

Millar. M. (2006). Responses to messages about health behaviors: The influence of

repressive coping. Psychology and Health, 21(2),231-

245.DOI:10.1080/14768320500105361

Millar M.G., & Millar, K. (1998) Processing messages about disease detection and health

promotion behaviors: The effects of anxiety. Health Communication, 10(3),211-226.

url:http://www.ncbi.nlm.nih.gov/pubmed/16370983

Miller,W.R., & Rollnick, S. (2002). Motivational interviewing: Preparing people for

change, 2nd ed, Guilford Press. New York, NY. ISBN:1-57230-563-0

Mullan, E., & NiGabhainn, S. (2002). Self-esteem and health-risk behaviors: Is there a link?

The Irish Journal of Psychology,23,1-2,27-

36.URL:http://repository.wit.ie/id/eprint/1333

Muraven, M., & Slessareva, E. (2003). Mechanisms of self-control failure: Motivation and

limited resources. Personality and Social Psychology Bulletin, 29(7), 894-906.

DOI:10.1177/0146167203029007008

Muraven, M., Tice, D.M., & Baumeister, B.F. (1998). Self-control as limited resource:

Regulatory depletion patterns. Journal of Personality and Social Psychology, 74(3),

774-789. DOI:/10.1037/0022-3514.74.3.774

Neighbors, C., Dillard, A.J., Lewis, M.A., Bergstrom, R.L., & Neil, T.A. (2006). Normative

misperceptions and temporal precedence of perceived norms and drinking. Journal

of Studies on Alcohol, 67(2), 290-299.

url:http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2443635/ MESSAGE FRAMING Framing 97

Neumann, C.A., Leffingwell, T.R.,Wagner, E.F., Mignogna, J.,& Mognogna, M. (2009).

Self-esteem and gender influence the response to risk information among alcohol

using college students. Journal of Substance Use, 14(6),353-

363.DOI:10.3109/14659890802654540

Newman, D.P. (1980). Prospect theory: Implications for information evaluation. Accounting,

Organization, and Society, 5(2), 217-230.

url:http://www.elsevier.nl/inca/publications/store/4/8/6/

Niederdeppe, J., Bu, Q.L., Borah, P., Kindig, D.A., & Robert, S.A. (2008). Message design

strategies to raise public awareness of social determinants of health and population

health disparities. The Milbank Quarterly,86(3),481-513.DOI:10.1111/j.1468-

0009.2008.00530.x

Ozubko, J.D., & Fugelsang, J. (2011). Remembering makes evidence compelling: Retrieval

from memory can give rise to the illusion of truth. Journal of Experimental

Psychology: Learning, Memory and Cognition, 37(1), 270-276.

DOI:10.1037/a0021323

O’Connor, D. B., Warttig, S., Conner, M., & Lawton, R. (2009). Raising awareness of

Hypertension risk through a web-based framing intervention: Does consideration of

Future consequences make a difference? Psychology, Health, & Medicine, 14(2),

213-219. DOI:10.1080/13548500802291618

Park, P., Prevost, A.T., & Griffin, S.J. (2010) A randomized evaluation of loss and gain

frames in an invitation to screening for type 2 diabetes: Effects on attendance, MESSAGE FRAMING Framing 98

anxiety and self-rated health. Journal of Health Psychology, 15(2),196-204.

DOI:10.1177/1359105309344896

Pelletier, L. G., Sharp, E. (2008). Persuasive communication and proenvironmental

behaviours: How message tailoring and message framing can improve the integration

of behaviours through self-determined motivation. Canadian Psychology, 49(3),210-

217.DOI:10.1037/a0012755

Petraglia, J., (2009). The importance of being authentic: Persuasion, narration, and dialogue

in health communication and education. Health Communication, 24, 176-185.

DOI:10.1080/10410230802076771

Putrevu, S. (2010). An examination of consumer responses toward attribute-and goal framed

messages. Journal of Advertising, 39(3), 5-24. DOI:10.2753/JOA0091-3367390301

Quick, B. L., & Bates, B.R. (2010).The use of gain- or loss- frame messages and efficacy

appealsto dissuade excessive alcohol consumption among college students: A test of

psychological theory, Journal of Health Communication, 15, 603-628.

DOI:10.1080/10810730.2010.499593

Quick, B.L, Scott, A.M., & Ledbetter, A.M. (2011). A close examination of trait reactance

and issue involvement as moderators of psychological reactance theory. Journal of

Health Communication, 16,660-679. DOI:10.1080/10810730.2011.551989

Rains, S. A., & Turner, M. M. (2007). Psychological reactance and persuasive health

communication: A test and extension of the intertwined model. Health

Communication Research,33,241-269.DOI:10.1111/j.1468-2958.2007.00298.x MESSAGE FRAMING Framing 99

Renger, R.(1993). Predicting athletic success: Issues related to analysis and interpretation of

study findings. The Sport Psychologist,7,262-274.

url: http://journals.humankinetics.com/tsp-back-

issues/tspvolume7issue3september/predictingathleticsuccessissuesrelatedtoanalysisa

ndinterpretationofstudyfindings

Rosenthal, P. I. (1972). The concept of the paramessage in persuasive communication.

Quarterly Journal of Speech, 58(1),15-30.DOI:10.1080/00335637209383097

Rosenberg, M. (1965). Society and the adolescent image. Princeton University Press:

Princeton, NJ.ISBN:10-0819562289

Rucker, D., Petty, R. E., & Briñol, P. (2008). What’s in a frame anyway?: A meta-cognitive

analysis of the impact of one versus two sided message framing on attitude certainty.

Journal of Consumer Psychology, 18, 137-149.DOI:10.1016/j.jcps.2008.01.008

Schmeichel, B.J.,Vohs, K.D., & Baumeister, R.F. (2003). Intellectual performance and ego

depletion: Role of the self in logical reasoning and other information processing.

Journal of Personality and Social Psychology, 85(1), 33-46. DOI:10.1037/0022-

3514.85.1.33

Seery, M.D., Blascovich, J., Weisbuch, M., & Vick, S.B. (2004). The relationship between

self-esteem level, self-esteem stability, and cardiovascular reactions to performance

feedback. Journal of Personality and Social Psychology, 87(1), 133-145.

DOI:10.1037/0022-514.87.1.133.

Seibel, C.A., & Dowd, E.T. (1999). Reactance and therapeutic noncompliance. Cognitive

Therapy and Research, 23(4),373-379. DOI:10.1023/A:1018751817046 MESSAGE FRAMING Framing 100

Seo, K, & Shen, F. (2009). The interactive effects of visual images and message framing on

health-related persuasion, presented in annual meeting of International

Communication of Association.

url: http://citation.allacademic.com/meta/p299275_index.html

Shamaskin, A.M., Mikels, J.A., & Reed, A.E.(2010). Getting the message across: Age

differences in the positive and negative framing of health care messages. Psychology

and Aging, 25, (3),746-751. DOI:10.1037/a0018431

Shen, L. (2010). The effect of message frame in anti-smoking public service announcements

on cognitive response and attitude toward smoking. Health Communication, 25, 11-

21. DOI:10.1080/10410230903473490

Silvia,P.J. (2006). Reactance and the dynamics of disagreement: Multiple paths from

threatened freedom to resistance to persuasion. European Journal of Social

Psychology, 36,673-685. DOI:10.1002/ejsp.309

Steginga, S. K., Occhipinti, S. (2004). The application of the heuristic-systematic processing

model to treatment decision making about prostate cancer, Medical Decision

Making, 24:573, 573-583.DOI:10.1177/2072989X04271044

Stein, J.A., Dixon, E.L., & Nyamathi, A.M. (2008). Effect of psychosocial and situational

variables on substance abuse among homeless adults. Psychology of Addictive

Behaviors, 22(3),410-416. DOI:10.1037/0893-164X.22.3.410

Trumbo, C.W. (2002). Information processing and risk perception: An adaption of the

heuristic-systematic model. Journal of Communication, June, 367-382.

DOI:10.1111/j.1460-2466.2002.tb02250.x MESSAGE FRAMING Framing 101

Tversky, A. & Kahneman, D. (1981). The framing decisions and the psychology of choice.

Science, New Series, 211(4481), 452-458. DOI:10.1126/science.7455683

Unger, A., & Stahlberg, D. (2011). Ego-depletion and risk behavior. Social Psychology,

42(1),28-38. DOI:10.1027./1864-9335/a00040

Uskul, A.K., & Oyserman, D. (2010). When message frame fits salient cultural-frame,

messages feels more persuasive. Psychology and Health, 25(3), 321-337.

DOI:10.1080/088704409020759156

Uy, P.J., Massoth, N., & Gottidiener, W.H. (2014). Rethinking male drinking: Traditional

masculine ideologies, gender-role conflict, and drinking motives. Psychology of Men

and Masculinity, 15(2), 121-128. DOI:10.1037/a0032239 van Brakel, W.H. (2006). Measuring health-related stigma-A literature review. Psychology,

health, & Medicine, 11(3),307-334. DOI:10.1080/1354850060059516

Vraga, E. K.,Carr, D. J., Nytes, J.P., & Shah, D.V. (2010). Precision vs. realism on the

framing continuum: Understanding the underpinnings of message effects. Political

Communication, 27, 1-19. DOI:10.1080/10584600903297927

Wakefield, M. A., Laken, B., Honik, R. C. (2010). Use of mass media campaigns to change

health behavior. Lancet,376,1276-1271. DOI:10.1016/S0140

Walker, M.J. (2010). Addiction and Self-deception: A method for self-control. Journal of

Applied Philosophy,27(3),305-319.DOI:10.1111/j.1468-5930.2010.00493.x MESSAGE FRAMING Framing 102

Wan, E.W., Rucker, D.D., Tormala, Z.L., & Clarkson, J.J. (2010). The effect of regulatory

depletion on attitude certainty. Journal of Marketing Research, 47,531-541.

url: http://hub.hku.hk/handle/10722/57671

Wang, K., & Dovidio, J.F. (2011). Disability and autonomy: Priming alternative identities.

Rehabilitation Psychology, 56(2), 123-127. DOI:10.1037/a0023039

Wang, X., Integrating the theory of planned behavior and attitude functions: Implications for

health campaign design. Health Communication, 24, 426-434. DOI:10.1080/104/

2030903023477

Wang, W., & Rachel, S. (2009). Does the trait-frame matter? An investigation of the

interactions between behavioral inhibition/approach system and message framing on

health message processing. In the International Communication Association’s

Annual Meeting, 1-35. http://www.allacademic.com/meta/p301066_index.html

Weisner, C., Ann Kaskutas, L., Hilton, M. E., & Barlie, A. L. (1999). ‘When you were

drinking’ vs. ‘in the past 12 months’: The impact of using different time frames in

clinical and general populations. Addiction, 94(5), 731-736.

url:http://www.wiley.com.ezproxy.lib.uh.edu/WileyCDA/

White, H.R. & Labouvie, E.W. (1989). Towards the assessment of adolescent problem

drinking. Journal of Studies on Alcohol, 50, 30-37. ISBN:9780803950795

Whitehead, D., Russel, G. (2004). How effective are health education programmes

resistance, reactance, rationality and risk? Recommendations for effective practice.

