UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Fear appeals in social marketing advertising Lavack, Anne Marie 1997-12-31

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
ubc_1997-196070.pdf [ 20.54MB ]
Metadata
JSON: 1.0088283.json
JSON-LD: 1.0088283+ld.json
RDF/XML (Pretty): 1.0088283.xml
RDF/JSON: 1.0088283+rdf.json
Turtle: 1.0088283+rdf-turtle.txt
N-Triples: 1.0088283+rdf-ntriples.txt
Original Record: 1.0088283 +original-record.json
Full Text
1.0088283.txt
Citation
1.0088283.ris

Full Text

FEAR APPEALS IN SOCIAL MARKETING ADVERTISING by ANNE MARIE LAVACK B.Sc., The University of Toronto, 1984 M.B.A., The University of Manitoba, 1988 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES Department of Marketing We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA February 1997 ® Anne Marie Lavack, 1997 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. 1 further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia Vancouver, Canada Date FgBeUAfcy 1*117 DE-6 (2/88) Fear Appeals in Social Marketing Advertising Abstract This thesis includes several studies on the use of fear appeals in social marketing advertising. The first study uses a content analysis to examine the use of fear appeals in a sample of 589 social marketing television ads. The social marketing ads represented five health-related behaviors (smoking, drinking, driving while impaired, drug abuse, unsafe sex) in five countries (Canada, United States, Great Britain, Australia, New Zealand), covering the period from 1980 through to 1994. The sample was content analyzed to examine the incidence of fear appeals, the adherence to the prescriptions of the Ordered Protection Motivation (OPM) model (Tanner, Hunt, and Eppright 1991), and whether fear appeals vary by country-of-origin, the types of behavior being targeted by social marketing advertising (smoking, drinking, driving while impaired, drug abuse, unsafe sex), and the choice of an intended target group (by age and/or sex). Findings suggest that ads generally adhere to the major tenets of the OPM model. In terms of incidence, the use of fear appeals is less common when the sponsor is a for-profit corporation, when the ads are targeted at a youthful target group, and when the behaviors being targeted are perceived to be less serious. Fear appeals appear to be more common in ads from Australia, as compared to the United States or Canada. To examine the idea that different target groups may respond differently to fear appeal ads, two experiments and a focus group were conducted. First, an exploratory experiment used drinking and driving (DUI) ads as a stimulus to examine the differential effectiveness of two different types of ads against different behavioral risk groups. This study compared an "OPM" social marketing print ad (i.e., one using fear appeals of the format prescribed by the OPM model), to a "MALADAPT" social marketing print ad (i.e., one which simply presents ii counter-arguments against maladaptive responses, beliefs, and behaviors). Individuals who differed in the extent to which they engaged in the targeted risky behavior (i.e., those who do engage in DUI versus those who do not engage in DUI) were exposed to either the "OPM" or "MALADAPT" social marketing ads, or to a control condition. It was expected that the non-DUI group would experience the greatest change in attitudes and behavioral intentions when exposed to the traditional "OPM" social marketing ad, while the DUI group would experience the greatest attitudinal/behavioral change when exposed to the "MALADAPT" social marketing ad. However, the results of the initial exploratory experiment were inconclusive, and further study of the DUI target group was warranted. Therefore, a focus group was conducted which examined the attitudes and beliefs of the DUI group. A key finding from this qualitative research was that DUI individuals are unconcerned about getting into an accident, but are instead primarily concerned with getting caught by the police. This suggests that some of the traditional high-fear appeals which feature bloody accidents may not be effective with this high-risk target group, and reinforces the idea that the MALADAPT ad which tries to attack maladaptive beliefs may be the most effective means of influencing this DUI target group. Insights from the focus group provided the means for improving the ad stimuli and questionnaire for a replication of the experiment. Pretests for the ad stimuli helped in developing ads which were compelling and interesting for all experimental conditions. Based on these inputs, the experiment was refined and replicated. Findings indicated that the "MALADAPT" ad (which attacked maladaptive coping responses) was actually more effective with the high risk DUI group than the traditional OPM fear-appeal type of ad. iii TABLE OF CONTENTS Abstract ii Table of Contents v List of Figures viList of Tables viii Acknowledgement xINTRODUCTION 1 Chapter One Examining Fear Appeals in Social Marketing Television Ads through Content Analysis: Problem Definition and Overview 6 Problem Definition 7 Theories and Models of Fear Appeals 9 Incidence of Fear Appeals in Advertising Messages 22 Other Dependent Variables 28 Factors Influencing the Use of Fear Appeals 33 Social Marketing Campaigns 46 AIDS Prevention Campaigns 50 DUI/Alcohol Moderation Campaigns 55 Anti-Smoking Campaigns 7 Anti-Drug Abuse Campaigns 58 Summary 59 Chapter Two Content Analysis Research Design 62 Content Analysis 63 Sampling Procedure 4 Data Collection Instrument 67 Coder TrainingData Collection Procedure 8 Composition of the Sample 70 Chapter Three Content Analysis Research Results 73 Protection Motivation Theory 4 Incidence of Fear Appeals 82 Country-of-Origin and Fear Appeals 8Sponsorship and Fear Appeals 7 Behavioral Category and Fear Appeals 93 Intended Target Group and Fear Appeals 10Exploratory Findings 109 Comparing This Study to Other Content Analyses 116 Discussion and Conclusion 127 iv Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses: Problem Definition and Overview 134 Problem Definition 135 Variables Studied in Fear Appeals Research 136 Use of Segmentation in Fear Appeals Research 138 Psychological Theories Relevant to the Use of Fear Appeals 145 An Alternative Approach: Counteracting Maladaptive Behavior in High-Risk Individuals 156 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads 164 Objectives 165 Experiment Methodology 16Research Results 176 Testing the Hypotheses 182 Discussion 185 Limitations of the Study 187 Future Research 9 Chapter Six Using Focus Group Research to Better Understanding the DUI Segment 191 Purpose of the Focus Group Research 2 Recruiting Focus Group Participants 193 Focus Group Procedure 194 Focus Group Findings 202 Implications/Discussion 210 Conclusion 21Chapter Seven The OPM Model and Market Segments: A Refined Replication Using DUI Print Ads 213 Ad Stimuli 214 Pretest 220 Experiment Overview 22Descriptive Statistics 237 Evaluation of the Ads 242 DUI Attitudes, Beliefs, and Behavioral Intentions 250 Discussion 25Chapter Eight Summary and Conclusion 265 Summary 266 Contributions of the Research 270 Limitations of the Research 1 Future Research 273 Conclusion 5 Bibliography 276 v Appendix 1 List of Organizations Contacted 303 Appendix 2 Letter Used for Contacting Organizations 339 Appendix 3 Content Analysis Data Collection Forms 344 Appendix 4 Pretest A & Pretest B Questionnaires 372 Appendix 5 Experiment #1 Survey Instruments 375 Appendix 6 Focus Group Materials 406 Appendix 7 Experiment #2 Survey Instruments 418 vi LIST OF FIGURES 1.1 The Fear-Drive Model 11 1.2 The Inverted-U Model 2 1.3 The Parallel Response Model 4 L4 The Protection Motivation Model 17 1.5 The Ordered Protection Motivation (OPM) Model 19 1.6 Selected Variables Affecting Use of Fear Appeals in Social Marketing Ads 34 1.7 Continuum in Sponsor Self-interest 40 4.1 The Health Belief Model 154.2 Protection Motivation Model 3 4.3 Adaptive and Maladaptive Coping Responses under the OPM Model 157 4.4 Ordered Protection Motivation Model 158 4.5 Structural Model of the Ordered Protection Motivation Schema 160 5.1 OPM Ad 165.2 MALADAPT Ad 175.3 CONTROL Ad 2 6.1 Testimonial Ad 197 6.2 High-fear "You" Ad 8 6.3 Black Humor Ad 200 6.4 MALADAPT Ad 1 7.1 OPM Ad 216 7.2 MALADAPT Ad 217 7.3 CONTROL Ad 9 7.4 REVISED-MALADAPT Ad 223 vii LIST OF TABLES 1.1 Comparison of Three Content Analysis Studies 25 1.2 Comparing Health Change Behaviors 49 2.1 Behavioral Categories of TV Commercials, By Country 70 3.1 Presence of Coping Advice in Ads Which Contain Threats, By Country 75 3.2 Presence of Coping Advice in Ads Which Contain Threats, By Behavioral Category 76 3.3 Presence of Coping Advice in Ads Which Contain Threats, By Time Period 77 3.4 Order of Threats and Coping Advice in Ads, By Country 78 3.5 Order of Threats and Coping Advice in Ads, By Behavioral Category 79 3.6 Does Commercial Contain Information About Response Efficacy?, By Country 80 3.7 Does Commercial Contain Information About Response Efficacy?, By Behavioral Category 83.8 Does Commercial Contain Encouragement for Self Efficacy?, By Country 81 3.9 Does Commercial Contain Encouragement for Self Efficacy?, By Behavioral Category 2 3.10 Overall Tone of the Commercial, By Country 84 3.11 ANOVA of Overall Tone of the Commercial, By Country 85 3.12 Any Specific Threats Mentioned or Shown?, By Country3.13 Amount of Fear Arousal in the Commercial, By Country 86 3.14 ANOVA of Amount of Fear Arousal in the Commercial, By Country 87 3.15 Sponsorship Categories, By Country 88 3.16 Sponsorship Category, By Behavioral Category 83.17 Any Specific Threats Mentioned or Shown?, By Sponsorship Category 89 3.18 Overall Tone of the Commercial?, By Sponsorship Category 90 3.19 ANOVA of Overall Tone of the Commercial, By Sponsorship Category 90 3.20 Amount of Fear Arousal in the Commercial, By Sponsorship Category 91 3.21 ANOVA of Amount of Fear Arousal in the Commercial, By Sponsorship Category 92 3.22 Is There an Attempt to be Humorous?, By Sponsorship Category 93 3.23 Mean Values of Seriousness of Behavior Consequences: Pretest A 4 3.24 Mean Values of Seriousness of Behavior Consequences: Pretest B 96 3.25 Overall Tone of the Commercial, By Behavioral Category 97 3.26 Overall Tone of the the Commercial, By Seriousness of the Behavior 97 3.27 ANOVA of Overall Tone of the Commercial, By Behavioral Category 98 3.28 Any Specific Threats Mentioned or Shown?, By Behavioral Category 99 3.29 Any Specific Threats Mentioned or Shown?, By Seriousness of the Behavior 100 3.30 Amount of Fear Arousal in the Commercial, By Behavioral Category 100 3.31 Amount of Fear Arousal in the Commercial, By Seriousness of the Behavior 101 3.32 ANOVA of Amount of Fear Arousal in the Commercial, By Behavioral Category 102 3.33 Is There an Attempt to be Humorous?, By Behavioral Category 103 3.34 Is There An Attempt to Be Humorous?, By Seriousness of the Behavior 103 3.35 Overall Tone of the Commercial, By Intended Target Group Age Category 104 3.36 Any Specific Threats Mentioned or Shown?, By Intended Target Group Age Category 10viii 3.37 Amount of Fear Arousal in the Commercial, By Intended Target Group Age Category 105 3.38 ANOVA of Amount of Fear Arousal in the Commercial, By Intended Target Group Age Category 103.39 ANOVA of Overall Tone of the Commercial, By Intended Target Group Age Category 106 3.40 Is There an Attempt to be Humorous?, By Intended Target Group Age Category 107 3.41 Summary of Hypotheses 108 3.42 Summary of Results of Logistic Regression, Indicating Percent of Commercials Correctly Classified 111 3.43 Frequency of Direction of Threats 112 3.44 Any Specific Threats Mentioned or Shown?, By Time Period 113 3.45 Overall Tone of the Commercial?, By Time Period 114 3.46 ANOVA of Overall Tone of the Commercial, By Time Period 113.47 Amount of Fear Arousal in the Commercial, By Time Period 115 3.48 ANOVA of Amount of Fear Arousal, By Time Period 113.49 Is There an Attempt to be Humorous?, By Time Period 6 3.50 Comparison of Content Analysis Studies 118 3.51 Comparison of Anti-Drug Content Analysis Studies 121 3.52 Comparison of Results for Key Variables of Anti-Drug Content Analysis Studies 122 3.53 Race of Main Characters in Anti-Drug Ads 3 3.54 Comparison of AIDS Prevention Content Analysis Studies 125 3.55 Comparison of Specific Variables in AIDS Prevention Content Analysis Studies 126 4.1 Conceptual Foundations for Designing and Implementing Media Campaigns 148 5.1 Ad Copy Design 171 5.2 Experimental Design 3 5.3 Distribution of Subjects in Experimental Cells 176 5.4 Rotated Factor Pattern 178 5.5 Cronbach Alpha Scale Reliability Measures 179 5.6a Cell Means: FEARAVG ANOVA 180 5.6b 2x3 ANOVA OF FEARAVG Scale5.7a Cell Means: CALMAVG ANOVA 181 5.7b 2x3 ANOVA OF CALMAVG Scale5.8a Cell Means: AAD ANOVA 182 5.8b 2x3 ANOVA OF AAD Scale5.9a Cell Means: ATTDUI2 ANOVA 183 5.9b 2x3 ANOVA OF ATTDUI2 Scale5.10 Summary of Hypotheses Tests on ATT_DUI2 Scale 183 5.11 Summary of Hypotheses 185 6.1 Profile of Focus Group Respondents 202 7.1 Ad Copy Design 218 7.2 Experimental Design 227.3 Cronbach Alpha Scale Reliability Measures 231 7.4a Mean Values for Fear Scale (FEARAVG) 4 7.4b 2x4 ANOVA of FEAR AVG Variable 23ix 7.5a Mean Values for Effectiveness of Maladaptive Behaviors Scale (MALADAPT) 236 7.5b 2x4 ANOVA of MALADAPT Variable 237.6a Mean Values for English Fluency Variable (FLUENCY) 237 7.6b 2x4 ANOVA of FLUENCY Variable 237.7 Number of Drinks Consumed in the Past Month (by Sex) 238 7.8 Number of Drinks Consumed in the Past Month (by DUI Category) 240 7.9 DUI within the past month/year/5 years? (by sex) 241 7.10a Mean Values for Number of Drinks to Exceed Legal Limit Variable (LEGALLIM) 242 7.10b 2x4 ANOVA of LEGALLIM Variable 247.11a Mean Values for Number of Drinks to Affect Reflexes • Variable (REFLEXES) 243 7.lib 2x4 ANOVA of REFLEXES Variable 247.12a Mean Values for Attitude toward the Ad Scale (AAD) 244 7.12b 2x4 ANOVA of AAD Variable 247.13a Mean Values for Attitude toward the PSA Scale (ATTPSA) 245 7.13b 2 x 4 ANOVA of ATTPSA Variable 247.14a Mean Values for Calm Scale (CALMAVG) 246 7.14b 2x4 ANOVA of CALMAVG Variable7.15a Mean Values for Involvement Scale (INVOLVE) 247 7.15b 2 x 4 ANOVA of INVOLVE Variable 247.16a Mean Values for Support Arguments Scale (SUPPORT) 248 7.16b 2 x 4 ANOVA of SUPPORT Variable 247.17a Mean Values for Counter-Arguments Scale (COUNTER) 249 7.17b 2x4 ANOVA of COUNTER Variable 247.18 Summary of 2 x 4 ANOVA Results on Ad Evaluation Scales and Variables 250 7.19 Summary of planned t-tests to Test Hypothesis 1 252 7.20 Summary of planned t-tests to Test Hypothesis 2 4 7.21 Summary of planned t-tests to Test Hypothesis 3a 255 7.22 Summary of planned t-tests to Test Hypothesis 3b 6 7.23 Summary of Hypotheses 257 x ACKNOWLEDGEMENT I am extremely grateful to Dr. Charles B. Weinberg and Dr. Gerald J. Gorn for giving untiringly of their advice, encouragement, and feedback during the preparation of this dissertation. I also appreciate the help of Dr. Lawrence W. Green and Dr. C. James Frankish, who provided a unique perspective on this research. My interest in social marketing advertising goes back to 1984 when I was employed at McKim Advertising, a major Canadian advertising agency, where I worked on a teen non smoking campaign for Health and Welfare Canada. In subsequent years I worked on behalf of many other government and nonprofit clients to help develop social marketing campaigns against driving while impaired, drug abuse, sexually transmitted diseases (STDs), AIDS, and wife abuse. Therefore, I would also like to acknowledge and thank the many clients and colleagues with whom I worked during these earlier years, who inspired my interest in social marketing. xi Introduction INTRODUCTION 1 Introduction INTRODUCTION Social marketing campaigns open people's minds; they expose people to choices and alternatives they otherwise might not have known existed. They change attitudes; sometimes, they save lives. - Wallach 1988, p. 118 The purpose of this dissertation is to examine the use of fear appeals in social marketing advertising. This introduction will briefly explain the idea of social marketing, as well as the role of fear appeals in social marketing. This will be followed by an outline of the research that has been conducted in preparing this dissertation. Advertising and marketing are most frequently applied to stimulating or changing purchasing behavior for products or services. Social marketing, however, differs from mainsteam marketing in that the intent is often to promote ideas, change attitudes, or alter health-related behaviors, rather than influencing purchasing behavior. The ultimate objective when promoting ideas through social marketing may still be a behavior change; however, the behavior that is being targeted is not necessarily an ordinary purchasing behavior. Social marketing is a growing field of increasing interest to both marketing practitioners and academics (Riggs 1990). One use of social marketing is to decrease consumption or attack maladaptive behaviors. When applied to advertising or communications, this approach is known variously as de-marketing, counter-advertising, or public service advertising. De-marketing is usually used in contexts where consumers are being persuaded or influenced to reduce their consumption of a particular product (Budden and Hossain 1986; Kotler and Levy 1971). Counter-advertising usually describes advertising whose purpose is to counter or oppose the advertising, promotion, and sales of potentially harmful substances (e.g., ads which try to 2 Introduction eliminate or reduce harmful consumption behavior, such as smoking or excessive alcohol consumption; Colford 1990). Public service announcements (PSA) are a generic name for social marketing advertising messages, but the use of the term PSA tends to imply that the airtime or space for the advertisement is donated rather than paid. In this dissertation the term social marketing advertising will be used to broadly refer to any type of persuasive media communication or message intended to promote social marketing goals. Fear appeals are used as a persuasive tool in some advertising messages. Arousing fear about a potential threat is used as a means of getting the consumer to take action and buy a product. Fears about bad breath, taking care of one's family, and automobile safety are appeals that have been effectively used in ads for mouthwash, life insurance, and tires, respectively. Fear appeals are also used in social marketing advertising as a means of persuading individuals to act. The action being encouraged in this case, however, is usually to change one's behavior in ways which will bring improved health, safety, or benefits to society. Social marketing ads have contained fear-arousing scenes of highway carnage, death by drug-overdose, blackened lungs, dying cancer patients, and family violence, all in the name of trying to change behavior to ultimately benefit individuals or society. Yet there is much that is not known about the use of fear appeals in social marketing advertising. First, we know little about the incidence of this type of appeal. All of us can recall ads of this type, but we do not actually know whether fear appeals are the norm, or whether fear appeals are actually used infrequently. Also, we do not know whether the use of fear appeals varies as a function of such factors as its intended target group, the type of behavior being targeted, the type of sponsor, or the country in which it is produced. And although there are prescriptive theories about fear appeals, we do not know whether social 3 Introduction marketing ads which contain fear appeals actually adhere to the tenets of these theories. Second, we know relatively little about the effectiveness of fear appeals with different target groups. Perhaps some individuals can be frightened into changing their behavior, but will this work for everyone? A key concern is that we do not really know whether fear appeals are effective with the target group of individuals who repeatedly engage in particular high-risk behaviors. For example, when we see a high-fear ad campaign credited with reducing the incidence of driving under the influence (DUI), we do not really know whether those who occasionally DUI have stopped doing it altogether, or whether those who frequently DUI are doing it less often. Either or both of these phenomena could account for an overall drop in DUI. This dissertation is comprised of two main types of studies which examine the use of fear appeals in social marketing advertising. Chapters One through Three describe a content analysis which uses a five-country sample of English-language social marketing TV commercials. It examines the incidence of fear appeals in social marketing ads, and attempts to determine how the use of fear appeals varies as a function of country of origin, type of ad sponsor, type of behavior being targeted, and attributes of the intended target group. As well, it examines whether social marketing ads which contain fear appeals adhere to the Protection Motivation model (Rogers 1975, 1983) or the Ordered Protection Motivation model (Tanner, Hunt, and Eppright 1991). Chapters Four through Seven describe an exploratory experiment, a focus group, and a subsequent replication of the experiment. The purpose of the experiments is to compare the effectiveness of fear appeal ads with a second type of ad which attacks maladaptive coping responses. Using these two types of ads for DUI and two groups of subjects who differ in the extent to which they engage in DUI (i.e., those who DUI versus those who do not DUI), it 4 Introduction will be possible to determine which type of ad is more effective with which type of behavioral target group. Finally, Chapter Eight will present a synthesis of the results of the studies, and present some guidelines for future research. While the research contained in this dissertation provides some insights into the role of fear appeals in social marketing ads, there are still many avenues for future research which can be fruitfully explored. 5 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis CHAPTER ONE EXAMINING FEAR APPEALS IN SOCIAL MARKETING TELEVISION ADS THROUGH CONTENT ANALYSIS: PROBLEM DEFINITION AND OVERVIEW 6 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis CHAPTER ONE EXAMINING FEAR APPEALS IN SOCIAL MARKETING TELEVISION ADS THROUGH CONTENT ANALYSIS: PROBLEM DEFINITION AND OVERVIEW Horrifying images of car crashes, bloody accident scenes, clanging jail cell doors, funerals, crying children, blackened lungs, broken families, death, and despair haunt our television sets. These scenes, and many others like them, seem to be frequently used by social marketers as a persuasive tool, arousing fear in order to change destructive behaviors, to encourage the adoption of healthier lifestyles, and to exhort individuals to obey the law. This research will examine these fear-arousing images in social marketing television commercials, in the context of previous research about fear appeal incidence and prescriptive models. Problem Definition Fear appeals have been relatively common in marketing communications in the twentieth century (Pollay 1985; Stern 1988), particularly in advertising products in which emotions play a key role in consumer behavior and decision-making. Social marketing ads in particular are thought to make fairly frequent use of fear appeals (Freimuth, Hammond, Edgar, and Monahan 1990). This is perhaps not surprising, since social marketing ads frequently deal with sensitive issues and emotional topics (Kotler and Andreasen 1996). Additionally, the unselfish intentions behind social marketing advertising render it less likely that the use of fear appeals would result in ethical concerns about exploitation. A number of issues and questions arise regarding the incidence of fear appeals in social marketing ads: • How frequently are fear appeals used in social marketing ads? 7 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis • Does the incidence of fear appeals in social marketing ads differ as a function of situational factors, such as country-of-origin, type of sponsor, type of behavior being targeted, or the intended target group? Additionally, questions arise about the intensity of fear arousal in social marketing ads: • Does the intensity of the fear appeal differ as a function of situational factors (country-of-origin, type of sponsor, type of behavior being targeted, or the intended target group)? Finally, questions exist concerning the content or format of the social marketing fear appeal ad, and its adherence to prescriptive theory: • Does the format of the ad follow the prescriptive advice suggested by theories of fear appeals, such as the Ordered Protection Motivation model (Tanner, Hunt, and Eppright 1991)? The content analysis described in Chapters Two and Three uses a comprehensive non-probability sample of English-language social marketing TV ads which were produced between 1980 and 1994 in five English-speaking countries (Canada, the United States, Great Britain, Australia, New Zealand), targeting five health-related behaviors (smoking, drinking, driving while impaired, drug abuse, unsafe sex). The use of this non-probability sample permits an in-depth and thorough study of social marketing advertising targeting these five specific types of behavior. While a primary goal of the content analysis study was to test some specific a priori hypotheses, it was also expected that the content analysis would be partially exploratory in nature, and might help to identify some interesting trends or lead to the development of new research ideas. The a priori hypotheses will be put forward where there is a basis in the literature to develop such hypotheses. Exploratory analyses will be used to examine interesting areas where there are few pointers in the literature to suggest an a priori hypothesis. 8 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis This Chapter will begin by examining the literature on theories of fear appeals, including an examination of the literature on the incidence of fear appeals. This will be followed by a brief review and discussion of cross-national1 differences in advertising, types of ad sponsorship, and types of intended target audiences, and how these all relate to the use of fear appeals in social marketing ads. Finally, this will be followed by an outline of the literature on social marketing campaigns, with particular reference to the literature on the five behavioral areas that will be examined in the content analysis. A review of content analysis techniques and methodology will be outlined later on in Chapter Two where the content analysis research methodology is discussed. Theories and Models of Fear Appeals Over the past 40 years a large literature has evolved on the use of fear appeals in persuasive messages, which has been reviewed extensively elsewhere (e.g., Boster and Mongeau 1984; Dillard 1994; Higbee 1969; Jahis 1967; Leventhal 1970; Miller and Hewgill 1966; Ray and Wilkie 1970; Rotfeld 1988; Sternthal and Craig 1974; Sutton 1982). The literature on fear appeals is full of contradictory results and conflicting models. Some of the early research indicated that mild fear appeals were more effective than strong fear appeals, suggesting an inverse relationship between level of fear and persuasiveness (e.g., Janis and Feshbach 1954). However, much of the later research has contradicted this finding. Other studies have indicated that moderate fear levels are most effective, suggesting a curvilinear or inverse-U relationship between level of fear and level of persuasiveness (e.g., Krisher, Darley, and Darley 1973; Schwarz, Servay, and Kumpf 1985). The majority of the studies, however, 1 The terms cross-cultural and cross-national will be used interchangeably. 9 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis have indicated that strong fear appeals are more persuasive than mild ones, suggesting a direct relationship between level of fear and level of persuasiveness (e.g., Miller and Hewgill 1966; Sternthal and Craig 1974). Essentially, there are five models of fear appeals which have gained prominence over the years in the literature: (1) the Fear Drive model, (2) the Parallel Response model, (3) the Inverted-U model, (4) the Protection Motivation model, and (5) the Ordered Protection Motivation model. Each of these models are reviewed here. The Fear-Drive Model The Fear-Drive model (Hovland, Janis, and Kelley 1953) is based on the idea that the information contained in a message evokes an emotional reaction, which in turn motivates a coping response (see Figure 1.1). This model assumes that the emotional response of fear functions as a drive which mediates belief change and behavior change. A fear appeal "is most likely to induce an audience to accept the communicator's conclusion if a) the emotional tension aroused during the communication is sufficiently intense to constitute a drive state; and b) silent rehearsal of the recommended belief or attitude is immediately followed by reduction of tension" (Hovland, Janis, and Kelley 1953, p. 62). The more serious the potential damage to one's physical or social self, the more likely the person will be to follow the recommended option. However, when the fearful drive state becomes so intense that the recommended actions are no longer sufficiently reassuring, individuals become motivated to reduce their fear by other means, such as defensive avoidance. People can defensively avoid the fear message by such things as calling into question the credibility of the communicator, or by mimmizing their perceived susceptibility to the threat. Consequently, this model suggests that moderate levels of fear arousal are most likely 10 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Threat Message Emotional Arousal High .Moderate. __Low .defensive avoidance acceptance of - recommended -action .arousal reduced .arousal reduced .no acceptance _^^_no arousal of recommended to be action reduced .reinforced avoidance responses .reinforced compliance responses .no reinforced compliance responses .protective actions not taken .protective actions taken .protective actions not taken Figure 1.1 The Fear-Drive Model (Source: Hovland, Janis, and Kelley 1953) to facilitate persuasion, whereas further increases in fear arousal are likely to encourage defensive avoidance and thus discourage persuasion. In essence, the Fear-Drive model proposes a curvilinear relationship between the level of negative arousal being experienced by the individual and the subsequent acceptance of the recommendations, since moderate levels of fear arousal are thought to be optimal. The Inverted-U Model The Inverted-U model (Janis 1967) also offers a theory of the curvilinear relationship between the amount of persuasion and the level of fear arousal in response to a threat communication. This model proposes that moderate fear levels are more effective than either low fear levels or high fear levels. Low fear levels are insufficient to motivate the individual to take preventive action, while high fear levels create a sense of paralysis, in which the individual becomes unable to respond. This model assumes that increases in fear have multiple effects, some of which facilitate persuasion and others of which have an interfering effect. As the level of fear arousal is increased, the facilitation increases more rapidly than the interference; however, at some point the interference starts to increase at a more rapid rate. 11 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis The resulting relationship between fear and acceptance takes the form of an inverted-U shaped curve with the optimal point occurring at that level of fear at which the interfering effects start to increase faster than the facilitating effects (see Figure 1.2). Opiimol A Opiimol 0 Optimal C Optimo) 0 Moiimjm stress tolerance \ Base level of acceptance induced by nonlKreat material 3 ^.d-5 6 7x-_-8 9 IC?^._ II Level of tear arousal Figure 1.2 The Inverted-U Model (Source: Janis 1967) Thus, when the message recipient is either extremely fearful or has very little fear, little attitude or behavioral conformity toward the message recommendation will occur. The optimal amount of conformity is produced when the listener is moderately fearful. At low levels of fear, the audience is unaffected because they dismiss all information as being inconsequential by means of blanket reassurances (e.g., we need not change because we are not convinced of the danger). At high levels of fear, the audience exhibits defense mechanisms (e.g., denial, detachment, and minimizing rationalization) which interfere with acceptance of the message. At moderate levels of fear, "the average person's vigilance and reassurance tendencies are stimulated which is the optimal condition for developing compromise attitudes of the type 12 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis required for acceptance of the recommended action" (Janis and Leventhal 1968, p. 1056).2 A practical limitation of the Inverted-U model is that, like the Fear-Drive model, it leaves researchers searching for an optimal level of fear. In practise it may be difficult to measure a moderate level of fear arousal, as distinct from a low or high level of fear arousal. A key reason for this is that the level of fear arousal may vary among individuals or among target segments. Therefore, practical testing of its predictions is extremely difficult. As well, numerous studies have not found any support for the curvilinear model (see Allen and Preiss 1990; Boster and Mongeau 1984; Sutton 1992). The Parallel Response Model Unlike the two afore-mentioned models which restrict threat coping to an emotional process, the Parallel Response model (Leventhal 1970) does not rely on fear arousal reduction as the principal mechanism underlying the effects of fear appeals. Instead, emotional and adaptive responses to threat messages are considered to be two independent and parallel processes (see Figure 1.3). Threat messages may make people fearful, but any protective actions that are taken result from a desire to control the danger, not to reduce the fear, according to this model. Essentially, the Parallel Response model suggests that two independent but parallel processes are at play when fear appeals are used. These two forces, danger control and fear control, are activated simultaneously in response to a fearful stimulus. 2 A two-dimensional (tension and energy) formulation of fear arousal effects has also been examined, in which stimulation of fear parallels a two-part continuum of increasing tension. Increasing tension generates energy up to a certain point, and beyond that threshhold, increasing tension arouses anxiety which begins to deplete energy (LaTour and Zahra 1989; LaTour and Pitts 1989; Henthorne, LaTour, and Nataraajan 1993). This mimics the results of the curvilinear model. 13 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis R Instrumental (Attitude, Action) R Instrumental feedback S Cognitive (Danger) Encoder S Cognitive (Danger) Encoder Message etc. R Fear R Fear Note: S=Stimulus and R=Response Figure 1.3 The Parallel Response Model (Source: Leventhal 1970) The first process, danger control, is an objective process which guides an individual's problem solving behavior and action. This objective-cognitive process generates a representation of the threat, and it also generates action plans to cope with the threat based on its representation. As problem solving proceeds, the external cues (such as information about the danger, and feedback regarding the adequacy of the action taken) influence subsequent adaptive behavior. The second process, fear control, is a subjective process which deals with the emotional component of the persuasive appeal. It is an emotion-coping process in which the message receiver strives to reduce the fear. Emotions are aroused by the message independent from thought, and behaviors are needed to cope with and control these emotions. These behaviors may be different than those behaviors needed to cope with the threats. For example, gory images and verbal statements regarding the severe consequences of noncompliance may heighten internal cues and induce behaviors such as denying the threat, resting, drinking, or eating to cope with the emotional response, even these do not affect the adaptive behavior that is needed to cope with the threat. 