The Relationship Between Advertising-Induced Anger and Self-Efficacy On

Home , Activism

The Relationship Between Advertising-Induced Anger and Self-efﬁcacy on Persuasive Outcomes A Test of the Anger Activism Model Using the Truth Campaign Vinu Ilakkuvan, MSPH; Monique Mitchell Turner, PhD; Jennifer Cantrell, DrPH, MPA; Elizabeth Hair, PhD; Donna Vallone, PhD, MPH

Turner’s Anger Activism Model (AAM) contends anger and efficacy interact in a unique way to determine message responses to campaign materials. This study tested the AAM using responses to 2 truth antismoking advertisements collected in August-October 2014 via an online, cross-sectional survey of 15- to 21-year-olds. Those aware of each of the truth advertisements (n = 319 for each) were organized into 4 anger/efficacy groups. Analysis of variance and regressions were conducted to understand group differences in message-related cognitions (persuasiveness, receptivity, conversation). Message cognitions were highest among the high anger/high efficacy group and lowest among the low anger/low efficacy group. Key words: Anger Activism Model, communication, emotion, theory, tobacco

T is considered obvious, perhaps, that the ob- stronger attitudes among the overall target audi- I jective of a public health media campaign is to ence. Effects of this ilk can happen when public modify the at-risk audience’s behaviors. In the to- health professionals leverage these strong attitudes bacco control domain, preventing tobacco use is a to develop the aligned audience into vehicles of be- common goal of media campaigns, as it was with havior change among the broader target audience the original truth campaign in the early 2000s. through their social influence. In other words, the The original version of the truth campaign targeted aligned audience can become influencers. youth who were open to smoking and/or sensation This precise the aim of the current iteration of the seeking in an attempt to prevent them from initiat- truth campaign, entitled Finish It, launched in Au- ing smoking cigarettes.1 gust 2014.3 Teen smoking rates have declined sig- Yet, another audience vital to the success of pub- nificantly since the early 2000s, with only 7%of lic health campaigns is the aligned audience—those teenagers in grades 8, 10, and 12 saying that they individuals whose attitudes and behaviors already smoked a cigarette in the past 30 days in 2015.4 align with the public health or antitobacco mes- Despite this decline, tobacco use remains the lead- sage at hand (ie, “preaching to the choir”). Such ing cause of preventable death in the United States,5 audiences may require attitudinal or behavioral and nearly 90% of cigarette smokers began smok- maintenance,2 but, perhaps more importantly, cam- ing before the age of 18 years.6 Thus, addressing paigns targeting these audiences can help develop the issue of tobacco use among teenagers remains a critical public health issue. Given that the over- Author Afﬁliations: Evaluation Science and Research, Truth whelming majority of teens—91%—do not use to- Initiative, Washington, District of Columbia (Ms Ilakkuvan and Drs bacco products,7 the Finish It campaign seeks to Cantrell, Hair, and Vallone); Department of Prevention and Com- build on this momentum and rally the current gener- munity Health, The George Washington University Milken Insti- tute School of Public Health, Washington, District of Columbia ation of young people to be the generation that ends 3 (Ms Ilakkuvan and Dr Turner); Department of Health, Behavior, and youth smoking. However, there are few, if any, be- Society, Johns Hopkins Bloomberg School of Public Health, Balti- havioral or communication theories that specifically more, Maryland (Drs Cantrell and Hair); and Global Institute of Pub- deal with (a) aligned audiences or (b) how to com- lic Health, New York University, New York (Dr Vallone). pel them into becoming behavior change agents, or Data collection was funded internally by Truth Initiative. There are activists. One novel, and fairly new, theory articu- no other funding sources for this work. lating the process by which aligned audiences can The authors have no conﬂicts of interest to report. be leveraged to take action for a particular cause is 8 Correspondence: Vinu Ilakkuvan, MSPH, Department of Preven- the Anger Activism Model (AAM). tion and Community Health, The George Washington University Milken Institute School of Public Health, 950 New Hampshire Ave ANGER ACTIVISM MODEL NW, Washington, DC 20037 ([email protected]). It has long been understood that emotions have a Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved. robust impact on attitudes, decisions, judgments, DOI: 10.1097/FCH.0000000000000126 and behaviors.9-11 Although most studies regarding

