Are Graphic Cigarette Warning Labels an Effective Message Strategy? A Test of Psychological Reactance Theory and Source Appraisal

Abstract

The purpose of this study is to examine psychological reactance in response to graphic cigarette warning labels and to strengthen and expand on the current literature by using validated measures. Young adults (N = 435) were randomly assigned to a cigarette package featuring a graphic image or a package featuring no image. Utilizing both structural equation modeling and multivariate analyses, the results indicate that graphic warning labels are associated with freedom threat perceptions directly and reactance indirectly. In addition, exposure to graphic cigarette warning labels resulted in higher freedom threat perceptions, negative cognitions, and source domineeringness. Our results are considered with an emphasis on the theoretical and practical implications for policy makers.

Keywords

psychological reactance graphic images tobacco fear appeals

Tobacco smoking is a costly health behavior, both for individuals who smoke and for the broader community who pays for tobacco-related diseases. According to the World Health Organization (WHO; 2013b), smoking is the number one contributor to preventable death in the world. Furthermore, for the United States alone, the financial cost associated with these preventable diseases is estimated at US$193 billion per year (Centers for Disease Control and Prevention [CDC], 2013a). Despite the financial and health costs, 19% of U.S. adults above the age of 18 smoke. This statistic, however, may be conservative, considering it is measured as adults who have smoked more than 100 cigarettes in their lifetime and who report smoking every day or most days (CDC, 2013a). This means that there are likely far more adults who smoke, but who consider themselves occasional smokers or social smokers. To reduce smoking prevalence, many countries have implemented new policies and laws for tobacco control including public smoking bans, increased taxes on tobacco products, and text warnings on cigarette packaging, which are often accompanied by graphic images (for the remainder of the article, whenever the term image is used, what is meant is a graphic image). This last measure has received a great deal of attention in international scholarly research in the past several years and is the focus of the current investigation.

Thus far, 168 countries have ratified the WHO’s Framework Convention on Tobacco Control, which requires large health warnings to be prominently placed on cigarette packaging (WHO, 2013a). Following the lead of other countries, the United States passed a law requiring the display of graphic images and warnings on all cigarette packaging by September 2012 (Pelofsky & Yukhananov, 2011). The U.S. Food and Drug Administration (FDA) selected the images, which include pictures of diseased lungs, a diseased mouth, a person dying of lung cancer, and a corpse, among others (Salahi, 2010). However, even after a federal court ruled that these images reach beyond the bounds of legal government involvement in private commerce, the case was heard by, and upheld by, an appellate court. According to U.S. District Judge Richard J. Leon, these labels are designed to intentionally elicit emotional responses, which goes beyond the government’s role in imparting factual information (R. J. Reynolds Tobacco Company, Lorillard Tobacco Company, Commonwealth Brands, Inc., Liggett Group, LLC, & Santa Fe Natural Tobacco Company, Inc. v. FDA, Hamburg, & Sebelius, 2012). The fate of this law may be tied up in litigation for some time (Pelofsky & Yukhananov, 2011), but in the interim, the tobacco industry and its supporters continue to fight the implementation of the law while antitobacco advocacy groups continue to cite the law’s effectiveness in other nations.

Numerous studies evaluating the results of graphic packaging claim that these images are effective in reducing smoking behaviors (Hammond et al., 2007; Hammond, Fong, McDonald, Brown, & Cameron, 2004; Hammond, Fong, McDonald, Cameron, & Brown, 2003; Hammond, McDonald, Fong, Brown, & Cameron, 2004; O’Hegarty et al., 2006; Peters et al., 2007; Willemsen, 2005). However, some scholars have questioned the legitimacy of these claims because the implementation of these warning labels has, in many cases, coincided with other tobacco control measures such as tax increases and smoking bans, thereby raising questions about the cause of declining smoking rates (Ruiter & Kok, 2005). Furthermore, research that espouses the success of these graphic labels also reports a percentage of the smoking population for which these warnings and images seem to have no effect or negative effects (Hammond et al., 2007; Hammond, Fong, et al., 2004; Hammond et al., 2003; Hammond, McDonald, et al., 2004; O’Hegarty et al., 2006; Peters et al., 2007; Willemsen, 2005). One possible explanation for these varying outcomes may be found in psychological reactance theory (PRT; J. W. Brehm, 1966). PRT predicts that these graphic warning images may help some people but inadvertently harm others by causing reactance and corresponding maladaptive responses.