International Journal of Nursing Studies, 41, 163-172.DOI:10.1016/S0020-

7489(03)0017-2 MESSAGE FRAMING Framing 103

Williams, T, Clarke, V., & Boland, R. (2001). Effects of message framing on breast-cancer

related beliefs and behaviors: The role of mediating factors. Journal of Applied

Social Psychology, 31(5), 925-950.

url:http://www.wiley.com.ezproxy.lib.uh.edu/WileyCDA/

Wilson, T.D., Lindsey, S,& Schooler, T.Y.(2000). A model of dual attitudes. Psychological

Review,107(1),101-126.DOI:10.1037/0033-295X.107.1.101

Wilson, D.K., Purdon, S.E., & Wallston, K. A. (1988). Compliance to health

recommendations: a theoretical overview of message framing. European Journal of

Social Psychology,24,25-3.DOI:10.1093/her/3.2/61

Wolburg, J. M. (2006). College students’ responses to antismoking messages: Denial,

defiance, and other boomerang effects. The Journal of Consumer Affairs, 40(2), 294-

323. DOI:10.1111/j.1745-6606.2006.00059.x

Yan, C., Dillard, J.P., & Shen, F. (2010). The effects of mood, message framing, and

behavioral advocacy on persuasion. Journal of Communication, 60, 344-363.

DOI:10.1111/j.1460- 2466.2010.01485.x

Yu, N., Ahern, L.A., Connolly-Ahern, C., & Shen, F.(2010). Communicating the risk of

fetal alcohol spectrum disorder: Effects of message framing and exemplification.

Health Communication, 25, 692-699. DOI:10.1080/10410236.2010.521910

MESSAGE FRAMING Framing 104

Tables

Table 1 Various Framing beside Gain vs. Loss

Research Frame

Niederdeppe, Bu, Borah, Kindig, & Robert, Internal Attribution vs. External Attribution

(2008) about cause of events.

Weisner, Ann Kaskutas, Hilton and Barlie “When you were drinking” vs. “in the past 12

(1999). months”

Kopfman, Smith, Ah Yun, & Hodges (1998) Narrative Evidence vs. Statistical Evidence

Uskul, & Oyserman (2010) Frame: Individualistic culture vs.

collectivistic culture

Durkin, Biener, & Wakefield (2009) Emotional frame vs. neutral

Personal testimonial vs. Comparison ads

Shen (2010) Frame: Health Consequence, Secondhand

smoke, and Industry manipulation

Kreuter, Green, Gappella, Slater, Wise, Narrative Forms: Entertainment education,

Storey, Clark, O’Keefe, Erwin, Holmes, Journalism, Literature, Testimonials, and

Leslie, Houston, and Wooley (2007) Storytelling

Pelletier and Sharp (2008) Framing: Intrinsic Goal vs. Extrinsic Goal

Putrevu (2010) Framing: Attribute vs. Goal

MESSAGE FRAMING Framing 105

Table 2

Framing (gain vs. loss) Researches in Miscellaneous Areas

Researchers Result Conclusion

Carnevale, (2007) Loss Frame X Positive Mood Framing is related to

concession or insistence in results in great concessions negotiation. Affect and self- in negotiation. relevance are also related

independent variables.

Vraga, Carr, Nytes and Shah, Gain Frame is linked to Framing is related to

(2010) support for a policy proposal; support or opposition to

Gain Frame X Credible Facts policy proposal; self-

increases support for a policy relevance and credible facts

proposal when the decision is are related

made for others. independent variables

McKenzie, (1994) Framing is related to Proponents can use message

persuasion about energy framing, attention-capturing

conservation. and commitment-gaining to

increase the persuasiveness

for energy conservation.

MESSAGE FRAMING Framing 106

Table 2 Continued

Framing (gain vs. loss) Researches in Miscellaneous Areas

Researchers Result Conclusion

Chang & Lee (2009) Framing affects intention Congruency between

to donate to charity picture and verbal message

matters of persuasion. .

Temporal framing plus

negative framing increases

persuasion.

MESSAGE FRAMING Framing 107

Table 3

Framing (gain vs. loss) Researches in Health Promotions.

Note. For the entire citations, please see the formal text.

Researchers Health Topic Conclusion

Latimer et al, Smoking By content analysis, the percentage of

2010 gain-framed, loss-framed, and non-framed

messages were reported.

Williams, Clarke, Breast Cancer Loss-framed messages increase the intention to and Borland attend breast self-examination.

(2001)

Armstrong, Colon Cancer Loss-framed messages leads to less

Fizgerald, Putt understanding of message and less intention to and Ubel (2002) undergo surgery.

Seo and Shen Dental Flossing and Loss-framed messages make the most

(2009) Sunscreen Use persuasion when paired with visual stimuli

O’Connor, Hypertension High consideration of future consequences

Warttig, Conner, (CFC) participants spent longer time reading and Cawton loss-framed message. Some association

between low CFC and Gain-framed messages. (2009)

MESSAGE FRAMING Framing 108

Table 3 Continued

Framing (gain vs. loss) Researches in Health Promotions.

Researchers Health Topic in Message Conclusion

Maguire et al Kidney Disease Framing (gain vs. loss) was not found to

(2010) affect the persuasion.

Latimer et al Exercise Gain-framed message is linked to higher

self-efficacy and higher participation (2008)

in exercise.

Gallopel-Morran, Tobacco Visual Warning is more effective in

Gabriel, Le Gall- production of avoidance; self-efficacy

Ely, Rieunier, Urien facilitates the effect.

(2009)

Banks, Salovey, Mammography Loss-framed videos were associated with

Greener, Rothman, higher rate of mammography check-ups and Moyer (1995)

Gerend, & Sias , HPV ((Human Loss-framed Message with background

(2009) Papillomavirus) red color was most threatening

MESSAGE FRAMING Framing 109

Table 4

Frame (Gain vs. Loss) and Emotion at Health Issues

Researchers Topic Conclusion

Cho and Choi (2010) Drug Gain Frame and positive

mood

can result in greater

persuasion.

Yan, Dillard and Shen (2010) Junk Food and Eye Exam Gain Frame and positive

mood

can result in greater

persuasion.

Loss Frame and negative

mood can result in greater

persuasion as well.

Keller, Lipkus, and Rimus Mammography Positive Mood X Loss Frame

(2003) and Negative Mood X Gain

frame were most persuasive.

MESSAGE FRAMING Framing 110

Table 4 Continued

Frame (Gain vs. Loss) and Emotion at Health Issues

Researchers Topic Conclusion

Millar, and Millar (1998) Vision, Cholesterol Level, State of anxiety moderates the

relationship between message and Skin Cancer framing and persuasion. High

state anxiety facilitated the

attractiveness of health

promotion-framed message;

Low state anxiety facilitated

the attractiveness of health

detection-framed message.

Millar (2006) Cholesterol Level, Blood Trait anxiety affects the

Pressure, and Skin Cancer cognitive processing of two

types of messages: health

detection vs. health

promotion messages.

MESSAGE FRAMING Framing 111

Table 5

Message Framing on Alcohol Problem

Researchers Topic Conclusion

Brown and Locker(2009) Drinking Loss-framed message with

significant emotional tone

decreased the participant’s

estimated risk.

Quick and Bates(2010) Drinking There is no relationship

between message framing and

defensiveness toward the

message

Yu, Ahen, and Shen (2010) Fetal Alcohol Spectrum Gain-framed message was

Disorder (FASD). more persuasive; Loss frame

plus exemplar frame

increases persuasion as well.

MESSAGE FRAMING Framing 112

Table 5 Continued

Message Framing on Alcohol Problem

Researchers Topic Conclusion

Brown and Locker(2009) Drinking Defensive and vulnerable

participants showed more

resistance to messages framed

with relative serious loss.

MESSAGE FRAMING Framing 113

Table 6 Correlation and Descriptive Statistics: Study 1:Ego-depletion Measure 1 2 3 4 5 7 1. Drinks per week --

2. Peak drinks .66*** --

3. Typical drinks .38*** .45*** --

4. Message Rating -.02 .03 .03 --

5. Intention-Week .79*** .54*** .27*** -.08 --

6. Intention-Typical .38*** .43*** .36*** -.03 .45*** --

Mean 6.16 4.82 4.06 4.91 4.55 2.53 SD 5.59 3.84 3.26 1.06 4.47 2.27 Note. *** p < .0001.

MESSAGE FRAMING Framing 114

Table 7

Message rating as a function of ego-depletion and drinks per week

b β SE t

Ego-depletion -.01 -.01 .15 -.12

Drinks per week -.01 -.05 .02 -.52

Ego X Drinks per .01 .04 .03 .38 week

MESSAGE FRAMING Framing 115

Table 8 Message rating as a function of ego-depletion and peak drinks

b β SE t

Ego-depletion -.01 -.00 .15 -.04

Peak drinks -.03 -.11 .03 -1.22

Ego X Peak drinks .10* .22* .04 2.44*

*p<.05

MESSAGE FRAMING Framing 116

Table 9

Message rating as a function of ego-depletion and typical drinks

b β SE t

Ego-depletion -.03 -.02 .15 -.22

Typical drinks .01 .05 .03 .36

Ego X Typical drinks -.01 -.02 .03 -.18

MESSAGE FRAMING Framing 117

Table 10

Message Rating as a Function of Drinks per week, Ego-depletion and Message Framing

b β SE t

Drinks per week .02 .12 .02 .98

Ego-depletion -.10 -.05 .20 -.53

Drinks X Ego -.01 -.04 .04 -.38

Frame -.04 -.02 .21 -.20

Drinks X Frame -.08* -.23* .04 -2.09*

Ego X Frame .23 .09 .31 .75

Drinks X Ego X .05 .08 .06 .78

Frame

Frame = Message Framing; * p<.05

MESSAGE FRAMING Framing 118

Table 11 Message Rating as a Function of Typical Drinks, Ego-depletion and Message Framing

b β SE t

Peak Drinks .04 .14 .04 .86

Ego-depletion -.12 -.06 .20 -.62

Peak X Ego .04 -.15 .05 -.95

Frame -.09 -.04 .22 -.39

Peak X Frame -.06 -.13 .06 .90

Ego X Frame .29 .11 .31 .93

Peak X Ego X Frame .14 .21 .08 1.64

Frame = Message Framing

MESSAGE FRAMING Framing 119

Table 12

Weekly Drinking Intentions a Function of Drinks per week, Ego-depletion and Message Framing

b β SE t

Drinks per week .75*** .90*** .06 11.82***

Ego-depletion -.37 -.04 .51 -.72

Drinks X Ego -.22* -.17* .09 -2.42*

Frame .37 .04 .55 .66

Drinks X Frame -.09 -.06 .10 -.93

Ego X Frame .29 .03 .80 .36

Drinks X Ego X Frame .30+ .12+ .16 1.85+

Frame = Message Framing; *** p<.0001; * p<.05 + Marginally significant

MESSAGE FRAMING Framing 120

Table 13

Typical Drinking Intentions a Function of Typical Drinks, Ego-depletion and Message Framing

b β SE t

Typical Drinks .60 1.12 .08 7.70***

Ego-depletion -.46 -.10 .37 -1.24

Typical X Ego -.51 -.76 .09 -5.54***

Frame -.71 -.16 .40 -1.77

Typical X Frame -.46 -.47 .12 -3.93***

Ego-depletion X Frame .32 .06 .57 .57

Typical X Ego X Frame .46 .34 .15 3.04**

Frame = Message Framing; *** p<.0001;** p<.01; * p<.05

MESSAGE FRAMING Framing 121

Table 14 Correlation and Descriptive Statistics: Study 2:Theory of Reactance Measure 1 2 3 4 5 6 7 8 9