72 Family and Community Health January–March 2017 ■ Volume 40 ■ Number 1

Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. V. Ilakkuvan et al. Relationship Between Advertising-Induced Anger and Self-efﬁcacy on Persuasive Outcomes 73

the role of emotion in health messaging have fo- ticipants had more negative thoughts about the mes- cused on the effects of fear, guilt, or humor, there sage (which had logical flaws embedded in it) and is reason to believe that anger can have beneficial were less supportive of message recommendations effects.8,11 To some degree, this is a surprising state- than the fear group. Notably, Nabi did not assess ment given that several studies reveal the deleteri- the effects of the intensity of the angry feelings expe- ous effects of anger, such as misjudging individuals rienced by participants. Yet, Frijda25 proposed that or events.8,12 In fact, Turner developed the AAM to action tendencies depend on an emotion’s intensity. predict when anger can lead to constructive versus The AAM posits that the intensity of the angry feel- deleterious effects on cognitions.8 Here, we focus on ings is a critical factor in determining cognitive and constructive outcomes. behavioral outcomes.8 Like health messaging studies broadly, research Whether feelings of anger lead to construc- on youth tobacco prevention campaigns has also tive outcomes (eg, systematic processing) or non- focused largely on the effects of fear, disgust, sad- constructive ones (eg, misjudging individuals or ness, and humor.13 While there is evidence that fear- events) is dependent upon the intensity of the inducing antismoking messages featuring health anger experienced.8 The AAM hypothesizes differ- consequences are effective among youth, much of ent cognitive and behavioral outcomes dependent the evidence for the effectiveness of fear appeals in upon how angry message recipients report being; antitobacco campaigns comes from campaigns fo- as Loewenstein and Lerner stated, “at sufficient cused on adult cessation, not youth prevention.14 levels of intensity, emotion can overwhelm cogni- There are many emotions beyond fear that can be tive processing and deliberative decision making evoked by health messages, and anger, in particular, altogether.”26(p627) is underexplored in the literature. The AAM pro- Turner argued that despite its potentially deleteri- vides a unique framework with which to understand ous outcomes, even intensely felt anger can lead to the impact of a tobacco prevention campaign such constructive results.8 Given that we are interested as truth. in persuasive situations, it would be a misstep to The AAM is an emerging theory that rests on sev- fail to recognize the role of self-efficacy, one’s be- eral key assumptions.8 First, anger is triggered when lief in his or her ability to accomplish a particu- one’s goals,15 particularly goals related to the preser- lar task, in affecting behavioral intentions.27 Peo- vation of ego, are threatened.16 Experiencing anger ple with a high degree of perceived self-efficacy are communicates that something in the environment able to translate knowledge into behavior.27 Hence, is wrong,17-19 motivating individuals to remove ob- the perceived ability to control a situation may be stacles blocking goal attainment.15 So, anger func- what distinguishes constructive from nonconstruc- tions to motivate people to regain or maintain con- tive anger. trol of a threatening situation.20 Second, anger can Turner argued that, taking an audience segmen- be caused by a campaign message or communica- tation perspective, we can use anger and efficacy to tion in general. Third, the effects of anger on indi- form 4 distinct audience segments that will uniquely viduals’ cognitions in response to a message depend respond to the anger-inducing campaign message.8 on 3 key moderating variables: (a) the intensity of The first group, activists, experience strong feelings the anger experienced (ie, communication can cause of anger and strong efficacy beliefs. The activist angry feelings ranging from mad to rage), (b) self- group will have the most positive attitudes regard- efficacy or perceptions of being able to control the ing the topic, will be most willing to engage in angering stimulus, and (c) whether the target au- higher commitment behaviors, and will engage in dience is aligned or not aligned with regard to the higher levels of systematic processing. The second angering stimulus (anger appeals are most effective group, empowered, experiences low levels of anger with aligned audiences).8 The model has thus far and strong perceptions of efficacy. Empowered been tested only with older adolescent populations.8 people feel that something can be done to “fix” the issue, but they do not perceive the issue to be anger- ANGER EFFECTS, ANGER INTENSITY, AND ing (ie, the topic is not of high importance). Thus, SELF-EFFICACY they will have positive attitudes but will only be Lazarus and Lazarus15 proposed that the coping re- willing to engage in midlevel behaviors. The third sponse of angry people is to remove barriers block- group, angry, is distinguished by strong feelings ing goal attainment, thereby motivating people to of anger and weak feelings of efficacy. Thus, these regain control of a threatening situation. Anger also people are angry about the current state of affairs, directs attention, memory, and judgments; in fact, it but they do not perceive that anything can be done. stimulates analytical thinking processes in threaten- Their anger is nonutilitarian and, therefore, they ing environments.21-23 Nabi24 found that angry par- will be the unlikely to intend to engage in high

Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. 74 Family and Community Health January–March 2017 ■ Volume 40 ■ Number 1 commitment behaviors. Finally, the disinterested study has been approved by the Human Subjects group is marked by weak feelings of anger and Review Committee of Chesapeake Institutional Re- weak perceptions of efficacy. These individuals will view Board. The online survey is fielded continu- have positive attitudes about the topic (given it is ously at a rate of approximately 140 participants a topic they feel aligned with), but these attitudes per week, and quotas are maintained by gender, will not translate into positively responding to a age, and race/ethnicity using US Census informa- message. Disinterested people will engage in the tion. Data used in this article include responses col- least amount of cognitive processing because they lected while the 2 advertisements used in this study do not care about changing the state of affairs. were airing on television, from mid-August to the The AAM posits that for all message-related out- end of October in 2014 (n = 1258). comes, the relative ordering of the 4 AAM groups The questionnaire includes items about youths’ is from most to least positive outcomes: activist, and young adults’ tobacco-related knowledge, atti- empowered, angry, and disinterested. tudes, beliefs, and behaviors, as well as demographic information, media use, and exposure and reaction TESTING THE AAM USING TRUTH to antitobacco advertising, particularly truth cam- CAMPAIGN ADVERTISEMENTS paign advertisements. In this study, secondary data analysis was conducted to test the AAM based on advertising- Measures induced anger and efficacy and their relationship to For each advertisement, all those whose responses key message-related cognitions, specifically message indicated that they were aware of the advertise- persuasiveness, message receptivity, and the level of ment were asked a series of items to determine their conversation prompted by the message. The specific message-related cognitions. Awareness of the adver- advertisements examined were 2 advertisements tisement included those answering yes to text-aided that are part of the truth youth antismoking brand’s awareness (in which individuals were given a text current campaign, entitled Finish It, launched in description of the advertisement and asked whether August 2014.3 The campaign targets 15- to 21-year- they had seen it in the last 6 months) and/or visu- olds with a message to be the generation that ends ally aided awareness (in which individuals were pro- youth smoking. The first advertisement, entitled vided screenshots from the advertisement and asked Finishers, calls on this generation of young people whether they had seen it in the last 6 months). to be the generation that ends youth smoking. The second advertisement, entitled Unpaid Spokesper- Aligned audience son, calls out celebrity smokers as being unpaid Because the AAM is built upon the underlying as- spokespeople for the tobacco industry given that sumption that the audience is aligned, the sample images of these celebrities smoking are widespread.3 used in this study was narrowed to only those who Both advertisements aired on television and online were aligned in relation to the Finish It campaign, from August through October 2014. as based on a scale of 3 items. Each item within the scale had a 5-point response scale from strongly disagree to strongly agree, and scores were averaged METHODS across items to create the scale. The 3 items were Data collection “I want my generation to be known as the one that The survey used in this study is the Legacy Me- ends smoking,” “I would be part of a movement to dia Monitoring Tracking Survey, which is adminis- end smoking,” and “Taking a stand against smok- tered online utilizing Research Now’s online panel. ing is important to me (α = .832). Those scoring The survey is fielded by Crux Research, Inc. for a below a 3 (ie, the midpoint) on the alignment scale truth Initiative study of the truth youth and young were excluded from the sample, further narrowing adult antismoking campaign. Panelists have been re- the sample to 808 individuals. cruited by Research Now through e-mail and online marketing with more than 300 diverse online affili- Creating groups in accordance with the AAM ate partners and targeted Web site advertising.28 Of these 808 respondents who were surveyed while To be eligible for this survey, panelists must be the Finishers and Unpaid advertisements were on within the target age range of the truth campaign, air, 319 were aware of the Finishers advertisement 15 to 21 years old. Before the survey is adminis- and 319 were aware of the Unpaid advertisement. tered, potential respondents are sent an e-mail invi- However, these were not the same 319 individu- tation with a link to begin the survey and informed als; only 209 respondents were aware of both ad- consent information. Participation is voluntary, and vertisements. Advertisement-induced anger was de- participants may end the survey at any time. This termined by agreement with the statement “Seeing