To isolate the effects of graphic warning images and investigate whether these labels contribute to negative outcomes, Erceg-Hurn and Steed (2011) examined exposure to graphic cigarette warnings and its possible arousal of psychological reactance. Their study found that smokers exposed to graphic warning images were more likely to experience anger than smokers who were exposed to text-only warnings. Although their findings supported the case against graphic tobacco packaging, there were several shortcomings in their study. First, although the study claimed to measure psychological reactance, the researchers failed to assess freedom threat perceptions and negative cognitions. In response to this limitation, the current study will reexamine freedom threat perceptions and psychological reactance following exposure to text warnings and graphic imagery on cigarette packaging. Specifically, this project will examine psychological reactance—as conceptualized by J. W. Brehm (1966) and operationalized by Dillard and Shen (2005) as an amalgamation of anger and negative cognitions—in response to a freedom threat initiated by graphic warning images. Furthermore, this study will examine three understudied endogenous variables associated with reactance: source domineeringness, expertise, and trustworthiness.

PRT

PRT posits that people value freedom, choice, and autonomy. Thus, perceived threats to or the elimination of a freedom may result in attempts to restore that threatened or lost freedom. J. W. Brehm (1966) theorized that a person’s reactance to a freedom threat is based on the strength of the threat and the amount of personal value attributed to that freedom. As reactance increases, people’s drive to restore their threatened or lost freedom subsequently increases, resulting in a variety of potentially negative outcomes. Messages perceived as threatening may be avoided, rejected, or result in boomerang effects, which restore freedom through the engagement in a threatened or proxy behavior (see J. W. Brehm, 1966). In addition, reactance may lead to derogation of the source (e.g., Quick & Bates, 2010; Miller, Lane, Deatrick, Young, & Potts, 2007), which is of particular interest in this study.

In the realm of health communication, PRT has enjoyed a wide variety of applications in recent years. The theory has been studied in the context of binge drinking (Dillard & Shen, 2005; Quick & Bates, 2010), condom use (Quick & Stephenson, 2007), drunk driving (Shen, 2010), exercise (Miller et al., 2007; Quick & Considine, 2008), organ donation (Reinhart, Marshall, Feeley, & Tutzauer, 2007), recycling (Bessarabova, Fink, & Turner, 2013), smoking (Grandpre, Alvaro, Burgoon, Miller, & Hall, 2003; Shen, 2010), and sunscreen use (Quick & Stephenson, 2008), as well as in political contexts such as gun control (Miller et al., 2013) and the legalization of marijuana (Miller et al., 2013). In addition, and most germane to this study, research applying PRT to tobacco use has shown that reactance proneness, or the ease with which a person experiences reactance when confronted with a freedom threat (Quick, Shen, & Dillard, 2013), is a positive predictor of tobacco use, as well as a moderator of perceived effectiveness of antismoking messages (Miller & Quick, 2010). That is, reactant prone individuals are less likely to perceive antismoking messages as effective compared with their nonsmoking counterparts. Even more concerning is that reactance may lead to an increase in smoking in response to strong antismoking messages (Grandpre et al., 2003; Henriksen, Dauphinee, Wang, & Fortmann, 2006; Miller, Burgoon, Grandpre, & Alvaro, 2006).

PRT research reveals that message features play an important role in the arousal of reactance (for a review, see Quick et al., 2013). Dogmatic, explicit, or forceful language in messages arouse more reactance than messages featuring nonopinionated, implicit, or nonforceful language (e.g., see Buller, Borland, & Burgoon, 1998; Miller et al., 2007; Quick & Considine, 2008), and controlling language arouses a greater freedom threat and elicits more negative cognitions and anger than noncontrolling language (Miller et al., 2007). In a related study, dogmatic and vivid language was perceived as more freedom threatening following exposure to exercise and sunscreen messages than nondogmatic and pallid language (Quick & Stephenson, 2008).

Because the literature has established that strong language choices have the ability to increase reactance, it seems reasonable that graphic and explicit message imagery would increase reactance as well. Wohlburg (2009) found partial evidence for this claim by examining college students’ responses to antismoking messages. Specifically, students reported that antismoking messages with imagery often made them angry, defensive, or defiant and, in some cases, served as cues to “light up” (Wohlburg, 2009). As cited earlier, Erceg-Hurn and Steed (2011) confirmed that graphic cigarette images also arouse anger. More specifically, they found that graphic imagery led to increased anger beyond text warnings. However, because their study did not measure freedom threat perceptions, the mechanism through which this anger occurred was theoretically unclear. J. W. Brehm’s (1966) conceptualization of PRT requires the audience to perceive a freedom threat. It is this freedom threat that mediates the relationship between the message and reactance, and freedom threat has been shown to be significant by a number of scholars (for a recent discussion, see Quick et al., 2013). The connection between these graphic tobacco warning images and psychological reactance remains unknown. However, considering previous findings on graphic message features and freedom threat perceptions (Quick & Stephenson, 2008), the first hypothesis is advanced:

Hypothesis 1 (H1): Cigarette warning labels with an image increase freedom threat perceptions compared with cigarette warning labels without an image.