14 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis "In the parallel-response model, emotional arousal is not a necessary antecedent of adaptive behavior ... both types of behavior, emotional and adaptive, are consequences of environmental stimulation. They do not cause one another" (Leventhal 1970, p. 124). In summary, the danger control process guides adaptive behavior while the fear control process guides emotional responses. There are several criticisms of the Parallel Response model. One problem is that Leventhal refers to aspects of the stimulus situation that can facilitate or inhibit the danger control and fear control processes, but he doesn't clearly specify what these aspects are or how they affect these two processes. Thus, it is a difficult model to test, since it is hard to predict the manifestation of danger control and fear control (Sutton 1982). Another criticism is that Leventhal initially assumes that danger control and fear control are independent processes, but he later suggested that these two processes may sometimes interact. It is clear that although these two processes are conceived as being independent, they may affect one another during the course of decision-making. A highly emotional response may disrupt adaptive behavior, or performance may disrupt the emotional response. For example, strong fear motivates avoidance behaviors, which in turn may disrupt danger control. The fear control process can arouse emotions that lead to behavior which may inhibit or lessen the persuasiveness of the message (McDaniel and Zeithaml 1984). Also, the Parallel Response model provided prescriptive guidelines that ultimately produce conflicting results (Beck and Frankel 1981; Boster and Mongeau 1984; Rogers 1975). Recently, the Parallel Response model has been further developed into the Extended Parallel Process Model (EPPM; Witte 1992), which specifies the relationship between threat and efficacy. In effect, the EPPM suggests that fear leads to message rejection, while cognitions (i.e., perceived threat and efficacy) lead to message acceptance. 15 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis The Protection Motivation Model The most recent of the fear models, the protection motivation (PM) model (Rogers 1975, 1983) has increasingly gained acceptance among academics (Roser and Thompson 1995). The protection motivation model suggests that fear arousing persuasive messages are comprised of three variables: (1) levels of noxiousness of an event, (2) the probability that the event will occur without behavioral change, and (3) the effectiveness of a coping response that can reduce or eliminate the noxious event (see Figure 1.4). Each of these crucial elements "initiates corresponding cognitive appraisal processes that mediate attitude change" by arousing what has been termed "protection motivation" (Rogers 1975, p. 93). Like Leventhal (1970), Rogers (1975) believes that emotional arousal is less important than a person's cognitive appraisal of the threat.3 Contrary to Leventhal (1970), however, Rogers (1975) is very specific about the cognitions involved in threat appraisal. His model focuses on the above-mentioned three factors, and he claims that people will accept or reject recommendations based on their assessment of these variables. Whether or not the viewer acts on the coping advice, or chooses a maladaptive coping response (e.g., ignoring the threat, discounting the probability of occurrence, etc.) is mediated by four cognitive appraisal processes (Rogers 1983). These four processes are appraisals of: (1) the available information about the perceived severity of the threat; (2) the perceived probability that the threat will occur; (3) the perceived ability of the recommended coping response to remove the threat (response efficacy); and (4) the individual's perceived ability to carry out the recommended coping response (self-efficacy). While both response efficacy and 3 Dillard (1994) has recently argued that some fear theories have become divorced from the notion of fear, and calls for using an emotional perspective on fear appeals to examine the role of fear appeals in persuasive communications. 16 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis self-efficacy are important, some authors have suggested that self-efficacy is somewhat more important than response efficacy in determining the likelihood of acting on coping advice (Bandura 1990, 1977a; Bandura, Adams, and Beyer 1977; Beck 1981; Beck and Frankel 1981; Beck and Lund 1981). A social marketing ad may provide information that is relevant to each of these four elements of the PM model. First, it is likely to give specific information about the potential threats, and may even suggest the degree to which each threat is serious or severe. Second, a social marketing ad may include information about the likelihood or probability that the threat will occur. Third, since the purpose of a social marketing ad is not simply to frighten, but rather, to mobilize action, a social marketing ad is also likely to give specific pieces of advice or information about actions that can or should be taken to prevent, ameliorate, reduce, eliminate, or cope with the threatened outcome, and may advise about the efficacy of the recommended action. Finally, a social marketing ad may provide a message of encouragement or reassurance in order to bolster the target group's sense of personal self-efficacy and confidence in their own ability to actually carry out the recommended actions (Maddux and Rogers 1983). Components of a Fear Appeals Cognitive Mediating Processes Magnitude of Noxiousness Appraised Severity Probability of Occurrence Expectancy of Exposure Protection Intention to Adopt Recommended Response Efficacy of Recommended Belief in Efficacy of Motivation . Response Coping Response Figure 1.4 The Protection Motivation Model (Source: Rogers 1975) 17 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis One criticism of the Protection Motivation model is that it does a poor job of accounting for the consequences of emotional arousal, and therefore leaves the receiver lost in thought since it does not predict what actions will take place (i.e., what is the actual relationship predicted to be?) (Leventhal et al. 1983). It has also been pointed out that in this model the relations between behavior and fear arousal are complex and multifactorial, and therefore it is unlikely that any single variable will explain a high proportion of the variance (Boster and Mongeau 1984). The Ordered Protection Motivation Model The Ordered Protection Motivation (OPM) model, a variant on the Protection Motivation model, has been developed recently in the marketing literature (Tanner, Hunt, and Eppright 1991). It seems likely that the PM or OPM models of fear appeals have the greatest applicability to social marketing advertising (Van der Velde and Van der Pligt 1991). Looking back over the past 40 years of fear appeals research, there appears to be significant support for the precepts of the OPM model (Hunt, Tanner, and Eppright 1995). The OPM model is shown in Figure 1.5. Under the Ordered Protection Motivation (OPM) model (Tanner, Hunt, and Eppright 1991), the ordered component of the model suggests that within a given communication or message, the threat information should precede the coping information. The cognitive processing of the threat information must occur prior to the cognitive processing of the coping information, in order for a given message to be effective. Research has indicated that the optimum position for the fear message is immediately prior to the recommendations (Skilbeck, Tulips, and Ley 1977). Therefore, the present content analysis will examine the order of threat and coping information in social marketing ads, to determine the extent to which social 18 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis SOURCES OF INFORMATION Verbal Persuasion (Fear Appeals) Observed Learning Experience -1 COGNITIVE MEDIATING PROCESS Severity ol Threat . Probability of Occurrence Behavior Repertory Appraisal INTERMEDIATE EMOTIONAL STATE COGNITIVE MEDIATING PROCESS INTERMEDIATE STATE COPING MODE Coping Response(s) Efficacy (maladaptive or adaptive) Threat Fear . Coping Appraisal Appraisal" Self-Efficacy . Protection Motivation Behavior (adaptive or maladaptive) Social Norms and Values Figure 1.5 The Ordered Protection Motivation Model (Tanner, Hunt, and Eppright 1991) marketing ads conform to this prescription from the OPM model. Some research has also examined the relative effectiveness of grouping multiple threats and recommendations (coping advice) within one ad, or whether it is more effective to alternate threats and coping advice, by following each individual threat with a meaningful recommendation (Boyd 1995). These two different patterns of presenting threats and coping advice may be differentially effective for different target groups, thereby increasing the importance of the "ordered" component of the OPM model. 19 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Summary of Theoretical Models of Fear Appeals The above five models present conflicting views of the impact of fear appeals on persuasiveness of a message. There are several reasons why there have been conflicting findings with regard to the relationship between level of fear and level of persuasiveness. First, there is a question as to whether much of the early research had even measured fear arousal, since the level of fear intended by the researcher may not match (or may be confounded with) the level of fear perceived by the subjects (Dillard 1994; Higbee 1969; Witte 1993). Second, a key problem with some of the studies is that fear is operationalized in different ways, and that variables called "fear" may variously also be thought of as being anxiety, nausea, worry, or even concern (Higbee 1974; Witte 1993). Third, it has been widely suggested that there may be mediating variables which have an impact on the persuasiveness of fear appeals, including source credibility (Dembroski, Lasater, and Ramirez 1978; Hewgill and Miller 1965), familiarity of the topic (Karlins and Abelson 1970), credulity of the subjects (Ray and Wilkie 1970), self-esteem of the subjects (Leventhal and Perloe 1962; Leventhal and Watts 1966), etc. These and other mediating variables may have an impact on how persuasive fear appeals are perceived to be. Finally, the fear-persuasion relationship might be segment specific, meaning that fear may be more effective for some individuals or groups than for others (Burnett and Oliver 1979; Burnett and Wilkes 1980). As well, some of the conflicting results may be due to: (1) the nature and object of fear studied; (2) the range of fear manipulations; (3) the heterogeneity of subject pools; (4) the variety of methods of presentation that have been employed; and (5) the diversity of measures used to establish persuasive effects (Rotfeld 1988). It has also been noted that much of the previous research may have limited applicability to social marketing advertising or public service announcements, since the persuasive communications utilized ranged from voice 20 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis recordings to personal interactions to written passages to films, and seldom used ads or promotional materials (Rotfeld 1988). Two important meta-analyses (Boster and Mongeau 1984; Sutton 1982) have demonstrated -that the weight of evidence in the literature supports the premise of a positive relationship between level of fear arousal and message persuasiveness.4 The Protection Motivation model and the Ordered Protection Motivation Model have gained widespread support, and are thought to be the most useful models, particularly for prescriptive purposes. For this reason, the content analysis study will examine adherence to the Ordered Protection Motivation (OPM) model within social marketing ads. Specifically, the following hypotheses with regard to the OPM model will be tested: Hla The majority of social marketing ads will contain one or more threats or negative consequences. Hlb Of the ads which contain one or more threats or negative consequences, the majority will also contain action recommendations or advice for coping with the threat. Hlc In social marketing ads which include both threat(s) and coping information, the threat(s) will tend to temporally precede the coping information. 4 Allen and Preiss (1990) conducted an interesting examination of 34 textbooks which dealt with persuasion and public speaking. In 25 of these 34 textbooks the use of fear appeals was discussed; 15 of the 25 disagreed with the hypothesis of a positive linear relationship between fear and persuasiveness, in spite of the fact that two major meta-analyses (Boster and Mongeau 1984; Sutton 1982) have found extensive support for this relationship. Instead, many textbooks continue to promulgate the Janis and Feshbach (1953) curvilinear hypothesis which recommends moderate fear appeals as being most effective. Allen and Preiss (1990) suggest that textbook authors may commonly rely on their own narrative review of the fear literature, rather than conducting a systematic meta-analysis of the literature in which effect sizes are taken into account. This use of a narrative review approach, rather than a meta-analytic approach, may lead some textbook authors to erroneous conclusions. 21 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Incidence of Fear Appeals in Advertising Messages Fear is an emotion which is thought to have considerable persuasive ability. Fear appeals generally create a sense of fear or dread, which appears to lend a high level of importance or credibility to the message (Zeitlin and Westwood 1986). The use of negative emotions can be a successful mediator of attitudes in advertising appeals (Edell and Burke 1987; Moore and Hoenig 1989). It appears that fear appeals have been a relatively common device in marketing communications during the twentieth century (Pollay 1985; Stern 1988), particularly in the areas of insurance marketing, political advertising (Calantone and Warshaw 1985), and other areas where emotions play a role in consumer decision making. Only one content analysis study has been uncovered which focuses on the prevalence of fear appeals (Unger and Stearns 1983). The Unger and Stearns (1983) study examined 675 TV commercials which aired during sampled day-parts during a 4-week period in July 1982. That study found that fear appeals were present in 16.6% of the ads. The most common fear types were social disapproval (in 45% of the fear ads) and physical threat (in 46%), followed by financial fear (15%) and functional fear5 (8%; totals equal more than 100% due to multiple fear types in some TV commercials). The Unger and Stearns (1983) study illustrated the prevalence of fear appeals in television commercials in general, but did not examine the frequency of fear appeals within social marketing television commercials in particular. (The present content analysis study will remedy that situation by focusing specifically on fear appeals in social marketing advertising.) The high prevalence of fear appeals found in the Unger and Stearns (1983) study is 5 Functional fear refers to "the threat of product malfunction" (Unger and Stearns 1983, p. 18). This would clearly be an important type of fear variable in product advertising, but is not relevant for social marketing advertising. 22 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis probably due to the successful impact of fear appeals on attitude change (Johnson 1991; Natarajan 1977; Smith, Frankenberger, and Kahle 1990). Fear appeals also appear to be relatively common in the narrower field of social marketing advertising, although there has been little research conducted which estimates their incidence of occurrence. A key reason for this lack of research is that there are relatively few social marketing ads on television, compared to the large number of television commercials representing for-profit products and services. Wallack and Dorfman (1992) conducted a content analysis of a composite day of television comprising 20 randomly selected hours of TV programming6 over a three week period in April/May 1989. They found that PSAs comprised only 5.8% of the commercials in this sample, and less than half of these PSAs were health-related messages (i.e., less than 7 minutes of health-related PSAs in the 20 hour sample of television programming). Somewhat surprisingly, not one of the PSAs in the sample addressed the issues of tobacco or alcohol, two of the leading behavioral risk factors for poor health. This low incidence of health-related PSAs indicates the difficulty that would be encountered in trying to obtain a probability sample of social marketing TV ads. A similar but larger scale content analysis was conducted by Hanneman, McEwen, and Coyne (1973) which covered four television stations in Hartford, Connecticut over a one-week period in June, 1972. This study made use of both data coded by 18 different observers covering over 500 hours of airtime (55% of the airtime hours during the week), as well as broadcast station logs. Because of the larger size of the database of TV commercials (500 hours of airtime, as compared to 20 hours of airtime in the Wallack and Dorfman (1992) 6 TV programming usually consists of approximately 12 minutes of TV commercials and 48 minutes of TV show per hour. This means that TV commercials comprise only 20% of TV programming. 23 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis study), it was possible to study PSAs in more detail in this study. A total of 1159 PSAs aired during the week, representing 747 minutes or 12.45 hours of public service advertising; this compares to 530 hours of total on-the-air time for the four stations. PSAs represented 2% of the total viewing time, while commercial advertising accounted for 20% of the total air time. One-third (33%) of the PSA commercials dealt with medical/health issues or social problems (drug abuse 5%, alcoholism 2%, venereal disease 2%, population growth 2%, and civil liberties/pollution/discrimination 7%). Fear appeals were used in 24% of the PSAs. A PSA was coded as employing a fear appeal if it showed or discussed the harmful or painful social or physical consequences of not doing what the message directed. The 24% incidence of fear appeals was relatively high in this study, compared to the 16.6% incidence of fear appeals found in the Unger and Stearns (1983) study, and indicates the importance of studying fear appeals in the context of social marketing commercials. A comparison of these three major content analysis studies is shown in Table 1.1. 24 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Table 1.1 Comparison of Three Content Analysis Studies Unger & Stearns (1983) Wallack & Dorfman (1992) Hanneman, McEwen, & Coyne (1973) Sampling Frame - sampled dayparts over a 4-week period across 3 TV networks - sampled 20 hours of TV airtime randomly selected over a 3-week period - sampled 500 hours of TV airtime during a 1-week period across 9 TV stations Timing July 1982 April/May 1989 June 1972 Type of Commercials all types all types all types but analysis limited to PSAs Total Number of Commercials 675 commercials 654 commercials 10,399 commercials Number of PSAs/ Social Marketing Commercials n/a 34 PSA commercials 1,159 PSA commercials Incidence of PSAs/ Social Marketing Commercials n/a 5.8% 11.1% Incidence of Fear Appeals 16.6% of all commercials n/a 24.0% of PSA commercials How Fear was Measured - coded as fear appeal if the threat of a negative consequence was used n/a - coded as fear appeal if it showed or discussed the harmful (or painful) social or physical consequences of not doing what the message directed Number of Coders 2 2 18 Other Variables Measured - guilt messages - degree of emotion - degree of vividness - # of "you" references - # of references to negative consequences - social/physical/ financial/functional types of fear - product/service types - types of health messages - celebrities - intended target group - topic of PSA - program/time at which PSA was shown 25 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Based on this evidence, it is anticipated that fear appeals will be present in 24% of the social marketing ads in the proposed sample. Since social marketing ads in this sample are aimed at changing negative behaviors, it seems likely that a substantial proportion of these ads will focus on the threats to health that are the result of the behaviors in question. Because of the inclusion of health threats in this type of ad, it is expected that this approach will be fear-arousing in a significant proportion of cases.7 Given the evidence in the Hanneman, McEwen, and Coyne (1973) study, this proportion is expected to be equivalent to that found in their study, or approximately 24%. There is no evidence to suggest that the incidence of fear appeals has changed since 1973; therefore, the null hypothesis that the incidence of fear appeals has stayed the same will be put forward. Therefore the following is hypothesized: H2 Approximately 24% of the social marketing ads will contain fear appeals. Ethical Concerns Regarding the Use of Fear Appeals The use of fear appeals in product advertising is sometimes thought to create ethical concerns (Benet, Pitts, and LaTour 1993; Duke, Pickett, Carlson, and Grove 1993; LaTour, Nataraajan, and Henthorne 1993; Spence and Moinpour 1972). The use of fear appeals in social marketing advertising, however, is deemed to be more socially acceptable (Benet, Pitts, and LaTour 1993; Duke, Pickett, Carlson, and Grove 1993). The general acceptability of fear appeals in social marketing advertising is another reason why fear appeals are expected to be used more commonly in a sample of social marketing ads than they might be in a sample of 7 Health threats are highly personal and suggest a vulnerability to disease or death that would tend to be fear-arousing. Therefore, the presence of health threats is believed to increase the likelihood that the message will be fear-arousing. 26 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis ads for products/services.8 Health Threats vs. Social Threats It has been suggested that threats in fear-arousing ads, while most commonly of the health-related variety, can also be related to a social context (Powell and Miller 1967; Tanner, Hunt, and Eppright 1991). For example, rather than using fairly distant threatening health-related consequences like lung cancer in social marketing ads against teenage smoking, more immediate negative social consequences like bad breath could be featured. As well, the literature suggests that most fear-arousing threats are related to consequences for oneself, although it is also possible to have threats which have consequences for close others (Hewgill and Miller 1965; Shelton and Rogers 1981; Stuteville 1970), or even for society in general (Hine and Gifford 1991). For example, DUI ads can threaten jail sentences for the driver, risk of accident or death for loved ones in the car, or risk of killing or maiming unknown others (King and Reid 1990). The present content analysis will code the nature of the threats found in social marketing ads, and identify whether they are health-related or social in nature, and whether the threats have consequences for oneself, close others, or for society in general. This part of the analysis will be exploratory, and no specific hypotheses will be proposed. The incidence of fear appeals in social marketing ads in the Hanneman, McEwen, and Coyne (1973) study was 24%, as compared to the 16.6% incidence of fear appeals in a sample of all types of ads in the Unger and Stearns (1983) study. 27 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Other Dependent Variables Positive Appeals The opposite of a fear appeal could be thought of as being a positive appeal. A social marketing ad could be considered positive if it either: (1) presents the message in an upbeat, light hearted, or humorous manner; or (2) presents the message in a neutral, factual manner without any mention of the negative consequences of failing to comply with the recommended behavior. Much of the literature on the use of persuasive appeals with regard to health-related behaviors has dealt with fear-arousing appeals, and only a few studies have examined "positive" appeals or compared negative (fear) and positive appeals (e.g., Evans et al. 1968, 1970; Robberson and Rogers 1988). Robberson and Rogers (1988) found that a negative fear appeal was more persuasive than a positive appeal; however, when self-esteem was involved, a positive appeal was superior to a negative appeal. In a study involving toothbrushing behavior, Evans et al. (1968) found that a message with positive affect arousal was just as effective as a negative fear appeal. Based on this evidence, some practitioners and academics have suggested that it may be preferable to use positive appeals, or even humorous appeals (Brooker 1981). There may be particular issues which lend themselves more readily to positive appeals than to fear appeals (Boyle 1984; Riggs 1990). As well, there has been some concern that fear appeals may be less effective with particular target segments (Burnett and Oliver 1979; Burnett and Wilkes 1980; Quinn, Meenaghan, and Brannick 1992). Some authors strongly believe that fear appeals are counterproductive (Steele and Southwick 1981) or ineffective (Boyle 1984; Singh 1974; Tripp and Davenport 1989), and that positive appeals should be used instead. The present content analysis will code for the presence of positive appeals and neutral appeals in addition to fear appeals. 28 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Humor The use of humor is common in many types of advertising (Kelly and Solomon 1975; Weinberger and Spotts 1989). There has been considerable controversy over the effect of using humor in advertising, although in general humor seems to improve liking for a commercial, reduce irritation with a commercial, and increase liking for the product advertised (Duncan and Nelson 1985). Two models attempt to describe how humor could influence persuasion (Gelb and Zinkhan 1986). The cognitive model suggests that humor can be used to gain attention for the substance of a message. Enhanced attention leads to more central processing which should result in greater persuasion (Petty and Cacioppo 1986). Alternatively, the classical conditioning model suggests that humor creates positive affect toward the ad, and this positive affect is then transferred to the brand. While these two models suggest that humor may aid persuasion, it is also thought that humor may harm recall and comprehension for complex messages because it acts as a distractor. However, humor is thought to have a greater persuasive impact when the message recipient has a favorable prior brand evaluation (Chattopadhyay and Basu 1990), although it is not clear how this would apply in the case of counter-advertising where brand names are not applicable9. Perhaps if the message recipient has a favorable prior evaluation for counter-ads against the specific targeted behavior (which might suggest a somewhat negative attitude toward the specific targeted 9 In social marketing advertising there is often an undesirable or negative behavior that is being addressed, and a recommended coping response that is provided. The message recipient might have a positive attitude toward the undesirable behavior and a negative attitude toward the recommended coping response. Alternatively, the message recipient might have a negative attitude toward the undesirable behavior and a positive attitude toward the recommended coping response. It is this latter situation which might be analagous to having a positive attitude toward a brand, since a brand is usually presented as a solution to a problem (i.e., a recommended coping response). Therefore, humor may have a greater persuasive impact in social marketing ads where the message recipient has a positive attitude toward the recommended coping response. 29 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis behavior), then the use of humor in the message would have a greater persuasive effect. Although some content analysis studies have examined the use of humor in advertising in general (e.g., Kelly and Solomon 1975; Weinberger and Sports 1989), little is known about the extent to which humor is used in counter-advertising. Gelb and Pickett (1983) used a non smoking ad in their experimental study on humor, remarking that humor has been used previously in non-smoking campaigns in the U.S.: "a Chicago physicians group, Doctors Ought to Care (DOC) began in 1980 to sponsor satiric print and radio advertising, such as an ad for 'Emphysema Slims.'10 Ads parodied cigarette promotions and were designed to make fun of cigarette advertisers" (p. 36). Gelb and Zinkhan (1985) point out that "Freudian theory ... interprets a joke as a way to reduce anxiety" (p. 13), which suggests that humor can be useful in defusing the tension or anxiety inherent in thinking about the types of issues that counter-advertising typically addresses. Therefore, humor may be able to play a role in persuasive health-related messages, although it is expected that it would not be used to a large extent. Weinberger and Campbell (1990-91) point out that humor is used least often in high-involvement product categories (only 10-15% of the time, compared to about 30% of the time for advertising in general). Since counter-ads tend to discuss high involvement topics, this suggests that counter-ads would not frequently make use of humor. While humor will not be used frequently in counter-advertising, when it is used it will tend to be under one or more of the following circumstances: where there is a youthful intended target group; where the consequences of the targeted behavior are less serious or less immediate (e.g., smoking); or where the sponsor of the counter-ad is a for-profit corporation using humor to defuse anxieties and avoid alienating their customers. The Emphysema Slims ad was a take-off on the Virginia Slims cigarette brand. 30 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Humor is ephemeral and difficult to define or measure; however, Sternthal and Craig (1973) suggest three methods. One method (which presents considerable methodological difficulties) is to measure the response of the subject, in terms of arousal, smiles, or laughter. A second method, most commonly used in experimental work on humor, simply requires subjects to give pencil and paper perceptual ratings of humor. These two methods are both suitable for laboratory experiments, where the success of a humor manipulation depends more upon whether the subject has actually perceived it to be humorous, rather than whether the advertiser intended it to be funny (Duncan and Nelson 1985). A third method for measuring humor is to define the type of humor being used, i.e., whether it is a double entendre, incongruity, irony, joke, pun, satire, slapstick, turn of phrase, or understatement (Sternthal and Craig 1973). This categorization of the humor residing within the communication measures whether there is an intention or attempt to be humorous, thus making it a suitable method for humor categorization in a content analysis study. Another humor typology used in content analysis is to define the nature of a humorous communication as being of either the 'aggressive,' 'sexual,' or 'nonsense' variety (Goldstein and McGhee 1972; Madden and Weinberger 1982). This latter typology may have less discriminatory power, however, since one study using this classification scheme found that all ads in their sample belonged to the 'nonsense' category (Madden and Weinberger 1982). A content analysis coding scheme which has previously been used successfully (Kelly and Solomon 1975; Weinberger and Spotts 1989) collected the following information: 1. Humorous intent (yes, no) 2. Prominence/role of humor (central, secondary) 3. Type of humorous device (irony, joke, ludicrous, pun, satire, understatement) 4. Integration of the use of humor with the product/issue; is humor related to the 31 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis product/issue (yes, no) In defining whether humor is integrated with, or related to the product, Weinberger and Campbell (1990-91) provide some definitions: "Unrelated humor is operationally defined as humor that is not closely linked to the product and is almost incidental to the product. Related humor is operationally defined as humor that is linked to the product and the fabric of the commercial" (p. 46-7). The above four coding points were all found to have high intercoder reliabilities of .95, .93, .82, and .89 respectively in the Weinberger and Spotts (1989) study, and therefore these four points present a reliable scheme for coding humor. Previous research has identified cross-cultural differences in the use of humor in advertising. Weinberger and Spotts (1989) have noted that there is a greater use of humor in British advertising than in American advertising, with 35.5% of British ads in their sample containing humorous attempts while only 24.4% of U.S. ads contained humorous attempts. British ads in their study gave humor a more central role, and were more likely to integrate the humor with the product. It is expected that these cross-national differences in the use of humor in general advertising would apply to counter-advertising as well. The present content analysis will code only for the presence or absence of humorous intent. Since the main purpose of the study is to examine the use of fear appeals, humor is of interest only insofar as it may exist when fear appeals are absent, or may be inversely correlated to the presence of fear appeals. Other Message Content Variables Most cross-cultural content analyses conducted previously have focused on general advertising rather than social marketing advertising. This means that some of the message 32 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis content variables coded in previous studies pertain specifically to brand name products or services, and are therefore not relevant to social marketing advertising. For example, Tse, Belk, and Zhou (1989) coded print ads for product-related key concepts, themes, and promises. Zandpour, Chang, and Catalano (1992) coded TV commercials for product categories, Simon's (1971) 10 Creative Strategies, and the Resnik-Stern (1977) Information Content Classification System (e.g., price-value, quality, and other variables relating to products). Because the present content analysis is limited to social marketing ads, the coding schemes used in previous content analysis studies about product advertising are largely irrelevant. The present study will gather data in some categories other than those reported within this thesis, for the purposes of future research. The data reported in this thesis will be that which relates specifically to the use of fear appeals. Factors Influencing the Use of Fear Appeals Little or nothing is known about the factors which facilitate or reduce the use of fear appeals in social marketing ads. Previous content analyses which have examined social marketing ads have not examined factors such as the country-of-origin, the ad's sponsor, the behavior it is targeting, or the assumed characteristics of the intended target audience. The premise of this research is that the use of fear appeals in social marketing advertising varies as a function of the ad's country-of-origin, its sponsor, the behavior it is targeting, and the characteristics of the intended target audience. This content analysis of television social marketing advertising is intended to examine the relationships outlined in Figure 1.6. It is expected that country-of-origin, type of sponsor, the behavior targeted, and the intended target group will each have an impact on the use of 33 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Type of Sponsor Intended Target Group Type of Behavior Figure 1.6 Selected Variables Affecting Use of Fear Appeals in Social Marketing Ads fear appeals in social marketing ads. This is not to say that there may not be other variables which influence the use of fear appeals. However, a thorough examination of the fear appeals literature suggests that these are four interesting variables which have not been examined extensively, and so they are areas that are ripe for research.11 Let us now examine each of these factors in turn. 11 Figure 1.6 should not be thought of as a structural model or a causal model in which all possible influencing relationships are included, but rather as a conceptual picture of four potential influencing factors. The reason that this should not be thought of as a structural model is that there may well be relevant variables which have been excluded. Such variables, if included in the model, could have a significant impact on the size and direction of other relationships in the model. Some of these variables could include factors which affect the decision-makers who are involved in producing these television commercials, and could not possibly be determined through a content analysis of the television commercials themselves. Since one goal of a model is to strive for simplicity, the fewest number of relevant explanatory variables should be chosen. The variables chosen in in Figure 1.6 have been selected as being the most relevant and interesting set of variables which will be investigated in this study. 