this ad made me feel angry” and categorized into erage score less than 4 were categorized as not re- 2 groups in accordance with the AAM (strongly ceptive). or somewhat agree was categorized as high anger; neither agree nor disagree, somewhat disagree, and Conversation. Conversation about the advertise- strongly disagree were categorized as low anger). ment was assessed by individuals’ agreement with Advertisement-induced efficacy was determined by the statement “I have talked to my friends about this agreement with the statement “Seeing this ad made ad.” On a 5-point scale from strongly disagree to me feel powerful” and categorized into 2 groups in strongly agree, the mean conversation score for the accordance with the AAM (strongly or somewhat Finishers advertisement was 3.10 (SD = 1.32) and agree was categorized as high efficacy; neither agree for the Unpaid advertisement was 3.02 (SD = 1.28). nor disagree, somewhat disagree, and strongly dis- This measure was dichotomized in regression anal- agree were categorized as low efficacy). For each yses, such that those who strongly or somewhat of the 2 advertisements, the 4 resultant groups— agreed with this statement were categorized as hav- (1) low anger, low efficacy (disinterested); (2) high ing conversed about the advertisement and all oth- anger, low efficacy (angry); (3) low anger, high effi- ers were categorized as not having conversed. cacy (empowered); and (4) high anger, high efficacy (activist)—were compared across 3 key outcomes. Data analysis For each advertisement, the 4 anger/efficacy groups Outcome measures were compared across outcomes using weighted Key outcome measures included 3 message-related 1-way analyses of variance. Then, weighted lo- cognitions: persuasiveness, receptivity, and conver- gistic regression analyses were used to assess the sation. odds of having message-related cognitions based on anger/efficacy group, adjusting for gender, age, Persuasiveness. Whether or not individuals were race/ethnicity, parental education, perceived finan- persuaded by an advertisement was assessed by cial situation, and smoking status. These analyses their agreement with the statement “This ad made used sampling weights to adjust for nonresponse me want to do something to end smoking,” since and sample design and were weighted to be nation- the key message of the overarching campaign that ally representative. encompasses both these advertisements is to acti- vate young people to be the generation to end smok- RESULTS ing. On a 5-point scale from strongly disagree to The demographic characteristics of the sample by strongly agree, the mean score for persuasiveness for advertisement and overall are displayed in Table 1. the Finishers advertisement was 3.91 (SD = 1.00) There were more females than males by advertise- and for the Unpaid advertisement was 3.76 (SD = ment and overall, and average age across all groups 1.07). This outcome measure was dichotomized in was approximately 18 years. Racial/ethnic break- regression analyses, such that those who strongly or downs were similar between those aware of each somewhat agreed with this statement were catego- advertisement, with a little less than half of the sam- rized as being persuaded by the advertisement, and ple being white, 18 or 19% African American, and all others were categorized as not being persuaded 20 or 23% Hispanic. In contrast, the racial/ethnic by the advertisement. breakdown of the overall sample was about 54% white and 16% African American. In the total sam- Receptivity. Receptivity to the advertisement was ple, 13% reported a parental education of less than determined on the basis of a scale of items: “This ad high school, 20% reported parental education at was worth remembering,” “This ad grabbed my at- the high school level, 30% reported a parental ed- tention,” “This ad was powerful,” and “This ad was ucation of some college/associate’s, and 36% re- convincing” (α = .908 for the Finishers advertise- ported a parental education of college or higher.The ment; α = .926 for the Unpaid advertisement). Each parental education breakdown was similar for those item within the scale had a 5-point response scale aware of the Finishers advertisement, but for those from strongly disagree to strongly agree, and scores aware of the Unpaid advertisement, there were more were averaged across items to create the scale. For individuals reporting a parental education of high the Finishers advertisement, the scale mean score school (about 24%) and fewer individuals report- was 4.03 (SD = 0.85) and for the Unpaid adver- ing a parental education of college or higher (about tisement, it was 3.91 (SD = 0.93). Receptivity was 34%). Across groups, financial situation skewed to dichotomized in regression analyses (those with an the higher end, with only about 11% or 12% say- average score of 4 or higher were categorized as re- ing that they do not meet basic expenses, 27% to ceptive to the advertisement and those with an av- 30% saying that they just meet basic expenses, 33%