Although S. S. Brehm and Brehm (1981) originally posited that psychological reactance could not be measured, Dillard and Shen (2005) provided a tenable measurement of the theory’s central construct by operationalizing reactance as the combination of negative cognitions and anger. The measurement properties of reactance were recently bolstered by research demonstrating the superiority of Dillard and Shen’s operationalization as compared with others’ methods, such as Lindsey’s (2005) measurement, with respect to the construct’s reliability and validity (Quick, 2012), as well as through a meta-analysis (Rains, 2013). Thus, to measure psychological reactance, researchers should assess freedom threat perceptions as well as anger and negative cognitions following message exposure. Erceg-Hurn and Steed (2011) acknowledged the importance of measuring reactance as a latent construct consisting of anger and negative cognitions, but they assessed only anger in their study. Therefore, the current research utilizes assessments of both anger and negative cognitions in order to draw conclusions that adhere more closely to PRT. Consistent with previous research, a positive association is hypothesized between freedom threat perceptions and psychological reactance.

Hypothesis 2 (H2): Freedom threat perceptions are positively associated with psychological reactance.

Hypothesis 3 (H3): Freedom threat perceptions mediate the relationship between warning labels (whether text or image) and psychological reactance.

Source Appraisal

A number of studies demonstrate that graphic, explicit, and/or overtly threatening language enhance freedom threat perceptions and, therefore, psychological reactance. What is missing from this literature, however, is how reactance affects source appraisal. Research has generally found a negative association between reactance and perceived source expertise, sociability, and trustworthiness within the contexts of exercise (Miller et al., 2007) and responsible drinking (Quick & Bates, 2010). To our knowledge, no study has examined how reactance is connected with the aforementioned source dimensions within the context of graphic cigarette imagery on labeling. In addition, no study has examined the relationship between reactance and the additional source dimension of source domineeringness. Dillard, Palmer, and Kinney (1995) conceptualized domineeringness as indicating interpersonal control. The authors contrasted interpersonal domineeringness as a forceful assertion compared with a respectful request. As noted earlier, previous PRT research demonstrates that forceful language is associated with freedom threat perceptions and reactance (Dillard & Shen, 2005). To be consistent with earlier work, it therefore seems plausible to hypothesize the following:

Hypothesis 4a (H4a): Reactance is negatively associated with perceived source expertise and trustworthiness.

Hypothesis 4b (H4b): Reactance is positively associated with perceived source domineeringness.

In addition to testing the hypothesized paths of the reactance process, examining mean differences among variables is also explored. Specifically, this study will identify mean differences between image conditions on freedom threat perceptions, on psychological reactance directly, and on source appraisal. In this spirit,

Hypothesis 5 (H5): Compared with warning labels without an image, exposure to cigarette warning labels with a graphic image results in (a) greater freedom threat perceptions, (b) greater reactance, and (c) less favorable source appraisal

Trait Reactance

Although the majority of PRT research conceptualizes reactance as a psychological state, a related literature perceives reactance as an individual difference variable (see S. S. Brehm & Brehm, 1981; Wicklund, 1974). Although such conceptualizations are not consistent with J. W. Brehm’s (1966) original depiction of reactance, he later posited that certain individuals are more prone to reactance than others (see S. S. Brehm & Brehm, 1981). Trait-reactant individuals are characterized as independent, autonomous, rebellious, mistrustful, and self-sufficient (for a review, see Quick et al., 2013). Research has found that reactant-prone individuals have a greater proclivity to consume tobacco products (Miller et al., 2006) and engage in illicit drug use and risky sex (Miller & Quick, 2010). Not surprisingly, given the characteristics of trait-reactant individuals, health communication researchers consider segmenting audiences based on this personality profile a fruitful endeavor (Miller et al., 2006; Miller & Quick, 2010; Quick, Bates, & Quinlan, 2009). With these studies in mind, the final hypothesis is advanced:

Hypothesis 6 (H6): High trait-reactant individuals experience (a) greater freedom threat perceptions and psychological reactance and (b) less favorable source appraisal compared with low trait-reactant individuals.

Method

Participants and Procedure

Participants were undergraduate students attending a large Midwestern university (N = 435), with an age range between 18 and 25 years, an average age of 19.88 years (SD = 1.44), and a median age of 20. The sample was ethnically diverse; 62.3% identified as Caucasian, 8.3% as African American, 7.6% as Latino, 17.0% as Asian, and 5.5% as other (including Multiracial, Native American, Pacific Islander and Other).¹ The sample was 34% male and 66% female. Regarding smoking status, we divided the sample by those who reported smoking at least once during the last month (smokers) and those who reported no smoking during the last month (nonsmokers). Within the sample, 82.5% reported being nonsmokers, whereas 17.5% reported smoking behaviors.

Participants were recruited through communication courses and offered extra credit for their participation. Participants were randomly assigned to either the graphic image or no-image condition. The stimulus consisted of a text warning alone (on a regular pack of cigarettes) or a text warning and graphic image together (on the government’s proposed new packaging). The text warning only, in this case, served as a control because it is the legal status quo. After proper consent was acquired, students were given envelopes and instructed not to open them until the researcher told them to do so. Students completed initial baseline surveys on demographics and smoking behaviors and then were given a short written paragraph explaining the researchers’ interest in their perceptions and opinions about antismoking messages. They were then instructed to open their envelopes and carefully examine the cigarette packages, after which they completed a thought-listing exercise and the remainder of the survey.