1. Drinks per __ week 2. Peak drinks .73*** __

3. Typical drinks .50*** .63*** __

4. Message -.16* -.12 -.09 __ Rating 5. Intention- .75*** .59*** .34*** -.12 __ Week 6. Intention- .44*** .58*** .42*** -.06 .70*** __ Typical 7. Self-esteem .03 -.05 -.08 .11 .09 .06 __

8. Instability of .08 .12 .13 .07 -.03 .08 -.50*** __ Self-esteem 9. Drinker .46*** .38*** .27*** -.05 .34*** .20 -.18* .14+ __ Identity Mean 6.42 4.65 3.83 4.96 4.73 2.32 3.97 1.44 .90

SD 7.82 4.36 4.33 1.10 6.35 2.14 .77 .87 1.26

Note. *** p < .0001; ** p<.001; *p<.05.; +: Marginally significant

MESSAGE FRAMING Framing 122

Table 15

Study 2: Message Rating as a Function of Framing, Statistical Information and Drinks per week.

b β SE t

Model A

Frame 0 0 .15 -.02

Stats -.08 -.04 .15 -.56

Drinks per week -.02* -.16* .01 -2.3*

Model B

Frame -.13 -.06 .21 -.60

Stats -.22 -.10 .22 -1.0

Frame*Stats .26 -.11 .31 .85

Drinks per week -.02* -.16* .01* -2.25*

Model C

Frame -.13 -.06 .21 -.61

Stats -.25 -.11 .22 -1.11

Frame*Stats .28 .12 .31 .92

Drinks per week -.03 -.19 .02 -1.54

Frame*Drinks 0 -.01 .03 -.05

Stats*Drinks .02 .12 .02 .91

Frame*Stats*Drinks -.03 -.10 .04 -.70

Frame = Message Framing; Stats= Existence of Statistical Information. * p<.05

MESSAGE FRAMING Framing 123

Table 16

Study 2: Message Rating as a Function of Framing, Statistical Information and Peak Drinks

b β SE t

Model A

Frame .03 .01 .15 .19

Stats -.11 -.05 .15 -.71

Peak Drinks -.03 -.12 .02 -1.79

Model B

Frame -.13 -.06 .21 -.59

Stats -.28 -.13 .22 -1.25

Frame*Stats .32 .13 .31 1.06

Peak Drinks -.03 -.12 .02 -1.76

Model C

Frame -.14 -.07 .21 -.60

Stats -.29 -.13 .22 -1.3

Frame*Stats .35 .14 .31 1.12

Peak Drinks -.03 -.10 .04 -.72

Frame*Peak Drinks -.03 -.08 .05 -.60

Stats*Peak Drinks .01 .04 .05 .25

Frame*Stats*Peak Drinks .01 .01 .07 .08

Frame = Message Framing; Stats= Existence of Statistical Information.

MESSAGE FRAMING Framing 124

Table 17

Study 2: Message Rating as a Function of Framing, Statistical Information and Typical Drinks

b β SE t

Model A

Frame .03 .01 .15 .20

Stats -.10 -.05 .16 -.64

Typical Drinks -.02 -.08 .02 -1.20

Model B

Frame -.13 -.06 .21 -.60

Stats -.27 -.12 .22 -1.22

Frame*Stats .33 .14 .31 1.08

Typical Drinks -.02 -.08 .02 -1.18

Model C

Frame -.15 -.07 .22 -.70

Stats -.27 -.12 .22 -1.18

Frame*Stats .34 .14 .32 1.10

Typical Drinks .03 .11 .03 .79

Frame*Typical Drinks -.06 -.15 .07 -.94

Stats*Typical Drinks -.06 -.20 .05 -1.38

Frame*Stats*Typical Drinks .06 .11 .08 .69

Frame = Message Framing; Stats= Existence of Statistical Information.

MESSAGE FRAMING Framing 125

Table 18

Study 2: Weekly Drinking Intention as a Function of Framing, Statistical Information and Drinks per week

b β SE t

Model A

Frame .16 .01 .59 .27

Stats .01 0 .59 .01

Drinks per week .61*** .75*** .04 16.13***

Model B

Frame 1.10 .09 .81 1.35

Stats 1.03 .08 .85 1.21

Frame*Stats -1.97 -.14 1.18 -1.61

Drinks per week .61*** .75*** .04 16.10***

Model C

Frame 1.19 .09 .80 1.49

Stats .88 .07 .84 1.05

Frame*Stats -2.03 -.14 1.16 -1.75

Drinks per week .58*** .71*** .07 8.73***

Frame*Drinks .19 .14 .12 1.59

Stats*Drinks .13 .12 .09 1.38

Frame*Stats*Drinks -.49** -.29** .16 -3.14**

Frame = Message Framing; Stats= Existence of Statistical Information. **p<.001; ***p<.0001.

MESSAGE FRAMING Framing 126

Table 19

Weekly Drinking Intention as a Function of Framing, Statistical Information and Peak Drinks

b β SE t

Model A

Frame -.03 0 .73 -.05

Stats -.02 0 .73 -.03

Peak Drinks .86*** .59*** .08 10.35***

Model B

Frame 1.10 .09 1.0 1.11

Stats 1.22 .10 1.04 1.17

Frame*Stats -2.39 -.17 1.44 -1.65

Peak Drinks .86*** .59*** .08 10.35***

Model C

Frame 1.23 .10 1.0 1.23

Stats 1.14 .09 1.04 1.09

Frame*Stats -2.45 -.17 1.44 -1.7

Peak Drinks .71*** .49*** .16 4.38***

Frame*Peak Drinks .42 .19 .25 1.67

Stats*Peak Drinks .25 .13 .22 1.13

Frame*Stats*Peak -.72* -.24* .34 -2.13*

Drinks

Frame = Message Framing; Stats= Existence of Statistical Information. ***p<.0001; *p<.05

MESSAGE FRAMING Framing 127

Table 20

Study 2: Typical Drinking Intention as a Function of Framing, Statistical Information and

Typical Drinks

b β SE t

Model A

Frame .08 .02 .28 .28

Stats -.09 -.02 .28 -.33

Typical Drinks .21*** .42*** .03 6.48***

Model B

Frame .53 .13 .38 1.41

Stats .41 .10 .40 1.03

Frame*Stats -.96 -.20 .55 -1.76

Typical Drinks .21*** .42*** .03 6.47***

Model C

Frame .75 .18 .38 1.97*

Stats .51 .12 .39 1.31

Frame*Stats -1.29* -.27* .54 -2.37*

Typical Drinks .31*** .63*** .06 5.19***

Frame*Typical Drinks .15 .18 .12 1.28

Stats*Typical Drinks -.20* -.33* .08 -2.57*

Frame*Stats*Typical Drinks -.11 -.12 .14 -.81

Frame = Message Framing; Stats= Existence of Statistical Information. ***p<.0001; *p<.05

MESSAGE FRAMING Framing 128

Table 21

Message Rating as a Function of Statistical Information, Message Framing, Self-esteem and

Instability of Self-esteem, and Drinks per week

b β SE t

Model A

Stats -.02 -.01 .13 - .16

Frame 0 0 .15 .01

Self-esteem .26* .19* .15 2.39*

Instability of Self-esteem .21* .17* .11 2.15*

Drinks per week -.03* .18* .10 -2.59*

*p<.05

MESSAGE FRAMING Framing 129

Table 22

Message Rating as a Function of Message Framing, Self-esteem and Instability of Self-esteem, with Statistical Information and Drinks per week controlled

b β SE t

Model B

Stats -.01 0 .15 -0.05

Frame 0 0 .15 0

Self-esteem .38* .27* .17 2.27*

Instability of Self-esteem .33* .27* .15 2.18*

Drinks per week -.03* -.18* .01 -2.6*

Frame*Self-esteem -.24 -.12 .23 -1.05

Frame*Instability of Self-esteem -.22 -.13 .20 -1.08

Self-esteem*Instability of Self-esteem -.22+ -.13+ .11 -1.89+

*p<.05;+: Marginally Significant

MESSAGE FRAMING Framing 130

Table 23

Message Rating as a Function of Message Framing, Self-esteem and Instability of Self-esteem, with Statistical Information and Drinks per week controlled (Continued)

b β SE t

Model C

Stats -.01 0 .15 -0.07

Frame -.06 -0.03 .17 -0.35

Self-esteem .36* 0.25* .17 2.12*

Instability of Self-esteem .31* 0.25* .16 2.01*

Drinks per week -.02* -0.18* .01 -2.57*

Frame*Self-esteem -.24 -0.12 .23 -1.04

Frame*Instability of Self-esteem -.21 -0.12 .20 -1.03

Self-esteem*Instability of Self-esteem -.09 -0.06 .20 -0.45

Frame*Self-esteem*Instability of Self-esteem -.19 -0.10 .24 -0.77

*p<.05

MESSAGE FRAMING Framing 131

Table 24

Message Rating as a Function of Statistical Information, Message Framing, Self-esteem and