TABLE 1. Unweighted Population Demographics Aware of Finishers Aware of Unpaid Advertisements Advertisements Demographic Characteristic (N = 319) (N = 319) Total (N = 808) Female (%) 57.68 58.62 57.05 Mean age (mean [SE]) 17.92 (0.11) 17.96 (0.11) 17.92 (0.07) Race/Ethnicity, % White 48.90 47.65 53.84 Black/African American 18.18 19.75 16.34 Hispanic 20.06 23.20 19.06 Other 12.85 9.40 10.77 Parental education, %

saying that they meet needs with little left, and 25% the Finishers advertisement and 38.24% conversed to 28% saying that they live comfortably. Approx- about the Unpaid advertisement. Because of some imately 15% of the overall sample was composed anger/efficacy groups being small in size and given of current smokers, in comparison to 19% of those high correlations between the 3 key outcomes and aware of the Finishers advertisement and 21% of anger/efficacy groups, some cell sizes were quite those aware of the Unpaid advertisement. small (eg only 5 individuals in the activist group for the Finishers advertisement found the advertisement Anger/efﬁcacy groups and group unpersuasive). differences by outcome As shown in Tables 2, across both advertise- To begin testing the model, those aware of the Fin- ments, there were significant differences among the ishers advertisement were split into anger/efficacy anger/efficacy groups for each of 3 outcomes (per- groups in accordance with the AAM. The groups suaded by advertisement, receptive to advertise- in order from largest to smallest (in terms of ment, and conversed about advertisement) as de- the percentage of individuals aware of the ad- termined by a weighted 1-way analysis of variance vertisement who fall into each group) were em- (P < .001 for each outcome for each advertise- powered (49.84%), disinterested (29.15%), activist ment). Across both advertisements for all 3 out- (13.79%), and angry (7.21%). For the Unpaid comes, the highest scoring group was the activist advertisement, groups in order from largest to group and the lowest scoring group was the disinter- smallest were disinterested (36.36%), empowered ested group (with 1 exception—the angry group had (27.59%), angry (18.81%), and activist (17.24%). a lower score than the disinterested group for mean Among the sample used in this analysis, 66.77% of conversing about the Unpaid advertisement). For of those asked about the advertisement found the each outcome for both advertisements, all high ef- Finishers advertisement persuasive and 59.87% ficacy groups scored higher than all low efficacy found the Unpaid advertisement persuasive. In addi- groups (with 1 exception—high anger/low efficacy tion, 65.52% were receptive to the Finishers adver- (angry) and low anger/high efficacy (empowered) tisement and 60.50% were receptive to the Unpaid groups had approximately equal means for convers- advertisement. Finally, 40.13% conversed about ing about the Finishers advertisement).

TABLE 2. Weighted Analysis of Variance: Message-Related Cognitions by Anger/Efficacy Groupa Disinterested Angry (High Empowered Activist (High One-Way (Low Anger, Anger, Low (Low Anger, Anger, High ANOVA F Low Efficacy) Efficacy) High Efficacy) Efficacy) statistic (P) Finishers advertisement (N = 319) Group size Weighted% 29.15% (93) 7.21% (23) 49.84% (159) 13.79% (44) (unweighted n)

Persuaded by advertisement < Mean (SD) 3.21 (0.96)a 3.98 (0.99)b 4.18 (0.90)b 4.27 (0.83)b 24.81 (P .001) Receptive to advertisement < Mean (SD) 3.41 (0.87)a 3.69 (0.96)a 4.3 (0.66)b 4.39 (0.76)b 32.55 (P .001) Conversed about advertisement < Mean (SD) 2.69 (1.13)a 3.12 (1.16)a,b,c 3.11 (1.35)a,b 3.92 (1.28)c 9.37 (P .001) Unpaid advertisement (N = 319) Group size Weighted% 36.63% (116) 18.81% (60) 27.59% (88) 17.24% (55) (unweighted n) Persuaded by advertisement < Mean (SD) 3.13 (0.86)a 3.66 (1.09)b 4.29 (0.86)c 4.39 (0.88)c 37.52 (P .001) Receptive to advertisement < Mean (SD) 3.42 (0.87)a 3.59 (1.00)a 4.42 (0.63)b 4.43 (0.73)b 35.32 (P .001) Conversed about advertisement < Mean (SD) 2.83 (1.04)a,c 2.51 (1.15)a,b 3.30 (1.40)c 3.67 (1.33)c 11.00 (P .001) Abbreviation: ANOVA, analysis of variance. aStatistical differences, based on a Scheffe’s test, are denoted by different subscripts. Groups sharing any of the same subscripts are not statistically different from each other.