Stimulus Materials

Erceg-Hurn and Steed’s (2011) study used online computer-generated images to present the stimuli to participants. However, as noted in their limitations section, viewing a flat image may not have the same effect as holding an actual cigarette package. For this reason, participants in the current study were provided with real cigarette packages and were able to hold and examine them. Half of the cigarette packages bore graphic images on the labels, and all packages of cigarettes were identical in every way with the exception of the presence or absence of the image. Seven different images were rotated equally but randomly among participants in the image condition. These images were chosen because they are some of the actual images that the FDA approved to appear on cigarette packages.

Marlboro Golds were the cigarette brand used for the stimuli. Marlboro is the most popular cigarette brand in the United States, making up 40% of the American market share (CDC, 2013b). This is more of the market share than five of Marlboro’s main competitors combined (CDC, 2013b). In addition, cigarettes labeled “light” or “ultra-light” (defined as 15g of tar or less) made up 94.7% of the cigarettes smoked in 2011 (Federal Trade Commission [FTC], 2013). Legislation introduced in 2010 prohibits referring to cigarettes as “light” or “ultra-light” on packaging, and Marlboro “Gold” was the name that replaced Marlboro Lights after this change (CDC, 2013b).

Measures

Smoking status

Smoking status was assessed using an item from the CDC’s (2010) National Adult Tobacco Survey Questionnaire. Specifically, the item asked participants how many days during the past month they had smoked. Participants reporting that they had smoked at least 1 day during the past month were coded as smokers (n = 76), and individuals reporting they had not smoked within the past month were coded as nonsmokers (n = 359).

Freedom threat

Freedom threat perceptions were measured using a four-item scale from Dillard and Shen (2005) on a 1 = strongly disagree to 7 = strongly agree scale. This measure has been validated and had respectable reliability in our analysis (M = 3.33, SD = 1.64, α = .83).

Reactance

Reactance was measured as a latent variable consisting of anger and negative cognitions. Two recent tests of Dillard and Shen’s (2005) operationalization support its continued use (Quick, 2012; Rains, 2013). Consistent with earlier work, anger was assessed on a 1 = not at all to 7 = a great deal of this feeling scale using the four-item scale created by Dillard and Shen (2005). Previous reliabilities have been consistently good with this scale, as was ours (M = 2.51, SD = 1.60, α = .93). Negative cognitions were assessed using the participant-as-coder technique. More specifically, participants were given 90 seconds to write out each thought that passed through their minds while holding and viewing the cigarette package. Next, participants coded each thought as relevant (pertaining to the cigarette package) or irrelevant (not pertaining to the cigarette package). Then, participants coded each of their thoughts as favorable, unfavorable, or neutral. Consistent with earlier research, only relevant negative thoughts were included in the analysis (M = 2.46, SD = 1.96; Quick, 2012; Rains, 2013).

Trait reactance

Trait reactance was measured using Hong and Faedda’s (1996) 14-item reactance proneness scale with participants indicating their reactance proneness on a 1 = strongly disagree to 7 = strongly agree scale (M = 3.87, SD = 0.80, α = .82).²

Source appraisal

Source appraisal (domineeringness, expertise, and trustworthiness) was measured using items adapted from Miller and colleagues (2007) and earlier conceptualizations regarding source domineeringness (Dillard et al., 1995). The items pertaining to source expertise and trustworthiness came from Miller and colleagues (2007) and were originally adapted from McCroskey’s (1966) scale. The questions asked participants to rate the message source using a 7-point scale with dichotomous descriptors anchoring each end of the scale for domineeringness (e.g., domineering/not domineering; M = 3.72, SD = 1.36, α = .80), expertise (e.g., knowledgeable/unknowledgeable; M = 4.71, SD = 1.32, α = .86), and trustworthiness (e.g., trustworthy/untrustworthy; M = 5.10, SD = 1.27, α = .86).

Data Analytic Strategy

In testing H1, H2, H3, and H4, structural equation modeling was employed using EQS 6.1. In doing so, a covariance matrix was computed and analyzed using full information maximum likelihood estimators. Specifically, the two-step approach (Kline, 2005) was performed by first examining the measurement properties of each variable to determine whether the measured items are consistent with the hypothesized latent variables followed by assessing the hypothesized paths using a structural model. A single item observed variable was used to assess the experimental message feature and negative cognitions. In addition, five latent composite variables were used including freedom threat, anger, source domineeringness, expertise, and trustworthiness. Each latent composite variable was identified with one indicator by setting the error term by subtracting the scale’s alpha reliability from 1 with the difference being multiplied by the measure’s variance (Holbert & Stephenson, 2002).