34 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Cross-national Differences in Advertising It is not known whether the incidence of fear appeals differs as a function of country-of-origin, or whether there may be cross-national differences in the intensity of fear appeals. Several studies have found evidence that advertising in general differs significantly between countries. However, the incidence of fear appeals between countries has not been examined. The study of cross-national differences in advertising is an area that has been increasingly gaining attention from academics, particularly during the past decade. Unique cultural, social, political, and historical factors within each country contribute to differences in cognition (Pick 1976; Triandis 1965), as well as differences in attitudes, values, and beliefs (Davidson and Thomson 1976; Hall 1990; Hofstede 1984). These differences in cognitions, attitudes, values, and beliefs are, in turn, likely to result in variations in the kinds of advertising that are produced across countries. For example, there may be cross-cultural differences in the types of persuasive approaches and message content that are used in social marketing ads. The extent to which fear appeals are used may differ between countries. Content analysis is a technique that has been of considerable value in examining cross-cultural differences in advertising messages. A number of cross-cultural content analyses of advertising have previously been conducted, including a longitudinal study of print ads from Hong Kong, the People's Republic of China, and Taiwan (Tse, Belk, and Zhou 1989); comparisons of U.S. and Japanese print advertising (Belk, Bryce, and Pollay 1985; Belk and Pollay 1985b; Hong, Muderrisoglu, and Zinkhan 1987; Ramaprasad and Hasegawa 1992); a comparison of French, Taiwanese, and U.S. television commercials (Zandpour, Chang, and Catalano 1992); a comparison of humor in U.S. and U.K. television commercials (Weinberger and Spotts 1989); and a comparison of sex roles in television advertising in Australia, Mexico, and the United States (Gilly 1988). The studies cited have all made use of a probability 35 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis sample of ads appearing in a particular advertising medium. While the use of a probability sample allows for a statistically testable comparison between countries, such a sample of ads is seldom large enough to examine a particular small genre of advertising like social marketing advertising. Therefore, it is not surprising that none of the above cross-cultural studies have specifically examined social marketing advertising. The research which is the subject of this dissertation is intended to fill this gap in the literature, by conducting a cross-cultural content analysis of television social marketing advertising. The television medium has been chosen because it is a common choice for carrying social marketing ads and PSAs, primarily due to its mass reach and dramatic impact. The use of a cross-cultural approach will illuminate similarities and differences in the use of fear appeals in social marketing TV advertising between the countries under study. It is anticipated that country-of-origin (i.e., the country in which the social marketing ad was produced) will have an impact on the use of fear appeals in social marketing ads. For example, previous studies have suggested that Australia uses fear appeals in their social marketing ads against AIDS to a greater extent than the United States does (Johnson and LaTour 1991; LaTour and Pitts 1989). Similarly, Australia is noted for its fear-arousing approach in its ads for the Traffic Accident Commission (TAC). The reason for the perhaps higher use of fear appeals in Australia may lie in the collective Australian psyche of machismo (Connell 1991; Moraes-Gorecki 1988). Australian society, particularly Australian men, comprise a rough and ready society, reminiscent of the old West in North America (Gilding 1992). It is thought that this toughness may necessitate a more hard-line approach in social marketing ads in Australia, whereby the message recipient must be jolted out of complacency by being scared or frightened into action (Loxley, Saunders, and Blaze-Temple 1990). Therefore, the following hypothesis is proposed: 36 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis H3 It is expected that a higher percentage of social marketing ads in Australia will use fear appeals, as compared to the United States. Sponsorship The sponsor of an ad is usually the party who is paying for the ad. For a traditional product ad, the sponsor is the manufacturer or company (e.g., Proctor and Gamble), and the clear-cut purpose of the ad is to promote that company's brand name product (e.g., Tide detergent). The purpose for advertising is definitely corporate self-interest; Proctor and Gamble uses advertising to increase sales of Tide detergent. In the case of social marketing advertising, sponsorship and purpose are not always so clear-cut. For example, social marketing advertising against DUI (driving under the influence of alcohol) is often sponsored by federal or provincial government departments, such as Health Canada or the Department of the Attorney General. In this case, these two sponsors are both promoting the same general social objective or "product," since they both wish to discourage DUI; however, they each have vested interests which can strongly influence the content of their social marketing advertising messages. Health Canada is likely more interested in the health aspects of DUI, so their advertising message may be more likely to emphasize the threatening health consequences; the coping response they would likely recommend would be abstention or reduced alcohol consumption. On the other hand, Department of the Attorney General is more concerned about the legal aspects of DUI, so they would likely emphasize the threatening legal consequences, such as jail sentences or suspension of one's driver's license; their recommended coping response would more likely include planning alternative transportation strategies, instead of reducing the amount of alcohol consumed. In this case, two government departments have the same goal of reducing DUI, but each department would 37 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis have a very different communications approach.12 When the sponsor of a social marketing ad against DUI is a for-profit corporation, such as a brewery or distiller, the situation becomes even more complex. Such corporations do not really want to ask their audience to drink less, because that would go against their corporate self-interest. When a brewery or distiller sponsors a social marketing ad against DUI, it seems likely that they would minimize the threat content of their DUI social marketing ad. Rather than emphasizing the negative health or legal consequences of DUI, their message would perhaps simply advise that drinking and driving don't mix. In essence, their coping advice would not be of the sort that would substantially reduce the amount of alcohol consumed. This point is reinforced in a review of alcohol moderation/DUI campaigns which were sponsored by Anheuser-Busch, Coors, and Miller, the three major brewing companies in the U.S. (DeJong, Atkin, and Wallack 1992). This study points out that some alcohol moderation TV ads are virtually indistinguishable from beer ads when the volume is turned off. Further, these ads make a point of promoting the social acceptability of alcohol consumption (DeJong, Atkin, and Wallack 1992). Sometimes brewers or distillers sponsor social marketing advertising which encourages drinkers to consume alcohol responsibly or in moderation. At first glance, advertising against over-drinking might seem to be an irrational act on the part of the corporate sponsor, since telling people to consume less of their alcoholic beverage product may have a negative impact on the sponsor's sales and profits. However, there are "ulterior commercial motives" (Oldenburg 1992, p. 23) involved in this type of advertising, since some corporate benefits 12 A further problem exists when governments earn revenue as a result of products such as tobacco or alcohol. Campaigns to reduce usage of these products among the public may be weighed against the amount of revenue lost from reduced sales of these products as consumption declines. 38 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis accrue to the sponsor. One benefit could be that the company is perceived to be more socially responsible or a good corporate citizen, descriptions which may be highly desirable in forestalling restrictive legislation or strengthening the case for industry self-regulation. A second benefit might be that such advertising actually reinforces the social acceptability of consuming alcohol in moderation, thereby enhancing the acceptability of alcohol consumption in general, and making moderate and light drinkers feel more secure about the social acceptability of their behavior. A third benefit of such advertising is that it puts the corporate sponsor's name in the public eye, as in the case of a Seagram's television campaign which highlighted the equivalency of various serving sizes of beer, wine, and spirits, thereby covertly advertising their brand on television in contravention of industry guidelines13 (Tucker, Hovland, and Wilcox 1987). Finally, depending on message content and style, this type of social marketing advertising against immoderate drinking may have relatively little impact on heavy drinkers or problem drinkers, so the sponsor's customer franchise would not necessarily be seriously threatened. Because a significant amount of time, money, and effort is involved in producing a television ad with a social marketing message, the sponsor inevitably expects some type of payback. In the case of a government or nonprofit organization, the desired payback is simply a reduction in the incidence of the targeted behavior. Sometimes governments may gain a long run financial benefit from reducing the incidence of the targeted behavior (e.g., reducing 13 On November 7, 1996 the U.S. liquor companies announced that they were lifting a 48-year old self-imposed ban on television and radio advertising. This significant change in the liquor industry's advertising guidelines was probably introduced in an attempt to stem a 15-year decline in consumption of hard liquor. However, the result has been a public backlash which may lead to legislation limiting the liquor industry's ability to advertise on Tv and radio (Leonhardt and France 1996). 39 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis publicly funded health care costs).14 More often, the payback is simply a reduction in social or human costs, so there is a relatively low degree of self-interest in government social marketing advertising. However, in the case of a self-interested for-profit corporate sponsor, there are likely to be other sorts of desired payoffs besides (or instead of) a reduction in the targeted behavior. It may be more important to the for-profit sponsor to be perceived to be taking action against the problem behavior; the degree of behavior change that actually occurs is secondary. The self-interest of the sponsor can be thought of as being placed somewhere along a continuum, with governments15 and non-profit organizations at the low end of the self-interest spectrum, and breweries or distillers at the high end of the self-interest spectrum (see Figure 1.7). Governments & Nonprofit For-profit Organizations Corporations LAW Self-interest High Self-interest Figure 1.7 Continuum in Sponsor Self-interest A sponsor's degree of self-interest is expected to influence the content of the social marketing ad messages that the sponsor is willing to put forward. Because sponsors with a high level of self-interest are likely to be more concerned about alienating their audience, one 14 Typically, the long term gain from reduced health care costs would more than offset the short term loss from reduced government tax revenues on products like tobacco or alcohol. 15 For the purposes of this model, governments will be assumed to be concerned only about the well-being of citizens, and unconcerned about the potential loss of tax revenues from reduced sales of tobacco or alcohol, and equally unconcerned about the potential loss of votes or campaign contributions from manufacturers of alcohol or tobacco products. 40 Chapter One Examining Fear Appeals in Social Marketing Television Ads through Content Analysis would expect to see a reduced incidence of fear appeals, and an increased use of humor or positive appeals in their social marketing ads. A second issue regarding the use of fear appeals in social marketing advertising by for-profit corporations concerns situations where fear appeals are used to encourage use of the sponsor's product as part of the recommended behavioral change (e.g., condom use to prevent disease). For example, high fear appeals in persuasive messages against unsafe sex (i.e., use condoms to reduce the spread of AIDS) have generally been found to be effective (LaTour and Pitts 1989; Plympton and Hibbard 1991; Rhodes and Wolitski 1990; Witte 1991). However, high fear appeals against unsafe sex which are sponsored by a particular brand of condoms have been found less effective than more moderate fear appeals (Hill 1988; Struckman-Johnson et al. 1990). In this case, the self-interest of the sponsor, or the trust that the audience has in the sponsor, may have an impact upon the perceived credibility of the sponsor's social marketing ads. This reduction in source credibility would have an impact on the persuasiveness of a fear appeal (Hewgill and Miller 1965; Stainback and Rogers 1983). However, if "there is a high degree of compatibility between the public good and advertisers' self-interests" (LaTour and Zahra 1989, p. 67), then fear appeals seem more acceptable. This provides further support for the idea that use of fear appeals among corporate sponsors is likely to be fairly limited. Therefore, the following hypothesis is proposed: H4a Corporate sponsors will be less likely than other organizations to use messages which incorporate threats or fear appeals. H4b Corporate sponsors will be more likely to use messages which are positive. H4c Corporate sponsors will be more likely to use messages which are humorous. 41 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Targeted Behavior It has been said that the content of advertising in general mirrors or reflects a society's cultural values (Belk and Pollay 1985a; Holbrook 1987; Pollay 1983, 1986; Pollay and Gallagher 1990). If this is so, then it seems likely that the content of social marketing advertising mirrors or reveals society's social concerns, including concerns about what has been referred to as the "dark side of consumer behavior" (Hirschman 1991, p. 1). On the other hand, the prevalence of a particular social problem may not necessarily be reflected in social marketing advertising. A study by Wallack and Dorfman (1992) indicates that none of the PSAs found in their content analysis sample were for tobacco, alcohol, or diet, three leading behavioral risk factors for poor health. However, their sample contained only 14 health message PSAs (out of a total of 654 TV commercials in a 20-hour sample of TV programming). The type of behavior which is targeted is expected to have an impact on the social marketing ad's message content variables. The potential threats or consequences associated with targeted behaviors will differ in their level of seriousness and probability of occurrence (Rogers 1983).16 Social marketing ads against behaviors with potentially serious consequences, such as unprotected sex which may result in the deadly spread of AIDS, may be more likely to use fear appeals as a means of reinforcing the seriousness of the outcome. On the other hand, social marketing ads against unhealthy behaviors such as smoking may be be less likely to use fear appeals and more likely to use humor or positive approaches. The seriousness of the threat of contracting AIDS through unprotected sex would be perceived to 16 There may be differences in perceptions of seriousness of behavioral consequences when comparing the social marketer's point of view to the general public's point of view (or a specific target group's point of view). For the purposes of this research, it will be assumed that seriousness is being measured from the general public's point of view. 42 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis be greater than the seriousness of the threat of contracting lung cancer through smoking, because AIDS virtually always results in death, while lung cancer victims may lose a lung but still survive.17 The more serious or threatening the potential consequences, the more likely it is that fear appeals could be used successfully in social marketing ads against that behavior. Sometimes the threatening consequences of engaging in a behavior are more social than physical. For example, anti-smoking ads might threaten either social rejection or lung cancer. The threat of social rejection from smoking may be much less serious than the health threat of lung cancer; however, the probability of suffering social rejection might seem much higher and more immediately relevant than the probability of contracting lung cancer. Harmful social consequences would seem to lend themselves more readily to a positive or humorous approach (Robberson and Rogers 1988). Therefore, the following is hypothesized: H5a The more serious the potential consequences of a targeted behavior, the more likely it is that fear appeals will be used in social marketing ads against that behavior. H5b The less serious the potential consequences of a targeted behavior, the more likely it is that positive appeals will be used in social marketing ads against that behavior. H5c The less serious the potential consequences of a targeted behavior, the more likely it is that humorous appeals will be used in social marketing ads against that behavior. 17 The perceived probability of threats can also differ (although this is not investigated in this study). The perceived probability of suffering harm from DUI might seem greater than the perceived probability of suffering harm from smoking cigarettes, because the chance of getting caught by police with a breathalyzer test or getting in a DUI car accident may seem greater than the chance of contracting lung cancer from smoking. 43 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Intended Target Audience Content analysis usually examines the manifest content of audio, video, or text. It catalogues the actual content, without making inferences about the communicator's intentions. However, in some cases it is worthwhile to contravene this content analysis convention in order to make inferences about the intentions of the ad sponsor. In the present content analysis, a variable known as "intended target audience" will be coded to indicate situations where the social marketing ad seems overtly aimed at a particular target group. Television is a mass medium, and so many of the social marketing TV ads will simply be aimed at a mass audience. However, some social marketing ads will appear to be overtly targeted at a particular audience segment, and it is important to recognize this when it occurs. For this reason, when a social marketing TV ad appears to be overtly targeted at a particular segment (e.g., women, teens, homosexuals, the elderly), then it will be coded as such. There have been some criticisms of target marketing. Targeting that involves consumers who are viewed as equal participants in transactions is regarded as acceptable, but objections are almost certain if targeting entails disadvantaged or vulnerable consumers participating in transactions involving potentially addictive products such as alcohol and cigarettes (Ringold 1995). However, targeting of social marketing campaigns has typically not been criticized, but instead is seen as a necessary means of conserving the scarce resources of governments and nonprofit organizations (Kotler and Andreasen 1996; Lovelock and Weinberg 1989). Targeting specific segments of the population can be used to address target groups which have a greater incidence of a given problem (Wyllie and Casswell 1993), or can be used in prevention efforts. Targeting of specific target audiences tells us something about how society views and treats those particular target groups, and about how society balances the special needs and 44 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis concerns of those groups against the needs and concerns of society as a whole. Cross-national differences in choice of intended target group will be illuminating, since this may reveal something about differences between countries regarding the way in which these specific groups are viewed. This analysis will be exploratory, and no specific hypotheses will be put forward regarding the interaction of intended target audience and country-of-origin. The choice of intended target group is expected to have an impact on the use of fear appeals. For example, social marketing ads directed specifically at youths should look substantially different than social marketing ads directed at adult target groups. Adolescents have different perceptions of risk than adults do, and are more likely to engage in high risk behavior (Burns, Hampson, Severson, and Slovic 1993; Severson, Slovic and Hampson 1993). Also, the beginnings of substance use (alcohol, drugs, cigarettes) and sexual behavior often begin in adolescence, and early patterns may have an impact on later behavior. The mass media has often been pointed to as being a prime source of information about sexuality, cigarettes, and alcohol (Davies 1993). Humor, positive approaches, or the use of celebrity spokespersons may be more appropriate or prevalent in social marketing ads directed at teens, while fear appeals may be less prevalent. It has been suggested that youth generally feel invulnerable to threats or negative occurrences, and that they discount both the seriousness of threats and their probability of occurrence (Lastovicka, Murry, Joachimsthaler, Bhalla, and Scheurich 1987). If this is so, then fear-arousing social marketing ads may be less effective with a youthful target group. Therefore, social marketing ads targeted at youth might be more likely to use positive/humorous appeals. For this reason, the following hypotheses are proposed: H6a Social marketing ads targeted at youth will be less likely to make use of fear appeals. 45 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis H6b Social marketing ads targeted at youth will be more likely to make use of positive appeals. H6c Social marketing ads targeted at youth will be more likely to make use of humorous appeals. Social Marketing Campaigns There have been many recent calls to marketers and consumer psychologists to become involved in social marketing research. For example, Hirschman (1991) recommends addressing research toward the "dark side of consumer behavior" (p.l), while Andreasen (1993) suggests that research into social marketing can address some of society's many ills and contribute to quality of life. Goldberg (1995) suggests taking a more radical approach in social marketing that emphasizes changing the negative social structural influences on individual behavior. Wallack (1984) emphasizes that prevention is a collective responsibility that must be shared, and not simply placed solely on the individual who is at risk. Social marketing advertising can be an important part of health promotion campaigns. Health promotion has been defined as the process of enabling people to increase control over their health, and to improve it (Kelly 1990; Ottawa Charter for Health Promotion 1986). A comprehensive approach to health promotion encourages individuals to adopt and maintain personal behaviors that would prevent disease and promote health, while discouraging health-damaging behaviors and eliminating health hazards from the environment (Breslow 1990; Brown 1991). In the United States, the Healthy People 2000 program is one such approach (Mason and McGinnis 1990; Mason 1990, 1991). 46 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Mass Media and Social Marketing Existing evidence indicates that mass media efforts to improve public health through social marketing campaigns can be highly influential (DeJong and Winsten 1990; Roberts and Bachen 1981; Solomon 1982). Mass media can be particularly effective in the early stages of a health promotion campaign due to its ability to expose large numbers of individuals to new ideas or behaviors (Green and McAlister 1984). Mass media campaigns can successfully accomplish some of the following tasks: (1) increase awareness of a health problem; (2) raise the level of information about health topics; (3) make a health topic or problem more salient, making the audience more receptive to other campaign efforts; (4) stimulate interpersonal influence via conversations with family, friends, and health professionals; (5) generate information seeking behavior; and (6) reinforce existing attitudes and behaviors (Schlinger 1976). Although mass media can be highly successful at raising awareness and knowledge, it is somewhat more difficult to effect attitude change and motivation to act (Novelli 1989). The power of mass media is sometimes overemphasized (Tomes 1981), and many mass media campaigns do not achieve their objectives in terms of changing attitudes or behavior (Budd and McCron 1981). It has been suggested that mass media is most effective in initiating behavior change if the change sought is relatively minor and if consumers have low information needs (Robertson and Wortzel 1978). Mass media campaigns are most likely to induce permanent attitude and behavior change if they are presented and repeated over long time periods via multiple sources, at different times of the day, using novel and involving techniques, using appeals to multiple motives, focusing on developing social support, and providing appropriate behavioral skills, alternatives, and reinforcement (Flay, DiTecco, and Schlegel 1980). In any PSA campaign it is important to ensure that commercial messages are given sufficient time, 47 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis frequency of exposure, and consistency in order to ensure their effectiveness (Wooden 1994).18 Choosing Social Marketing Topic Areas In developing this content analysis, it was necessary to choose social marketing topic areas which would fit the following criteria: • were important public health issues • were the subject of a sizeable number of television commercials • were relatively similar to one another in terms of psychology, motivations, and behavior change processes. The five behavioral areas that were selected as the topic areas in this content analysis are smoking cessation, illegal drug use, DUI, alcohol moderation, and AIDS prevention. These five areas fit the criteria of being important public health issues, and have each been the subject of a relatively large number of television commercials. Also, based on a typology of health change behaviors presented in Table 1.2 (drawn from Cooper-Martin and Stephens 1990), it appears that the behaviors under study in this content analysis are fairly similar to each other, in terms of their frequency, convenience, etc., as explained below. 18 Recall and persuasiveness of an ad decay with the passing of time, hence the need for repetition to keep the message top-of-mind. 48 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Table 1.2 Comparing Health Change Behaviors Characteristics of Health Change Behaviors Health Change Behavior Frequency Convenience Individual/ Passive Private •' Public Consequences Smoking cessation Every time Low Individual Private & public Others & self Illegal drug use cessation Every time Low Individual Private Others & self DUI cessation Every time Low Individual Private & public Others & self Alcohol moderation Every time Low-moderate Individual Private & public Others & self Condom usage Every time Low Individual Private Others & self Source: Based on Cooper-Martin and Stephens 1 [9% All of the five health change behaviors shown in Table 1.2 are high involvement19, and changing them is a process that must be repeated over and over again. It is relatively inconvenient to adopt the recommended remedial behaviors, and all five problem behaviors can be characterized as being an individual's choice, rather than being passive situations in which individuals do not make choices. All five behaviors have a large private component (i.e., they are often conducted in private), and four out of the five also have public components (only AIDS prevention is a completely private act). All five health change behaviors have consequences not only for the individual who engages in the problem behavior, 19 High-involvement behavior can be thought of as existing under one or more of the following conditions: (1) the behavior will reflect upon the individual's self-image; (2) there are high economic or personal costs for behaving incorrectly; (3) there are high personal or social risks of making a wrong decision about the behavior; and (4) outside reference group pressures to act in a particular way are strong and one's motivation to comply is strong (Kotler and Andreasen 1996). 49 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis but also for others around them. An additional problem entailed in smoking, illegal drug use, and alcohol use is the addictive component, which makes it much more difficult for individuals to voluntarily control or change their behavior (Hirschman 1995). The recommended coping response for four of these behaviors is to simply quit or stop engaging in the behavior; however, for alcohol moderation the appropriate recommended coping response is to reduce alcohol consumption, not necessarily to quit drinking completely. Because there is a large degree of overall similarity in these five behaviors, they were selected as topic choices in this content analysis. The following sections will briefly review some of the literature in each of these five topic areas, and highlight any content analyses or other studies which may be relevant. ADDS Prevention Campaigns When social marketing campaigns for AIDS prevention started in North America and other parts of the world in 1985-86, they were considered to be highly controversial (Bush and Davies 1989; Sherr 1987). Prior to that time, advertising for condoms was seldom aired on television, and was not accepted by the major US networks. However, TV networks had to decide whether their responsibility was to inform, or simply not to offend (Litman and Bain 1987), with the ultimate result that most television stations now accept advertising for condoms and welcome ads for AIDS prevention. This trend has also followed in other countries: for example, South Africa okayed condom ads in 1993, and the South Africa Broadcasting Corp. received very few complaints in spite of their fears about public outrage (Barnes 1993). It seems apparent that the nature and role of AIDS campaigns has changed over a period of several years. For example, in the U.S. the "America Responds to AIDS" advertising campaign progressed from building an awareness of the facts in 1987, to building 50 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis worry and fear in 1988, to providing a coping response in 1989 (Bush and Boiler 1991). Critics have charged that this campaign used minimal segmentation or targeting (Winett, Altman, and King 1990). However, in reality, many targeted commercials were actually produced by the Centre for Disease Control (CDC) to aim at specific targets such as blacks, Hispanics, families, etc. Unfortunately, the choice of which commercials to use was left up to individual states, and most chose to use commercials which had bland general messages (Bush and Davies 1989). Public service announcements on AIDS prevention tend to have a high level of credibility or believability among college students, but college students say that they are somewhat less likely to act upon or use the information provided in PSAs as compared to interpersonal sources (Cline and Engel 1991). Pamphlets appear to have a high degree of credibility, although when a pamphlet was sent out to every home in the United States in 1988 only half the sample in one study remembered receiving it (Snyder 1991). Australian AIDS prevention ads have been some of the most explicit and forceful (Pritchard 1993). In Australia the AIDS campaign which began in April 1987 was launched with the "Grim Reaper" commercial, which "showed a series of death-like images ('Grim Reapers' with skeletal faces, each carrying scythes) wreaking destruction at bowling alleys with bowls aimed with deadly intent at the 'pins' in the form of men, women and children unable to avoid the relentless onslaught.... The Australian campaign was distinctive in that there was particularly heavy emphasis upon the use of shock tactics to arouse public interest and to shake people out of a supposed state of apathy about AIDS.... It was decided that the Australian campaign should begin with a strong message giving an emotional shock to the general community in order to raise awareness and public concern" (Ross, Rigby, Rosser, Anagnostou, and Brown 1990, p. 340). This ad was clearly intended to be fear-arousing, given its strong emphasis on the negative consequences of unsafe sex, as well as its attempt 51 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis to arouse fear through the use of frightening images and gloomy atmosphere (LaTour and Pitts 1989). The "Grim Reaper" commercial was highly memorable, with 90% of respondents recalling the commercial even after a period of 5 months had passed since its last airdate. The campaign was successful in changing attitudes and beliefs, even though there was no change in knowledge levels (Ross et al. 1990; Rigby et al. 1989). One problem with AIDS prevention campaigns may be a reluctance on the part of campaign sponsors to provide explicit advice to viewers (Prewitt 1989). For example, a Norwegian campaign which used the slogan, "Talk about sex, about being in love, and about love," referred to AIDS only indirectly, and did not generate higher condom usage among those adolescents who were aware of the campaign (Traeen 1992). This campaign might have been more successful if it had been more direct and provided more specific instructions about condom use (Traeen 1992). A campaign produced by the Health Education Authority in Great Britain in Spring 1988 used two television commercials aimed at 16-30 year olds featuring "boy meets girl" scenarios. These two commercials were evaluated in qualitative research conducted by the Scottish Health Education Group: "Both films emphasized that the risk of AIDS from heterosexual contact should influence the decision of either sex to sleep with a new partner. The final slogan of each film was 'AIDS: you know the risks, the decision is yours.'... The films were also severely criticized [by subjects in the research study] because they offered no help on how to reduce the risk of heterosexual transmission. They simply highlighted the problem and passed it on to the audience without attempting to suggest solutions. In short, they ended where they should have begun." (Hastings, Eadie, and Scott 1990) On the other hand, a successful campaign in the Netherlands was directed at young people and targeted inappropriate beliefs about the transmission of HIV and STDs which were considered to function as rationalizations and excuses for individuals and their sexual partners not to take preventive measures (de Vroome, Sandfort, de Vries, Paalman, and Tielman 1991). 