Adjusted odds of reacting positively to group of conversing about the Unpaid advertise- advertisement across anger/efﬁcacy groups ment). To better understand the differences in outcomes Examining message-related cognitions for the across anger/efficacy groups, weighted and adjusted Finishers advertisement, for the high anger/high logistic regression analyses were conducted, con- efficacy (activist) group compared with the low trolling for gender, age, race/ethnicity, parental ed- anger/low efficacy (disinterested) group, odds of ucation, perceived financial situation, and smok- being persuaded by the advertisement were 19.97 ing status. Results are shown in Table 3. For both (95% confidence interval [CI], 6.51-61.23) times advertisements across each of the 3 outcomes, in higher, odds of being receptive to the advertisement comparison to the low anger/low efficacy (disin- were 23.82 (95% CI, 8.31-68.27) times higher, and terested) group, most other anger/efficacy groups odds of conversing about the advertisement were had significantly higher odds of having message- 8.60 (95% CI, 3.39-21.82) times higher.For the Un- related cognitions, with only 2 exceptions (the high paid advertisement, for the high anger/high efficacy anger/low efficacy [angry] group did not have sig- (activist) group compared with the low anger/low nificantly higher odds than the low anger/low effi- efficacy (disinterested) group, odds of being per- cacy [disinterested] group of being receptive to the suaded by the advertisement were 27.07 (95% CI, Finishers advertisement, and the high anger/low effi- 10.17-72.03) times higher, odds of being receptive cacy [angry] group did not have significantly higher to the advertisement were 16.87 times higher (95% odds than the low anger/low efficacy [disinterested] CI, 6.26-45.48), and odds of conversing about

TABLE 3. Regression Models: Adjusted Odds of Message-Related Cognitions by Anger/Efficacy Group Persuaded by Receptive to Conversed About Advertisement Advertisement Advertisement Adjusteda Adjusteda Adjusteda OR (95% CI) OR (95% CI) OR (95% CI) Finishers advertisement (N = 313) Disinterested Referent Referent Referent Angry 5.22 (1.69-16.10) 2.32 (0.84-6.39) 1.96 (0.71-5.41) Empowered 8.73 (4.42-17.27) 11.31 (5.81-22.02) 2.57 (1.30-5.08) Activist 19.97 (6.51-61.23) 23.82 (8.31-68.27) 8.60 (3.39-21.82) Unpaid advertisement (N = 312) Disinterested Referent Referent Referent Angry 4.65 (2.19-9.84) 2.03 (1.01-4.04) 0.88 (0.39-2.00) Empowered 16.30 (7.43-35.77) 17.07 (7.59-38.38) 3.69 (1.90-7.20) Activist 27.07 (10.17-72.03) 16.87 (6.26-45.48) 4.92 (2.26-10.73) Abbreviations: CI, confidence interval; OR, odds ratio. aWeighted regression models control for gender, age, race/ethnicity, parental education, perceived financial situation, and smoking status. the advertisement were 4.92 (95% CI, 2.26-10.73) smoking, a call to action that may be particularly times higher. influenced by how angry and efficacious the audience feels, in contrast to the Unpaid advertisement, CONCLUSIONS which focused on highlighting the negative impact of images of celebrities smoking tobacco. However, Summary the odds of activist group in comparison to the dis- It is evident from the results that anger/efficacy interested group being persuaded by the advertise- group status impacts message-related cognitions. ment are higher for the Unpaid advertisement than As predicted by the AAM, these message-related for the Finishers advertisement. cognitions were generally highest among the high In general, across both advertisements, high ef- anger/high efficacy (activist) group and lowest ficacy groups had higher odds of message-related among the low anger/low efficacy (disinterested) cognitions than low efficacy groups. Similarly, group. These differences were substantial, with within each efficacy strata, high anger groups often odds ratios of expressing message-related cogni- had higher odds of message-related cognitions than tions ranging from 4.92 (95% CI, 2.26-10.73) times those with lower anger, although this difference was higher to 27.07 (95% CI, 10.17-72.03) times higher less pronounced for the Unpaid advertisement than for the activist versus the disinterested group, de- for the Finishers advertisement. pending on the exact advertisement and outcome. The difference between groups varied on the basis of the advertisement and outcome, which suggests Limitations that the type of messaging found in advertisements Limitations of this study include the use of an opt- can alter how strongly anger and efficacy impact in online sample of youth and young adults, which message-related cognitions. The odds of the activist may be subject to biases resulting from undercov- group in comparison to the disinterested group are erage and nonresponse, as well as the use of a higher across both advertisements for being per- cross-sectional sample. Following individuals longi- suaded by or receptive to the advertisement than tudinally would provide a better understanding of for conversing about the advertisement. In addition, the effect of advertisement-induced anger and ef- the odds of the activist group in comparison to the ficacy on response to the advertisement over time. disinterested group being receptive to or conversing In addition, this would also allow for actual ac- about the advertisement are higher for the Finishers tions taken in response to the message to be as- advertisement than for the Unpaid advertisement. sessed, as opposed to relying solely on individ- This may be due to the fact that the Finishers ad- uals’ self-reported perceptions of whether or not vertisement focused explicitly on activism and mo- the advertisement made them want to do some- tivating the audience to be the generation that ends thing. The sample, while weighted to be nationally