Instability of Self-esteem, with Peak Drinks controlled

b β SE t

Model A

Stats -.05 -.02 .15 -.32

Frame .03 .02 .15 .22

Self-esteem .26* .18* .11 2.30*

Instability of Self-esteem .21* .16* .10 2.08*

Peak Drinks -.03*+ -.14*+ .02 -1.97*+

* p<.05; *+ p =.05

MESSAGE FRAMING Framing 132

Table 25

Message Rating as a Function of Message Framing, Self-esteem and Instability of Self-esteem, with Statistical Information and Peak Drinks controlled

b β SE t

Model B

Stats -.03 -.01 .15 -.19

Frame .03 .01 .15 .19

Self-esteem .39* .27* .17 2.3*

Instability of Self-esteem .33* .26* .16 2.12*

Peak Drinks -.04* -.14* .02 -2.07*

Frame*Self-esteem -.27 -.14 .23 -1.16

Frame*Instability of Self-esteem -.21 -.13 .20 -1.05

Self-esteem*Instability of Self-esteem -.23* -.14* .12 -2.02*

* p<.05

MESSAGE FRAMING Framing 133

Table 26

Message Rating as a Function of Message Framing, Self-esteem and Instability of Self-esteem, with Statistical Information and Peak Drinks controlled (Continued)

b β SE t

Model C

Stats -.03 -.01 .15 -.21

Frame -.03 -.01 .17 -.17

Self-esteem .37* .26* .17 2.15*

Instability of Self-esteem .31*+ .25*+ .16 1.96*+

Peak Drinks -.04* -.14* .02 -2.05*

Frame*Self-esteem -.26 -.14 .23 -1.16

Frame*Instability of Self-esteem -.20 -.12 .20 -1.00

Self-esteem*Instability of Self-esteem -.11 -.07 .20 -.53

Frame*Self-esteem*Instability of Self-esteem -.19 -.10 .25 -.76

* p<.05; *+: p=.05

MESSAGE FRAMING Framing 134

Table 27

Message Rating as a Function of Statistical Information, Message Framing, Self-esteem and

Instability of Self-esteem, with Typical Drinks controlled

b β SE t

Model A

Stats -.04 -.02 .16 -.26

Frame .03 .02 .15 .23

Self-esteem .25* .18* .11 2.25*

Instability of Self-esteem .20* .16* .10 2.00*

Typical Drinks -.02 -.09 .02 -1.33

* p<.05; *+ p =.05

MESSAGE FRAMING Framing 135

Table 28

Message Rating as a Function of Message Framing, Self-esteem and Instability of Self-esteem, with Statistical Information and Typical Drinks controlled

b β SE t

Model B

Stats -.02 -.01 .16 -.15

Frame .03 .02 .15 .22

Self-esteem .38* .27* .17 2.26*

Instability of Self-esteem .33* .26* .16 2.08*

Typical Drinks -.02 -.09 .02 -1.29

Frame*Self-esteem -.27 -.14 .23 -1.15

Frame*Instability of Self-esteem -.22 -.13 .20 -1.08

Self-esteem*Instability of Self-esteem -.22+ -.13+ .12 -1.85+

* p<.05; +: Marginally Significant

MESSAGE FRAMING Framing 136

Table 29

Message Rating as a Function of Message Framing, Self-esteem and Instability of Self-esteem, with Statistical Information and Typical Drinks controlled (Continued)

b β SE t

Model C

Stats -.03 -.01 .16 -.18

Frame -.02 -.01 .17 -.10

Self-esteem .37* .26* .17 2.13*

Instability of Self-esteem .31*+ .24*+ .16 1.93*+

Typical Drinks -.02 -.09 .02 -1.20

Frame*Self-esteem -.26 -.14 .23 -1.15

Frame*Instability of Self-esteem -.21 -.13 .20 -1.03

Self-esteem*Instability of Self-esteem -.10 -.06 .20 -.51

Frame*Self-esteem*Instability of Self-esteem -.17 -.09 .25 -.67

* p<.05; *+ p =.05

MESSAGE FRAMING Framing 137

Table 30

Weekly Drinking Intention as a Function of Statistical Information, Message Framing, Self- esteem and Instability of Self-esteem, with Drinks per week controlled

b β SE t

Model A

Stats -.07 -.01 .60 -.12

Frame .08 .01 .59 .13

Self-esteem .26 .03 .43 .61

Instability of Self-esteem -.51 -.07 .38 -1.34

Drinks per week .61 .76 .04 16.20***

*** p<.0001;

MESSAGE FRAMING Framing 138

Table 31

Weekly Drinking Intention as a Function of Message Framing, Self-esteem and Instability of

Self-esteem, with Statistical Information and Drinks per week controlled

b β SE t

Model B

Stats -.01 0 .59 -.01

Frame .06 .01 .59 .11

Self-esteem -.76 -.09 .65 -1.16

Instability of Self-esteem -.62 -.09 .60 -1.04

Drinks per week .61*** .76*** .04 16.28***

Frame*Self-esteem 1.81* .16* .88 2.05*

Frame*Instability of Self-esteem .15 .02 .78 .19

Self-esteem*Instability of Self-esteem -.21 -.02 .45 -.48

*** p<.0001; *p <.05

MESSAGE FRAMING Framing 139

Table 32

Weekly Drinking Intention as a Function of Message Framing, Self-esteem and Instability of

Self-esteem, with Statistical Information and Drinks per week controlled (Continued)

b β SE t

Model C

Stats 0 0 .60 0

Frame .18 .01 .66 .27

Self-esteem -.72 -.09 .66 -1.09

Instability of Self-esteem -.58 -.08 .61 -.95

Drinks per week .61*** .75*** .04 16.23***

Frame*Self-esteem 1.81* .16* .89 2.04*

Frame*Instability of Self-esteem .13 .01 .78 .16

Self-esteem*Instability of Self-esteem -.46 -.05 .78 -.59

Frame*Self-esteem*Instability of Self-esteem .37 .03 .95 .39

*** p<.0001; *p <.05

MESSAGE FRAMING Framing 140

Table 33

Weekly Drinking Intention as a Function of Statistical Information, Message Framing, Self- esteem and Instability of Self-esteem, with Peak Drinks controlled

b β SE t

Model A

Stats -.05 0 .73 -.07

Frame -.14 -.01 .72 -.19

Self-esteem .75 .09 .53 1.44

Instability of Self-esteem -.42 -.06 .47 -.90

Peak Drinks .88*** .60*** .08 10.55***

*** p<.0001

MESSAGE FRAMING Framing 141

Table 34

Weekly Drinking Intention as a Function of Message Framing, Self-esteem and Instability of

Self-esteem, with Statistical Information and Peak Drinks controlled

b β SE t

Model B

Stats -.02 0 .73 -.03

Frame -.14 -.01 .72 -.19

Self-esteem -.16 -.02 .80 -.20

Instability of Self-esteem -.36 -.05 .73 -.49

Peak Drinks .87*** .60*** .08 10.51***

Frame*Self-esteem 1.70 .15 1.08 1.57

Frame*Instability of Self-esteem -.11 -.01 .96 -.12

Self-esteem*Instability of Self-esteem .16 .02 .55 .29

*** p<.0001

MESSAGE FRAMING Framing 142

Table 35

Weekly Drinking Intention as a Function of Message Framing, Self-esteem and Instability of

Self-esteem, with Statistical Information and Peak Drinks week controlled (Continued)

b β SE t

Model C

Stats -.02 0 .73 -.02

Frame .04 0 .81 .05

Self-esteem -.10 -.01 .81 -.12

Instability of Self-esteem -.30 -.04 .75 -.40

Peak Drinks .87*** .60*** .08 10.48***

Frame*Self-esteem 1.69 .15 1.09 1.56

Frame*Instability of Self-esteem -.14 -.01 .96 -.15

Self-esteem*Instability of Self-esteem -.24 -.03 .96 -.25

Frame*Self-esteem*Instability of Self-esteem .59 .05 1.16 .51

* p<.05;

MESSAGE FRAMING Framing 143

Table 36

Weekly Drinking Intention as a Function of Statistical Information, Message Framing, Self- esteem and Instability of Self-esteem, with Typical Drinks controlled

b β SE t

Model A

Stats -.15 -.01 .86 -.18

Frame -.26 -.02 .85 -.31

Self-esteem .86 .10 .61 1.40

Instability of Self-esteem -.19 -.03 .55 -.35

Typical Drinks .52*** .35*** .10 5.25***

*** p<.0001

MESSAGE FRAMING Framing 144

Table 37

Weekly Drinking Intention as a Function of Message Framing, Self-esteem and Instability of

Self-esteem, with Statistical Information and Typical Drinks controlled

b β SE t

Model B

Stats -.08 -.01 .86 -.09

Frame -.28 -.02 .85 -.33

Self-esteem -.13 -.02 .94 -.13

Instability of Self-esteem -.24 -.03 .86 -.28

Typical Drinks .52*** .35*** .10 5.20***

Frame*Self-esteem 1.74 .16 1.27 1.37

Frame*Instability of Self-esteem .04 .00 1.12 .04

Self-esteem*Instability of Self-esteem -.26 -.03 .64 -.41

*** p<.0001

MESSAGE FRAMING Framing 145

Table 38

Weekly Drinking Intention as a Function of Message Framing, Self-esteem and Instability of

Self-esteem, with Statistical Information and Typical Drinks week controlled (Continued)

b β SE t

Model C

Stats -.08 -.01 .87 -.09

Frame -.25 -.02 .95 -.27

Self-esteem -.12 -.01 .95 -.12

Instability of Self-esteem -.23 -.03 .88 -.26

Typical Drinks .52*** .35*** .10 5.15***

Frame*Self-esteem 1.74 .16 1.27 1.37

Frame*Instability of Self-esteem .04 .00 1.12 .03

Self-esteem*Instability of Self-esteem -.32 -.03 1.12 -.28

Frame*Self-esteem*Instability of Self-esteem .08 .01 1.37 .06

*** p<.0001

MESSAGE FRAMING Framing 146

Table 39

Typical Drinking Intention as a Function of Statistical Information, Message Framing, Self- esteem and Instability of Self-esteem, with Typical Drinks controlled

b β SE t

Model A

Stats -.02 -.01 .28 -.09

Frame .07 .02 .28 .27

Self-esteem .37+ .13+ .20 1.84+

Instability of Self-esteem .21 .09 .18 1.19

Typical Drinks .21*** .42*** .03 6.42***

*** p<.0001; + Marginally Significant

MESSAGE FRAMING Framing 147

Table 40

Typical Drinking Intention as a Function of Message Framing, Self-esteem and Instability of

Self-esteem, with Statistical Information and Typical Drinks controlled

b β SE t

Model B

Stats -.02 0 .28 -.07

Frame .07 .02 .28 .25

Self-esteem -.07 -.02 .30 -.22

Instability of Self-esteem -.01 -.01 .28 -.05

Typical Drinks .21*** .42*** .03 6.38***

Frame*Self-esteem .79+ .21+ .41 1.91+

Frame*Instability of Self-esteem .38 .12 .36 1.05

Self-esteem*Instability of Self-esteem .12 .04 .21 .55

*** p<.0001; + Marginally Significant

MESSAGE FRAMING Framing 148

Table 41

Weekly Drinking Intention as a Function of Message Framing, Self-esteem and Instability of

Self-esteem, with Statistical Information and Typical Drinks week controlled (Continued)

b β SE t

Model C

Stats -.03 -.01 .28 -.09

Frame -.02 0 .31 -.05

Self-esteem -.10 -.03 .31 -.31

Instability of Self-esteem -.05 -.02 .29 -.16

Typical Drinks .21*** .42*** .03 6.40***

Frame*Self-esteem .79+ .21+ .41 1.91+

Frame*Instability of Self-esteem .40 .12 .37 1.09

Self-esteem*Instability of Self-esteem .31 .10 .37 .84

Frame*Self-esteem*Instability of Self-esteem -.28 -.08 .45 -.64

*** p<.0001; + Marginally Significant

MESSAGE FRAMING Framing 149

Table 42

Message Rating as a Function of Statistical Information, Framing, Self-relevance and Drinks per week

b β SE t

Model A

Stats -.09 -.04 .15 -.58

Frame .00 .00 .15 0

Self-relevance .02 .03 .07 .36

Drinks per week -.02* -.17* .01 -2.21*

Model B

Stats -.24 -.11 .22 -1.06

Frame -.13 -.06 .21 -.59

Stats*Frame .28 .11 .31 .89

Self-relevance -.04 -.05 .13 -.34

Stats*Self-relevance .14 .13 .18 .80

Frame*Self-relevance .04 .03 .19 .20

Stats*Frame*Self-relevance -.10 -.07 .25 -.40

Drinks per week -.02* -.18* .01 -2.22*

Frame = Message Framing; Stats= Existence of Statistical Information.