Prior to assessing model fit, the distribution of the data was examined. The data were not heavily skewed as indicated by a Mardia’s normalized estimate of 2.32. That said, the assumption that the sample data are drawn from a nonnormal population cannot be rejected. In evaluating each model, the chi-square distributed goodness-of-fit statistic (χ²), the comparative fit index (CFI), and the standardized root mean square residual (SRMR; see Bentler, 2007) were employed. We controlled for smoking status for both the measurement and structural models. In doing so, smoking status was used to create residualized scores for the variables in the model, thus controlling for smoking status. With respect to controlling for smoking status, we partitioned out the variance from smoking status prior to entering the variables into the model. All reported probabilities are one-tailed. The means, standard deviations, and correlations among the measured variables are shown in Table 1.

Table 1.

Means, Standard Deviations, and Correlation Matrix Among Measured Variables.

	M	SD	1	2	3	4	5	6	7
1. Image	—	—	1.0
2. Freedom threat	3.33	1.64	.52***	1.0
3. Negative cognitions	2.46	1.96	.27***	.22***	1.0
4. Anger	2.51	1.60	.04	.23***	.14**	1.0
5. Domineering	3.72	1.36	.16**	.23***	.02	.09	1.0
6. Expert	4.71	1.32	.08	.02	.06	−.06	.28***	1.0
7. Trustworthiness	5.10	1.27	.03	−.03	.01	−.02	.10*	.67***	1.0

Note. Image was coded with 0 = no image, 1 = image. All results were two-tailed.

p < .05. **p < .01. ***p < .001.

Results

We found acceptable fit indices for the measurement and structural models. The measurement model demonstrated good model fit, χ²(4, N = 435) = 25.33, p < .001, SRMR = .03, and CFI = 1.0. However, we also ran a confirmatory factor analysis to ascertain that all the items loaded on their respective latent variables. The results of this analysis support their unidimensionality, CFI = .96, SRMR = .05, root mean square error of approximation (RMSEA) = .06 (90% confidence interval [CI] = [.05, .07]), χ²(109, N = 435) = 271.17, p < .001. In addition, the hypothesized structural model featuring the latent composite variables demonstrated good fit, χ²(10, N = 435) = 31.09, p < .001, SRMR = .04, and CFI = 1.0. The unstandardized regression coefficients are presented in Figure 1.

Figure 1.

Unstandardized regression coefficients for hypothesized structural model.

H1: Cigarette Warning Labels and Freedom Threat Perceptions

As evidenced by the structural model presented in Figure 1, in support of H1, cigarette packages containing a graphic image were positively associated with freedom threat perceptions. In other words, cigarette packages containing a graphic image resulted in stronger freedom threat perceptions than cigarette packages without an image.

H2: Freedom Threat and Psychological Reactance

The second hypothesis, as shown in Figure 1, predicted that freedom threat perceptions are positively associated with psychological reactance. H2 was supported. A significant positive association between freedom threat perceptions and reactance emerged.

H3: Mediating Role of Freedom Threat Perceptions

The third hypothesis predicted that freedom threat perceptions mediate the cigarette warning label (both with and without images) to reactance relationship. In testing this association, PRODCLIN was employed to obtain the asymmetric 95% CI (MacKinnon, Fritz, Williams, & Lockwood, 2007). Results revealed that freedom threat perceptions mediated the image to reactance relationship, 95% CI = [0.342, 1.027]. Therefore, the third hypothesis was supported. It should be noted that a main effect between graphic cigarette warning labels and reactance was not significant.

H4: Psychological Reactance and Source Appraisal

It was hypothesized that reactance is significantly associated with source appraisal. Specifically, it was hypothesized that reactance is negatively associated with source expertise and trustworthiness but positively associated with source domineeringness. The results revealed that reactance was not significantly associated with source expertise or trustworthiness but was significantly associated with source domineeringness in the hypothesized direction. Thus, H4b received support, although H4a did not.

Overall, the model accounted for moderate variance in the measured variables. Specifically, the model accounted for variance in a freedom threat (R² = .41), psychological reactance (R² = .75), and source domineeringness (R² = .10).

H5: Mean Differences Among Measured Variables by Warning Images

H5 predicted that exposure to cigarette warning labels featuring a graphic image would arouse greater freedom threat perceptions and reactance and less favorable source appraisal compared with labels without a graphic image. Smoking status was not a significant covariate and therefore was removed from the analysis. A multivariate main effect was found for image, Wilks’s lambda = .69, F(6, 417) = 19.81, p < .001, η² = .31. Univariate differences revealed that individuals exposed to warning labels with an image (M = 4.16, SD = 1.45) reported greater freedom threat perceptions than those exposed to labels without an image (M = 2.46, SD = 1.32), F(1, 422) = 158.80, p < .001, η² = .27. Similarly, participants exposed to warning labels featuring a graphic image (M = 3.00, SD = 2.24) had more negative cognitions than those receiving labels without an image (M = 1.90, SD = 1.46), F(1, 422) = 35.74, p < .001, η² = .08. In addition to these differences, exposure to warning labels with a graphic image (M = 3.92, SD = 1.44) aroused stronger perceptions of source domineeringness than labels without an image (M = 3.51, SD = 1.23), F(1, 422) = 9.97, p < .05, η² = .02.