representative, is a nonprobability sample and thus Implications for advancing theory in health may not be generalizable. In addition, dividing the promotion and community health anger/efficacy groups on the basis of their reac- This study tests an emerging theory in health pro- tion to the advertisement led to small cell sizes, so motion, specifically health communication, shed- some estimates may be unstable. Despite this lim- ding light on the role of anger in persuading itation, statistically significant differences between audiences to change health-related attitudes and the groups were demonstrated, but further analy- behaviors. Results suggest that health promotion ses with a larger sample could help make estimates practitioners should explore the use of anger and more precise and stable. other emotions to relay their messages and that re- searchers should measure anger and other emotions Next steps in response to health messages in order to test the While this analysis examined advertisement- AAM and other such communication theories fo- induced anger and efficacy, further analyses could cused on emotion. This research in turn could bet- examine individuals’ anger and efficacy in relation ter inform practitioners about which emotions in to the topic of the advertisement more broadly. which contexts are most powerful in persuading au- In addition, while this analysis examines the 4 diences to change health-related attitudes and be- anger/efficacy groups as the key independent haviors. variable, future analyses could examine anger and efficacy as separate independent variables, REFERENCES providing a more nuanced understanding of how 1. Allen JA, Vallone DM, Vargyas E, Healton CG. The truth® these 2 factors, both independently and together, campaign: using countermarketing to reduce youth smok- impact message-related cognitions. Testing the ing. In: Healey B, Zimmerman R, eds. The New World of Health Promotion. New Program Development, Implemen- anger/activism model using public health and tation and Evaluation. Sudbury, MA: Jones and Bartlett; other messages on different topics with a variety 2009:195-215. of target audiences could also help determine the 2. Rothman AJ. Toward a theory-based analysis of be- degree to which the model holds in a variety of havioral maintenance. Health Psychol. 2000;19(1):6. doi:10.1037//0278-6133.19.Suppl1.64. contexts. 3. Legacy. Proven-Effective Youth Smoking Prevention Campaign Launches New Chapter, Challenges Teens Implications to—“Finish It” [press release]. http://legacyforhealth.org/ Understanding how to motivate individuals to act newsroom/press-releases/the-end-of-tobacco-truth- is critical to public health communication. Public r-rallies-teens-to-harness-their-online-influence-to- become-the-first-smoke-free-generation. Published health mass media campaigns, such as truth,areof- August 25, 2014. Accessed August 2015. ten attempting to motivate specific behaviors and 4. Miech RA, Johnston LD, O’Malley PM, Bachman JG, actions.29 The results of this analysis indicate that Schulenberg JE. National Press Release, “Teen Cigarette messages that incite high anger and high efficacy in Smoking Drops to Historic Low in 2015.” Ann Arbor, MI: University of Michigan News Service. Published December combination can have powerful effects. While the 16, 2015. http://monitoringthefuture.org/pressreleases/ high anger/high efficacy (activist) group in this anal- 15cigpr_complete.pdf. Accessed May 2016. ysis was smaller than the low anger/low efficacy 5. Centers for Disease Control and Prevention. Vital (disinterested) and low anger/high efficacy (empow- signs: current cigarette smoking among adults aged > ered) groups, this group experienced notably higher 18 years—United States, 2005-2010. MMWR Morb Mortal Wkly Rep. 2011;60(35):1207-1212. odds of being receptive to the message, finding the 6. U.S. Department of Health & Human Services. The Health message persuasive, and speaking to others about Consequences of Smoking—50 Years of Progress: A Re- the message. This suggests that while the activist port of the Surgeon General. Atlanta, GA: U.S. Department group may be small, they can be a critical group in of Health & Human Services, Centers for Disease Con- trol and Prevention, National Center for Chronic Disease responding to a message and activating others to do Prevention and Health Promotion, Office on Smoking and so as well. Health; 2014. In addition, these results suggest that antito- 7. Johnston LD, O’Malley PM, Bachman JG, Schulenberg bacco campaigns should seek to explore using anger JE, Miech RA. Monitoring the Future National Survey Re- and other emotions to relay their message. While sults on Drug Use, 1975–2013: Overview, Key Findings on Adolescent Drug Use. Ann Arbor, MI: Institute for Social fear, sadness, and disgust aroused by messages of Research, University of Michigan; 2014. health consequences have been shown to be ef- 8. Turner MM. Using emotion in risk communication: The fective, this study suggests that a wider range of Anger Activism Model. Public Relat Rev. 2007;33(2):114- emotions should be investigated to assess the va- 119. doi:10.1016/j.pubrev.2006.11.013. 9. Mitchell M. Motivated, but not able? The effects of riety of roles such emotions might play in elicit- positive and negative mood on persuasive message ing the desired cognitive, emotional, and behavioral processing. Commun Monogr. 2000;67:215-225. responses. doi:10.1080/03637750009376505.