MESSAGE FRAMING Framing 150

Table 43

Message Rating as a Function of Statistical Information, Framing, Self-relevance and Peak

Drinks

b β SE t

Model A

Stats -.11 -.05 .15 -.71

Frame .03 .01 .15 .20

Self-relevance .01 .01 .07 .09

Peak Drinks -.03 -.13 .02 -1.69

Model B

Stats -.29 -.13 .22 -1.30

Frame -.13 -.06 .22 -.61

Stats*Frame .35 .14 .31 1.10

Self-relevance -.05 -.06 .13 -.39

Stats*Self-relevance .13 .11 .18 .71

Frame*Self-relevance .02 .02 .20 .11

Stats*Frame*Self-relevance -.09 -.06 .25 -.34

Peak Drinks -.03 -.13 .02 -1.66

Frame = Message Framing; Stats= Existence of Statistical Information.

MESSAGE FRAMING Framing 151

Table 44

Message Rating as a Function of Statistical Information, Framing, Self-relevance and

Typical Drinks

b β SE t

Model A

Stats -.10 -.04 .16 -.61

Frame .03 .01 .16 .19

Self-relevance -.02 -.02 .06 -.29

Typical Drinks -.02 -.08 .02 -1.08

Model B

Stats -.28 -.13 .23 -1.25

Frame -.14 -.06 .22 -.64

Stats*Frame .36 .14 .31 1.14

Self-relevance -.07 -.08 .13 -.53

Stats*Self-relevance .11 .10 .18 .63

Frame*Self-relevance .01 .01 .20 .04

Stats*Frame*Self-relevance -.06 -.04 .25 -.23

Typical Drinks -.02 -.08 .02 -1.05

Frame = Message Framing; Stats= Existence of Statistical Information.

MESSAGE FRAMING Framing 152

Table 45

Weekly Drinking Intention as a Function of Statistical Information, Message Framing, Self- relevance and Drinks per week

b β SE t

Model A

Stats .02 .00 .60 .03

Frame .15 .01 .60 .26

Self-relevance -.06 -.01 .26 -.23

Drinks per week .62*** .76*** .04 14.44***

Model B

Stats .82 .06 .85 .96

Frame 1.12 .09 .81 1.38

Stats*Frame -1.76 -.12 1.18 -1.49

Self-relevance -.65 -.13 .50 -1.31

Stats*Self-relevance 1.70* .26* .67 2.53*

Frame*Self-relevance .74 .11 .73 1.01

Stats*Frame*Self-relevance -2.22* -.25* .95 -2.33*

Drinks per week .59*** .73*** .04 13.98***

Frame = Message Framing; Stats= Existence of Statistical Information. ***p<.0001; *p<.05

MESSAGE FRAMING Framing 153

Table 46

Weekly Drinking Intention as a Function of Statistical Information, Framing, Self-relevance and Peak Drinks

b β SE t

Model A

Stats -.15 -.01 .72 -.21

Frame .04 0 .72 .06

Self-relevance .66* .13* .31 2.14*

Peak Drinks .79*** .54*** .09 8.89***

Model B

Stats .76 .06 1.01 .75

Frame 1.23 .10 .97 1.26

Stats*Frame -2.06 -.14 1.41 -1.47

Self-relevance -.46 -.09 .60 -.77

Stats*Self-relevance 2.73*** .42*** .80 3.43***

Frame*Self-relevance 1.17 .17 .88 1.33

Stats*Frame*Self-relevance -3.25* -.37* 1.13 -2.87*

Peak Drinks .77*** .53*** .09 8.86***

Frame = Message Framing; Stats= Existence of Statistical Information.

MESSAGE FRAMING Framing 154

Table 47

Weekly Drinking Intention as a Function of Statistical Information, Framing, Self-relevance and Typical Drinks

b β SE t

Model A

Stats -.38 -.03 .82 -.46

Frame -.02 .00 .82 -.02

Self-relevance 1.33 .26 .33 3.99

Drinks per week .40 .28 .10 4.12

Model B

Stats .66 .05 1.15 .57

Frame 1.36 .11 1.10 1.23

Stats*Frame -2.37 -.17 1.59 -1.49

Self-relevance .08 .02 .68 .12

Stats*Self-relevance 3.03*** .47 .90*** 3.36***

Frame*Self-relevance 1.39* .20 1.00* 1.39*

Stats*Frame*Self-relevance -3.83* -.44* 1.28 -2.98*

Typical Drinks .40*** .27 .10 4.19***

Frame = Message Framing; Stats= Existence of Statistical Information.

MESSAGE FRAMING Framing 155

Table 48

Typical Drinking as a Function of Statistical Information, Framing, Self-relevance and

Typical Drinks

b β SE t

Model A

Stats -.12 -.03 .28 -.43

Frame .10 .02 .28 .35

Self-relevance .16 .10 .11 1.44

Typical Drinks .20*** .40*** .03 5.92***

Model B

Stats .33 .08 .40 .83

Frame .60 .14 .38 1.58

Stats*Frame -.94 -.20 .55 -1.72

Self-relevance .10 .06 .23 .40

Stats*Self-relevance .30 .14 .31 .98

Frame*Self-relevance .27 .12 .34 .79

Stats*Frame*Self-relevance -.76 -.26 .44 -1.72

Typical Drinks .19*** .39*** .03 5.84***

Frame = Message Framing; Stats= Existence of Statistical Information.

MESSAGE FRAMING Framing 156

Figure 1 Power Analysis Diagram

MESSAGE FRAMING Framing 157

Figure 2

Two-way Interaction Predicting Message Rating by Both Ego-depletion and Peak Drinking on Table 8

5.3 Low Peak Drinking 5.2 High Peak Drinking

5.1 .17† 5

4.9

4.8

4.7

4.6 -.18†

4.5

4.4 Control Group Ego-depletion

+: Marginally significant

MESSAGE FRAMING Framing 158

Figure 3

The Graph for the Two-way interaction between Drinks per week and Frames on Table 10

5.2 .17 5.1 DV=Message Rating 5

4.9 Low Drink 4.8 High Drink

4.7 -.10

4.6

4.5 Loss Frame Gain Frame

MESSAGE FRAMING Framing 159

Figure 4 The Graph for the Two-way Interactions between Drinks per week and Ego-depletion on

Table 12

Outcome =Weekly Dirnking Intentions 10 .85*** 8 .74*** 6

4

2

0 Low Drink High Drink

Low Ego High Ego

*** p <.0001

MESSAGE FRAMING Framing 160

Figure 5 The Graph Depicting the Three-way interaction on Table 13. The Major Dimensions are

Typical Drinks, Ego-depletion and Message Framing with Outcome of Typical Drinking

Intention

6 1.12***

5

4

3 Loss Frame low ego .18* Loss Frame high ego 2

1

0 low drink high drink

6

5

4

3 .26 Gain Frame low ego Gain Frame high ego 2 .18

1

0 low drink high drink

*** p <.0001; * p<.05

MESSAGE FRAMING Framing 161

Figure 6 The Graph Depicting the Three-way interaction on Table 18.The Major Dimensions are

Stats, Frame and Drinks per week, with Outcome of Weekly Drinking Intention

12 .50*** DV= Weekly 10 Intention

8

6 No-stats .18* Stats 4

2 Gain Frame 0 Low Drink High Drink -2

12 .87*** 10

8 .71***

6 No-stats

4 Stats

2

0 Loss Frame Low Drink High Drink -2

*** p <.0001; * p<.05

MESSAGE FRAMING Framing 162

Figure 7

The Graph Depicting the Three-way interaction on Table 19.The Major Dimensions are

Stats, Frame and Peak Drinks, with Outcome of Weekly Drinking Intention

DV=Weekly Drinking Intention Gain Frame 12 .77*** 10

8

6 .45***

4

2

0 Low Peak Drinks High Peak Drinks

No-stats Stats

DV=Weekly Drinking Intention Loss Frame 12 .66*** 10

8

6 .49*** 4

2

0 Low Peak Drinks High Peak Drinks

No-stats Stats

*** p <.0001

MESSAGE FRAMING Framing 163

Figure 8

The Significant Pair of Two-way Interactions on Table 20 with the outcome of Typical

Drinking Intention

DV=Typical Drinking Intention 4.5 4 3.5 .12 3 2.5 2 1.5 -.18* 1 0.5 0 No Stats Stats

Gain Loss

DV=Typical Drinking Intention 4.5 4 .78*** 3.5 3 2.5 .26**+ 2 1.5 1 0.5 0 Low Typical Drinks High Typical Drinks

No Stats Stats

**+ p=.001; ***p<.0001; **p<.001

MESSAGE FRAMING Framing 164

Figure 9

The Two-way Interaction between Instability of Self-esteem and Self-esteem with Outcome of

Message Rating on Table 25

DV =Message Rating 5.5 .03 5.3 5.1 4.9 .32* 4.7 4.5 4.3 4.1 3.9 3.7 3.5 Low Self-esteem High Self-esteem

Low Instability of Self-esteem High Instability of Self-esteem

* p<.05

MESSAGE FRAMING Framing 165

Figure 10 The Two-way Interaction between Self-esteem and Message Framing with Outcome of