52 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis In a study of 21 international AIDS TV commercials which were evaluated by an audience sample of 56 laypersons, those announcements perceived as the most educationally effective were those presenting hard facts about AIDS in a simple, straightforward manner (Baggaley 1988). Interestingly, however, three of the five PSAs which were perceived to be the most effective were never sanctioned for broadcast use in their respective countries, presumably because they were considered to be too explicit. Conversely, those PSAs perceived by the audience to be least effective had been broadcast in their respective countries much more frequently, undoubtedly because they were relatively bland and inoffensive. Some research suggests that while college students appear to be factually knowledgeable about AIDS, some still report engaging in high risk sexual behaviors (Golden and Suder 1994). Also, men and women feel different emotions when they think about AIDS, indicating that there may be a need for separately targeted health promotion efforts for the two sexes (Golden and Suder 1994). Some research has suggested that the most important determinant of condom use behavior is perceived normative pressure, which suggests that mass media campaigns should focus on this aspect (Fishbein, Middlestadt, and Trafimow 1993).20 One concern about mass media, however, may be that those target groups most at risk for contracting and spreading AIDS distrust the accuracy and objectivity of the media (Reardon and Richardson 1991). Media monitoring and analysis is one method that has been advocated for determining key issues and information that need to be communicated publicly with regards to AIDS (Schechter, Middlestadt, and Doner 1993). Freimuth, Hammond, Edgar, and Monahan (1990) conducted a content analysis of 127 20 In terms of creating normative beliefs, it is interesting to note that Sapolsky and Tabarlet (1991) compared the amount of sex shown in primetime television programming in 1979 versus 1989, and found that issues of safe sex or sexually transmitted disease were rarely discussed on television programs. 53 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis AIDS PSAs in the United States. Results show that these PSAs were generally not targeted according to risk behavior, but were instead directed to general audiences. The messages tended to use rational rather than emotional appeals, provided some facts to the audience, and encouraged the audience to seek more information rather than adopt a specific preventive behavior. Fear appeals were used in 26% of the spots. For example, a PSA entitled "Dance With Death" portrayed a young woman dancing with various partners, the last of whom is a specter. This particular study is similar in methodology to the present research, and specific comparisons can be made between the two studies with regard to a few of the variables. Calvert, Cocking, and Smrcek (1991) conducted a content analysis of 31 AIDS prevention PSAs using a sample which included 10 German spots, 9 Austrian spots, and 12 American spots. This study found that the commercials from these three countries were remarkably similar in terms of content, but this was primarily because they all presented very little factual information. No specific target audience was selected for any of the commercials, making it difficult for viewers to feel that the commercials might be personally relevant to them. Lupton (1992) conducted a content analysis of articles mentioning AIDS published in the Australian press between July 1986 and June 1988, and found that the focus of reporting of AIDS changed during that time from representing AIDS as a risk to only homosexuals and intravenous drug users, to generating panic-stricken articles suggesting that everyone was now threatened in support of the "Grim Reaper" PSA campaign. Articles in the popular press may parallel the types of messages that would be seen in PSAs, since PR campaigns may serve as an alternative to paid or PSA media campaigns. 54 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis DUI/Alcohol Moderation Campaigns Alcohol consumption appears to be widespread on television, according to content analyses of television programming (Breed and Defoe 1972; Gerbner, Morgan, and Signorielli 1982). Content analysis of alcohol ads has demonstrated that these ads present favorable portrayals of alcohol, including benefits such as social camaraderie, escape, refreshment, relaxation, social approval, romance, and elegance (Atkinand Block 1984). Watching alcohol advertising can actually prime further alcohol consumption among those who have already had a drink prior to viewing (McCarty and Ewing 1983). It has been suggested that the alcohol industry damages the black community by targeting them with specific products and promotional programs (Hacker, Collins, and Jacogson 1987). Furthermore, marketers introduce new products which specifically target and appeal to young drinkers either by their flavor (Goldberg, Gorn, and Lavack 1994) or high alcohol content (Vancouver Sun 1993). Of particular concern is the fact that these alcohol messages reach children and teens. Children spend more time watching television than they spend in school (Singer 1983), so there is a significant opportunity for exposure to beer ads and alcohol images. Even if children are not presently consumers of a product class, exposure to TV advertising for adult products can influence their perspective (Gorn and Florsheim 1985), and television advertising can play a dominant role in shaping children's product preferences (Atkin 1982). Even very young children aged three to six years old can be aware of the intent of television commercials (Donohue, Henke, and Donohue 1980). The exposure of children and teens to these glamorized images and attractive new alcohol products is an important reason why social marketing ads with messages about alcohol moderation are needed. Several authors have argued for increasing the number of alcohol counteradvertising 55 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis and PSAs, as well as for increasing their effectiveness (Andrews, 1995; Mazis, 1990). Hudson and Bloom (1984) have tried to demonstrate how eight social change strategies (Sheth and Frazier 1982) can be used by social marketers to create more effective communications campaigns in order to combat the drinking and driving problem. These eight strategies are: (1) informing and education; (2) persuasion and propaganda; (3) social controls; (4) delivery systems; (5) economic incentives; (6) economic disincentives; (7) clinical counseling and behavior modification; and (8) mandatory rules and regulations. Social marketing advertising and mass media can play a role in all eight of these strategies (Murry 1991). An interesting aspect of DUI campaigns has been the use of segmentation and target marketing (Wyllie and Casswell 1993). Some anti-drinking and driving campaigns have been successful in reducing the incidence of drinking and driving among youths (Murry, Lastovicka, and Stam 1991). While most DUI campaigns are directed at young male drinkers (Lastovicka, Murry, Joachimsthaler, Bhalla, and Scheurich 1987), or at the general public, there is some evidence that there may be a need for campaigns which are directed at young women (Celis 1994). There seem to be many recent DUI campaigns which have elected to take a fear-arousing approach, including those in the Canadian provinces of Quebec (Chiasson 1994), Alberta (Bullick 1994), and Ontario (Kryhul 1994). It has been suggested that fear appeals may not be effective among young men, since a young man's desire to maintain a macho image might preclude allowing himself to be influenced by fear (Lastovicka et al. 1987). However, other researchers have called for the use of stronger fear arousal arguments, presumably because of their effectiveness with other target audiences (King and Reid 1990). 56 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Anti-Smoking Campaigns Why do youths continue to smoke despite the well-known health risks? The uptake of smoking is associated with beliefs or stereotypes that picture smokers as having socially desirable attributes (Chapman 1985; Pechmann and Ratneshwar 1993), and cigarette advertising helps to perpetuate these positive beliefs (Pollay and Lavack 1993; Pollay, Siddarth, et al. 1996). The impact of cigarette advertising and sponsorship on young children is of particular concern (Aitken, Leathar, and O'Hagan 1985). It appears that counteradvertising and PSAs have played an important role in facilitating a sharp reduction in cigarette consumption (Schuster and Powell, 1987; Warner, 1977), although it is difficult to accurately estimate the magnitude of this effect (McAuliffe, 1988). While some anti-smoking campaigns such as the California campaign have been credited with reducing the incidence of smoking (Tyrer 1994), others like Health Canada's Break Free campaign have been deemed to be rather ineffective (Sutter 1994). There has been criticism of some anti-smoking ads which have had the unintended side-effect of offending certain societal groups (Chapman 1988). For example, there are two commercials produced in Australia by the New South Wales Department of Health which: show amputation as a consequence of the peripheral vascular disease that is caused by smoking. In one, a youth is seen in an artificial limb factory saying "some people think that smoking is attractive.... I don't think artificial legs are very attractive". The other is a poster of a line of one-legged people which is captioned "Gangrene — one of the least attractive aspects of smoking".... After the screening of this advertisement in Britain, the parents of a child with an artificial leg complained that the child was distressed that artificial legs were being described as unattractive. The advertisement subsequently was withdrawn from broadcasting. (Chapman 1988, p. 259). Another example of smoking prevention ads which have been singled out for criticism are those which focus on bad breath as a motivation for young women to quit smoking. These have been criticized as being trivial and belittling of women's health concerns. However, the 57 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis reality is that young women are much more motivated to quit smoking due to social concerns about bad breath rather than because of fears about lung cancer or other future health effects (Chapman 1988). Perhaps social marketers and health promoters need to think more about what types of messages will change the behavior of their target markets, and worry less about what the general public will think (Willard 1985). Anti-Drug Abuse Campaigns Much effort goes into developing and testing effective anti-drug abuse campaigns (Ray and Ward 1976; Freimuth, Greenberg, Romano, and Wagner 1981), and some, such as the campaign run by Florida in the late 1970s, have been highly effective (Wotring, Heald, Carpenter, and Schmeling 1979). However, not all anti-drug abuse campaigns are effective. For example, Smart and Fejer (1974) found that anti-marijuana messages had little impact on high school students because they already had well-developed orientations toward the drug. Further, Ray and Ward (1976) found that there was a high counterargument rate among older teenagers watching anti-drug abuse television PSAs. Feingold and Knapp (1977) found that anti-drug abuse radio PSAs were not only ineffective with teenagers, but led to a boomerang effect by stimulating interest in drugs. As well, Hanneman and McEwen (1973) found that drug abuse appeals were often telecast during low audience viewing times, and were directed at a general audience rather than a specific target group. Drug abuse campaigns of the late 1960s and early 1970s seemed to be dominated by negative messages that relied on anxiety or fear-arousing content (Capalaces and Starr 1973). Even a more recent campaign from 1989/1990 used fear tactics to frighten viewers into feeling that drugs cause harm: "This is drugs," the announcer says as butter burns in the frying pan. A broken egg drops into the frying pan and as the egg sizzles, the announcer continues, "This is your brain on drugs ~ Any questions?" (Calvert, Cocking, and Smrcek 1991) 58 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Some of the recent drug campaigns have relied on positive themes, rather than fear appeals, as evidenced by this description of a voice-over from a recent TV commercial produced by the Partnership for a Drug-Free America: "Last night, Lisa Watson got higher than she's ever been," the spot's voice-over says, showing blurred pictures of a girl. It continues, as Lisa comes into focus, completing a gymnastics dismount: "And the only thing she took was first place." (Cleland 1994) The Partnership for a Drug-Free America, a coalition of ad agencies and the media, has been spending over $250 million per year in donated media time and space for the last several years (Cutler and Thomas 1994). Hanneman and McEwen (1973) conducted a content analysis of anti-drug abuse commercials. Among their findings was the fact that drug abuse appeals are often telecast late at night or during other low audience viewing times, rather than during prime time, which tends to limit the viewership of these commercials. Another finding was that many of these types of spots are directed at a general audience rather than a specific target group, which may also limit their effectiveness. Walker (1990) conducted a content analysis of 46 anti-drug abuse TV commercials and 60 print ads produced between 1980-89. A key finding was that 52% of ads made use of fear appeals, and 84% of the PSAs used at least one fear-inducing element in the ad. Fear appeals seemed to be used more frequently when targeting children or teens, or when addressing parents. Interestingly, 56% of spots in this sample specifically mentioned cocaine or crack, rather than simply discussing drugs in general. Summary This Chapter has attempted to outline the background literature that is relevant to the content analysis study that will be described in Chapters Two and Three. Several models of 59 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis fear appeals were outlined, and justification was provided for coding the characteristics of the Ordered Protection Motivation model in the content analysis. As well, some previous content analysis studies which have examined incidence or use of fear appeals were outlined. In addition, some of the literature on the structural characteristics that may affect the use of fear appeals was discussed, including sections on country-of-origin, type of sponsor, intended target audiences, and types of targeted behaviors. Finally, literature on each of the five types of targeted behaviors was described, with particular reference to any content analysis studies which had been conducted in these areas. Chapter Two will describe the methodology used in conducting the content analysis, as well as the characteristics of the sample. Chapter Three will then present the results of the content analysis, with particular reference to the hypotheses which were presented earlier in this Chapter. For ease of use, these hypotheses are summarized on the following page. 60 Chapter One Examining Fear Appeals in Social Marketing Television Ads Through Content Analysis Summary of Hypotheses Hla The majority of social marketing ads will contain one or more threats or negative consequences. Hlb Of the ads which contain one or more threats or negative consequences, the majority will also contain action recommendations or advice for coping with the threat. Hlc In social marketing ads which include both threat(s) and coping information, the threat(s) will tend to temporally precede the coping information. H2 Approximately 24% of the social marketing ads will contain fear appeals. H3 It is expected that a higher percentage of social marketing ads in Australia will use fear appeals, as compared to the United States. H4a Corporate sponsors will be less likely than other organizations to use messages which incorporate threats or fear appeals. H4b Corporate sponsors will be more likely to use messages which are positive. H4c Corporate sponsors will be more likely to use messages which are humorous. H5a The more serious the potential consequences of a targeted behavior, the more likely it is that fear appeals will be used in social marketing ads against that behavior. H5b The less serious the potential consequences of a targeted behavior, the more likely it is that positive appeals will be used in social marketing ads against that behavior. H5c The less serious the potential consequences of a targeted behavior, the more likely it is that humorous appeals will be used in social marketing ads against that behavior. H6a Social marketing ads targeted at youth will be less likely to make use of fear appeals. H6b Social marketing ads targeted at youth will be more likely to make use of positive appeals. H6c Social marketing ads targeted at youth will be more likely to make use of humorous appeals. 61 Chapter Two Content Analysis Research Design CHAPTER TWO CONTENT ANALYSIS RESEARCH DESIGN Chapter Two Content Analysis Research Design CHAPTER TWO CONTENT ANALYSIS RESEARCH DESIGN Content Analysis Content analysis is "the study of the message itself, and not the communicator or the audience" (Kassarjian 1977, p. 8). Content analysis has frequently been used to study advertising as well as many other forms of written documents or transcripts, visual or graphic materials, and verbal protocols, and it seems a particularly appropriate technique for studying social marketing advertising. Content analysis has been used to examine a wide range of advertising message or content variables, such as information content or claims (Healey and Kassarjian 1983; James and Vanden Bergh 1990; Kassarjian and Kassarjian 1988; Resnik and Stern 1977; Rice and Lu 1988; Ringold and Calfee 1989), use of the problem-resolution format in television commercials (Marlowe, Selnow, and Blosser 1989), use of time-oriented appeals in magazine ads (Gross and Sheth 1989), use of color (Lee and Barnes 1989), ecology themes (Peterson 1991), social responsibility themes (Lill, Gross, and Peterson 1986), use of humor in television commercials (Weinberger and Spotts 1989), and many other areas. As outlined in Chapter 1, only a handful of content analysis studies have dealt with fear appeals (Unger and Stearns 1983) or social marketing advertisements in general (e.g., Hanneman, McEwen, and Coyne 1973; Wallack and Dorfman 1992). Several content analyses have dealt with specific types of social marketing commercials, such as AIDS prevention commercials (Calvert, Cocking, and Smrcek 1991; Freimuth, Hammond, Edgar, and Monahan 1990), and anti-drug abuse commercials (Hanneman and McEwen 1973; Walker 1990). Still other content analysis studies have touched on subjects which are of more peripheral interest to social marketing, such as AIDS press coverage (Lupton 1992), sex in primetime television 63 Chapter Two Content Analysis Research Design programming (Sapolsky and Tabarlet 1991), alcohol ads (Atkin and Block 1984), and alcohol use shown in television programming (Breed and Defoe 1972; Gerbner, Morgan, and Signorielli 1982). While some of the above studies examine some aspects of social marketing ads, none of them have the scope or breadth of the social marketing content analysis study that was carried out for this dissertation. For this reason, the present study presents a useful contribution to the literature on social marketing. The remainder of this chapter will outline the methodology and procedures employed for the content analysis study, including collecting the sample of TV commercials, the development of the data collection instrument, the training of coders, the process of conducting the coding, the methods used for analyzing the resulting data, and the sample characteristics. Chapter 3 will then outline the content analysis results. Sampling Procedure The sample for this content analysis was a comprehensive snowball sample of TV social marketing ads which met the following criteria: 1. English-language social marketing TV ads from one of the following five countries: Canada, United States, Great Britain, Australia, New Zealand 2. Targeted at one of the following five behaviors: smoking, illicit drug use, DUI, excessive alcohol consumption, unsafe sex. 3. Produced between 1980 and 1994. Looking at five different behaviors allows for an in-depth analysis of each behavioral category, with the ability to make comparisons between the categories. The five behaviors are also different enough from each other that they represent a good cross-section of social 64 Chapter Two Content Analysis Research Design marketing commercials. The study is limited to five English-speaking countries to facilitate the analysis of social marketing ads. Eliminating non-English speaking countries avoids the problem of trying to compare accurately ads from different languages. Finally, the 15-year time span covers a broad enough period that potential changes in communication philosophy may be uncovered by comparing the early and latter parts of the sample. Social marketing ads were collected from industry sources including national television bureaus (e.g., Television Bureau of Canada), government agencies (e.g., Health Canada, U.S. Department of Health and Human Services, etc.), and relevant nonprofit organizations (e.g., Mothers Against Drunk Driving, Addiction Research Foundation, Cancer Society, etc.). A complete list of all organizations contacted is included in Appendix 1. The letter that was used to contact these individuals is contained in Appendix 2. The sampling methodology used was a type of nonprobability sample called snowball sampling, where those individuals who were initially asked to provide relevant TV social marketing ads were also asked to identify their counterparts in other agencies or countries who would be in a position to provide additional relevant social marketing ads. Individuals and agencies were contacted in successive waves, until few new or previously unknown social marketing ads were emerging, and it was deemed that the collection of social marketing ads from each of the five countries was relatively complete. The process started with obtaining five videotapes from the Television Bureau of Canada, one on each of the five behavioral areas (i.e., social marketing ads against smoking, illegal drug use, DUI, excessive alcohol consumption, and unsafe sex), which contained over 100 television commercials from over 50 organizations in the five countries. These organizations formed part of the initial list, and their addresses were sought using various directories and other resources. As well, these directories and other resources were used to 65 Chapter Two Content Analysis Research Design obtain additional names of organizations and contact persons, bringing the initial mailing list to 163 organizations. When organizations replied to this mailing indicating that they did or did not have any TV relevant commercials, they often provided the names of other organizations. In many cases, these names were duplicates of names that were initially on the list; however, an additional 75 organizations were uncovered in this way, bringing the total number of organizations to 238. Because the initial mailing list was so exhaustive, it is not surprising that relatively few additional names were found through the snowball approach. Many of the organizations were able to provide not only their own television commercials, but also the television commercials which they had in their libraries which had been produced by other organizations. For this reason, a total of 589 commercials were obtained, a much larger number than the total number of organizations contacted. Therefore, this method of snowball sampling through industry sources was successful in providing a relatively large pool of commercials in each of the five specific targeted behavior areas (i.e., social marketing ads against smoking, illegal drug use, DUI, excessive alcohol consumption, and unsafe sex). A sample size of 589 is considered to be quite large for a content analysis of TV commercials (Kolbe and Burnett 1991). Collecting specific categories of social marketing ads through industry sources provides a relatively complete population of social marketing ads which have aired since 1980. This longitudinal sample may allow for a limited exploratory comparison of social marketing ads over time (e.g., early 1980s versus early 1990s). Having a large pool of each type of social marketing ad also allows for relatively more in-depth analysis and comparison. Differences in incidence of fear appeals between social marketing ad groups can more readily be attributed to differences in the grouping category under 66 Chapter Two Content Analysis Research Design consideration (i.e., country of origin, sponsor type, targeted behavior, intended target group). Data Collection Instrument The data collection instrument used in this study was refined through a series of stages. First, a draft instrument was prepared. The author and an assistant used the draft instrument to code 20 commercials chosen randomly from the sample. Based on this initial pretest, the data collection instrument was modified. This modified version was tried out on an additional 15 commercials randomly selected from the sample. Based on the second pretest, a few further refinements were made to the instrument. These refinements included additional categories for some variables, to ensure that there would not be large numbers of commercials under a miscellaneous type of category. The data collection instrument included many questions that were common to all types of social marketing TV commercials, but there were also five sections that were unique to each of the five behavioral categories of the commercials in the sample (i.e., smoking, illegal drug use, DUI, excessive alcohol consumption, and unsafe sex). In total, over 200 variables were coded. The data collection instrument was refined through this process to answer all potential questions considered relevant to the study. A copy of the finalized data collection instrument (five versions, customized for each of the five targeted behaviors in the set of commercials being examined) is included in Appendix 3. Coder Training Four undergraduate students were trained to effectively and consistently code the TV commercials. Their training procedure consisted of about two hours of practice at coding 10 social marketing TV commercials which were not part of the sample (two commercials from 67 Chapter Two Content Analysis Research Design each behavioral category). For each of these 10 commercials, the coders followed the same procedure. They would view a commercial several times, while individually coding their responses with the finalized data collection instrument. After they had finished their individual coding for a given commercial, they would discuss their responses as a group. This discussion helped the coders to reach a concensus about the appropriate way to code some of the items which required a small degree of judgment. At the conclusion of coding the tenth training commercial, the coders felt well-prepared to handle the main coding task. (Had they not felt well-prepared at that point, the training session would have continued until the coders did perceive themselves to be competent, and confident in their abilities to carry out the task accurately.) This procedure is in accordance with that recommended by Kassarjian (1977) in his seminal paper on content analysis. Data Collection Procedure Each of the 589 social marketing TV commercials was analyzed and coded by the four independent coders using the data collection instrument. During the coding process, each social marketing ad was viewed several times by each coder to ensure that all of the relevant features were rated or recorded. Although all four coders watched the television commercials together in the same room, each of the coders filled out their data collection instrument independently for each TV commercial. The coders were supervised to ensure that they were, indeed, filling out their coding sheets independently without discussion or peeking at each others' work. As well, there was no real incentive for coders to cheat at this task, since there was no right or wrong answer. After the coders had completed filling out their data collection instrument for a given commercial, the four coding sheets would be compared. All items of agreement would be 68 Chapter Two Content Analysis Research Design recorded on a master coding sheet. Items of disagreement were recorded separately, through the procedure described here. While there was a very high degree of agreement on the majority of the variables (in excess of 95%), coders sometimes disagreed on some items.1 These items would be discussed, and the TV commercial would be replayed in order to attempt to resolve the differences. This procedure was generally sufficient to resolve differences in judgment between coders, and the concensus agreement would then be recorded on the master coding sheet for that TV commercial. In the few cases where this procedure was followed yet coders still failed to come to an agreement, the principal investigator would make a judgment for that variable. These judgments would then be recorded on the master coding sheet for that particular commercial. The result of this process was a master coding sheet for each of the 589 TV commercials. The entire coding process took approximately 90 hours spread over 3 weeks. This means that it took approximately 9 minutes per coder to code each commercial. The 9-minute average would include viewing the commercial 3-7 times, comparing coding sheets, and discussion and agreement on any items where the coders had failed to code identically. Longer commercials and more complex commercials would take longer to code, primarily due to the need for repeated viewings and a somewhat greater level of disagreement among coders, while shorter and simpler commercials would take less time to code. 1 Differences of opinion most often occurred on items which required some judgment on the part of the coders. The two questions on which this happened most often were those dealing with intended target group, and coding of the level of fear (no/low/medium/high). Coders were given guidelines for rating level of fear and had practised assigning commercials to fear categories, but this was still a difficult area where individual coder opinion sometimes played a role. The rate of inter-coder agreement on these two questions, where all four coders agreed on the rating, is approximately 82%. If these two questions are excluded from the analysis of inter-coder agreement, the rate of agreement is in excess of 98%. 69 Chapter Two Content Analysis Research Design Composition of the Sample Of the 589 commercials collected, a majority came from the U.S. (57%) and Canada (27%). The distribution of commercials in each of the five behavioral categories is shown in Table 2.1. Table 2.1 Behavioral Categories of TV Commercials By Country Canada U.S.A. Britain Australia New Zealand Total AIDS 34 21.66% 48 14.16% 8 17.02% 1 4.00% 6 28.57% 97 16.47% Alcohol Moderatio n 14 8.92% 29 8.55% 1 2.13% 2 8.00% 3 14.29% 49 8.32% DUI 63 40.13% 97 28.61% 4 8.51% 13 52.00% 11 52.38% 188 31.92% Drugs 21 13.38% 75 22.12% 9 19.15% 0 0.00% 0 0.00% 105 17.83% Smoking 25 15.92% 90 26.55% 25 53.19% 9 36.00% 1 4.76% 150 25.47% N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 % The sparsity of the sample from Australia and New Zealand, and even to some extent from Great Britain, makes it difficult to conduct cross-national comparisons within specific behavioral categories. However, the sample size as a whole is sufficient to test the general hypotheses about fear appeals. It is perhaps a limitation of this study that there is a paucity of commercials gathered from Great Britain, Australia, and New Zealand. One explanation for this might simply be 70 Chapter Two Content Analysis Research Design that there were relatively few relevant commercials produced in these areas that were suitable for inclusion in this study. The reason for the lesser amount of TV commercial production might be due to a number of factors, including differences in organization of government departments, and differences in budgetary levels. In North America many public health concerns appear to be relegated to the state and provincial level. Given the 50 states in the United States and 10 provinces in Canada, the result is many more social marketing TV ads in North America, particularly in categories such as DUI and smoking. The overall result is that Canada and the U.S. produce many more social marketing TV commercials, given that many health promotion efforts are regional in nature. However, an alternative explanation for having collected fewer commercials from these three countries might be the relative difficulty involved in communicating with overseas countries (real or perceived barriers such as time zone differences, delivery distances, etc.). Because of delays in communicating with overseas countries, it is more difficult to conduct a snowball sample which has an implicit need for making contacts on an iterative basis. However, there is little reason to believe that having a collection of commercials from some countries which is possibly less complete will systematically change the nature of the results. There was no systematic bias in the types of organizations contacted, or the relative levels of response from different types of organizations, so while the sample from some countries is smaller, these smaller samples do not necessarily introduce systematic biases. It is the belief of the author that, although the effort to collect commercials from all five countries involved was an intensive one, the greater number of commercials in the sample from Canada and the U.S. reflects a combination of these two factors: greater numbers of commercials produced in Canada and the United States, and increased difficulty in obtaining commercials from outside Canada and the United States. The result is that the results will be 71 Chapter Two Content Analysis Research Design fairly reliable for Canada and the United States, but the results for other countries may need to be treated with caution. While every effort was made to ensure that the sample of commercials was as complete as possible, it is apparent that it does not include every commercial ever produced in these countries. For example, the present study collected a total of 48 AIDS prevention TV commercials in the United States; however, a study by Freimuth, Hammond, Edgar, and Monahan (1990) collected 127 AIDS prevention TV commercials. Attempts to obtain this videotape of 127 AIDS TV commercials from these authors for inclusion in the present study were unsuccessful. The longitudinal nature of the collection of commercials in this study means that it is more difficult to obtain commercials from earlier years, relative to more recent years. This difficulty means that it is possible only to make observations about longitudinal comparisons, but it is not possible to test longitudinal hypotheses. 