10. Turner MM. Emotional branding: what, when and why. In: Families. New York, NY: WW Norton & Company; Evans WD, ed. The Psychology of Branding. Hauppauge, 1991. NY: Nova Science Publishers; 2013:15-37. 20. Pfau M, Szabo A, Anderson J, Morrill J, Zubric J, 11. Turner MM. Discrete emotions and the design and evalua- H-Wan HH. The role and impact of affect in the tion of health communication messages. In: Cho H, ed. De- process of resistance to persuasion. Hum Commun signing Messages for Health Communication Campaigns: Res. 2001;27(2):216-252. doi:10.1111/j.1468-2958.2001 Theory and Practice. Thousand Oaks, CA: Sage; 2011: .tb00781.x. 59-71. 21. Izard CE. The Psychology of Emotions.NewYork,NY: 12. Lazarus RS. Emotion and Adaptation.NewYork,NY:Ox- Basil Blackwell; 1991. ford University Press; 1991. 22. Schwarz N. Feelings as information: informational and 13. Farrelly MC, Niederdeppe J, Yarsevich J. Youth tobacco motivational functions of affective states. In: Higgins ET, prevention mass media campaigns: past, present, and Sorrentino RM, eds. Handbook of Motivation and Cogni- future directions. Tob Control. 2003;12 (suppl 1):i35-i47. tion: Vol. 2. Foundations of Social Behavior.NewYork,NY: doi:10.1136/tc.12.suppl_1.i35. Guilford Press; 1990:527-561. 14. National Cancer Institute. The Role of the Media in 23. Thomas SP. Anger and its manifestations in women. In: Promoting and Reducing Tobacco Use. Bethesda, MD: Thomas SP, ed. Women and Anger: Vol. 15. Focus on U.S. Department of Health & Human Services, Na- Women. New York, NY: Springer; 1993:40-67. tional Institutes of Health, National Cancer Institute; 24. Nabi RL. Anger, fear, uncertainty, and attitudes: a test 2008. Tobacco Control Monograph No. 19. NIH Pub. of the cognitive-functional model. Commun Monogr. No. 07-6242. 2002;69:204-216. 15. Lazarus RS, Lazarus BN. Passion and Reason: Making 25. Frijda NH. The Emotions. Cambridge, England: Cambridge Sense of Our Emotions. New York, NY: Oxford University University Press; 1986. Press; 1996. 26. Loewenstein G, Lerner JS. The role of affect in decision- 16. Reiser C. Reﬂections on Anger: Women and Men in a making. In: Davidson R, Scherer K, Goldsmith H, eds. Changing Society. Westport, CT: Greenwood Publishing Handbook of Affective Science. Oxford, England: Oxford Group; 2001. University Press; 2003:619-642. 17. Denham G, Bultemeier K. Anger: targets and triggers. 27. Bandura A. Self-efﬁcacy mechanism in human agency. Am In: Thomas SP, ed. Women and Anger. New York, NY: Psychol. 1982;37:122-147. Springer; 1993:68-90. 28. Research Now. Recruitment. http://www.researchnow. 18. Lerner HG. The Dance of Anger: A Woman’s Guide to com/en-US/Panels/Recruitment.aspx. Accessed August Changing the Patterns of Intimate Relationships.New 2015. York, NY: Harper & Row; 1985. 29. Hornik RC. Public Health Communication Evidence for Be- 19. Potter-Efron RT, Potter-Efron PS. Anger, Alcoholism, havior Change. Mahwah, NJ: Lawrence Erlbaum Asso- and Addiction: Treating Individuals, Couples, and ciates; 2002.