H6: Mean Differences Among Measured Variables by Trait Reactance

The sixth and final hypothesis predicted that high trait-reactant (HTR) individuals would experience greater freedom threat perceptions, psychological reactance, and maintain less favorable source appraisal compared with low trait-reactant (LTR) individuals. We performed a median split to determine high (n = 221, M = 4.50, SD = 0.45) versus low (n = 214, M = 3.22, SD = 0.52) trait-reactant individuals. Once again smoking status was not a significant covariate and as a result was removed from the analysis. A multivariate main effect was found for trait reactance, Wilks’s lambda = .94, F(6, 417) = 4.14, p < .001, η² = .06. Univariate differences were found for freedom threat perceptions with HTR individuals (M = 3.56, SD = 1.64) reporting greater freedom threat perceptions than LTR persons (M = 3.08, SD = 1.59), F(1, 422) = 9.14, p = .01, η² = .02. Similarly, HTR (M = 2.75, SD = 1.63) individuals experienced more anger than their LTR counterparts (M = 2.23, SD = 1.53), F(1, 422) = 11.85, p = .001, η² = .03. A similar difference appeared for source domineeringness, F(1, 422) = 5.20, p < .05, η² = .01, with HTR (M = 3.87, SD = 1.31) individuals perceiving the source as more domineering than LTR (M = 3.57, SD = 1.39) individuals. No other differences were discovered.

Discussion

The intention of the pending federal law requiring graphic images on warning labels on cigarette packaging is clearly designed to deter citizens from the dangers of smoking. Although previous studies have hailed these efforts as successes in other countries (e.g., Hammond et al., 2007; O’Hegarty et al., 2006; Peters et al., 2007), results from the current study indicate using graphic imagery could result in unintended effects that potentially do more harm than good. Specifically, we found that freedom threat perceptions occurred to a greater degree in response to the graphic image packaging than to warning labels without an image. This, in turn, was associated with heightened psychological reactance and the perception that the source was domineering. In addition, we found mean differences for both high and low trait-reactant individuals across the measured variables, with HTR individuals experiencing stronger freedom threat perceptions, anger, and source domineeringness than LTR individuals.

Our study points to several important theoretical and practical considerations. First, as hypothesized, graphic images led to freedom threat perceptions and psychological reactance. This study lends credence to the notion that reactance can be triggered via an assortment of message types. Previous research has largely focused on, and manipulated, language in testing reactance (see Quick et al., 2013). However, moving beyond verbal or textual accounts, the current study shows that primarily visual messages can invoke strong freedom threat perceptions and corresponding reactance above and beyond text alone. Our confidence in this finding is strengthened given that 75% of the variance explained in reactance was accounted for by the image and freedom threat perceptions. In addition, our finding that graphic imagery leads to reactance, which has an emotional component (anger), may provide some support to one of the Supreme and Appellate Courts’ reasons for rejecting this bill (R. J. Reynolds Tobacco Company, Lorillard Tobacco Company, Commonwealth Brands, Inc., Liggett Group, LLC, & Santa Fe Natural Tobacco Company, Inc. v. FDA, Hamburg, & Sebelius, 2012). That is, these graphic images may galvanize strong emotional reactions more than inform the public of factual information.

Although there is no surprise in the finding that freedom threat perceptions led to reactance, the resulting source appraisal deserves further attention. Trustworthiness and expertise were unaffected by reactance to these images; however, negative evaluations of source domineeringness were heightened. Participants not only viewed the source of these warnings to be knowledgeable and trustworthy but also perceived the source as domineering. It is possible that people interpreted these messages as credible but found the way in which the messages were delivered to be problematic. It is also possible that participants did not dispute the claims on these cigarette packages, but that the tone in which these claims were presented was not well received. Taken together, these findings illustrate how an image can indirectly influence source perceptions. The finding that participants derogate the source as a response to reactance has been supported in other recent PRT studies as well (e.g., Bessarabova et al., 2013). In their study, they discovered that freedom threat perceptions led to reductions in source credibility and increases in source derogation. The reasons for discrepancies in the different dimensions of source appraisal as outcomes (i.e., credibility) are deserving of greater attention in future research. Perhaps source attribution or attitudes toward the source before message exposure may be variables to consider in the future.