72 Chapter Three Content Analysis Research Results CHAPTER THREE CONTENT ANALYSIS RESEARCH RESULTS Chapter Three Content Analysis Research Results CHAPTER THREE CONTENT ANALYSIS RESEARCH RESULTS This chapter will outline the results obtained from analyzing the data from the content analysis study. For ease of presentation, the results will generally be provided in the order that the hypotheses were presented. Protection Motivation Theory (Hypotheses la, lb, and lc) Of the 589 commercials in the sample, 436 contain one or more threats, while 153 contain no threats. This provides support for Hla, which suggested that the majority of ads would contain one or more threats or negative consequences. Hlb deals with the degree to which the TV commercials adhere to the tenets of the Protection Motivation model, stating that the majority of social marketing ads which contain threats should also contain action recommendations or advice for coping with the threat. Analyses for this Hlb were limited to those 436 commercials which included one or more threats. In general, Hlb appears to be supported in the majority of the commercials. In total, 69.27% of the 436 ads that contain a threat also contain some coping advice; 30.73% of the ads that contain a threat do not contain any coping advice (see Table 3.1). This provides significant support for Hlb. In exploring the data by country in Table 3.1, there do not appear to be significant differences between countries with regard to adherence to the Protection Motivation model's recommendation to couple threats with coping advice (chi-square=2.194, d.f. =4, p=0.700). 74 Chapter Three Content Analysis Research Results Table 3.1 Presence of Coping Advice in Ads Which Contain Threats By Country Canada U.S.A. Britain Australia New Zealand Total Threat but No Advice 29 27.36% 81 32.40% 11 27.50% 6 26.09% 7 41.18% 134 30.73% Threat and Advice 77 72.64% 169 67.60% 29 72.50% 17 73.91% 10 58.82% 302 69.27% No Threat 51 89 7 2 4 153 N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 % When exploring the data by behavioral category (Table 3.2), it becomes apparent that the degree to which the commercials conform to the tenets of the Protection Motivation model differs, depending on the behavior in question. DUI, drug, and smoking commercials are much more likely to offer threats with no advice, while AIDS commercials are the least likely to offer a threat without also offering advice (chi-square=24.057, d.f. =4, p = .000). 75 Chapter Three Content Analysis Research Results Table 3.2 Presence of Coping Advice in Ads Which Contain Threats By Behavioral Category AIDS Alcohol Moderation DUI Drugs Smoking Total Threat but No Advice 8 10.13% 8 22.86% 53 40.77% 29 35.80% 36 32.43% 134 30.73% Threat and Advice 71 89.87% 27 77.14% 77 59.23% 52 64.20% 75 67.57% 302 69.27% No Threat 18 14 58 24 39 153 N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% It is not entirely clear why the behavioral category should have an impact on the degree to which the commercials offer advice in situations where they also offer a threat. One explanation could have to do with the age of the commercial; perhaps more recent commercials are more likely to conform to the tenets of the PM model, by including coping advice when threats are used.1 Table 3.3 examines this possibility using the 350 TV commercials in the sample for which the year of production was available2. Using a median split on the data, the first 7 years (1980-1986) were compared to the last 8 years (1987-1994).3 The results 1 This presumes that there has been diffusion of the PM and OPM models among social marketing practitioners. It is entirely possible that many social marketing practitioners do not know this model, but there is likely to be a sub-set of practitioners who are interested in theoretical advances and who would be knowledgeable about the model. 2 Only 350 of the 589 commercials had a confirmed date of production. Numerous commercials for the sample were provided with a range of possible years which might have been the original production date, but unless a definitive year was provided, the production date was left blank during the data input phase. 3 The Protection Motivation model was first published by Rogers in 1975 (revised Rogers 1983). Therefore, one might expect to see the tenets of this model applied to even the earliest social marketing commercials in this sample which covers 1980-1994. However, one would 76 Chapter Three Content Analysis Research Results appeared to be directionally supportive of the idea that the tenets of the PM model were adhered to, but this outcome was not statistically significant (chi-square=2.209, d.f. = 1, p=.137). Table 3.3 Presence of Coping Advice in Ads Which Contain Threats By Time Period 1980-1986 1987-1994 Total Threat but No Advice 46 32 78 35.38% 26.67% 31.20% Threat and Advice 84 88 172 64.62% 73.33% 68.80% No Threat 40 60 100 N 170 180 350 Total 100.00% 100.00% 100.00% Hypothesis lb suggests that in social marketing ads which include both threat(s) and coping information, the threat(s) will tend to temporally precede the coping information. Table 3.4 demonstrates that this appears to be true in the sample as a whole, since 57.11 % of all ads containing threats do have the threat coming prior to the advice, thereby confirming Hlc. As well, when exploring the data for each country individually in Table 3.4, there appear to be only relatively small differences in the degree to which different countries adhere to this principle in their ads (chi-square=7.860, d.f. =4, p=0.097). expect that with the passage of time, there would be greater acceptance of the model, and greater adherence to its precepts [particularly since no other well-known fear appeals model has come along to replace this one; the OPM model (Tanner, Hunt, and Eppright 1991) provides only an extension to the PM model]. Therefore, if there were to be a difference in adherence to the tenets of the PM model, one would expect that there would be more adherence in more recent years. 77 Chapter Three Content Analysis Research Results Table 3.4 Order of Threats arid Coping Advice in Ads By Country Canada U.S.A. Britain Australia New Zealand Total Threat first, then Advice 63 59.43% 133 53.20% 28 70.00% 17 73.91% 8 47.06% 249 57.11% Other Order 43 40.57% 117 46.80% 12 30.00% 6 26.09% 9 52.94% 187 42.89% No Threat4 51 89 7 2 4 153 N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 % When exploring the data by behavioral category, Table 3.5 demonstrates that there are some differences between behavioral categories in terms of the extent to which threats precede coping advice in commercials (chi-square=12.515, d.f. =4, p=0.014). Overall, however, there is relatively high adherence to the principle of following threats with coping advice in commercials which contain both, regardless of behavioral category.5 4 This category includes ads which offer advice only, as well as ads where there is neither a threat nor advice. 5 It seems possible that in some campaigns, the threats and advice may appear in different commercials (i.e., the threats in some of the early commercials appearing in the campaign, and the advice in later commercials in the campaign). Because the unit of analysis in this study is the individual commercial and not campaigns, it is not possible to measure whether this phenomenon may be occurring. However, it presents an interesting topic for future research. 78 Chapter Three Content Analysis Research Results Table 3.5 Order of Threats and Coping Advice in Ads By Behavioral Category AIDS Alcohol Moderation DUI Drugs Smoking Total Threat first, then Advice 55 69.62% 26 74.29% 67 51.54% 43 53.09% 58 52.25% 249 57.11% Other Order 24 30.38% 9 25.71 63 48.46% 38 46.91% 53 47.75% 187 42.89% No Threat 18 14 58 24 39 153 N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% Other important aspects of the Protection Motivation model concerned response efficacy and self efficacy. Although no hypotheses specifically addressed this question, it is worthwhile to explore this aspect of the data.6 Table 3.6 explores the issue of whether or not the social marketing commercials contain information about response efficacy. In fact, the vast majority (84.04%) of the commercials do not include any elements of response efficacy; they may give advice, but they do not give any information about how effective acting on the advice will be in preventing the negative outcomes. Table 3.6 also explores whether there are differences in the use of response efficacy between countries, but there appear to be few significant differences (chi-square=2.298, d.f.=4, p=0.681). 6 The PM and OPM models would suggest that response efficacy and self efficacy are important elements of the ad message, so a hypothesis could have been put forward stating that ads should contain these elements. However, there was nothing in the literature which indicated that actual ads would include these elements. Therefore, this was treated as an exploratory variable rather than one for which a priori hypotheses were put forward. 79 Chapter Three Content Analysis Research Results Table 3.6 Does the Commercial Contain Information About Response Efficacy? By Country Canada U.S.A. Britain Australia New Zealand Total No 129 82.17% 284 83.78% 40 85.11% 23 92.00% 19 1 495 90.48% 1 84-04% Yes 28 17.83% 55 16.22% 7 14.89% 2 8.00% 2 1 94 9.52% § 15.96% N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 % Table 3.7 explores the use of response efficacy within social marketing commercials belonging to different behavioral categories. There do appear to be significant differences between categories of commercials, with-the AIDS commercials being most likely to include elements of response efficacy (chi-square=62.670, d.f. =4, p<0.001). Table 3.7 Does the Commercial Contain Information About Response Efficacy? By Behavioral Category AIDS Alcohol Moderation DUI Drugs Smoking Total No 58 58.79% 43 87.76% 174 92.55% 99 94.29% 121 80.67% 495 84.04% Yes 39 40.21% 6 12.24% 14 7.45% 6 5.71% 29 19.33% 94 15.96% N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% It is not clear why these differences exist between types of commercials. One explanation is that the kind of advice most often given in AIDS prevention commercials is to use a condom, a relatively straightforward behavior, which is highly effective in preventing 80 Chapter Three Content Analysis Research Results AIDS. This effectiveness of condoms is often stressed in AIDS prevention commercials as a means of encouraging people to adopt the behavior. The other four behaviors examined (drugs, smoking, DUI, alcohol moderation) all have potentially addictive aspects to the behavior (i.e., drug addiction, nicotine addiction, alcoholism), which make it relatively infeasible to prescribe any advice or solution except for abstention. Such advice regarding abstention is so obvious that it is usually implicit rather than explicit. If advice to abstain was not mentioned explicitly in the commercials, then it was not coded as being present. Table 3.8 examines the issue of including elements of self efficacy in social marketing commercials. In general, few commercials (6.96%) actually include any self efficacy message. When exploring whether different countries have a different prevalence of including self efficacy messages, it appears that there are no significant differences (chi-square=6.839, d.f.=4, p= 0.145). Table 3.8 Does the Commercial Contain Encouragement for Self Efficacy? By Country Canada U.S.A. Britain Australia New Zealand Total No 140 89.17% 321 94.69% 43 91.49% 23 92.00% 21 100.00% 548 93.04% Yes 17 10.83% 18 5.31% 4 8.51% 2 8.00% 0 0.00% 41 6.96% N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 % In exploring whether different behavioral categories display differential usage of self efficacy messages, we see in Table 3.9 that smoking commercials are somewhat more likely to include self efficacy messages (13.33%), while DUI and drug abuse commercials are least 81 Chapter Three Content Analysis Research Results likely to include self efficacy messages (3.72% and 2.86% respectively; chi-square=15.297, d.f. =4, p=0.004). Table 3.9 Does the Commercial Contain Encouragement for Self Efficacy? By Behavioral Category AIDS Alcohol Moderation DUI Drugs Smoking Total No 90 92.78% 45 91.84% 181 96.28% 102 97.14% 130 86.67% 548 93.04% Yes 7 7.22% 4 8.16% 7 3.72% 3 2.86% 20 13.33% 41 6.96% N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% A qualitative examination of the 20 smoking commercials which did contain encouragement for self efficacy revealed that these commercials had a "you can do it" type of message. This message was intended to make the smoker feel more confident in his/her ability to quit, and in some cases, took the form of explaining how many people had already quit and encouraging the smoker to join these successful quitters. Incidence of Fear Appeals (Hypothesis 2) H2 states that 24% of the social marketing ads will be intended to be fear arousing (i.e., ads which stress negative outcomes or threats). This was intended to be in line with Hanneman, McEwen, and Coyne (1973), who coded a commercial as containing a fear appeal if it showed or discussed the harmful (or painful) social or physical consequences of not doing what the message directed. However, in the present study, there are three different ways of measuring the presence or absence of fear. First, it is possible to measure whether there were 82 Chapter Three Content Analysis Research Results one or more specific threats.7 Second, the commercial was rated as having a low, medium, or high level of fear arousal.8 Finally, the commercial was rated as having an overall tone of fear-arousal (i.e., giving an overall impression of being fear-arousing).9 In fact, 74.02% of the commercials in the sample contained at least one or more specific threats or negative consequences (reported and discussed later on by country in Table 3.12). However, some of these threats or negative consequences were very mild, so that only 53.99% of the commercials were perceived to have low, medium, or high levels of fear arousal (reported and discussed later on by country in Table 3.13). Thus, 46.01% of the commercials contain either no or minimal fear-arousing elements. Most importantly, though, in examining the overall tone of the commercial, only 21.39% of the commercials gave an overall impression of being fear-arousing (reported and discussed later on by country in Table 3.10). This suggests that while slightly more than half the commercials contained some fear-arousing elements, slightly less than one-quarter were judged to leave an overall impression of being fear-arousing. This latter percentage is comparable to that identified by Hanneman, McEwen, and Coyne (1973), who found that fear appeals were used in 24% of PSAs in their 7 A threat or negative consequence may be relatively minor, in which case, it might not be perceived as a fear appeal according to the definition given by Hanneman, McEwen, and Coyne (1973). 8 A commercial could contain a low level of fear arousal, yet not be perceived as having an overall impression or overall tone of being fear-arousing. Many commercials with low or medium levels of fear arousal in the first part of the commercial tried to alleviate this emotion by lightening the tone or atmosphere in the latter part of the commercial. This would result in an overall tone or overall impression of the commercial which was not fear-arousing. 9 The overall tone of the commercial is simply the coders' impression of whether or not the ad was fear-arousing. For example, it is possible for an ad to contain a mild threat and to have a low level of fear arousal, but if that fear is dismissed near the end of the commercial by the introduction of humor, then the overall tone of the commercial might be rated as neutral or positive. 83 Chapter Three Content Analysis Research Results sample. Therefore, using a t-test of proportions, H2 is supported, since the proportion of social marketing ads in the present study which were of the fear-arousing variety was not significantly different from the 24% reported by Hanneman, McEwen, and Coyne (1973)(t= 1.4833, p>.05). Country-of-Origin and Fear Appeals (Hypothesis 3) H3 suggests that there will be differences between the countries under study with regard to the frequency of the use of fear appeals. Specifically, it is expected that Australia will use fear appeals to a greater extent than the United States does, as measured as a proportion of social marketing ads collected. Table 3.10 demonstrates that in terms of the overall tone of the commercials, Australia is indeed the most likely to use an overall tone that is fear-provoking (chi-square=13.865, d. f. = 4, p =.008). Table 3.10 Overall Tone of the Commercial By Country Canada U.S.A. Britain Australia New Zealand Total Neutral/ Positive 120 76.43% 275 81.12% 34 72.34% 14 56.00% 20 95.24% 463 78.61% Fear 37 23.57% 64 18.88% 13 27.66% 11 44.00% 1 4.76% 126 21.39% N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 % In a one-way ANOVA conducted on the "overall tone" variable, there were significant differences between the means of the countries, with Australia clearly being the country most 84 Chapter Three Content Analysis Research Results likely to have an overall tone that was fear-provoking (xAustraiia=l-56, xBritain=1.74, *USA=1.88, xCanada=1.92, xNewZealand=2.05; F=3.783, p<0.005; see Table 3.11). A Newman-Keuls post hoc test indicated that there was a significant difference between Australia and the U.S.A., but that other countries were not significantly different from the U.S.A. (t=4.1641, p<.05). Table 3.11 One-way Analysis of Variance Table Overall Tone of the Commercial By Country Source Sum-of-Squares DF Mean-Square F-Ratio P Country 4.165 4 1.041 3.783 0.005 Error 160.769 584 0.275 Australia is also the country most likely to mention or show one or more specific threats (chi-square=11.199, d.f. =4, p=0.024). These results are shown in Table 3.12. Table 3.12 Any Specific Threats Mentioned or Shown? By Country Canada U.S.A. Britain Australia New Zealand Total No 51 32.48% 89 26.25% 7 14.89% 2 8.00% 4 19.05% 153 25.98% Yes 106 67.52% 250 73.75% 40 85.11% 23 92.00% 17 80.95% 436 74.02% N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 -% Finally, Australia is also the most likely to use medium or high levels of fear arousal, 85 Chapter Three Content Analysis Research Results and the least likely to use no fear arousal (chi-square=20.544, d.f. =8, p=0.008). These results are shown in Table 3.13. Table 3.13 Amount of Fear Arousal in the Commercial By Country Canada U.S.A. Britain Australia New Zealand Total None 79 50.32% 157 46.31% 18 38.30% 7 28.00% 10 47.62% 271 46.01% Low 60 38.22% 156 46.02% 19 40.43% 11 44.00% 10 47.62% 256 43.46% Medium/ High 18 11.46% 26 7.67% 10 21.28% 7 28.00% 1 4.76% 62 10.53% N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 % This last result was confirmed with a one-way ANOVA which examined the "amount of fear arousal" variable, and determined that there were significant differences in the means for the various countries, with commercials from Australia containing the largest degree of fear arOUSal (.*Australia=l'12> X Britain = 0- 83, ACTJSA = 0-6T, Canada = 0-°T, ^NewZealand1^-^; F=4.375, p< .002; see Table 3.14). A Newman-Keuls post hoc test indicated that Australia was significantly different from the USA (t=5.194, p<.01). 86 Chapter Three Content Analysis Research Results Table 3.14 One-way Analysis of Variance Table Amount of Fear Arousal in the Commercial By Country Source Sum-of-Squares DF Mean-Square F-Ratio P Country 7.854 4 1.964 4.375 0.002 Error 262.098 584 0.449 All of these results provide significant support for H3, indicating that there are differences between the countries in terms of the use of fear appeals. Australia appears to make significantly more use of fear appeals than any of the other countries. Sponsorship and Fear Appeals (Hypotheses 4a, 4b, and 4c) To begin with, it is interesting to examine the distribution of sponsorship types10 by country (see Table 3.15). This table shows that corporate sponsorships of the social marketing commercials were limited almost exclusively to Canada and the United States (chi-square=13.475, d.f. =4, p=0.009). This suggests that sponsorship of social marketing ads may be largely a North American phenomenon. 10 The Traffic Accident Commission (TAC) in Australia is considered to be a nonprofit organization for the purposes of this analysis. 87 Chapter Three Content Analysis Research Results Table 3.15 Sponsorship Categories By Country Canada U.S.A. Britain Australia New Zealand Total Non-Corp. 133 84.71% 309 91.15% 47 100.00% 23 92.00% 21 100.00% 533 90.49% Corporate 24 15.29% 30 8.85% 0 0.00% 2 8.00% 0 0.00% 56 9.51% N Total 157 100.00% 339 100.00% 47 100.00% 25 100.00% 21 100.00% 589 100.00 % Further, it is interesting to look at the distribution of sponsorship types between the various types of targeted behaviors. Table 3.16 shows this distribution, and indicates that DUI and alcohol moderation ads are much more likely to have corporate sponsors than ads targeted at other types of behaviors (chi-square=27.619, d.f. =4, p>0.001). Table 3.16 Sponsorship Categ By Behavioral Cate ory gory AIDS Alcohol Moderation DUI Drugs Smoking Total Non corporate 90 92.78% 42 85.71% 155 82.45% 99 94.29% 147 98.00% 533 90.49% Corporate 7 7.22% 7 14.29% 33 17.55% 6 5.71% 3 2.00% 56 9.51% N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% To turn now to testing the hypotheses, H4a suggested that corporate sponsors would be less likely than nonprofit sponsors to use messages which incorporate threats or fear appeals. Table 3.17 examines whether any specific threats were mentioned or shown, 88 Chapter Three Content Analysis Research Results according to whether or not the sponsor was a for-profit corporation. Only 53.57% of for-profit corporations had any specific threats mentioned or shown in their commercials, compared to 75.80% of nonprofit or government organizations (chi-square=11.214, d.f. =1, p=0.001). It appears that for-profit corporations are less likely to mention or show threatening or fear-arousing elements in their TV commercials, which provides partial support for H4a. Table 3.17 Any Specific Threats Mentioned or Shown? By Sponsorship Category Non-Corporate Corporate Total No 128 25 153 24.02% 44.64% 25.98% Yes 405 31 436 75.98% 55.36% 74.02% N 533 56 589 Total 100.00% 100.00% 100.00% However, when examining the overall tone of the commercial there was no significant difference between for-profit corporations and other organizations in terms of whether positive, neutral, or fear-arousing appeals were used (chi-square= 0.176, d.f. =2, p=0.916). These results are shown in Table 3.18. 89 Chapter Three Content Analysis Research Results Table 3.18 Overall Tone of the Commercial? By Sponsorship Category Non-Corporate Corporate Total Positive 44 5 49 8.26% 8.93% 8.32% Neutral 376 38 414 70.54% 67.86% 70,29% Fear 113 13 126 21.20% 23.21% 21.39% N 533 56 589 Total 100.00% 100.00% 100.00% This result was confirmed with a one-way ANOVA which examined the "overall tone" variable, and determined that there were no significant differences in the means for the corporate versus non-corporate categories (*Non_corp=1.87, xCorp = 1.86; F=0.032, p = .857; see Table 3.19). Table 3.19 One-way Analysis of Variance Table Overall Tone of the Commercial By Sponsorship Category Source Sum-of-Squares DF Mean-Square F-Ratio P Sponsorship 0.009 1 0.009 0.032 0.857 Catgory Error 164.925 587 0.281 Similarly, when examining the level of fear arousal (none, low, medium, high), there were no significant differences between for-profit corporations and other organizations (chi-square=2.421, d.f. =2, p=0.298). However, corporations did seem less likely to use medium/high levels of fear arousal (5.36%) as compared to non-corporate organizations 90 Chapter Three Content Analysis Research Results (11.07%). These results, while not statistically significant, are directionally supportive of H4a. These results are shown in Table 3.20. Table 3.20 Amount of Fear Arousal in the Commercial By Sponsorship Category Non-Corporate Corporate Total None 241 30 271 45.22% 53.57% 46.01% Low 233 23 256 43.71% 41.07% 43.46% Medium/ 59 3 62 High 11.07% 5.36% 10.53% N 533 56 589 Total 100.00% 100.00% 100.00% However, a one-way ANOVA was used to examine the "amount of fear arousal" variable, and determined that there were no significant differences in the means for the corporate versus non-corporate categories (*Non-corp=0.66, xCorp=0.52; F=2.368, P=.124; see Table 3.21). 91 Chapter Three Content Analysis Research Results Table 3.21 One-way Analysis of Variance Table Amount of Fear Arousal in the Commercial By Sponsorship Category Source Sum-of-Squares DF Mean-Square F-Ratio P Sponsorship Catgory 1.085 1 1.085 2.368 0.124 Error 268.868 587 0.458 It should be noted that slightly less than 10% of the commercials in the sample included a corporate sponsor. This small sample size means that the chi-square statistics do not have a great deal of power. While there was a significantly greater likelihood of having specific threats mentioned or shown in the corporate television commercials, there was not a significantly greater incidence of having an overall tone of fear arousal, nor was there a significantly greater incidence of having elevated levels of fear arousal. Therefore, there is only partial support for H4a. H4b suggested that corporate sponsors would be more likely to use messages which are positive. As mentioned above, there was no significant difference in the degree of use of positive appeals, as was shown in Table 3.18. Therefore, there is no support for H4b. H4c suggested that corporate sponsors would be more likely to use messages which contain humor. However, there was no significant difference in the use of humorous appeals (chi-square=2.468, d.f. = 1, p=0.116). These results are shown in Table 3.22. Therefore, there is no support for H4c. 92 Chapter Three Content Analysis Research Results Table 3.22 Is There an Attempt to be Humorous? By Sponsorship Category Non-Corporate Corporate Total No 454 52 506 85.18% 92.86% 85.91% Yes 79 4 83 14.82% 7.14% 14.09% N 533 56 589 Total 100.00% 100.00% 100.00% Behavioral Category and Fear Appeals (Hypotheses H5a, H5b and H5c) H5a states that the more serious the potential consequences of a targeted behavior, the more likely it is that fear appeals will be used in social marketing ads against that behavior. A cursory examination of the categories would suggest that smoking would likely be perceived as having the least serious consequences, particularly since many people smoke without any apparent negative consequences, and in any case, if negative consequences do occur it is almost always after many years of smoking. Unsafe sex, on the other hand, might be perceived as being the most serious category, since the likelihood of getting AIDS is perceived to be high, the consequences of getting AIDS are almost always deadly, and AIDS as a disease has been highly publicized. To confirm this ranking, two pretests were conducted. The purpose of the pretests was to examine how the behaviors were ranked in terms of the seriousness of their consequences. Pretest A uses a rating scale, while Pretest B uses a ranking method. The objective of using two separate pretests is to look for some convergence of results, so it is expected that the two pretests will lead to the same relative ranking of the five behaviors. 93 Chapter Three Content Analysis Research Results Pretest A The questionnaire for Pretest A listed the five behaviors (smoking, driving while impaired, engaging in unprotected11 sex, heavy alcohol consumption, and use of illegal drugs), and asked subjects to rate how serious they considered the consequences of each behavior to be, using a scale from 1 to 7 (where 1 = not very serious, and 7 = extremely serious). The questionnaire for Pretest A is given in Appendix 4. The pretest was administered to a sample of 28 students enrolled in a business diploma program. The mean values for the five behaviors are shown in Table 3.23: Table 3.23 Mean Values of Seriousness of Behavior Consequences Pretest A Behaviors Rank Mean Standard Deviation Most Serious Behaviors: Driving while imparied 1 6.571 0.879 Engaging in unprotected sex 2 6.556 0.698 Least Serious Behaviors: Heavy alcohol consumption 3 6.107 0.994 Smoking 4 5.821 1.307 Use of illegal drugs 5 5.667 1.271 Using the Tukey method of multiple comparisons, it was found that the five behaviors could be divided into two groups, as shown in Table 3.23. There was no statistically significant difference between the two highest means, which can be thought of as being the two most serious behaviors (driving while impaired and engaging in unprotected sex). There were 11 The pretest questionnaires used the wording, "Engaging in unprotected sex." In debriefing the subjects after the pretest, it was clear that this was generally understood to mean engaging in sex without a condom. The implication that was generally understood was that the protection, a condom, was needed as a barrier against AIDS rather than against pregnancy. 94 Chapter Three Content Analysis Research Results also no statistically significant differences between any of the three lowest means, which can be thought of as being the three least serious behaviors (heavy alcohol consumption, smoking, and use of illegal drugs). However, each of the three least serious behaviors have means which are significantly different (p<.05) from the means of each of the two most serious behaviors. Pretest B The questionnaire for Pretest B listed the five behaviors (smoking, driving while impaired, engaging in unprotected sex, heavy alcohol consumption, and use of illegal drugs), and asked subjects to rank the five behaviors from 1 to 5 in terms of how serious they considered the consequences of each behavior to be (where a ranking of 1 = most serious consequences, 5 = least serious consequences). The questionnaire for Pretest B is also given in Appendix 4. The pretest was administered to a sample of 19 students enrolled in an undergraduate marketing class. The mean values for the five behaviors are shown in Table 3.24. 95 Chapter Three Content Analysis Research Results Table 3.24 Mean Values of Seriousness of Behavior Consequences Pretest B Behaviors Rank Mean Standard Deviation Most Serious Behaviors: Driving while imparled 1 1.842 1.015 Engaging in unprotected sex 2 2.579 1.346 Least Serious Behaviors: Use of illegal drugs 3 3.000 1.333 Heavy alcohol consumption 4 3.579 0.902 Smoking 5 4.000 1.453 Using Tukey multiple comparisons, the ranking of the five behaviors in Pretest B (Table 3.24) was found to be similar to that found in Pretest A. "Driving while impaired" and "Engaging in unprotected sex" were considered to be the two behaviors with the most serious consequences in both Pretest A and Pretest B. However, in Pretest B "use of illegal drugs" switched to 3rd ranking position, whereas it had been in the 5th ranking position in Pretest A. Smoking continued to be viewed as having less serious consequences than heavy alcohol consumption in both Pretest A and Pretest B. From these two pretests we can readily conclude that "Driving while impaired" and "Engaging in unprotected sex" are the two most serious behaviors, being ranked #1 and #2 respectively in both studies. In averaging the ranking of the other 3 behaviors, we would have "Heavy alcohol consumption" ranked 3rd, "Use of illegal drugs" ranked 4th, and "Smoking" ranked 5th (i.e., smoking is the behavior with the least serious consequences). The information from Pretests A and B is useful in testing Hypotheses H5a, H5b, and H5c. In examining the overall tone of the commercials by behavioral type in the content analysis study for the purposes of testing H5a and H5b, it can be seen that smoking commercials are the least likely type of commercial to use fear/threat appeals, and are the most 96 Chapter Three Content Analysis Research Results likely to use positive appeals (chi-square=47.987, d.f. =8, p<0.001). These results are shown in Table 3.25. Table 3.25 Overall Tone of the Commercial By Behavioral Category AIDS Alcohol Moderation DUI Drugs Smoking Total Positive 3 3.09% 5 10.20% 10 5.32% 6 5.71% 25 16.67% 49 8.32% Neutral 83 85.57% 32 65.31% 121 64.36% 68 64.76% 110 73.33% 414 70.29% Fear 11 11.34% 12 24.49% 57 30.32% 31 29.52% 15 10.00% 126 21.39% N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% Examining these data in a somewhat different manner by grouping them into the more serious versus less serious behaviors is shown in Table 3.26. This grouping still shows that the more serious behaviors have an overall tone that is significantly more fear-provoking than the less serious behaviors (chi-square=11.012, d.f. =2, p<.005), which supports H5a. 97 Chapter Three Content Analysis Research Results Table 3.26 Overall Tone of the the Commercial By Seriousness of the Behavior More Serious Behaviors Less Serious Behaviors Total (AIDS/DUI) (Drugs/Ale. Mod. / Smoking) Positive 13 36 49 4.56% 11.84% 8.32% Neutral 204 210 414 71.58% 69.08% 70.29% Fear 68 58 126 23.86% 19.08% 21.39% N 285 304 589 Total 100.00% 100.00% 100.00% A one-way ANOVA was conducted on the "overall tone" variable, and there were significant differences found in the means for the different behaviors, with smoking having the most positive overall tone and DUI having the most negative (fear-provoking) overall tone (^smoking=2.07, xAIDS = 1.92, xAicMod=1-86, xDrugs = 1.76, xDUI = 1.75; F=9.386, p<0.001; see Table 3.27). A Newman-Keuls post hoc test indicated that Smoking was significantly different from DUI (t=6.693, p> .001). Table 3.27 One-way Analysis of Variance Table Overall Tone of the Commercial By Behavioral Category Source Sum-of-Squares DF Mean-Square F-Ratio P Behavioral 9.963 4 2.491 9.386 0.000 Category Error 154.971 584 0.265 When examining whether any specific threats were mentioned or shown, AIDS 98 Chapter Three Content Analysis Research Results commercials showed the highest incidence (81.44%). Threats were most notably absent from DUI, alcohol moderation, and smoking ads; however, this result was not statistically significant (chi-square=5.803, d.f. =4, p=0.214). These results are shown in Table 3.28. Table 3.28 Any Specific Threats Mentioned or Shown? By Behavioral Category AIDS Alcohol Moderation DUI Drugs Smoking Total No 18 18.56% 14 28.57% 58 30.85% 24 22.86% 39 26.00% 153 25.98% Yes 79 81.44% 35 71.43% 130 69.15% 81 77.14% 111 74.00% 436 74.02% N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% ) Collapsing this data again into the two categories of more serious behaviors and less serious behaviors does not change the results, since we see in Table 3.29 that this does not result in a significant difference between these two groups (p > .05). Therefore, this particular result does not support H5a. 99 Chapter Three Content Analysis Research Results Table 3.29 Any Specific Threats Mentioned or Shown? By Seriousness of the Behavior More Serious Behaviors Less Serious Behaviors Total (AIDS/DUI) (Drugs/Ale. Mod. /Smoking) No 76 77 153 26.57% 25.32% 25.98% Yes 209 227 436 73.33% 74.67% 74.02% N 285 304 589 Total 100.00% 100.00% 100.00% Finally, fear arousal is most likely to be absent in smoking commercials. Medium/high levels of fear arousal are most likely to be found in DUI commercials (chi-square=18.170, d.f. =8, p=0.020). These results are shown in Table 3.30. Table 3.30 Amount of Fear Arousal in the Commercial By Behavioral Category AIDS Alcohol Moderation DUI Drugs Smoking Total None 50 51.55% 24 48.98% 79 42.02% 34 32.38% 84 56.00% 271 46.01% Low 40 41.24% 21 42.86% 84 44.68% 58 55.24% 53 35.33% 256 43.46% Medium/ High 7 7.22% 4 8.16% 25 13.29% 13 12.38% 13 8.67% 62 10.53% N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% However, when these results were analyzed by more serious behaviors versus less serious behaviors, there was again no significant difference between the two groups of 100 Chapter Three Content Analysis Research Results behaviors (p>.05; see Table 3.31). Table 3.31 Amount of Fear Arousal in the Commercial By Seriousness of the Behavior More Serious Behaviors Less Serious Behaviors Total (AIDS/DUI) (Drugs/Ale. Mod./Smoking) None 129 142 271 45.26% 46.71% 46.01% Low 124 132 256 43.51% 43.42% 43.46% Medium/ 32 30 62 High 11.22% 9.87% 10.53% N 285 304 589 Total 100.00% 100.00% 100.00% A one-way ANOVA was conducted on the "amount of fear arousal" variable, and there were significant differences found in the means for the different behaviors, with smoking commercials having the least amount of fear arousal, while drug commercials had the greatest amount of fear arousal (xSmOking=0.53, xAIDS=0.57, xAlcMod=0.59, xDUI=0.72, xDmgs=0.80; F=3.597, p< .007; see Table 3.32). A Newman-Keuls post hoc test of the means indicates that there is a significant difference between the two highest means (for DUI and Drugs) versus the three lowest means (Smoking, AIDS, Alcohol Moderation; t=4.4685, p<.05); however, the ranking of these means does not support H5a. 101 Chapter Three Content Analysis Research Results Table 3.32 One-way Analysis of Variance Table Amount of Fear Arousal in the Commercial By Behavioral Category Source Sum-of-Squares DF Mean-Square F-Ratio P Behavioral Category 6.491 4 1.623 3.597 0.007 Error 263.462 584 0.451 Overall, there is only partial support for H5a, which suggests a positive correlation between the seriousness of the targeted behavior and the use of fear appeals. The overall tone of the commercials is significantly more fear arousing for the more serious behaviors; however, there is not a significant relationship between the seriousness of the targeted behavior and either the level of fear arousal or the mention of threats/negative consequences. H5b states that the less serious the potential consequences of a targeted behavior, the more likely it is that positive appeals will be used in social marketing ads against that behavior. As shown above in Table 3.25, smoking commercials were the most likely type of commercial to use positive appeals (16.67% compared to 8.32% of the sample; chi-square=47.987, d.f. =8, p < .001). As well, Table 3.26 showed that the less serious behaviors were more likely to have a positive overall tone (chi-square=11.012, d.f. =2, p < .005). This provides support for H5b. H5c states that the less serious the potential consequences of a targeted behavior, the more likely it is that humorous appeals will be used in social marketing ads against that behavior. There is support for H5c, since 30.00% of the smoking commercials made use of humorous appeals, a much greater percentage than for any other behavioral category (chi-square=42.825, d.f. =4, p<0.001). These results are shown in Table 3.33: 102 Chapter Three Content Analysis Research Results Table 3.33 Is There an Attempt to be Humorous? By Behavioral Category AIDS Alcohol Moderation DUI Drugs Smoking Total No 90 92.78% 46 93.88% 169 91.43% 96 91.43% 105 70.00% 506 85.91% Yes 7 7.22% 3 6.12% 19 10.11% 9 8.57% 45 30.00% 83 14.09% N Total 97 100.00% 49 100.00% 188 100.00% 105 100.00% 150 100.00% 589 100.00% This result is now broken out by the seriousness of the behavior in Table 3.34. Table 3.34 Is There An Attempt to Be Humorous? By Seriousness of the Behavior More Serious Behaviors Less Serious Behaviors Total (AIDS/DUI) (Drugs/Ale. Mod. /Smoking) No 259 247 506 90.88% 81.25% 85.91% Yes 26 57 83 9.12% 18.75% 14.09% N 285 304 589 Total 100.00% 100.00% 100:00% Table 3.34 clearly shows that there is a greater tendency to use humor in commercials for the less serious behaviors (18.75%) as compared to commercials for the more serious behaviors (9.12%) (chi-square = 11.2617, d.f. = l, p<.001). Therefore, H5c is supported. Intended Target Group and Fear Appeals (Hypotheses H6a, H6b, and H6c) H6a stated that social marketing ads targeted at youth would be less likely to make use 103 Chapter Three Content Analysis Research Results >• of fear appeals. This seems to be supported, since ads directed at youth were less likely to make use of fear appeals as shown in Table 3.35. Table 3.35 Overall Tone of the Commercial By Intended Target Group Age Category Non- Youth Youth Total Fear 120 6 126 23.53% 7.59% 21.39% Neutral 369 45 414 72.35% 56.96% . 