Message Rating on Table 31

DV=Weekly Drinking Intention 6 .14*+ 5

4

3 -.10

2

1

0 Low Self-esteem High Self-esteem

Gain Frame Loss Frame

*+ p=.05

MESSAGE FRAMING Framing 166

Figure 11

The Two-way Interaction between Self-esteem and Message Framing with Outcome of

Message Rating on Table 32

DV=Weekly Drinking Intention 6

5 .14*+

4

3 -.10

2

1

0 Low Self-esteem High Self-esteem

Gain Frame Loss Frame

*+ p=.05

MESSAGE FRAMING Framing 167

Figure 12

The Three-way Interaction, Stats X Frame X Self-relevance with Outcome of Weekly

Drinking Intention on Table 45

DV=Weekly Drinking Intention; Gain Frame 7

6 .02 5

4

3 -.09

2

1

0 Low Self-relevance High Self-relevance

No Stats Stats

DV=Weekly Drinking Intention; Loss Frame 7

6 .21*

5

4 3 -.13 2

1

0 Low Self-relevance High Self-relevance

No Stats Stats

*p<.05

MESSAGE FRAMING Framing 168

Figure 13

The Three-way Interaction, Stats X Frame X Self-relevance with Outcome of Weekly

Drinking Intention on Table 46

DV=Weekly Drinking Intention; Gain Frame 8

7 .14 6 5 4 .04 3 2 1 0 Low Self-relevance High Self-relevance

Stats No Stats

DV=Weekly Drinking Intention; Loss Frame 8 7 .46*** 6 5 4 -.09 3 2 1 0 Low Self-relevance High Self-relevance

Stats No Stats

***p<.0001

MESSAGE FRAMING Framing 169

Figure 14

The Three-way Interaction, Stats X Frame X Self-relevance with Outcome Weekly Drinking

Intention on Table 47

DV=Weekly Drinking Intention; Gain Frame 9 8 .29* 7 6 5 4 .13 3 2 1 0 Low Self-relevance High Self-relevance

Stats No stats

DV=Weekly Drinking Intention; Loss Frame 9 8 .62*** 7 6 5 4 .02 3 2 1 0 Low Self-relevance High Self-relevance

Stats No stats

***p<.0001; *p<.05

MESSAGE FRAMING Framing 170

Appendix A: Health Messages

Citation: Exemplary Messages

From Meyerowitz and Chaiken (1987)

Gain-framed Message

By doing BSE (Breast Self-Examination) now, you can learn what your normal, healthy breasts feel like so that you will be better prepared to notice any small, abnormal changes that might occur as you get older.

Research shows that women who do BSE have an increased chance of finding a tumor in the early, more treatable stage of the disease.

Loss-framed Message.

By not doing BSE (Breast Self-Examination) now, you will not learn what your normal, healthy breasts feel like so that you will be ill-prepared to notice any small, abnormal changes that might occur as you get older.

Research shows that women who do not do BSE have a decreased chance of finding a tumor in the early, more treatable stage of the disease.

MESSAGE FRAMING Framing 171

Appendix A: Tested Health Messages

Message Source

To serve the experimental stimulation, the messages are taken from website NIHHH and

American Heart Association et al. Study 1 (Ego Depletion) used the messages with statistical information only while Study 2 (Reactance) adopted all the messages, while http://rethinkingdrinking.niaaa.nih.gov/WhatsTheHarm/WhatAreTheRisks.asp http://www.heart.org/HEARTORG/Conditions/Alcohol-and-Cardiovascular-

Disease_UCM_305173_Article.jsp http://www.dosomething.org/tipsandtools/11-facts-about-teens-and-alcohol http://www2.potsdam.edu/hansondj/AlcoholAndHealth.html http://www.sciencedirect.com/science/article/pii/0376871696012604 http://www.cdc.gov/Features/dsAlcoholChildbearingAgeWomen/ http://www.cspinet.org/booze/fas.htm

MESSAGE FRAMING Framing 172

Appendix A: Tested Health Messages

Messages Used in the Current Dissertation

Loss Frame Message (Text Only)

You may have heard that regular light to moderate drinking can be good for the heart. With heavy or at-risk drinking, however, any potential benefits are outweighed by greater risks, including:

Injuries. Drinking too much increases your chances of being injured or even killed. Alcohol is a factor in, for example, fatal burn injuries, drowning, and homicides; as well as severe trauma injuries and sexual assaults, and in fatal motor vehicle crashes, suicides, and fatal falls.

Health problems. Heavy drinkers have a greater risk of liver disease, heart disease, sleep disorders, depression, stroke, bleeding from the stomach, sexually transmitted infections from unsafe sex, and several types of cancer. They may have problems managing diabetes, high blood pressure, and other conditions. For additional information, every year, many college students require medical treatment after overdosing on alcohol.

Birth defects. Drinking during pregnancy can cause brain damage and other serious problems in the baby. Because it is not yet known whether any amount of alcohol is safe for a developing baby, women who are pregnant or may become pregnant should not drink.

Alcohol use disorders. Generally known as alcoholism and alcohol abuse, alcohol use disorders are medical conditions that doctors can diagnose when a patient's drinking causes distress or harm. In the United States, about 18 million people have an alcohol use disorder.

MESSAGE FRAMING Framing 173

Beyond these physical and mental health risks, frequent heavy drinking also is linked with personal problems, including losing a driver's license and having relationship troubles. Full- time college students were as likely to have an alcohol use disorder as nonstudents, but were more likely than part-time college students and non-college students.

MESSAGE FRAMING Framing 174

Appendix A: Tested Health Messages

Messages Used in the Current Dissertation

Loss Frame Message (Text + Statistical Information)

You may have heard that regular light to moderate drinking can be good for the heart. With heavy or at-risk drinking, however, any potential benefits are outweighed by greater risks, including:

Injuries. Drinking too much increases your chances of being injured or even killed. Alcohol is a factor, for example, in about 60% of fatal burn injuries, drowning, and homicides; as well as in 50% of severe trauma injuries and sexual assaults, and 40% of fatal motor vehicle crashes, suicides, and fatal falls.

Health problems. Heavy drinkers have a greater risk of liver disease, heart disease, sleep disorders, depression, stroke, bleeding from the stomach, sexually transmitted infections from unsafe sex, and several types of cancer. They may have problems managing diabetes, high blood pressure, and other conditions. Additionally, every year, an estimated 30,000 college students require medical treatment after overdosing on alcohol.

Birth defects. Drinking during pregnancy can cause brain damage and other serious problems in the baby. In the United States, the prevalence of fetal alcohol syndrome is estimated at 0.5 to 2.0 cases per 1,000 births, but other fetal alcohol spectrum disorders

(FASDs)* are believed to occur approximately three times as often as fetal alcohol syndrome.

MESSAGE FRAMING Framing 175

Because it is not yet known whether any amount of alcohol is safe for a developing baby, women who are pregnant or may become pregnant should not drink.

Alcohol use disorders. Generally known as alcoholism and alcohol abuse, alcohol use disorders are medical conditions that doctors can diagnose when a patient's drinking causes distress or harm. In the United States, about 18 million people have an alcohol use disorder.

Full-time college students (21%) were as likely to have an alcohol use disorder as nonstudents (19%), but were more likely than part-time college students (15%) and non- college students (12%). Beyond these physical and mental health risks, frequent heavy drinking also is linked with personal problems, including losing driver's licenses and having relationship troubles.

MESSAGE FRAMING Framing 176

Appendix A: Tested Health Messages

Messages Used in the Current Dissertation

Gain Frame Message (Texts Only)

You may have heard that regular light to moderate drinking can be good for the heart.

However, by avoiding heavy or at-risk drinking, you are able to increase and improve your potential benefits.

Health. You can maintain your health by keeping away from alcohol. Individuals who drink lightly or never drink have healthier hearts, livers, more optimistic emotions, efficient digestion, safer sex life, and sleep better. They have lower risk of problems managing diabetes, high blood pressure, and other conditions. By avoiding drinking, your body is less likely to have these health problems. Drinking alcohol in moderation was found to reduce the risk of mortality significantly according to the meta-analysis studies of alcohol and total mortality for men and women around the world.

Prevention of Birth Defects. Would you like to raise healthy children? Abstinence of alcohol during pregnancy can facilitate the healthy development of the baby’s brain.

Because it is not yet known whether any amount of alcohol is safe for a developing baby, women who are pregnant or may become pregnant should not drink.

Avoid Alcohol use disorders. You can be safe from alcohol use disorder by not drinking at all. Generally known as alcoholism and alcohol abuse, alcohol use disorders are medical conditions that doctors can diagnose when a patient's drinking causes distress or harm. In the United States, many people have an alcohol use disorder. Not drinking can prevent physical and mental health diseases, and abstinence of alcohol is linked with better

MESSAGE FRAMING Framing 177 life, including keeping one’s driver's license safe and having satisfactory relationships. The prevalence of alcohol dependence and abuse is a decreasing trend. It is worth abstaining from alcohol to keep your social life positive. If you really intend to drink and keep healthy, please keep your drinking within the most appropriate intake amount.

MESSAGE FRAMING Framing 178

Appendix A: Tested Health Messages

Messages Used in the Current Dissertation

Gain Frame Message (Texts + Statistical Information)

You may have heard that regular light to moderate drinking can be good for the heart.

However, by avoiding heavy or at-risk drinking, you are able to amplify your potential benefits.

Health. You can maintain your health by keeping away from alcohol. Individuals who drink lightly or never drink have healthier hearts and livers, more optimistic emotions, efficient digestion, safer sex life, and sleep better. They may have lower risk of problems managing diabetes, high blood pressure, and other conditions. By avoiding drinking, your body is less likely to have these health problems. Drinking alcohol in moderation (1-2 drinks per day for women and 2-4 for men) was found to reduce the risk of mortality significantly according to meta-analysis of 34 studies of alcohol and total mortality among 1,015,835 men and women around the world.

Prevention of Birth Defects. Would you like to raise healthy children? Abstinence of alcohol during pregnancy can facilitate the healthy development of the baby’s brain.

Because it is not yet known whether any amount of alcohol is safe for a developing baby, women who are pregnant or may become pregnant should not drink. A national survey found that more than half of women age 15-44 drank while pregnant. It is important for women to avoid drinking during pregnancy.

Avoid Alcohol use disorders. You can be safe from alcohol use disorder by not drinking at all. Generally known as alcoholism and alcohol abuse, alcohol use disorders are

MESSAGE FRAMING Framing 179 medical conditions that doctors can diagnose when a patient's drinking causes distress or harm. In the United States, about 18 million people have an alcohol use disorder. Not drinking can prevent potential physical and mental health diseases, and abstinence of alcohol is linked with better life, including keeping one’s driver's license safe and having satisfactory relationships. In 1996, the prevalence of alcohol dependence and abuse had decreased by 21.7%. It is worth abstaining from alcohol to keep your social life positive. If you really intend to drink and keep healthy, an average of 1 to 2 drinks per day for men and

1 drink per day for women is the most appropriate intake amount.

MESSAGE FRAMING Framing 180

Appendix B: Consent Forms

Study 1

UNIVERSITY OF HOUSTON CONSENT TO PARTICIPATE IN RESEARCH

Study 1: Alcohol Message Framing and Cognitive Resource Limitation

Primary Investigator Mr. Chun-Han Chen Faculty Sponsor: Dr. Clayton Neighbors Department of Psychology University of Houston Email: [email protected] or [email protected]

You are being invited to participate in a research project conducted by Mr. Chun-Han Chen (M.A.) in the Department of Psychology at the University of Houston.