With specificity to our findings, Dillard and Shen (2005) stated, “Roughly parallel to forcefulness or authoritarianism, dominance captures the extent to which a message reveals that the source believes he or she can control the message recipient” (p. 163). Furthermore, research has established that negative source appraisals lead to a reduction in message persuasiveness (Grandpre et al., 2003; Quick & Considine, 2008). Thus, experiencing reactance toward a message and, therefore, perceiving the source as more domineering, could potentially create more negative attitudes toward the message or the source. Negative attitudes toward the source may reduce behavioral intentions, an operationalization of persuasiveness, of graphic image warning labels as an antitobacco strategy.

Two other dimensions, trustworthiness and expertise, were not affected to the same degree; individuals still viewed the source (presumably the government) as an expert and rated the source as credible. There is little explanation for why these results would emerge; however, observing reactance’s effect on only the source appraisal of domineeringness leaves questions as to the relationship between these variables. Therefore, future research should consider whether manipulating the source changes the source appraisal as an outcome of reactance or whether reactance is consistently only associated with domineeringness.

Our project also considered the potential moderating effect of trait reactance. The high and low trait-reactance models were very similar. In both HTR and LTR models, a freedom threat was positively associated with reactance, and reactance was positively associated with source domineeringness. Although we collapsed these models due to no statistical differences, the similarity of these two models indicates that the hypothesized relationships were not unique to individuals based on their reactance proneness. In fact, these findings indicate that the reactance process is experienced similarly in response to these cigarette graphic images regardless of this individual difference. Put differently, images too graphic or shocking may create freedom threat perceptions to a much broader audience, not just among individuals with a proclivity to resist freedom-threatening messages. This is important information for the FDA and lawmakers who are debating the future of this persuasive strategy because reactance to these messages may lead to unintended consequences and a variety of maladaptive freedom restoration behaviors among different audience segments. Furthermore, highly freedom-threatening messages, whether textual or visual, may reduce the persuasiveness of the intended message (Bessarabova et al., 2013), thereby thwarting the primary goal of this tobacco control measure.

Although the unstandardized regression coefficients were the same for both HTR and LTR individuals, our analyses did reveal significant mean differences. Specifically, HTR individuals experienced greater freedom threat perceptions, anger, and source domineeringness compared with their LTR counterparts. Thus, although the associations among the measured variables were experienced similarly across the entire sample, the magnitude of these variables was heightened among individuals with a tendency to experience reactance. That said, caution should be used when interpreting these findings given the relatively small effect sizes observed. Smokers are the audience most targeted by graphic image cigarette packaging, and previous research indicates that trait reactance is more prevalent in smokers (Miller et al., 2006). However, if the smoking population is more trait-reactant, the freedom threat and the resulting reactance may be magnified for this group. This highlights the potential unintended consequences this antitobacco message strategy may have on smokers and those most at risk to initiate smoking in the first place. It may be the case that the group that needs the most help would be the least likely to find support through these graphic image messages.

Research has begun to consider the role of genetics in addiction to nicotine, and there is evidence that for many smokers no message, by itself, can overcome a genetic or physical dependence (e.g., Hoft et al., 2009; Weiss et al., 2008). Many graphic image campaigns have been successful in reaching many smokers who can quit easily (e.g., Hammond et al., 2007; Hammond, Fong, et al., 2004; Hammond et al., 2003; Hammond, McDonald, et al., 2004; O’Hegarty et al., 2006; Peters et al., 2007; Willemsen, 2005), “but for hard-core smokers, there may be something else going on” (Marcus, 2011, p. 1). In addition, medication and addiction counseling are likely to be more helpful to addicted smokers than campaigns and graphic imagery (Marcus, 2011).

Between the two variables composing psychological reactance (anger and negative cognitions), only anger was of greater magnitude for highly trait-reactant individuals. In contrast, only negative cognitions were heightened across the entire sample in response to the graphic image condition. Future studies should assess whether trait-reactant individuals are more prone to anger as a reaction to freedom threat. If this were the case, it would underscore the need for addressing trait reactance when designing persuasive messages. Furthermore, how might anger, for HTR individuals, or negative cognitions, for the overall population, wax or wane when exposed to these graphic images on cigarette packaging? In other words, future research should investigate if anger and negative thoughts would increase in intensity and number with each subsequent exposure or would they diminish as suggested in theories of desensitization (Salahi, 2010).

In the current study, we also looked at the differences between exposure to the graphic image and no-image conditions. Participants exposed to a graphic image experienced significantly greater freedom threat perceptions, negative cognitions, and perceived source domineeringness. Interestingly, anger was not significantly different across message conditions. Miller and colleagues (2013) recently argued that negative cognitions might be longer lasting than anger, which is short-lived following exposure to a message. Additional research should investigate why anger arousal did not differ between image conditions. This may be useful information for scholars looking to advance PRT, whether in harnessing the motivational power of reactance or in looking to minimize it.