70.29% Positive 21 28 49 4.12% 35.44% 8.32% N 510 79 589 Total 100.00% 100.00% 100.00% Also, ads directed at youth have fewer threats mentioned or shown (chi-square=6.831, d.f.=l, p=0.009), as outlined in Table 3.36. Table 3.36 Any Specific Threats Mentioned or Shown? By Intended Target Group Age Category Non-Youth Youth Total No 123 30 153 24.12% 37.97% 25.98% Yes 387 49 436 75.88% 62.03% 74.02% N 510 79 589 Total 100.00% 100.00% 100.00% The amount of fear arousal in the ads was also examined. Ads that are directed at 104 Chapter Three Content Analysis Research Results youth had lower levels of fear arousal, when compared to ads that were not directed at youth (chi-square=25.187, d.f. =2, p<0.001), as shown in Table 3.37: Table 3.37 Amount of Fear Arousal in the Commercial By Intended Target Group Age Category Non-Youth Youth Total None 214 57 271 41.96% 72.15% 46T01% Low 239 17 256 46.86% 21.52% 43.46% Medium/ 57 5 62 High 11.18% 6.33% 10.53% N 510 79 589 Total 100.00% 100.00% 100.00% A one-way ANOVA was conducted on the "amount of fear arousal" variable, and there were significant differences found in the means for the non-youth versus youth target groups, with youth-targeted commercials having a significantly lower level of fear arousal (xNon. Youth=0.70, ^Youth=o.34; F = 19.507, p<.001; see Table 3.38). Table 3.38 One-way Analysis of Variance Table Amount of Fear Arousal in the Commercial By Intended Target Group Age Category Source Sum-of-Squares DF Mean-Square F-Ratio P Age Category 8.682 1 8.682 19.507 0.000 Error 261.271 587 0.445 105 Chapter Three Content Analysis Research Results Therefore, H6a is fully supported. H6b stated that social marketing ads targeted at youth would be more likely to make use of positive appeals. In examining the overall tone of the commercial, commercials that were directed at youth were, indeed, more likely to make use of positive appeals, and less likely to make use of fear appeals (chi-square=91.110, d.f. =2, p< 0.001), as shown earlier in Table 3.35. A one-way ANOVA was conducted on the "overall tone" variable, and there were significant differences found in the means for the non-youth versus youth target groups, with youth-targeted commercials having the most positive overall tone (* Non-Youth=1-81, xYouth=2.28; F=59.925, p<0.001; see Table 3.39). Therefore, H6b was supported. Table 3.39 One-way Analysis of Variance Table Overall Tone of the Commercial By Intended Target Group Age Category Source Sum-of-Squares DF Mean-Square F-Ratio P Age Category 15.278 1 15.278 59.925 0.000 Error 149.656 587 0.255 H6c stated that social marketing ads targeted at youth would be more likely to make use of humorous appeals. However, in examining whether there was an attempt to be humorous in the ad, there was no significant difference between ads directed at youth and those that were not directed at youth (chi-square=0.002, d.f. = 1, p=0.963), as shown in Table 3.40: 106 Chapter Three Content Analysis Research Results Table 3.40 Is There an Attempt to be Humorous? By Intended Target Group Age Category Non-Youth Youth Total No 438 68 506 85.88% 86.08% 85.91% Yes 72 11 83 14.12% 13.92% 14.09% N 510 79 589 Total 100.00% 100.00% 100.00% As shown in Table 3.40, youth-targeted ads did not use humor more frequently than ads aimed at non-youth, so there is no support for H6c. This completes the testing of the hypotheses which had been proposed in Chapter 1. A summary of the key findings is shown in Table 3.41. 107 Chapter Three Content Analysis Research Results Table 3.41 Summary of Hypotheses Hla The majority of social marketing ads will contain one or more threats or negative consequences. Supported Hlb Of the ads which contain one or more threats or negative consequences, the majority will also contain action recommendations or advice for coping with the threat. Supported Hlc In social marketing ads which include both threat(s) and coping information, the threat(s) will tend to temporally precede the coping information. Supported H2 Approximately 24% of the social marketing ads will contain fear appeals. Supported H3 It is expected that a higher percentage of social marketing ads in Australia will use fear appeals, as compared to the United States. Supported H4a Corporate sponsors will be less likely than other organizations to use messages which incorporate threats or fear appeals. Partially Supported H4b Corporate sponsors will be more likely to use messages which are positive. Not Supported H4c Corporate sponsors will be more likely to use messages which are humorous. Not Supported H5a The more serious the potential consequences of a targeted behavior, the more likely it is that fear appeals will be used in social marketing ads against that behavior. Partially Supported H5b The less serious the potential consequences of a targeted behavior, the more likely it is that positive appeals will be used in social marketing ads against that behavior. Supported H5c The less serious the potential consequences of a targeted behavior, the more likely it is that humorous appeals will be used in social marketing ads against that behavior. Supported H6a Social marketing ads targeted at youth will be less likely to make use of fear appeals. Supported H6b Social marketing ads targeted at youth will be more likely to make use of positive appeals. Supported H6c Social marketing ads targeted at youth will be more likely to make use of humorous appeals. Not supported 108 Chapter Three Content Analysis Research Results Exploratory findings Logistic Regression The diagram in Figure 1.6 in Chapter One indicated that the use of fear appeals in social marketing television commercials would vary according to the four independent variables shown. These four independent variables were sponsorship, intended target group age, type of behavior being targeted, and country-of-origin. To examine the relative impact of these four independent variables on the dependent variables which measure the use of fear appeals, a regression analysis was used. Given the dichotomous and categorical nature of most of the content analysis variables, a logistic regression was deemed to be the most appropriate method of analysis. Logistic regression demands that the dependent variable being predicted must be a dichotomous variable. Two of the dependent variables were already dichotomous variables (the variable measuring HUMOR, and the variable measuring the presence or absence of a THREAT or negative consequence). For the purposes of this logistic regression analysis, two other dependent variables were modified to become dichotomous variables, as described here. The overall TONE2 of the commercial was recoded into two categories: (1) negative, and (2) positive/neutral. The level of fear arousal (FEARLEV2) intended in the commercial was also recoded into two categories: (1) none, and (2) low/medium/high. Two of the four independent (predictor) variables were also dichotomous variables: sponsorship (SPONSOR) was indicated as being corporate or non-corporate, and intended target group age category (TARGET AGE) was indicated as being youth or non-youth. The other two independent variables each contained five categories, which could optionally be reduced to two categories. Country-of-origin consisted of Canada, United States, Britain, 109 Chapter Three Content Analysis Research Results Australia, New Zealand (i.e., C0UNTRY(5)), but could potentially be reduced to two categories of North-American versus non-North-American (i.e., COUNTRY(2)). Targeted behavior consisted of AIDS prevention, DUI, alcohol moderation, smoking, and drug abuse (i.e., BEHAVIOR(5)), but could be reduced to two categories according to the seriousness of the behavior (i.e., BEHAVIOR(2)) as discussed earlier in this Chapter. The full regression model which describes the model in Figure 1.6 uses a full set of predictor variables: five countries, five targeted behaviors, two sponsorship categories, and two targeted age categories. It is expected that this full model would perform the best in the logistic regression in terms of classifying commercials to the correct categories for each of the four independent variables. However, it is interesting to compare this full model with various reduced versions of the model, in order to see whether a more parsimonious model can make predictions of equal accuracy. Table 3.42 indicates the results of running a logistic regression using several different combinations of these predictor variables. 110 Chapter Three Content Analysis Research Results Table 3.42 Summary of Results of Logistic Regression Indicating Percent of Commercials Correctly Classified THREAT TONE2 FEARLEV2 HUMOR SPONSOR(2) TARGET AGE(2) BEHAVIOR(5) COUNTRY(5) 75.04% 79.29% 63.33% 86.93% SPONSOR(2) TARGET AGE(2) BEHAVIOR(2) COUNTRY(5) 73.85% 78.61% 61.12% 85.91% SPONSOR(2) TARGET AGE(2) BEHAVIOR(2) COUNTRY(2) 73.85% 78.61% 61.12% 85.91% SPONSOR(2) TARGET AGE(2) BEHAVIOR(5) 73.85% 78.61% 61.46% 85.91% SPONSOR(2) TARGET AGE(2) BEHAVIOR(2) 73.85% 78.61% 60.78% 85.91% SPONSOR(2) TARGET AGE(2) 73.85% 78.61% 60.78% 85.91% SPONSOR(2) 74.02% 78.61% 54.67% . 85.91% TARGET AGE(2) 74.02% 78.61% 59.95% 85.91% BASE RATE: 74.02% 78.61% 53.99% 85.91% The "base rate" in Table 3.42 refers to the percentage of correct classifications that would be obtained by the logistic regression model if it had no real predictive ability, but simply assigned the same value to all commercials in the sample. In predicting the correct classification of the dependent variables THREAT, TONE2, and HUMOR, the use of the full model containing all four independent variables [SPONSOR(2), TARGET AGE(2), BEHAVIOR(5), and COUNTRY(5)] provides only a little improvement over the base rates (i.e., an improvement of approximately 1 % or less). The full model is more useful in making predictions about the FEARLEV2 dependent variable, since its rate of correct classifications is 63.33%, nearly 10 percentage points higher than the base rate of 53.99%. Ill Chapter Three Content Analysis Research Results Types of Fear Appeals One area of interest was whether the type of threats which suggest harm to oneself will be more common than those which suggest harm to close others and/or to society in general. Table 3.43 lists the frequency of occurrence of the various types of threats. Because some ads list more than one type of threat, the percentages total to more than 100%. Table 3.43 Frequency of Direction of Threats Type of Threat All Ads Threat Ads N % N % No Threat 153 26.0% Threat toward oneself 365 62.0% 365 84.1% Threat toward close others 107 18.2% 107 24.7% Threat toward unspecified others or strangers 64 10.9% 64 14.7% Threat toward society in general 28 4.8% 28 6.5% TOTAL: 589 100.0% 434 100.0% It is clear that the majority of threat ads (84.1%) contain one or more threats toward oneself. Also, the majority of all ads (62.0%) contain one or more threats toward oneself. Date of Origin and Fear Appeals It is interesting to explore whether there is any difference in the use of fear appeals in the earlier years of the sample (1980-1986) versus the later years (1987-1994). Data regarding the year in which the commercial was produced was available for only 350 out of the 589 commercials in the sample. In analyzing these 350 commercials, the data indicate that there was, indeed, a significant difference between the earlier and later time periods in terms of 112 Chapter Three Content Analysis Research Results whether any specific threats were mentioned or shown (chi-square=4.118, d.f. = 1, p=0.042). As shown in Table 3.44, there was a larger proportion of commercials with specific threats mentioned or shown in the earlier part of the sample (76.47%) when compared with the latter part of the sample (66.67%). Table 3.44 Any Specific Threats Mentioned or Shown? By Time Period 1980-86 1987-94 Total No 40 60 100 23.53% 33.33% 28.57% Yes 130 120 250 76.47% 66.67% 71.43% N 170 180 350 Total 100.00% 100.00% 100.00% However, in terms of the overall tone of the commercial, there was no significant difference in the use of fear-arousing commercials during the two time periods (chi-square= 0.642, d.f. = l, p=0.423; see Table 3.45). 113 Chapter Three Content Analysis Research Results Table 3.45 Overall Tone of the Commercial? By Time Period 1980-86 1987-94 Total Positive/ 126 140 266 Neutral 74.12% 77.78% 76.00% Fear 44 40 84 25.88% 22.22% 24.00% N 170 180 350 Total 100.00% 100.00% 100.00% A one-way ANOVA was conducted on the "overall tone" variable, examining whether the group means varied by time period, but the differences in mean values were not significant 1980-86= 1-83 vs. *1987_94=1.89, F = 1.138, p=0.287; see Table 3.46). Table 3.46 One-way Analysis of Variance Table Overall Tone of the Commercial By Time Period Source Sum-of-Squares DF Mean-Square F-Ratio p Time Period 0.370 1 0.370 1.138 0.287 Error 113.047 348 0.325 There was also no statistically significant difference in the perceived level of fear arousal when comparing the earlier years versus the later years (chi-square=1.721, d.f. =2, p=0.423), as shown in Table 3.47. 114 Chapter Three Content Analysis Research Results Table 3.47 Amount of Fear Arousal in the Commercial By Time Period 1980-86 1987-94 Total None 75 92 167 44.12% 51.11% 47.71% Low 74 69 143 43.53% 38.33% 40.86% Medium/ 21 19 40 High 12.35% 10.56% 11.43% N 170 180 350 Total 100.00% 100.00% 100.00% Similarly, in a one-way ANOVA conducted on the "amount of fear arousal" variable, the group means were examined by time period, but the differences in mean values were not significant (*i98o-86=0.69 vs. JC1987.94=0.61, F = 1.213, p=0.272; see Table 3.48). Table 3.48 One-way Analysis of Variance Table Amount of Fear Arousal By Time Period Source Sum-of-Squares DF Mean-Square F-Ratio P Time Period 0.598 1 0.598 1.213 0.272 Error 171.471 348 0.493 While there do appear to be more threats or negative consequences present in ads from the earlier years of the sample (as shown above in Table 3.44), there does not appear to be a significantly greater degree of fear arousal in the earlier years of the sample. Therefore, it cannot be conclusively stated that there are any differences between the earlier and later years of the sample with regard to the use of fear appeals. 115 Chapter Three Content Analysis Research Results In examining the relationship of humor to time period in the sample, Table 3.49 shows that humor was more prevalent in the latter years of the sample (chi-square=8.633, d.f. = 1, p=0.003). Table 3.49 Is There an Attempt to be Humorous? By Time Period 1980-86 1987-94 Total No 162 155 317 95.29% 86.11% 90.57% Yes 8 25 33 4.71% 13.89% 9.43% N 170 180 350 Total 100.00% 100.00% 100.00% Also, looking back at Table 3.45, we see that positive/neutral approaches were somewhat more prevalent in the latter part of the sample, although that result was not statistically significant (chi-square=0.642, d.f. = 1, p = .423). Comparing this Study to other Content Analyses It is interesting to compare the results of the present content analysis to the results obtained in other content analysis studies. As mentioned earlier in Chapter 1 in Table 1.1, there are three general content analysis studies which examine either fear appeals or social marketing ads (Hanneman, McEwen, & Coyne 1973, Unger & Stearns 1983, Wallack & Dorfman 1992). Additionally, there are two content analysis studies which specifically examine anti-drug commercials (Hanneman & McEwen 1973; Walker 1990), and two which specifically examine AIDS prevention commercials (Calvert, Cocking, and Smrcek 1991; 116 Chapter Three Content Analysis Research Results Freimuth, Hammond, Edgar, and Monahan 1990). This section will be devoted to comparing the results of the present content analysis study to these other content analysis studies (Hanneman, McEwen, & Coyne 1973, Unger & Stearns 1983, Wallack & Dorfman 1992). Comparing Three General Content Analyses Table 3.50 compares the present content analysis to three other general content analysis studies which dealt with either fear and/or social marketing TV commercials. It can be seen that the present study differs from the previous studies in a number of ways. First, the present study used snowball sampling, a non-probability method of sampling. This meant that commercials could be collected over a longer period of time in the present study (1980-1994), rather than being limited to a single short period as in the other three studies. Like the Hanneman, McEwen, and Coyne (1973) study, the present study was limited to PSAs and social marketing ads; however, the present study was limited to ads for only 5 particular types of behaviors (anti-smoking, DUI, alcohol moderation, AIDS prevention, and anti-drug abuse). In terms of incidence of fear appeals, in the present study 21.39% of the TV commercials had an overall tone which was considered to be fear-arousing; this compared to 24% of the PSA commercials in Hanneman, McEwen, and Coyne (1973) which used fear appeals, and only 16.6% of the general sample of product/service TV commercials in Unger and Stearns (1983) which used fear appeals. The present used four coders to evaluate every TV commercial in the content analysis, while Unger and Stearns (1983) and Wallack and Dorfman (1992) used two coders to evaluate every TV commercial. The Hanneman, McEwen, and Coyne (1973) study used a total of 18 coders, but that was spread out over their entire sample, and each TV commercial was actually viewed by only one coder. 117 Chapter Three Content Analysis Research Results Table 3.50 Comparison of Content Analysis Studies Unger & Stearns (1983) Wallack & Dorftnan (1992) Hanneman, McEwen, & Coyne (1973) Present Study Sampling Frame - sampled dayparts over a 4-week period across 3 U.S. TV networks - sampled 20 hours of U.S. TV airtime randomly over a 3-week period - sampled 500 hours of TV airtime during a 1-week period across 9 U.S. TV stations - used snowball sampling method, contacting relevant organizations in 5 countries Time period covered July 1982 April/May 1989 June 1972 1980-1994 Type of Commercials all types all types all types but analysis limited to PSAs only 5 types of PSAs: AIDS, smoking, DUI, alcohol moderation, drugs Total Number of Commercials 675 commercials 654 commercials 10,399 commercials 589 PSA commercials Number of PSAs/ Social Marketing Commercials n/a 34 PSA commercials 1,159 PSA commercials 589 PSA commercials Incidence of PSAs/ Social Marketing Commercials n/a 5.8% 11.1% n/a Incidence of Fear Appeals 16.6% of all commercials n/a 24.0% of PSA commercials 21.39% of PSA commercials 118 Chapter Three Content Analysis Research Results Table 3.50 Comparison of Content Analysis Studies Unger & Stearns Wallack & Hanneman, Present Study (1983) Dorfman McEwen, (1992) & Coyne (1973) How Fear - coded as fear n/a - coded as fear - coded as fear was appeal if the appeal if it showed appeal if it had Measured threat of a or discussed the overall tone negative harmful or painful which was fear-consequence was social or physical arousing used consequences of - also coded not doing what the whether there message directed were any threats or negative consequences, and level of fear as none/ low/med/hi Number of 2 2 18 4 Coders Other - guilt messages - product/service - celebrities - celebrities Variables - degree of types - intended target - intended target Measured emotion - types of health group group - degree of behaviors (topic) - topic of PSA - topic of PSA vividness - program/time at (type of -# "you" which PSA was behavior) references shown - sponsorship - # references to - country-of-negative origin consequences - humor - social/physical/ - positive financial/ appeals functional types of fear In terms of examining the other variables which were measured, since the Hanneman, McEwen, and Coyne (1973) study bears the closest resemblance to the present study in terms of scope and purpose (i.e., limited to social marketing commercials), some of the results from the present study will be compared to that study. In particular, that study examined the use of celebrities, and the intended target group. 119 Chapter Three Content Analysis Research Results Hanneman, McEwen, and Coyne (1973) found that 15 % of their PSA commercials used celebrities, while the present study found that 13.41% (79 out of 589 PSAs) were directed toward youth. Using a z-test of proportions, it was calculated that these two proportions are not significantly different (z=1.0809, p > .05). Hanneman, McEwen, and Coyne (1973) also found that 17% of their PSA commercials were explicitly directed toward youth. The present study found that 15.96% (94 out of 589 PSAs) were directed at youth. Using a z-test of proportions, it was again calculated that these two proportions are not significantly different (z=0.6720, p> .05). It is interesting that there is no significant difference between the Hanneman, McEwen, and Coyne (1973) study and the present study in terms of the presence of celebrities and the percentage of PSA commercials targeted at youth. These two studies took place at very different points of time using very different sampling methods, yet the results on two variables are very similar. This suggests that the results in the present study are fairly robust. Comparing Studies on Anti-Drug Abuse Commercials The two content analyses which deal with anti-drug commercials (Hanneman and McEwen 1973; Walker 1990) are summarized in Table 3.51. 120 Chapter Three Content Analysis Research Results Table 3.51 Comparison of Anti-Drug Content Analysis Studies Hanneman & McEwen (1973) Walker (1990) Present Study Sampling Frame - U.S. networks and independent TV stations over a 2-week period - collected ads through 3 nonprofit organizations - used snowball sampling method, contacting relevant organizations in 5 countries Timing December 1971 1985-90 1980-1994 Type of Commercials Anti-drug abuse Anti-drug abuse Anti-drug abuse Total # Anti-drug Abuse Commercials 32 PSAs 46 PSAs 105 PSAs How Fear was Measured n/a - whether fear was the primary appeal - coded as fear appeal if it had overall tone which was fear-arousing Number of Coders 22 coders in total (1-2 coders per ad) 1 coder 4 coders Other Variables Measured - intended target group - sponsor - race of characters - program/time of airing - physicial and social effects of drug abuse - drug paraphernalia - intended target group - celebrities - race of characters - intended target group - celebrities - race of characters - topic of PSA (type of behavior) - sponsorship - country-of-origin - humor - positive appeals As noted in Table 3.51, Hanneman and McEwen (1973) did examine sponsorship (i.e., they mentioned that 63.5% of the ads were sponsored by the National Clearinghouse for Drug Abuse Information, 7% were sponsored by the Advertising Council, and 6% were sponsored by State Governments), but they did not mention specifically whether the other 24% were sponsored by nonprofit or for-profit organizations or corporations. Therefore, this sponsorship variable cannot be compared with the present content analysis. 121 Chapter Three Content Analysis Research Results Table 3.52 Comparison of Results for Key Variables of Anti-Drug Content Analysis Studies Hanneman & McEwen (1973) Walker (1990) Present Study Use of Celebrities 43.75% had celebrities 21.3% had celebrities 8.0% had celebrities Intended Target Group 18% targeted youth 21.2% targeted youth 18.1% targeted youth Fear Appeals 42% used fear appeals 84% had one or more fear-inducing elements; 52.3% used primarily a fear appeal 72.0% had one or more threats mentioned or shown; 61.3% had a low/ medium/high level of fear arousal; 18.7% had fear-arousing overall tone As well, Hanneman and McEwen (1973) measured the intended target audience, and noted that 18% of the anti-drug ads were considered to be targeted at youth. This compares with 18.10% of anti-drug ads (i.e., 19 out of 105 anti-drug ads) in the present study which were targeted at youth. These two proportions are not significantly different from each other, using a z-test of proportions (p> .05) (see Table 3.52 for comparisons). Walker (1990) also examined intended target group, and found that 21.2% of the commercials had a youthful intended target group; this is not significantly different than the present study, in which 18.1 % of the anti-drug U.S. PSAs targeted youth (z=0.6187, p>.05). Celebrities were used in 14 out of the 32 ads in the Hanneman and McEwen (1973) study (43.75%); in the present study celebrities were used in only 6 out of the 75 U.S. anti drug ads (8.00%), which is significantly different (z=6.2411, p< .001). Walker (1990) also measured the incidence of celebrities, and found that celebrities were present in 21.3% of his sample (see Table 3.52). This is still a significantly greater incidence than that found in the 122 Chapter Three Content Analysis Research Results present study (z=2.6499, p<.05). Finally, Hanneman and McEwen (1973) measured the incidence of black characters in their PSAs. Of the main characters in their ads, 12% of the PSAs had only black characters, while 15% of the PSAs had both black and white main characters. Walker (1990) also measured the race of the main characters in the ads; the comparable figures are shown in Table 3.53. Table 3.53 Race of Main Characters in Anti-Drug Ads Hanneman & McEwen (1973) Walker (1990) Present Study - U.S.A. Anti-drug commercials White only not reported 51.4% 45.3% Black only 12% 7.5% 16.0% Hispanic only not reported 1.9% 12.0% White and Black 15% 7.5% 1.3% White and Hispanic not reported 5.3% Unknown (e.g., voice-over) not reported 31.7% 20.0% Table 3.53 indicates that, like the Hanneman and McEwen (1973) study which reported 12% of the commercials having only black characters, the present study also found that 16.0% of the U.S. commercials had only black characters; using a z-test of proportions, these two figures are not significantly different (z=1.0660, p> .05). However, unlike the Hanneman and McEwen (1973) study which reported that 15% of commercials had white and black characters, the present study showed a significant difference in that only 1.3% of the U.S. PSAs had both white and black characters (z=3.3227, p< .01). What is of interest to note is that 17.3% of the U.S. commercials in the present study had Hispanic main characters, either alone or in combination with white characters. The incidence of Hispanic characters 123 Chapter Three Content Analysis Research Results was not measured in the Hanneman and McEwen (1973) study, and the strong level of Hispanic characters in the present study probably reflects the increase in the Hispanic population in the United States during the past 20 years. Comparing Studies on AIDS Prevention Commercials There are two content analysis studies which examined AIDS prevention commercials (Calvert, Cocking, and Smrcek 1991; Freimuth, Hammond, Edgar, and Monahan 1990). These two studies are summarized in Table 3.54. 124 Chapter Three Content Analysis Research Results Table 3.54 Comparison of AIDS Prevention Content Analysis Studies Calvert et al. (1991) Freimuth et al. (1990) Present Study Total Number of AIDS PSAs 31 AIDS PSAs 127 AIDS PSAs 97 AIDS PSAs Sampling Method Collection by contacting specific nonprofit organizations Collection by contacting state and city health departments Snowball sample collected by contacting many organizations, govts., etc. Timing of Sample 1987-88 1988 1986-1994 Countries in Sample U.S., Germany, Austria U.S. only U.S., Canada, Britain, Australia, New Zealand How Fear was Measured n/a - whether fear was used as a primary appeal - coded as fear appeal if it had overall tone which was fear-arousing - also coded whether there were any threats or negative consequences, and level of fear as none/low/med/hi Number of Coders 4 coders 2 teams of 2 coders per team 4 coders Other Variables Measured - amount of medical information - information about transmission of the disease - prevention information - emotional tone - intended target group (by race) - humor - guilt - execution format - testimonials - preventive behaviors - celebrities - intended target group - topic of PSA (type of behavior) - sponsorship - country-of-origin - humor - positive appeals As seen in Table 3.54, the Calvert et al. (1991) study used a relatively small sample (31), which was divided between three countries. The Freimuth et al. (1990) study had a large sample of 127 AIDS PSAs which were all from the United States. None of the three studies 125 Chapter Three Content Analysis Research Results used probability sampling methods, but relied instead on collecting commercials from government or health department sources. Table 3.55 compares the two studies on the results of variables which are comparable to the present study. Table 3.55 Comparison of Specific Variables in AIDS Prevention Content Analysis Studies Calvert et al. (1991) Freimuth et al. (1990) Present Study Race: Blacks Hispanics n/a 14% 10% 13% 1% Incidence of Humor n/a 2% used humor 7% used humor Incidence of Fear Appeals n/a 26% had fear appeals 81% mentioned a threat or negative consequence; 48% had low/ medium/high levels of fear; 11% had fear-arousing overall tone Preventive Behaviors: - use a condom - abstain - avoid multiple partners 26% 6% 16% 18% 6% 4% 45% 9% 8% Table 3.55 indicates that the Calvert et al. (1991) study did not examine the incidence of fear appeals, humor, or race of main characters, but rather focused on preventive behaviors which were mentioned in the ads. It is difficult to compare the fear variable between the Freimuth et al. (1990) study and the present study because of the different methods used to code the fear variable. 