In order to participate, you must be at least 18 years of age.

NON-PARTICIPATION STATEMENT

Your participation is voluntary and you may refuse to participate or withdraw at any time without penalty or loss of benefits to which you are otherwise entitled. You may also refuse to answer any question. A decision to participate or not or to withdraw your participation will have absolutely no effect on you.

PURPOSE

The purpose of this study is to evaluate the relationships among health communication, cognitive fatigue, and alcohol drinking. In addition, this study may help identify specific ways through which universities and health personnel can help reduce excessive undergraduate alcohol consumption. Although we will be collecting data for several weeks, your participation in the study will last less than approximately 60 minutes.

PROCEDURES

A total of approximately 400 subjects will be asked to participate in this project.

MESSAGE FRAMING Framing 181

You will be asked to complete a survey. The survey will ask you to provide basic demographic information and will also ask you to answer questions pertaining to your current and past alcohol use and related experiences. In this study you will also be asked to make imaginations and write according the instructions. Also, you would be asked to complete the relevant questionnaires, reviewing a message, and rate that health message.

CONFIDENTIALITY Every effort will be taken to protect your identity and keep your answers completely confidential. Your participation in this project is only identified by the researchers. The data will be stored according to the guidelines of the American Psychological Association in a secure computer for a minimum of 5 years, after which the data will be destroyed. Data will be stored digitally in a password-protected digital format. Your participation and record in this project will be confidential. All the material of data, i.e., flash drives, will be stored in the campus of the University of Houston.

RISKS/DISCOMFORTS Occasionally the survey questions ask about personal experiences that you may feel uncomfortable answering (e.g., How many times have you “Felt that you had a problem with alcohol?” or “Felt that you were going crazy?”). You do not have to answer every question and may discontinue the survey at any time by closing your browser window. Should you experience any feelings of discomfort, and wish to speak with someone about them, you may contact the Crisis Hotline at 1- 800-273-8255 or the UH Counseling and Psychological Services Center at (713) 743-5454.

BENEFITS While you will not directly benefit from participation, your participation may help investigators better understand the relationships among memory for alcohol-related experiences, drinking behavior, and attitudes toward drinking. In exchange for participation in this study, you will receive two hours of extra credit to be used according to your instructor’s policy. Findings could potentially be used by universities, health educators, and medical personnel to make suggestions to undergraduates entering in the college atmosphere.

ALTERNATIVES Participation in this project is voluntary and the only alternative to this project is non-participation.

PUBLICATION STATEMENT The results of this study may be published in professional and/or scientific journals. They may also be used for educational purposes or for professional presentations. However, no individual subject will be identified publicly.

OFFER TO ANSWER QUESTIONS You have the opportunity to ask and to have answered any question about this research at any point during the study. If you have questions, please email [email protected]

MESSAGE FRAMING Framing 182

SUBJECT RIGHTS 1. I understand that consent is required of all persons participating in this project.

2. All procedures have been explained to me and all my questions have been answered to my satisfaction.

3. Any risks and/or discomforts have been explained to me.

4. Any benefits have been explained to me.

5. I understand that, if I have any questions, I may contact Chun-Han Chen at [email protected]. I may also contact Dr. Clayton Neighbors at (713) 743-2616.

6. I have been told that I may refuse to participate or to stop my participation in this project at any time before or during the project. I may also refuse to answer any question.

7. ANY QUESTIONS REGARDING MY RIGHTS AS A RESEARCH SUBJECT MAY BE ADDRESSED TO THE UNIVERSITY OF HOUSTON COMMITTEE FOR THE PROTECTION OF HUMAN SUBJECTS (713-743- 9204). ALL RESEARCH PROJECTS THAT ARE CARRIED OUT BY INVESTIGATORS AT THE UNIVERSITY OF HOUSTON ARE GOVERNED BY REQUIREMENTS OF THE UNIVERSITY AND THE FEDERAL GOVERNMENT.

8. All information that is obtained in connection with this project and that can be identified with me will remain confidential as far as possible within legal limits. Information gained from this study that can be identified with me may be released to no one other than the principal investigator and his faculty sponsor. The results may be published in scientific journals, professional publications, or educational presentations without identifying me by name.

I have read the information provided above and voluntarily agree to participate in this study.

I HAVE READ (OR HAVE HAD READ TO ME) THE CONTENTS OF THIS CONSENT FORM AND HAVE BEEN ENCOURAGED TO ASK QUESTIONS. I HAVE RECEIVED ANSWERS TO MY QUESTIONS. I GIVE MY CONSENT TO PARTICIPATE IN THIS STUDY. I HAVE RECEIVED (OR WILL RECEIVE) A COPY OF THIS FORM FOR MY RECORDS AND FUTURE REFERENCE.

MESSAGE FRAMING Framing 183

Study 2

UNIVERSITY OF HOUSTON CONSENT TO PARTICIPATE IN RESEARCH

Study 2: Message Framing and Theory of Reactance on Drinking Reduction Campaign

Primary Investigator Chun-Han Chen Faculty Sponsor: Dr. Clayton Neighbors Department of Psychology University of Houston Email: [email protected] or [email protected] You are being invited to participate in a research project conducted by Chun-Han Chen (M.A.) in the Department of Psychology at the University of Houston. In order to participate, you must be at least 18 years of age.

NON-PARTICIPATION STATEMENT

Your participation is voluntary, and you may refuse to participate or withdraw at any time without penalty or loss of benefits to which you are otherwise entitled. You may also refuse to answer any question. A decision to participate or not, or to withdraw your participation, will have absolutely no effect on you.

PURPOSE

The purpose of this study is to evaluate the relationships among emotions, health promotion messages, drinking behavior, self-relevance and self-esteem. In addition, this study may help identify specific ways through which universities and health personnel can help reduce excessive undergraduate alcohol consumption. Although we will be collecting data for several weeks, your participation in the study will last less than approximately 60 minutes.

PROCEDURES

A total of approximately 200 subjects will be asked to participate in this project.

MESSAGE FRAMING Framing 184

You will be asked to complete a survey. The survey will ask you to provide basic demographic information and will also ask you to answer questions pertaining to your current and past alcohol use and related experiences. In this study you will also be asked to complete the relevant questionnaires, review a message, and rate that health message.

CONFIDENTIALITY Every effort will be taken to protect your identity and keep your answers completely confidential. Your participation in this project is only identified by the researchers. The data will be stored according to the guidelines of the American Psychological Association in a secure computer for a minimum of 5 years, after which the data will be destroyed. Data will be stored digitally in a password-protected digital format. Your participation and record in this project will be confidential. All the material of data, i.e., flash drives, will be stored in the campus of the University of Houston.

RISKS/DISCOMFORTS Occasionally the survey questions ask about personal experiences that you may feel uncomfortable answering (e.g., How many times have you “Felt that you had a problem with alcohol?” or “Felt that you were going crazy?”). You do not have to answer every question and may discontinue the survey at any time by closing your browser window. Should you experience any feelings of discomfort, and wish to speak with someone about them, you may contact the Crisis Hotline at 1- 800-273-8255 or the UH Counseling and Psychological Services Center at (713) 743-5454.

BENEFITS While you will not directly benefit from participation, your participation may help investigators better understand the relationships among memory for alcohol-related experiences, drinking behavior, and attitudes toward drinking. In exchange for participation in this study, you will receive two hours of extra credit to be used according to your instructor’s policy. Findings could potentially be used by universities, health educators, and medical personnel to make suggestions to undergraduates entering in the college atmosphere.

ALTERNATIVES Participation in this project is voluntary and the only alternative to this project is non-participation.

PUBLICATION STATEMENT The results of this study may be published in professional and/or scientific journals. They may also be used for educational purposes or for professional presentations. However, no individual subject will be identified publicly.

MESSAGE FRAMING Framing 185

OFFER TO ANSWER QUESTIONS You have the opportunity to ask and to have answered any question about this research at any point during the study. If you have questions, please email [email protected]

SUBJECT RIGHTS 1. I understand that consent is required of all persons participating in this project.

2. All procedures have been explained to me and all my questions have been answered to my satisfaction.

3. Any risks and/or discomforts have been explained to me.

4. Any benefits have been explained to me.

5. I understand that, if I have any questions, I may contact Mr. Chun-Han Chen at [email protected]. I may also contact Dr. Clayton Neighbors at (713) 743-2616.

6. I have been told that I may refuse to participate or to stop my participation in this project at any time before or during the project. I may also refuse to answer any question.

7. ANY QUESTIONS REGARDING MY RIGHTS AS A RESEARCH SUBJECT MAY BE ADDRESSED TO THE UNIVERSITY OF HOUSTON COMMITTEE FOR THE PROTECTION OF HUMAN SUBJECTS (713-743- 9204). ALL RESEARCH PROJECTS THAT ARE CARRIED OUT BY INVESTIGATORS AT THE UNIVERSITY OF HOUSTON ARE GOVERNED BY REQUIREMENTS OF THE UNIVERSITY AND THE FEDERAL GOVERNMENT.

8. All information that is obtained in connection with this project and that can be identified with me will remain confidential as far as possible within legal limits. Information gained from this study that can be identified with me may be released to no one other than the principal investigator and his faculty sponsor. The results may be published in scientific journals, professional publications, or educational presentations without identifying me by name.

I have read the information provided above and voluntarily agree to participate in this study.

I HAVE READ (OR HAVE HAD READ TO ME) THE CONTENTS OF THIS CONSENT FORM AND HAVE BEEN ENCOURAGED TO ASK QUESTIONS. I HAVE RECEIVED ANSWERS TO MY QUESTIONS. I GIVE MY CONSENT TO PARTICIPATE IN THIS STUDY. I HAVE RECEIVED (OR WILL RECEIVE) A COPY OF THIS FORM FOR MY RECORDS AND FUTURE REFERENCE.