126 Chapter Three Content Analysis Research Results Discussion and Conclusion This content analysis study examined the use of fear appeals in a sample of 589 social marketing ads which were collected from Canada, the United States, Britain, Australia, and New Zealand. The social marketing ads dealt with five specific behavioral categories: smoking, illegal drug use, DUI, excessive alcohol consumption, and unsafe sex. Over one-quarter of the ads contained no threats of any kind, and as hypothesized (H2), only about one-quarter of the ads gave an overall impression of being fear arousing. Only a very small proportion of the ads actually used very strong fear appeals. The use of fear appeals did seem to be tied to the seriousness of the behavior that was being portrayed. Behaviors which were deemed to be more serious (or to have more immediate negative consequences) seemed to be more likely to be linked with fear appeals in social marketing advertising (H5a). Behaviors which were deemed to have less serious consequences seemed to be more likely to be linked with positive or humorous appeals in social marketing advertising (H5b). The use of fear appeals appears to vary according to the age group of the target market, since social marketing ads with youthful target groups were less likely to make use of fear appeals (H6a). The flip-side of this was that ads with youthful target groups were more likely to make use of humor or positive appeals (H6b). The resistance of youth to threats or media pressure tactics may lie behind this phenomenon. There appeared to be consistency with some of the tenets of the Ordered Protection Motivation (OPM) model. Perhaps not surprisingly, given the limitations of the 30-second or 60-second television medium, the results found that the social marketing ads tended to have consistency with only to the less restrictive tenets of the OPM model. The majority of the ads contained one or more threats or negative consequences (support for Hla). Of those ads which 127 Chapter Three Content Analysis Research Results contained one or more threats or negative consequences, the majority also contained some coping advice, as recommended by the OPM model (Hlb). The majority of the ads also followed the temporal order prescribed by the OPM model, wherein the threat precedes the coping advice (Hlc). However, few of the ads included elements of coping response efficacy and self efficacy. These elements of coping response efficacy and self efficacy are recommended by the OPM model as being important elements for effective communication, but the time restrictions of the TV medium appear to preclude their widespread inclusion (Hunt, Tanner, and Eppright 1995). It is clear that there are definite differences between countries in the use of fear appeals. As suggested by the amount of media publicity which shocking Australian campaigns have garnered, the Australian ads were more likely to make use of fear appeals (H3). One might conjecture that this can be attributed to the macho nature of the Australian psyche, whereby a great deal of fear or threat must be used to extract behavioral change from the general public. Given that there was a relatively small sample of corporate sponsors of social marketing ads, the analyses involving corporate sponsors had a small degree of statistical power. Based on these analyses, there seemed to be only very limited evidence that corporate sponsors made less use of fear appeals than did other types of sponsors (H4a). Also, corporate sponsors were not more likely to make use of humor or positive approaches in their social marketing ads, contrary to what had been hypothesized (H4b). This perhaps suggests that corporate sponsors do not "water down" their social marketing messages. Given the self-interested nature of corporate sponsors, one might have expected that they would try to avoid making viewers uncomfortable by avoiding fear appeals, or might try to portray their product category more favorably by avoiding fear appeals. However, this does not seem to have necessarily been the 128 Chapter Three Content Analysis Research Results case. In terms of exploratory findings, more recent years of the sample have seen a trend toward a more light-hearted approach in social marketing ads, as evidenced by the increased use of humour and positive approaches. However, there was only limited evidence of a parallel trend in the reduction of the use of fear appeals in the latter part of the sample. One factor which should be kept in mind is that at least some proportion of the TV commercials or PSAs which were examined in this study may have been part of an integrated marketing communication (IMC) program (Schultz, Tannenbaum, and Lauterborn, 1993). This means that the television commercials that were examined may be conveying only part of the intended message from their respective campaigns. A 30- or 60-second television commercial cannot convey the entire spectrum of information or messages that a social marketing campaign would be trying to convey; therefore, examining the television commercials in isolation is necessarily only part of the whole picture. Limitations of the Research This content analysis research had several limitations. A key concern was that the use of the snowball sampling methodology yielded a relatively small sample of commercials in Australia, New Zealand, and Great Britain. As discussed earlier, this leads to the question of whether the commercials that were obtained are really a full set of the commercials that are actually available. If they are not a full set of the commercials that should actually be available, then there is an additional concern that this set may be biased in some way, or unrepresentative of the full sample of commercials that should be available. Future research which uses this type of methodology should undertake special additional efforts to gather all possible commercials from the countries being studied, in order to ensure that the sample is 129 Chapter Three Content Analysis Research Results as complete as possible. One possibility would be share this sample information with key individuals from each specific country, who might be in a better position to determine whether or not the sample of commercials from their particular country seems complete. Having a small sample of commercials from Australia, New Zealand, and Great Britain made it difficult to obtain sufficient power for more detailed statistical analyses. The low number of ads which had corporate sponsors also resulted in the same type of statistical analysis problem. Therefore, redoubling efforts to gather commercials would enhance the power of the statistical tests, and ensure that the study results had greater reliability and validity. A further limitation of this study relates to the coding of the commercials in the sample. While the coding scheme developed for this study was intended to be comprehensive, it is possible that some important variables may have been inadvertently left out. Developing a coding scheme requires that an initial subset of commercials be analyzed, in order to provide coding categories. It is possible that the choice of this initial subset of commercials could influence the choice of coding categories, by framing the category choices in a biased fashion. The commercials which formed the initial subset were randomly selected, but this does not rule out the possibility that a different initial subset of commercials might have resulted in a slightly different coding scheme. When coding the commercials using the established coding scheme, it was necessary for the four coders to rely on their own judgment regarding the measurement of the fear variable. These four coders generally agreed (>95%) about whether a commercial should be coded as high, medium, low, or no fear. The four coders learn from each other as they practice the coding task, and over time their judgments become more alike. This provides greater reliability within the group of coders. However, it is entirely possible that another set 130 Chapter Three Content Analysis Research Results of four coders might have rendered different judgments about the particular placement of some of the commercials into these categories, given that there might be different interpersonal and group dynamics within a different set of coders. However, it is unlikely that such an effect would be substantial. Contributions of the Research This content analysis study has made a number of useful contributions to knowledge about the use of fear appeals in social marketing advertising. Being the only study of its kind that has encompassed several behavioral categories across several countries, looking specifically at the phenomenon of fear appeals, this study is unique and rich in data. The study provided the first and only attempt to examine the extent to which fear appeals are used in several types of social marketing ads across several countries. This knowledge is expected to be of interest to academics as well as policymakers. The study also provided the only known attempt to examine the extent to which the message elements in social marketing ads are consistent with the tenets of the OPM model. This is expected to be of significant interest to researchers who study fear appeals, as well as to practitioners who may try to incorporate elements of this communication theory into their social marketing television commercials. The study has provided some insight into cross-national differences in the use of fear appeals, and validated a belief that Australia has tended to make greater use than the U.S. does of fear appeals in social marketing ads. Finally, the study provided an examination of the issue of corporate sponsorship of social marketing ads. Somewhat surprisingly, there is only limited evidence to suggest that corporate sponsors make less use of fear appeals, and corporate sponsors do not seem to use 131 Chapter Three Content Analysis Research Results humor or positive approaches more often.12 This type of information is expected to be of interest to academics as well as policymakers. Future Research There are many opportunities for future research which stem from this content analysis study. Future research opportunities are expected to be of two types: (1) content analysis studies which could provide an interesting comparison to the present study; and (2) laboratory studies or experiments which examine consumer reaction to some of the areas examined within the present content analysis. Future content analysis studies could extend the comprehensive snowball sampling technique to collect social marketing ads from non-English speaking countries, either in Europe or the Pacific Rim. The newly gathered social marketing ads from the non-English speaking countries could be compared with the results for the present study, in order to examine similarities and differences between countries and cultures. Other types of potential content analysis studies could examine different behavioral categories of social marketing ads, from the same or different countries, in order to make comparisons. The cross-sectional probabilistic variety of content analysis study could also be conducted, in order to investigate the incidence and types of social marketing ads within the television, radio, magazine, or newspaper media. While typically the incidence of social marketing ads is very low, necessitating a very large sample of TV/radio programming, this type of study would provide an interesting comparison to the present study. 12 Because of the relatively small size of the sample of corporate-sponsorship ads, the null results should not be thought of as definitive. Further study of this issue is needed in future research. 132 Chapter Three Content Analysis Research Results A wide variety of experimental work could extend out of the present study. One potential study would examine consumer response to varying the sponsorship of social marketing ads (i.e., corporate sponsor versus government or nonprofit organization). Different groups of consumers would be exposed to the same social marketing ads, either sponsored by corporations or nonprofit organizations, and their responses to the ads (behavioral intentions, AAd, etc.) would provide interesting indications about the effectiveness of corporate social marketing advertising. Other studies could expose subjects to ads which either adhere to, or depart from, the tenets of the OPM model. One purpose of such studies would be to demonstrate which of the elements of the OPM model have the greatest impact on subjects, and whether the exclusion of some of the elements seriously impedes communication or persuasion attempts. Another purpose of such studies could be to examine whether adherence to, or departure from, the OPM model has a potentially different impact on various target groups. Relatively little experimental work has been conducted on the OPM model, and this would be a fruitful area for future research. The use of fear appeals, while apparently somewhat in decline in recent years as measured in this study, still comprise a substantial enough segment of social marketing ads to warrant being an important area of research, particularly due to the public policy implications. There are still many interesting avenues to explore in this field. 133 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses CHAPTER FOUR USING SOCIAL MARKETING ADS TO ATTACK MALADAPTIVE COPING RESPONSES: PROBLEM DEFINITION AND OVERVIEW 134 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses CHAPTER FOUR USING SOCIAL MARKETING ADS TO ATTACK MALADAPTIVE COPING RESPONSES: PROBLEM DEFINITION AND OVERVDEW Problem Definition As we saw in Chapter One, fear arousal is a technique that is commonly used by social marketers in their advertising messages. The content analysis described in Chapters Two and Three examined the incidence of fear arousal in five types of social marketing TV commercials (i.e., those for AIDS prevention, smoking, drug abuse, drinking and driving, and alcohol moderation). The content analysis results indicated that some threats or negative consequences are mentioned in a majority of social marketing ads, and actual fear arousal is used in nearly a quarter of all social marketing messages. However, in spite of an extensive history of fear appeals research in the psychology literature, relatively little is known about the effectiveness of fear-arousing communications. Most of the studies in the fear literature have focused on manipulating the characteristics of the fear appeal message, such as high versus low fear, probability of occurrence, providing reassuring coping advice, positioning of coping advice, level of response efficacy or reassurance, comparing different communication modes, high versus low vividness of the information, source effects, message repetition, etc. The assumption in these types of studies is that fear appeals have the same effect on everyone. However, it seems possible, and even likely, that fear appeals could vary substantially among different groups, yet only a few studies have examined the impact of fear appeals on different target groups. It is proposed in this Chapter that there can be substantial differences in target groups regarding their propensity to respond to fear appeals. Specifically, it is proposed that 135 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses individuals who have been engaging in a particular negative behavior will have built up defensive mechanisms over time, which include maladaptive responses1. It is proposed that fear appeals will be unsuccessful in motivating these individuals to change their behavior, since they will respond by becoming defensive and engaging in maladaptive responses. In order to motivate these individuals to change their behavior, it is suggested that it would be necessary to use social marketing ads which attack the maladaptive responses. This Chapter will lay the groundwork for studies described in Chapters Five through Seven which are intended to test the usefulness of including counterarguments against maladaptive coping responses in social marketing ads targeting negative health behaviors. The efficacy of such counterarguments in changing attitudes and behavioral intentions is examined in the context of two behavioral groups: (1) those who are heavily committed to, or involved with, the behavior, and (2) those who are less involved or uninvolved with the behavior). This research is intended to test the hypothesis that social marketing ads containing counterarguments against maladaptive coping responses will be most effective with the high-involvement behavioral group. Therefore, this Chapter will outline the relevant literature, and explain the basis for the formation of the hypotheses. Variables Studied in Fear Appeals Research Over forty years of fear appeals research in the psychology literature has still resulted in relatively little being known about the effectiveness of fear-arousing communications. Most 1 Maladaptive coping responses are responses which reduce the level of fear, without appreciably reducing the level of danger (Rippetoe and Rogers 1987). Maladaptive coping responses can include simply ignoring the threat, minimizing the severity of the threat, minimizing the probability of the threat occurring, or engaging in coping actions whose efficacy is questionable. 136 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses of the studies in the fear literature have focused on manipulating the characteristics of the fear appeal, such as high fear versus low fear (e.g., Insko, Arkoff, and Insko 1965; Leventhal and Watts 1966; Ramirez and Lasater 1976; Rogers, Deckner, and Mewborn 1978; Smart and Fejer 1974; Steele and Southwick 1981), probability of occurrence (Chu 1966; Hass, Bagley, and Rogers 1975; Rogers and Mewborn 1976), providing reassuring coping advice (Cope and Richardson 1972; Leventhal, Watts, and Pagano 1967; Rogers and Mewborn 1976; Rogers and Thistlethwaite 1970; Tanner, Day, and Crask 1988, 1989), position of coping advice (Cecil, Weiss, and Feinberg 1978; Leventhal and Singer 1966), level of response efficacy (Natarajan 1979; Rogers and Deckner 1975) or reassurance (Mewborn and Rogers 1979; Natarajan 1978), comparing oral versus written communication modes (e.g., Krishnamurthy 1986), high versus low vividness of the information (Sherer and Rogers 1984), source effects (e.g., Dembroski, Lasater, and Ramirez 1978; Fritzen and Mazer 1975; McCroskey and Wright 1971; Ramirez and Lasater 1977; Stainback and Rogers 1983), importance/involvement of the topic (Beck and Davis 1978), attributing arousal effects to some other source (e.g., Schwarz, Servay, and Kumpf 1985), and message repetition (Horowitz 1969; Treise, Weigold, and Stankey 1995). The assumption in these types of studies is that the impact of the fear appeal is a universal one, and that it is only necessary to determine which type of fear appeal format or which level of fear-arousal is optimal (Strong and Dubas, 1993). Yet it seems possible, and even likely, that the effectiveness of fear appeals could vary substantially among different target groups. There is little reason to believe that everyone should respond to a fear appeal in the some way. Some authors have already pointed out that some target groups may be less likely to respond well to fear appeals. For example, Lastovicka et al. (1987) has suggested that young men may not respond well to fear appeals in DUI advertising. In spite of this, only a few studies have examined the impact of fear 137 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses appeals on different types of target audiences. There have been some studies which have examined how characteristics of the individual can act as variables which mediate the effectiveness of fear appeals. Among the characteristics examined are credulity of the subjects (Ray and Wilkie 1970), and self-esteem of the subjects (Leventhal and Perloe 1962; Leventhal and Watts 1966). These individual characteristics can shed light on why fear appeals might not work as well on some people as compared to others; however, they do not necessarily provide a practical basis for segmentation. The following section will highlight studies which have specifically examined how different target groups can vary in their response to fear appeals. Use of Segmentation in Fear Appeals Research While it seems possible that the effectiveness of fear appeals could vary among different target groups, only a few studies have examined the impact of fear appeals on different types of target audiences. Before reviewing these studies, it is necessary to first outline the different types of segmentation approaches that are possible. There are three basic segmentation approaches that are commonly used in marketing: demographic/socioeconomic segmentation, personality/values/lifestyle segmentation, and product usage segmentation (Darmon and Laroche 1991). The relevance of each of these to the use of fear appeals messages will be examined in turn. Demographic/Socioeconomic Segmentation Two particular studies which have examined the impact of fear appeals on different demographic groups found that different target audiences reacted differently to the same fear appeal (Burnett and Oliver 1979; Burnett and Wilkes 1980). These two studies examined the 138 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses response of different demographic groups to varying levels of fear appeals, and found that fear appeals used in the context of selling the services of a health maintenance organization (HMO) tended to work better in target segments that consisted of blue collar workers, lower income groups, and those with lower educational attainment. However, these results appear to have limited generalizability. It is unlikely that one can generalize and say that blue collar workers and lower income groups are always more persuaded by fear appeals. It may in fact be the case that it is only for this particular HMO product that these specific target groups were more susceptible to fear appeals. Because these two studies were conducted in the context of selling specific HMO services, rather than social marketing, it seems clear that the results of these two studies cannot be generalized to the social marketing context. Several other studies on the effectiveness of fear appeals have broken their results down by demographics such as age or sex (e.g., Leventhal, Jones, and Trembly 1966), although their findings in this regard tend to be ambiguous or inconclusive (Higbee 1969). For example, it has been suggested that for AIDS commercials men and women may need to be treated as distinct target groups, with separate messages (Golden and Suder 1994); however, it is well known that men and women view sex differently so it is not surprising that these two target groups might require different approaches for AIDS prevention messages. Yet knowing that men and women need to be targeted separately for AIDS prevention messages does not necessarily provide any guidance for the usefulness of this segmentation approach when using fear appeals in other types of social marketing campaigns. For this reason, it is not expected that useful generalizations can be made regarding the reactions of specific demographic groups to fear appeals. If the effectiveness of fear appeals differs between different target groups, this will most likely be related to personality/values/lifestyle variables, or by product usage/behavioral variables. 139 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses Personality/Values/Lifestyle Segmentation There are a number of studies in the psychology literature which have examined fear appeals which are segmented by psychological/personality variables such as self-esteem (e.g., Leventhal 1962; Rosen, Terry, and Leventhal 1982; Zemach 1966), need for approval (Rozelle et al. 1973), need for cognition (Stout and Sego 1995), internal/external locus of control (Burnett 1981), vulnerability (Dziokonski and Weber 1977; Niles 1964), personal level of anxiety (Wheatley and Oshikawa 1970). These studies provide some psychological insights into which personality variables may influence the success of fear appeals, but the findings tend to be inconclusive and of limited usefulness for social marketers. For example, knowing that someone with internal locus of control responds differently to fear appeals than someone with external locus of control does not provide social marketers with a useful tool for segmentation. These variables may explain why some individuals respond somewhat differently to fear appeals, but they may be confounded with other variables. The likelihood of engaging in the behavior in question may vary between people with these different personality traits; if that is the case, the behavioral category will be confounded with the personality variable. While there are several common lifestyle segmentation schemes (e.g., VALS2), there is no known research which has measured the effectiveness of fear appeals among different lifestyle segments. Therefore, it is not known whether segmentation according to lifestyles or values can provide a useful typology for predicting the effectiveness of fear appeals. 2 Values and Lifestyles (VALS) is an annual study which is conducted by SRI. Its web site is located at http://future.sri.com/vals/vals.description.html. 140 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses Segmentation by Product Usage or Behavioral Category The type of segmentation approach which has not been examined much in the existing fear appeals literature is what marketers would call product experience or usage variables. An exception to this is a study by Wheatley (1971) which found distinct differences between owners and non-owners of life insurance in their reactions to fear-arousing life insurance ads. Non-owners had a postive reaction toward the fear appeal ad, while owners of life insurance had a negative reaction to the fear appeal ad. Unfortunately, there was little in the way of theory which might explain why there might be this differential response between the two product usage groups. However, the point is underlined that product marketers recognize that it may be necessary to tailor messages differently for heavy product users as compared to light product users or non-users (Darmon and Laroche 1991). The reason for this is that the different target groups may differ significantly in their knowledge about the product and the extent to which they are committed to using the product. As well, they may differ significantly in how easily they could be persuaded to change their product usage behavior (i.e., switch brands). The social marketing situation is analagous to this: a person who is currently engaging in a negative health behavior may be either heavily committed to the behavior (i.e., like a "heavy" user of a brand) or may only engage in the behavior occasionally (i.e., like a "light" user of a brand). Depending on their level of knowledge about the issue and their level of commitment to the negative health behavior, individuals may differ in how easily they can be persuaded to change their behavior or switch to the behavior being recommended by the social marketer. This suggests that social marketers using fear appeals may need to use a segmentation approach (Quinn 1992), in order to communicate most effectively with different "product usage" groups. 141 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses Social marketers are starting to recognize that different "product usage" groups may require different messages (Golden and Johnson 1991). For example, the youth non-smoking campaign run in California does not assume that all young smokers are alike; instead, it uses a six-part classification system to represent a continuum (Mintz 1996). Included in this continuum are the following six teen-age groups: (1) regular smokers who intend to continue, (2) regular smokers, (3) occasional smokers, (4) quitters, (5) susceptible non-smokers, and (6) emphatic non-smokers. Recognizing that not all smokers are equally committed to their habit makes it easier to develop goal-directed campaigns that are aimed at specific segments (e.g., preventing susceptible non-smokers from starting to smoke; reinforcing the efforts of quitters; encouraging occasional smokers to become quitters). Different psychological approaches may work best for each of these target groups. Other studies have used psychographic or lifestyle typologies to segment smokers (Hornik 1989). Fear appeals would likely meet with a different degree of success in different target groups, because some groups might be more responsive to fear appeals. In a similar vein, Prochaska and DiClemente (e.g., DiClemente, Prochaska, et al. 1991; DiClemente, Prochaska, and Gibertini 1985; Prochaska and DiClemente 1986; Prochaska et al. 1992, 1991, 1985) posit a revolving door model involved in trying to quit smoking. This cyclical model suggests that there are several stages involved in quitting smoking (or other addictive behaviors): (1) pre-contemplation; (2) contemplation; (3) preparation; (4) action; and (4) maintenance. The types of messages that would be most appropriate would differ, depending on which part of the cycle one is targeting, with the general goal being to move individuals along to the next stage in the process. Those in the pre-contemplation stage would be urged to contemplate giving up smoking; those in the contemplation stage would be urged to act; and those in the action stage would be urged to 142 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses keep up the good work and avoid relapse. The presence of five stages in this model suggests that a segmentation approach to communications would be useful. Again, while fear appeals might be an appropriate psychological approach for individuals who are engaged in one or more of these stages, it seems quite unlikely that fear appeals would be universally appropriate for individuals at every stage. Individuals at different stages of the addictive process are influenced by different motives, social groups, and other factors. A six-stage model of addiction posits the following stages: (1) usage not contemplated; (2) usage contemplated; (3) trial; (4) light use; (5) moderate use; and (6) heavy use, and suggested that the impact of PSAs would depend upon the stage of the addictive process, with higher ad recall and perceived ad effectiveness among non-users than among users (Bozinoff, Roth, and May 1989). Another model of stages of involvement in the addiction process suggests that the stages may include (1) trial; (2) light use; (3) transition to addictive use; (4) addictive use; (5) cessation of addictive use; (6) post-cessation; and possibly, although not inevitably, (7) relapse-repeat dependence (DePaulo, Rubin, and Milner 1987). Anti-drug and alcohol campaigns can differentially impact upon individuals, depending upon the stage in the addictive process that they belong to. Different behavioral target groups are likely to react differently to fear appeals. For example, one study which examined the use of a fear appeal found that smokers who viewed a fear-arousing anti-smoking film had a greater increase in anxiety than did non-smokers (Watson, Pettingale, and Goldstein 1983). The reason for this may be that non-smokers had little to fear if they did not smoke, but smokers had much to fear. The opposite result was found in a study which examined an AIDS prevention campaign in the UK. This campaign, which used fear messages regarding the link between IV drug use and AIDS, was found to significantly increase anxiety in low/no risk individuals, but had no 143 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses impact on those at high risk group of IV drug users (Sherr 1990). This seems surprising, since those who were not using IV drugs had no reason to fear anything; however, because AIDS is transmitted in ways other than through IV use, the thought of contracting AIDS through other means may have generated the anxiety in the non-IV drug user group. Interestingly, the willingness to change sexual behavior did not change in either group, nor did drug-using behavior change in those in the high risk group. Research for a drinking and driving campaign in Quebec found that there were three groups of drivers: 55% who recognize the risks and won't drive after drinking; 23% who do DUI but are oblivious to the DUI message; and 22% who are receptive to the DUI message but who still sometimes take risks (Chiasson 1994). It is the latter group that a graphic fear-provoking DUI campaign was aimed at, since this is the group that is most willing to change its behavior. This approach aims to change the behavior of those who are most willing to change, rather than trying to change the behavior of those whose behavior is most in need of change. Those who are oblivious to the DUI message account for 23% of the population, but probably account for a much greater proportion of the DUI behavior (i.e., the 80/20 rule3). Given that "product usage" segmentation is an approach that has been commonly used by marketers, and that it seems to be gaining some credence among social marketers [e.g., California non-smoking campaign (Mintz 1996), APPLAUSE program (Lavigne, Albert, and Simmons 1985)], it seems logical to state that communications and advertising messages should be tailored to the specific "product usage" group being targeted. This tailoring of advertising messages suggests that fear appeals (or any appeals) will not be universally successful with all 3 It is often said that 20% of the customers account for 80% of the business, etc. It seems possible that the 20/80 rule of thumb may hold for DUI, with 20% of the population accounting for 80% of the DUI incidents. 144 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses target groups or all "product usage" groups. So if fear appeals are to be successful, for which target groups are they appropriate? There is virtually no research to date which answers this specific question. Psychological Theories Relevant to the Use of Fear Appeals It seems likely that fear appeals would be more successful at persuading those who are less committed to the behavior. The conceptual reasoning goes like this: If an individual is heavily committed to a negative health behavior, then fear appeals are likely to make them react defensively to shut out the message. However, if one's involvement with the negative health behavior is "light" or occasional, rather than habitual, then fear appeals may be more likely to motivate action rather than raise defensive mechanisms.4 This section will use a variety of psychological and behavioral theories to support this contention, and explain why fear appeal messages are less likely to be successful with high-risk individuals who are already heavily committed to a negative behavior. According to psychological reactance theory (Brehm 1966), threatening to restrict a person's behavior, or actually eliminating a person's freedom to act, motivates the individual to reestablish the lost or threatened behavior or attitude. Thus, when heavy drinkers are told that they should abstain in certain situations (e.g., when driving), or when smokers are told that their behavior is likely to create health problems, these individuals are likely to see their freedom to indulge in these behaviors as being threatened. This threat may make the individual react negatively or defensively, so that they increase their attachment to the behavior 4 The Quebec DUI campaign, mentioned above, is an example of using a high-fear campaign that is directed at those who are less committed to the risk behavior (i.e., who only occasionally engage in DUI). 145 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses in question. This could help explain why individuals who frequently engage in DUI behavior may not be receptive to messages which threaten their freedom. These messages may actually have the reverse effect of increasing their attachment to the negative behavior. If a persuasive message inadvertently arouses emotions other than fear, the effectiveness of the fear appeals may be reduced (Dillard et al. 1996). For example, some fear appeals may also arouse anger, which can inhibit rather than enhance persuasion (Dabbs and Leventhal 1966; Leventhal and Singer 1966; Leventhal, Singer, and Jones 1965). As well, these additional emotions such as anger may be aroused selectively in only some individuals, resulting in different segments reacting differently to the same fear appeal message. It is relatively difficult to design a message with only a fear-arousing component which does not arouse any other emotions in potential message recipients, so this may help explain why fear appeal messages meet with limited success among some target groups. The theory of perceptual defense (McGinnies, 1949; Schuster and Powell, 1987) offers another explanation about why fear appeals may be ineffective with some segments of people. Consumers either ignore or do not pay attention to messages that are contrary to their own beliefs or that do not interest them (Kline, Miller, and Morrison 1976). This defensiveness, along with the subsequent increases in the individual's anxiety level, is especially present when the consumer is not provided with a method to cope with or help solve the problem (Leventhal, Watts, and Pagano, 1967). Fear about a health topic can reduce the ability to process messages about that topic, suggesting that defensive avoidance is a fear-reducing response (Jepson and Chaiken 1990). The use of moderate fear appeals can have an impact on argument processing, depending on subjects' expectations of the reassurance that would be provided by the message (Gleicher and Petty 1992). This may help to explain why it is extremely difficult to persuade high-risk groups to adopt appropriate coping responses 146 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses (Leventhal, Singer, and Jones, 1965). High-risk groups are already committed to a negative behavior, and may ignore high-fear messages that are contrary to their beliefs. This then leads to the following question: if fear appeals are not successful (or less successful) among those heavily committed to a negative health behavior, then what approach would work better with this target group? One possible answer would be that in order for a message to work well with a target group that is heavily committed to a negative health behavior, the message must avoid raising defensive mechanisms. A second answer would be that such messages must promote the adoption of appropriate coping responses, and successfully discourage maladaptive coping responses. Designing Media Campaigns Designing and implementing effective media campaigns is a difficult task, which can incorporate many different conceptual perspectives. Table 4.1 indicates the most commonly used conceptual foundations for media campaigns (Winett, Altman, and King 1990). 147 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses Table 4.1 Conceptual Foundations for Designing and Implementing Media Campaigns Theory/Framework' Perspective Significance Example of an Application Social Cognitive Theory Principles and stratgies of individual behavior change Use of modeling principles in media Communication Theory Principles for information processing, delivery system design Organize information for retention enhancement Process of Change Schema Ordering of principles and strategies for stages of behavior change Successive change steps in a campaign to change individual behavior Community Organization Principles and strategies for change in community systems Organization of community resources to facilitate a media campaign Diffusion of Innovation Principles and strategies for adoption and spread of innovations Augment a media campaign with an interpersonal component Public Health Approach Principles and strategies for preventive interventions Combining "passive" and "active" prevention methods in one campaign Developmental/ Ecological Perspective Focus on critical periods, roles, and settings Target people at critical transition points in life in particular settings Social Marketing Overarching elements of a campaign Product design, price, promotion, place, and positioning Behavioral Systems Integration and ordering of principles and strategies Optimize design, community aspects, delivery, and evaluation of a campaign Source: Winett, Altman, and King 1990 There are also many models that have been developed to explain the workings of health communication messages (Albert 1981). Among these are Social Learning theory (Bandura 1977b), the Health Belief model (Beck 1974; Rosenstock 1974), the PRECEDE model (Green et al. 1980), and the Fishbein/Ajzen Theory of Reasoned Action model (Fishbein and Ajzen 1975; Ajzen and Fishbein 1980). These models have been compared with one another elsewhere (Hecker and Ajzen 1983; Mullen, Hersey, and Iverson 1987; Oliver and Berger 148 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses 1979; Prentice-Dunn and Rogers 1986), and are often used in combination with one another to improve explanatory power (e.g., Gonzalez 1989; Rosenstock, Strecher, and Becker 1988; Sutton 1987). These models will be outlined here briefly, in the context of DUI as a problem behavior. Social Learning Theory Social learning theory has been advocated to help change addicted behavior and reorient the processing of risk information in a more objective fashion (Bandura 1977b; Petty, Baker, and Gleicher, 1991). This can be accomplished through developing new skills, actions, and enhanced self-perceptions in the modeling of behavioral consequences. In the DUI context, this would suggest that individuals who engage in DUI behavior might need to develop a new repertoire of skills to handle potential DUI situations (e.g., taking a bus or taxi home, asking a friend to drive them home, asking a friend if they can stay overnight, etc.). However, in order to convince an individual to undertake these new behavioral skills, it might be necessary to first convince them that their old method of dealing with the situation is not effective. For someone who frequently engages in DUI, this would mean that it would be necessary to convince them that the maladaptive coping responses that they have used in the past, such as taking side streets home or driving more slowly, are relatively ineffective in avoiding the negative consequences of DUI. The Health Belief Model The Health Belief Model (Beck 1974; Rosenstock 1974) is often used to examine and predict responses to health-related issues. Key elements in this model include perceived susceptibility, perceived severity, perceived benefits of action, and perceived barriers to action, 149 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses INDIVIDUAL PERCEPTIONS Perceived Susceptibility to Disease "X" Perceived Seriousness (Severity) of Disease "X" MODIFYING FACTORS Demographic Variables (age, sex, race, ethnicity, etc.) Sociopsychological variables (personality, social class, peer and reference group pressure, etc.) Structural Variables (knowledge about the disease, prior contact with the disease, etc.) Cues to Action Mass media campaigns Advice from others Reminder postcard from physician or dentist Illness of family member or friend Newspaper or magazine article LIKELIHOOD OF ACTION Perceived benefits of preventive action Perceived barriers to preventive action 1 1 Perceived Threat of Disease "X" Likelihood of Taking Recommended Preventive Health Action L Figure 4.1 The Health Belief Model (Source: Rosenstock 1974) all of which combine to determine the likelihood of taking the recommended preventive health action (see Figure 4.1). The Health Belief Model posits that individuals who are more likely to adopt preventive behaviors are those who believe that they are personally susceptible to the target disease, that the consequences of having the disease would be severe, and that the benefits of engaging in preventive actions outweigh the barriers. This suggests that it would be necessary to convince those that engage in DUI behavior that they are susceptible to being injured in a DUI-related accident, that this would be a serious occurrence, and that there are substantial benefits to be gained from preventing DUI from occurring. However, it would clearly be easier to convince those who only occasionally engage in DUI than it would be to convince individuals who repeatedly engage in DUI. Therefore, the Health Belief Model can predict which segments will be more likely to adopt the preventive behavior, but is not 150 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses necessarily useful in designing campaigns which will persuade the most at-risk segment. The Health Belief Model has been applied to many health-related behaviors, including AIDS prevention (Brown, DiClemente, and Reynolds 1991; Montgomery et al. 1989; Petosa and Jackson 1991; Petosa and Wessinger 1990, 1989-90; Rosenthal, Hall, and Moore 1992; Walter et al. 1993; Wilson, Manual, and Lavelle 1991), contraceptive use (Lowe and Radius 1978), children's drug use attitudes and behaviors (Almarsdottir and Bush 1992), alcohol and drug education (Gonzalez 1988), drinking and driving prevention (Albert and Simpson 1985), smoking prevention (Ferraro 1990), and preventive dental behavior (Chen and Land 1986), among others. However, a meta-analysis of studies which used the Health Belief Model found weak effect sizes and lack of homogeneity, which indicate that it would be premature to draw conclusions about the predictive validity of the Health Belief Model (Harrison, Mullen, and Green 1992). The PRECEDE Model The PRECEDE model of health education (Green et al. 1980) suggests that the behavioral antecedents of health problems are predisposing factors, enabling factors, and reinforcing factors. The predisposing factors include one's beliefs, attitudes, values and perceptions, which can facilitate or hinder personal motivation for change. Enabling (or disenabling) factors are barriers created by societal forces or systems, such as limited facilities, inadequate personnel, or restrictive laws. Enabling factors include the structure of the environment or community and an individual's situation that facilitates or presents obstacles to change. Reinforcing factors are those related to the feedback which the learner receives, which will either encourage or discourage behavioral change. Reinforcing factors include the positive or negative effects of adopting the behavior (including social support) that influence 151 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses continuing the behavior. Behavioral change will not proceed until the attitudes which underlie the motivation to perform (or not perform) health-related behaviors are changed. These three factors all require that the individual be considered in the context of their community and social structures, and not in isolation, when planning communication or health education strategies. In the DUI context, this again suggests that changing DUI behavior will be easier among individuals who only occasionally engage in DUI than among those who frequently or repeatedly engage in DUI. The Theory of Reasoned Action The Theory of Reasoned Action model (Fishbein and Ajzen 1975; Ajzen and Fishbein 1980) has been advocated for predicting health behavior, and as a means of planning public service messages (Evans 1978). This model suggests that one's behavioral intentions and actions are determined by one's attitudes toward the behavior. This suggests that rather than behavior being predicted by broad attitudes, behavior can be better predicted by more specific attitudes toward a specific behavior. In a health promotion context, this would suggest that one's specific attitudes toward one's own nutritional habits would be a better predictor of one's eating habits than one's more general attitude toward healthy behaviors. The specificity of the attitude must match the specific behavior, in order to provide predictive power. In the DUI context, this model suggests that there would be a substantial correlation between an individual's intention to DUI and their subsequent DUI behavior. To change the DUI behavior, it would be necessary to somehow change the individual's intention to engage in DUI behavior. The model is not specific about how this attitude change is best accomplished, but it is clear that the specific attitude must change in order for the behavior 152 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses to change. Review of Protection Motivation Theory The Protection Motivation (PM) model (Rogers 1975, 1983) was reviewed in detail in Chapter One when various theories of fear appeals were introduced. However, because the tenets of the model are important in outlining the proposed alternative theoretical approach, the PM/OPM models will be reviewed here briefly. Components of a Fear Appeals Magnitude of Noxiousness Probability of Occurrence Efficacy of Recommended Response Cognitive Mediating Processes Appraised Severity Expectancy of Exposure. Belief in Efficacy of Coping Response .Protection Motivation Intention to - Adopt Recommended Response Figure 4.2 Protection Motivation Model (Rogers 1975, 1983) The Protection Motivation model (see Figure 4.2) suggests that fear-arousing persuasive messages are comprised of: (1) the fear-arousing information about the threat, and (2) advice about how to best cope with the threat. Whether or not the viewer acts on the coping advice, or chooses a maladaptive coping response (e.g., ignoring the threat, discounting the probability of occurrence, etc.) is mediated by four cognitive appraisal processes. These four processes are appraisals of: (1) the available information about the perceived severity of the threat; (2) the perceived probability that the threat will occur; (3) the perceived ability of the recommended coping response to remove the threat (response efficacy); and (4) the individual's perceived ability to carry out the recommended coping response (self efficacy). 153 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses It is common for fear-arousing communications, such as social marketing ads, to include both threat information and coping information. The threat is used to arouse a level of fear in the message recipient, and then the coping information is presented as the recommended behavior which the message recipient should undertake in order to reduce that level of fear (Rogers 1975, 1983). However, if the message recipient perceives that the recommended coping response will be ineffective, or that he/she will personally be unable to effectively carry out the recommended coping response, then he/she may either begin or continue to engage in maladaptive coping responses instead. In the DUI context, the protection motivation model suggests that the individual must first be presented with some fear-arousing information about DUI. It is clear from this that the type of threat will be important, since it must be considered to be fear-arousing by the message recipient.5 For example, in the DUI context there are two possible types of threats: (1) getting into an accident and harming oneself/others/one's car; and (2) being caught by the police.6 The first threat (getting into an accident) also implies being caught by the police7. The second threat, being caught by the police, has negative legal consequences which may range from losing one's driver's license, to a fine, to a jail sentence. While both of these may be considered to be fear-arousing, it is possible that different subjects may react differently to 5 Different people may experience different levels of fear-arousal from the same stimuli. 6 The focus group research undertaken in Chapter Six shows that young men who engage in DUI behavior are far more concerned about being caught by the police than they are about getting into an accident. Their reason for this is that if they felt they were not capable of driving safely and avoiding an accident, they would not get into the car at all. Having gotten into their car, they have decided that they are sufficiently sober to drive, and therefore their only concern is about being caught by the police. 7 If a DUI is involved in an accident, the police will inevitably be involved and the DUi will be caught (unless it is a hit-and-run accident where the DUI flees the scene). 154 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses these two threats. Next, the protection motivation model suggests that the message recipient should be provided with some information about how best to cope with the threat (i.e., some adaptive coping advice). Whether or not the individual acts on the coping advice, or whether they instead choose to undertake a maladaptive coping strategy, will depend on the four mediating processes described above. The individual will make an assessment of the available information about the perceived severity of the threat. The individual will also assess the perceived probability that the threat will occur. These two assessments will influence the degree of fear-arousal that is experienced. Further, the individual will assess the perceived ability of the recommended coping response to remove the threat. For example, if the message recommends taking a taxi home, how effective a strategy is this for coping with the threatening consequences of DUI? Generally, most coping advice is considered to be fairly effective at removing the threatening consequences of DUI; however, the problem arises with the individual's perceived ability to carry out the recommended coping response. If the individual is short of money, taking a taxi home may not be a feasible alternative, and the individual may not be able to carry out the recommended coping response. The main problem with using fear-arousing communications lies in the fact that individuals in the high-risk category are already committed to their maladaptive coping responses, having found them effective many times in the past. They do not believe that new adaptive coping responses will be more effective than their maladaptive coping responses. Therefore, it seems necessary to show these individuals that their maladaptive coping responses do not, in fact, work as well as they might think. 155 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses An Alternative Approach: Counteracting Maladaptive Behavior in High-Risk Individuals Maladaptive coping responses reduce the level of fear, without actually reducing the level of danger (Rippetoe and Rogers 1987). Maladaptive coping responses can include simply ignoring the threat, minimizing the severity of the threat, niininiizing the probability of the threat occurring, or engaging in coping actions whose efficacy is questionable. For example, Tanner, Hunt, and Eppright (1991) provide a scale of STD (sexually transmitted disease) maladaptive coping responses which includes items such as: (1) developing a belief that there is nothing one can really do to prevent it; (2) nurturing a belief that all STDs can be cured; (3) believing that the possibility of getting an STD is minimal; or (4) believing in an ability to choose STD-free partners. What these maladaptive coping responses have in common is that they either reduce the perceived size of the threat, or they inflate the perceived efficacy of a questionable coping action. Previous studies have shown the difficulty of convincing the most at-risk group to adopt the appropriate adaptive coping response8 (e.g., Leventhal, Singer, and Jones 1965). This difficulty may be due to strong maladaptive coping responses which the high-risk behavioral group (e.g., heavy smokers, heavy drinkers) have adopted previously in similar threatening situations (Tanner, Hunt, and Eppright 1991). These previous threatening situations might include times when they have experienced some fear arousal upon being advised of the negative consequences of their behavior. In order to cope with the fear which is aroused, maladaptive coping responses may have been learned and repeated many times. Since the 8 The greatest social benefit can, however, be gained from persuading this most at-risk group to change their behavior. For example, there is a much greater social benefit that can be won by changing the behavior of 100 people who frequently engage in DUI behavior than there would be from changing the behavior of 100 people who only occasionally engage in DUI behavior. On the other hand, is undoubtedly easier to change the behavior of those who only occasionally engage in DUI, because they are less committed to the behavior. 156 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses dangers associated with many negative health behaviors are well known to those who engage in such behaviors, it seems likely that members of the high-risk behavioral group are very adept at using maladaptive coping responses to negate the level of threat. <ffionvenie"nT^ Adaptive coping response: Take a bus. Maladaptive belief: "Nothing I can do about it; need to get home." ignore threat Adaptive coping response: Choose a designated driver Adaptive coping response: (^Too expensive) QnconvenTent) Maladaptive belief: j "Chances of getting caught are minimal." = minimize perceived probability of threat Maladaptive behavior: Drive slowly or take side streets. = inflate perceived efficacy of questionable coping strategy Maladaptive belief: Penalties for DUI are not large" = minimize perceived seriousness of threat a Figure 4.3 Adaptive and Maladaptive Coping Responses under the OPM Model Figure 4.3 illustrates a case of a social marketing ad for DUI. The adaptive behaviors prescribed by the social marketing ad might be to "avoid dririking and driving" (or "don't drink," particularly for teenagers), "choose a designated driver", or "take a bus," or "take a taxi." However, individuals who have the highest behavioral risk (i.e., those who regularly engage in DUI) are unlikely to be persuaded to adopt these recommended adaptive coping responses, and are likely to counterargue against them by saying that the recommended adaptive coping responses are "boring/no fun," or that "it's too difficult to plan ahead," or that 157 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses "taking a taxi is too expensive." This high-risk behavioral group will have developed a number of maladaptive coping responses over time, to deal with their fears surrounding DUI. These maladaptive coping responses might include beliefs like, "There is nothing I can do about it because I need to get home," "Chances of getting caught are minimal," or "The penalties for DUI are small." Another maladaptive coping response might be a behavior which the individual believes he/she can engage in successfully as a means of avoiding being caught for DUI, such as driving more slowly, or taking the side streets home.9 Figure 4.3 illustrates how a person can choose from a number of adaptive coping responses and maladaptive coping responses (beliefs or behaviors) to deal with their fears related to getting caught for DUI. SOURCES OF INFORMATION Verbal Persuasion (Fear Appeals) Observed Learning COGNITIVE MEDIATING PROCESS INTERMEDIATE EMOTIONAL STATE COGNITIVE MEDIATING PROCESS INTERMEDIATE STATE COPING MODE Experience -1 Severity of Threat Probability of Occurrence Behavior Repertory Appraisal Coping Response(s) Efficacy (maladaptive or adaptive) Threat ^ Fear = Coping Appraisal Appraisal' Self-Efficacy . Protection Motivation Behavior (adaptive or maladaptive) Social Norms and Values Figure 4.4 Ordered Protection Motivation model (Tanner, Hunt, and Eppright 1991) In order to effectively persuade the high-risk behavioral group, it might be useful for 9 The individual may recognize the risk of DUI, but simply thinks that the maladaptive coping response is the most appropriate response for reducing the risk. 158 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses social marketing ads to present counterarguments against the most common types of maladaptive coping responses (i.e., a 'MALADAPT' ad). If presenting such counterarguments against maladaptive coping responses enhances the persuasiveness of social marketing ads (as measured by changes in attitudes and behavioral intentions), particularly in the case of the high-risk behavioral group, this would provide an interesting segmentation perspective for the Protection Motivation (PM) model (Rogers 1975, 1983) and the Ordered Protection Motivation (OPM) model (Tanner, Hunt, and Eppright 1991; reproduced in Figure 4.4). Understanding the impact of using counterarguments against maladaptive coping responses would also enhance the prescriptive abilities of the PM and OPM models, and be useful in providing guidelines to social marketers. Why would giving counterarguments against maladaptive coping responses be useful in changing attitudes and behavior of the high-risk behavioral group? By counterarguing against maladaptive coping responses, it may be possible to reduce reliance on these responses in favor of adopting the recommended adaptive coping response. The structural model outlined in Figure 4.5 highlights the relationships between some of the components of the OPM schema. Although this structural model will not be examined in the present research, a future study could be developed which would test the strength of the relationships in this model. As shown in Figure 4.5, it is expected that Attitude toward the high-risk behavior would be the result of three other attitudinal factors: Attitude toward the negative consequences resulting from the behavior, Attitude toward the recommended coping response (s), and Attitude toward the maladaptive coping response (s). Factors correlated with 159 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses Attitude toward Negative Consequences Attitude toward Attitude toward Involvement with Recommended Coping $ High-risk Behavior ^ High-risk Behavior Responses Attitude toward Maladaptive Coping Responses Figure 4.5 Structural Model of the Ordered Protection Motivation Schema having a negative Attitude toward the high-risk behavior10 are: (1) having a negative attitude toward (i.e., being highly concerned about) the negative consequences; (2) having a positive attitude toward the recommended coping responses; and (3) having a negative attitude toward the maladaptive coping responses. Based on this structural model, it can be seen that in order to change Attitude toward the high-risk behavior, it is necessary to change one of the three contributing Attitudes. Social marketers can attempt to make the Attitude toward the negative consequences increasingly 10 Having a negative attitude toward the risky behavior implies that the individual is not likely to personally engage in the behavior, and does not condone the behavior in others. In the case of DUI, the individual who has a negative attitude toward DUI would be unlikely to personally engage in DUI, would not have intentions to engage in DUI, and would not condone DUI behavior in others. 160 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses negative (via the use of fear appeals). Alternatively, social marketers could improve Attitude toward the recommended coping response, by making the coping response seem more socially acceptable or easier to do. Finally, social marketers could attempt to reduce the favorability of Attitude toward the maladaptive coping responses, making these alternatives seem less viable as coping mechanisms. Fear appeals attempt to increase the salience of the negative consequences, thereby making the Attitude toward the negative consequences even more negative. This strategy is likely to meet with success in individuals who already have a strong negative attitude toward the consequences (i.e., are already very concerned about the negative consequences). However, in individuals who use strong defensive mechanisms to deny the negative consequences, this fear appeal communication strategy is unlikely to be effective. For this reason, fear appeals are unlikely to work well on individuals who already repeatedly engage in high-risk behavior. It is posited in this thesis that a more successful strategy with individuals in the high-risk behavioral group would result from using social marketing ads to attack the maladaptive coping responses. If the favorability of their Attitude toward the maladaptive coping responses can be reduced, then it is likely that this will reduce the favorability of their Attitude toward the high-risk behavior, and ultimately reduce the incidence with which they engage in the risky behavior. Based on the preceding discussion, the following hypotheses are proposed: HI Among the high-risk behavioral group, MALADAPT ads will be more effective than OPM ads in changing attitudes in the desired direction. 161 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses This first hypothesis suggests that ads using what will be termed the MALADAPT format (i.e., containing arguments that counteract maladaptive coping responses) will be more effective with the high-risk behavioral group. The reason for this is that ordinary fear appeal ads are tuned out by this target group through defensive mechanisms. Since the high-risk target group has repeatedly engaged in their risky behavior and uses maladaptive coping responses to avoid the negative consequences of engaging in this behavior, one way to reduce the incidence of the behavior is to attack the maladaptive coping responses in the advertising message. H2 Among the low-risk behavioral group, OPM ads (which follow the OPM format of threat, coping advice, self efficacy information) will be more effective in changing attitudes in the desired direction than MALADAPT ads (which contain counterarguments against maladaptive coping responses). This second hypothesis suggests that the traditional OPM ads, which follow the OPM format of providing a threat, coping advice, and self efficacy information, will be more effective in changing attitudes11 and behavioral intentions in the low-risk behavioral group than the MALADAPT ads would be. The reason for this is that the MALADAPT ads contain counterarguments against maladaptive coping responses; the low-risk behavior group rarely or never engages in the risky behavior, and so has not developed a repertoire of maladaptive coping responses. Therefore, an ad counteracting these maladaptive coping responses will be ineffective with the low-risk behavioral group. It should be noted that the low-risk group may already have very negative attitudes toward DUI. If this is the case, then there may be little room for the attitudes and behavioral intentions of this group to change, creating a potential ceiling effect. 162 Chapter Four Using Social Marketing Ads to Attack Maladaptive Coping Responses H3a Among the high-risk group, the MALADAPT ad treatment group will show a greater degree of change in attitudes than the control group. H3b Among the high-risk group, the OPM ad treatment group will show a greater degree of change in attitudes than the control group. In order to show that both the OPM and MALADAPT treatments are truly effective, a control group will be used. Results of changes in attitudes and behavioral intentions for the two experimental groups will be compared with that obtained for the control group, in order to show the effectiveness of the treatments. The above-stated hypotheses will be tested in an exploratory experiment, which makes use of DUI ads and paper-and-pencil measures. The methodology and results of this experiment are outlined in detail in Chapter Five. 163 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads CHAPTER FIVE THE OPM MODEL AND MARKET SEGMENTS: AN EXPLORATORY STUDY USING DUI PRINT ADS 164 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads CHAPTER FIVE THE OPM MODEL AND MARKET SEGMENTS: AN EXPLORATORY STUDY USING DUI PRINT ADS At the conclusion of Chapter Four several hypotheses were outlined. Chapter Five will outline the objectives, methodology, and results of the exploratory study that was conducted using DUI print ads, which was intended to test these hypotheses. Objectives The main goal of this research is to examine the effectiveness of including counterarguments against maladaptive coping responses within a social marketing ad message, particularly in terms of its impact on a high-risk target group. The reasoning behind this is that different behavioral segments may react differently to fear appeals. It is thought that a high-risk segment might be particularly prone to react negatively or defensively when faced with a fear appeal message, so a different approach (i.e., counteracting maladaptive coping responses) may be more effective with the high-risk segment. It is expected that this research will be of academic interest, as well as being of practical significance. Experiment Methodology This section will outline the experimental design and procedures that will be used to test the hypotheses outlined at the conclusion of Chapter Four. 165 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads Experiment Overview The design of this experiment was a 3 (ad types) x 2 (risk groups) between-subjects experiment. The three ad types (OPM, MALADAPT, and CONTROL) represented the three experimental conditions. The two risk groups were determined on the basis of their answers to a particular questionnaire item.1 This experimental design resulted in a total of six cells, two of which were CONTROL group cells. Subjects were asked to view one ad, and then were asked to fill out a questionnaire. The entire procedure took approximately 15 minutes. Analysis of the data made use of ANOVAs. Targeted behavior The targeted behavior for this study was driving under the influence (DUI). DUI was chosen as the behavior for several reasons: (1) drinking and driving is considered to be a relatively common problem among college aged youth; (2) there was likely to be a reasonably high incidence of drinking and driving in the sample, so that it would not be too difficult to fill the high-risk behavior cells; (3) there are well-known maladaptive coping responses that are associated with drinking and driving; (4) there is a large literature on drinking and driving; and (5) there is a large body of advertising that has been developed to combat drinking and driving. 1 Determining DUI status on the basis of a single questionnaire item proved to be somewhat problematic, as will be shown later on. The questionnaire item asked about DUI behavior during the past month, which meant that those who hadn't engaged in DUI during the last month were classified as DUI-NO (even if it was their usual practise to engage in DUI). Furthermore, anyone who had engaged in DUI during the past month was classified as DUI-YES (even if it was their first time, or an unusual occurrence for them). These are two situations where individuals may have been inappropriately classified into DUI categories. As well, the criterion of having engaged in DUI during the past month proved to be a fairly stringent one, which severely limited the sample size. This problem was addressed in the follow-up study in Chapter Seven, where a multi-item indicator of DUI behavior was used. 166 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads Social Marketing Advertising Stimuli The social marketing advertising stimuli that were used for this experiment consisted of print ads against drinking and driving. Print ads were chosen (rather than television or radio ads) for several reasons: (1) print ads are much less costly to create than television or radio ads; (2) print ads are readily adaptable to each of the experimental conditions; (3) much of the previous work on fear appeals has dealt with printed materials, so the use of print ads makes the results more readily comparable with previous work. Two different versions of the DUI ad were developed, corresponding with the OPM and MALADAPT conditions. As well, a CONTROL group ad version was produced which dealt with AIDS prevention. These ads were loosely based on other DUI and AIDS prevention ads which were available. The ads were developed through discussion with colleagues. They were pretested as part of the questionnaire, but were not pretested separately prior to the questionnaire pretest. The OPM ad followed the tenets of the Ordered Protection Motivation model, and included the following four prescribed elements: (1) information about the threatening consequences of the targeted behavior (i.e., fear-arousing information); (2) advice about the recommended coping behavior (i.e., what to do to avoid the threatened consequences); (3) reassurance that the recommended coping behavior would be effective (i.e., efficacy message); and (4) reassurance that the message recipient could carry out the coping behavior effectively (i.e., self efficacy message). The OPM ad is shown in Figure 5.1. 167 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads DRINKING & DRIVING KILLS If you drink and drive, you put yourself at risk. Risk of losing your license or getting a jail sentence. But it's not just yourself you put at risk... there are passengers, other drivers, and pedestrians to consider. Think about how you would feel if you killed someone... So avoid drinking and driving. Take a bus or a taxi instead. Or choose a designated driver. It's that easy. Do your part. You can beat the drinking and driving problem. Figure 5.1 OPM Ad 168 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads The MALADAPT ad was designed to be similar in appearance and content to the OPM ad, in order to ensure equivalency of the two experimental conditions. However, the threat information was replaced with counterarguments against engaging in maladaptive coping responses. All other elements of the ad remained the same. The MALADAPT ad is shown in Figure 5.2. 169 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads DRINKING & DRIVING If you drink and drive, you probably have some strategies that reduce the risk - you hope... Like driving down side streets. Or driving more slowly. The risks are small, you think. But is the risk really so small? And is it worth the risk at all? Maybe not... So avoid drinking and driving. Take a bus or a taxi instead. Or choose a designated driver. It's that easy. Do your part. You can beat the drinking and driving problem. Figure 5.2 MALADAPT Ad 170 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads Table 5.1 gives a brief summary of the contents of these two ads. Table 5.1 Ad Copy Design Ad Elements "OPM" Ad "MALADAPT1' Ad MANIPULATION: Fear appeal vs. counter-arguing against maladaptive behavior Information about the threatening consequences of the targeted behavior Counter-arguments against maladaptive coping responses COPING RESPONSE Advice about the recommended coping behavior (i.e., what to do to avoid the threatened consequences) EFFICACY OF COPING RESPONSE Reassurance that the recommended coping behavior will be effective SELF EFFICACY REASSURANCE Reassurance that the message recipient can carry out the coping behavior effectively The CONTROL group ad was designed to appear similar to the OPM and MALADAPT ads; however, the CONTROL group ad concerns a behavior which is unrelated to DUI. For the purpose of this experiment, the CONTROL group ad was an ad for AIDS prevention. This ad had the same graphic as the DUI ads, and appeared to be quite similar. The CONTROL group ad is shown in Figure 5.3. 171 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads AIDS If you engage in unprotected sex with new partners, you probably have some strategies that reduce the risk - you hope... Like choosing partners who seem "nice" or "clean." Or dating someone a few times before engaging in sex. The risks are small, you think. But is the risk really so small? And is it worth the risk at all? Maybe not... So avoid unprotected sex. Always use a latex condom. It's that easy. Do your part. You can beat the AIDS problem. Figure 5.3 CONTROL Ad 172 Chapter Five The OPM Model and Market Segments: An Exploratory Study Using DUI Print Ads Treatment Groups The three ads (OPM, MALADAPT, and CONTROL) represent the three experimental conditions: 1. "OPM" ad for DUI (threat, coping advice, self efficacy reassurance) 2. "MALADAPT" ad for DUI (counterarguments against maladaptive coping responses) 3. CONTROL Group - ad for AIDS prevention The experimental design, including the number of participants in each experimental cell, is shown in Table 5.2. Table 5.2 Experimental Design Targeted Behavior OPM Ad MALADAPT Ad CONTROL Group Total High-risk behavioral group (DUI-YES) 10 8 5 23 Low-risk behavioral group (DUI-NO> 42 42 31 115 Total: 52 50 36 138 Whether or not an individual was deemed to be a member of the DUI group depended upon their answer to Question #4 on the data c