MESSAGE FRAMING Framing 186

Appendix C: Measurement Tools

Demographic Information

Age:______years What is your year in school? __1st year __2nd year __3rd year __ 4th year __5th year __6th year __7th year __ more Height:______ft.______in

Class Standing: Sex: __ Freshman __ Male __ Sophomore __ Female __ Junior

__ Senior Weight: (for Blood Alcohol Content):

______lbs. Student Status:

__ Part-time (1-11 credits) Racial Background: __ Full-time (12+ credits) __ White/Caucasian

__ Native American/American Indian Where you are living this quarter: __ Black/African American __ Residence Halls/Dorm Room __ Asian __ Fraternity/Sorority House __ Native Hawaiian/ Pacific Islander __ Off-Campus Housing/Apartment/House __ Multi-Ethnic __ With Parents __ Other

Most recent Quarter’s GPA (Write N/A if this Ethnic Background: does not apply to you:______Hispanic/Latino __ Nonhispanic Are you currently a Fraternity or Sorority Member? __ Yes __ No

MESSAGE FRAMING Framing 187

Relationship Status? __ Single, not dating __ Single, casual dating __ Single, exclusively dating __ Engaged __ Married/Life partner

If you are currently in a relationship, do you live with your partner? __ Yes __ No __ Not applicable

Work Status: __ I do not work __ Working part-time __ Working full-time

MESSAGE FRAMING Framing 188

The Quantity-Frequency-Peak Alcohol Use Index by Baer (1993); Marlatt, Baer, & Larimer

(1995)

For all questions, one drink equals:

For all questions, one drink equals: - 5oz. wine - 12oz. wine cooler - 12oz. beer (10oz. of Microbrew; 8-9 oz. Malt Liquor, Canadian beer or Ice beer) - 6oz. Ice Malt Liquor - 1 Cocktail with 1 oz. of 100 proof liquor or 1 ½ oz. (single jigger) of 80 proof liquor.

1. Think of the occasion you drank the most this past month. How much did you drink?

 0 drinks  5 drinks  10 drinks  15 drinks  20 drinks  1 drink  6 drinks  11 drinks  16 drinks  21 drinks  2 drinks  7 drinks  12 drinks  17 drinks  22 drinks  3 drinks  8 drinks  13 drinks  18 drinks  23 drinks  4 drinks  9 drinks  14 drinks  19 drinks  24 drinks  25+ drinks

2. Think of the occasion you drank the most this past month. How many HOURS did you spend drinking on that occasion?

 0-1  3-4  6-7  9-10  1-2  4-5  7-8  10+  2-3  5-6  8-9

3. On a given weekend evening during the past month, how much alcohol did you typically drink? Estimate for the past month.

MESSAGE FRAMING Framing 189

 0 drinks  5 drinks  10 drinks  15 drinks  20 drinks  1 drink  6 drinks  11 drinks  16 drinks  21 drinks  2 drinks  7 drinks  12 drinks  17 drinks  22 drinks  3 drinks  8 drinks  13 drinks  18 drinks  23 drinks  4 drinks  9 drinks  14 drinks  19 drinks  24 drinks  25+ drinks

4. On a given weekend evening during the past month, how many HOURS did you spend drinking? Estimate for the past month.

 0-1  3-4  6-7  9-10  1-2  4-5  7-8  10+  2-3  5-6  8-9

5. How many days of the week did you drink alcohol during the past month?

 I do not drink at all  About once a month  Once a month  Two times a month  Three times a month  Once a week  Twice a week  Three times a week  Four times a week  Five times a week  Six times a week  Every day

MESSAGE FRAMING Framing 190

Daily Drinking Questionnaire (Collins, Parks, & Marlatt, 1985)

This section asks you to report on your drinking over the past three months.

For all questions, one drink equals: - 5oz. wine - 12oz. wine cooler - 12oz. beer (10oz. of Microbrew; 8-9 oz. Malt Liquor, Canadian beer or Ice beer) - 6oz. Ice Malt Liquor - 1 Cocktail with 1 oz. of 100 proof liquor or 1 ½ oz. (single jigger) of 80 proof liquor.

1. Consider a typical week during the last three months. How much alcohol, on average (measured in number of drinks), do you drink on each day of a typical week?

Monday Tuesday Wednesday Thursday Friday Saturday Sunday

2. Consider a typical week during the last three months. Over how many hours do you drink the above number of drinks?

Monday Tuesday Wednesday Thursday Friday Saturday Sunday

3. On average, during the last three months, how often have you consumed alcohol?

 Never  Three times a month  Four times a week  Less than once per month  Once a week  Five times a week  Once a month  Twice a week  Six times a week  Two times a month  Three times a week  Every day 4. During the last three months, when you have consumed alcohol, how many drinks on average did

MESSAGE FRAMING Framing 191

you typically consume on a given occasion?

 0 drinks  9 drinks  18 drinks  1 drink  10 drinks  19 drinks  2 drinks  11 drinks  20 drinks  3 drinks  12 drinks  21 drinks  4 drinks  13 drinks  22 drinks  5 drinks  14 drinks  23 drinks  6 drinks  15 drinks  24 drinks  7 drinks  16 drinks  25 or more drinks  8 drinks  17 drinks

MESSAGE FRAMING Framing 192

Rutgers Alcohol Problem Index (White & Labouvie, 1989)

Directions: How many times did the following things happened to you while you were drinking or because of your alcohol use during the last 3 months?

1 2 3 4 5

Never 1 to 2 Times 3 to 5 Times 6 to 10 Times More than 10

Times

1. Not able to do your homework or study for a test?

2. Got into fights, acted badly, or did mean things?

3. Missed out on other things because you spent too much money on alcohol?

4. Went to work or school high or drunk?

5. Caused shame or embarrassment to someone?

6. Neglected your responsibilities?

7. Relative avoided you?

8. Felt that you needed more alcohol than you used to use in order to get the same effect?

9. Tried to control your drinking by trying to drink only at certain times of the day or in certain

places?

10. Had withdrawal symptoms, that is, felt sick because you stopped or cut down on drinking?

11. Noticed a change in your personality?

12. Felt that you had a problem with alcohol?

MESSAGE FRAMING Framing 193

13. Missed a day (or part of a day) of school or work?

14. Tried to cut down or quit drinking?

15. Suddenly found yourself in a place that you could not remember getting to?

16. Passed out or fainted suddenly?

17. Had a fight, argument or bad feelings with a friend?

18. Had a fight, argument or bad feelings with a family member?

19. Kept drinking when you promised yourself not to?

20. Felt you were going crazy?

21. Had a bad time?

22. Felt physically or psychologically dependent?

23. Was told by a friend or neighbor to stop or cut down drinking?

24. Drove shortly after having more than two drinks?

25. Drove shortly after having more than four drinks?

MESSAGE FRAMING Framing 194

Drinking Identity Scale. ( Lindgren, Neighbors, Teachman, Wiers, Westgate, & Greenwald,

2012)

Please indicate the extent to which you agree or disagree with the following statements: 1 2 3 4 5 6 7 Strongly Strongly Disagree Agree

1. Drinking is part of my self-image. 2. Drinking is part of “who I am”. 3. Drinking is part of my personality. 4. Drinking is a large part of my daily life. 5. Others view drinking as part of my personality

MESSAGE FRAMING Framing 195

Daily Drinking Questionnaire—Intention from Collins, Parks, & Marlatt, (1985).

This section asks you to report on your intentions for drinking over the next three months. For all questions, one drink equals: - 5oz. wine - 12oz. wine cooler - 12oz. beer (10oz. of Microbrew; 8-9 oz. Malt Liquor, Canadian beer or Ice beer) - 6oz. Ice Malt Liquor - 1 Cocktail with 1 oz. of 100 proof liquor or 1 ½ oz. (single jigger) of 80 proof liquor.

1. Consider a typical week over the coming three months. How much alcohol, on average (measured in number of drinks), do you intend to drink on each day of a typical week over the next month? Monday Tuesday Wednesday Thursday Friday Saturday Sunday

2. Consider a typical week over the coming month. Over how many hours do you intend to drink the above number of drinks? Monday Tuesday Wednesday Thursday Friday Saturday Sunday

3. On average, during the next month, how often do you intend to consume alcohol?  Never  Three times a month  Four times a week  Less than once per month  Once a week  Five times a week  Once a month  Twice a week  Six times a week  Two times a month  Three times a week  Every day

4. During the next month, when you consume alcohol, how many drinks on average do you intend to typically consume on a given occasion?  0 drinks  9 drinks  18 drinks  1 drink  10 drinks  19 drinks  2 drinks  11 drinks  20 drinks  3 drinks  12 drinks  21 drinks  4 drinks  13 drinks  22 drinks  5 drinks  14 drinks  23 drinks  6 drinks  15 drinks  24 drinks  7 drinks  16 drinks  25 or more drinks  8 drinks  17 drinks

5. For Women: During the next month, how many times do you intend to consume 4 or more drinks on a single occasion? (No. 5 not used)

MESSAGE FRAMING Framing 196

For Men: During the next month, how many times do you intend to consume 5 or more drinks on a single occasion?  1 time  2 times  3 times  4 times  5 times  6 times  7 times  8 times  9 times  10 or more times

MESSAGE FRAMING Framing 197

Rosenberg Self-esteem Scale (Rosenberg, 1965) Following are ten statements about you. Indicate how much you agree with each of the statements using the following scale. strongly moderately neutral moderately strongly disagree disagree agree agree 1 2 3 4 5

_____ 1. I feel that I’m a person of worth, at least on an equal basis with others. _____ 2. I feel that I have a number of good qualities. _____ 3. All in all, I am inclined to feel that I am a failure. _____ 4. I am able to do things as well as most other people. _____ 5. I feel I do not have much to be proud of. _____ 6. I take a positive attitude with myself. _____ 7. On the whole, I am satisfied with myself. _____ 8. I wish I could have more respect for myself. _____ 9. I certainly feel useless at times. _____ 10. At times I think I am no good at all.

MESSAGE FRAMING Framing 198

Instability of Self-esteem Scale (Charbol, Rorsseau, & Callahan, 2006)

0 1 2 3

Disagree Strongly Agree Strongly

Pick and number from one to four for the following statements to fit with your real status.

1. Sometimes I feel worthless; at other times I feel that I am worthwhile.

2. Sometimes I feel happy with myself; at other times I feel very unhappy with myself.

3. Sometimes I feel useless; at other times I feel very useful.

4. Sometimes I feel very bad about myself; at other times I feel very good about myself.

MESSAGE FRAMING Framing 199

The Realism/believability Scale (Flanagin, & Metzger, 2000)

1 2 3 4 5 6 7

Not at All Extremely

After reading the health message which had been displayed, please rate the messages according to these dimensions:

1. Believable

2. Accurate

3. Trustworthy

4. Biased

5. Complete

MESSAGE FRAMING Framing 200

Appendix D: Experimental Manipulations for Study 1 Ego Depletion Task Adopted and modified from Fischer et al. (2008)

Note. These messages will display on the computer screen as the participants take the internet experiment and survey.

Experimental Condition---Major Ego Depletion

We would like you to express your attention and imagination in a brief writing task. Specifically, we would like you to spend three minutes writing about a zoo trip.

You can either make up a story or present one based on your personal experience. Please type about this “Zoo Visit” in the space provided on the next screen.

However, the restriction is that you must not use the words “I”, “like”, “and”, “or” in your description. Please use around 2 minutes to think about your ideas, and begin to type for 1 minute.

Control Condition---Waiting

We would like you to pause now for three minutes. Please sit in front of the computer and wait for 3 minutes. You are welcome to feel relaxed or use your cell phone without making a sound.

When three minutes have passed please click continue and complete the remaining questionnaires.

Thank you very much.

MESSAGE FRAMING Framing 201