Despite gaining greater insight into the role of reactance within the context of cigarette warning labels, we are aware of the limitations of this research project. First, this study was cross-sectional. Although it may illuminate some interesting processes and important variables within the context of graphic smoking warning labels, the results cannot speak to any lasting positive or negative effects on smoking behaviors, specifically behavioral outcomes.

Although we are sympathetic to the perception that including attitudes and behavioral intentions in studies is an accepted standard or proxy, we see not including these measures in the current study as a limitation of studying this particular health behavior, not a weakness of our study. We did not measure these outcomes toward smoking because research has shown that these measures are not that helpful in discriminating smokers from nonsmokers (Huijding, de Jong, Wiers, & Verkooijen, 2005). Further addressing this particular health behavior, affective reactions to (attitudes toward) smoking have little effectiveness in providing explanation for smoking behavior (Svanson, Rudman, & Greenwald, 2001). Behavioral intentions can be equally problematic with this topic; studies have shown that the effects between intentions and behaviors are not as strong as once thought, especially when behaviors are risky, lack perceived controllability, or are habitual (Webb & Sheeran, 2006; Wood, Quinn, & Kashy, 2002).

Furthermore, because 80% of our sample were nonsmokers, asking whether they intend to smoke is an odd question, because most people do not anticipate trying cigarettes as adults. Research shows that “more than 80 percent of all adult smokers begin smoking before the age of 18”; and more than 90% do so by age 20 (National Survey on Drug Use and Health, 2012). Although our college-student sample is only slightly older, it is unlikely that they would feel like they intended to start smoking. In our case, asking nonsmokers if they have a positive attitude toward smoking as well as an intention to smoke would be better evaluated with a younger sample, because smoking initiation happens much earlier on average, and asking smokers if they intend to quit or smoke would very likely be misleading. Overall, a longitudinal study with a younger sample would be needed to determine the lasting psychological, attitudinal, and behavioral effects of these images.

Another possible limitation was the use of a college-student sample. Although there are good reasons supported by the literature that this population was more than adequate (see Arnett, 2000), generalizability is still restricted. Young adults are no longer adolescents but may not have reached the maturity of adulthood (Arnett, 2000), making them a sample that has characteristics of both younger and older adults. Correspondingly, the emergent adult population has been determined to be a good segment to examine in reactance studies. According to Cho and Salmon (2006), “perceived threats to freedom may be amplified among college-age individuals—who tend to be more defiant to persuasion regarding behavior change than older adults” (p. 97). Second, one limitation of the Erceg-Hurn and Steed (2011) study was that they only sampled smokers. However, college students may have engaged in a wider variety of smoking behaviors. After all, college is a time in which, as Arnett (2007) argued, the emerging adult population is at “the peak age period for many behaviors most societies try to discourage” (p. 72). Furthermore, student status and drinking alcohol have both been related to the frequencies of smoking behaviors as well as identification as a smoker (Berg et al., 2009). Thus, a college-student population offered the opportunity to attain a sample of both smokers and nonsmokers to test our hypotheses.

Despite the rationale for using a college sample, however, a sample more representative of the general population may illuminate important moderators such as sex, education, race, and socioeconomic status. A more general sample may also capture more variation in smoking behaviors or yield greater numbers of smokers than a relatively homogeneous college sample. Our sample did not yield a sufficient number of smokers to adequately divide the sample by smoking status. We did run MANOVAs to determine if smokers experienced any of the measured variables in question to a greater degree than nonsmokers; however, there were no significant differences. The proportion of smokers in our sample was consistent with the current population-wide smoking rates; 18.9% of the general U.S. population between the ages of 18 and 24 currently smoke (CDC, 2013b). However, we would need a larger sample to yield a sufficient number of smokers for the division of the sample by smoking status.

What our study may have overlooked, though, is that previous research has shown that smoking behaviors do not necessarily predict identification as a smoker, and identification may be an important predictor in a number of outcomes related to smoking behaviors, including psychological reactance (Berg et al., 2009; van den Putte, Yzer, Willemsen, & de Bruijn, 2009). We defined smoker as anyone who had smoked in the previous month, thus grouping social smokers with daily smokers, two groups that may have different identities with respect to their smoking behavior. Identification has not been examined in the context of cigarette warning labels or reactance to antismoking messages, although future research should aim to determine whether smoking frequency correlates with smoking identity. Furthermore, addressing how frequency and identity may affect reactions to and outcomes from graphic antismoking warning labels would be a worthy undertaking because of the role identification performs in PRT (Berg et al., 2009; van den Putte et al., 2009). In essence, including more smokers in the sample would allow for a more finessed analyses of smoker identity based on overtime measures of smoking behaviors.

Finally, our study establishes a pattern of psychological reactance to warning labels with graphic images, but its scope did not exhaust the variety of outcomes linked to reactance. We acknowledge that ideally we would have measured attitudes, intentions, and behaviors had we recruited a different sample. When individuals attempt to restore a threatened freedom, they may be more likely to engage in the threatened behavior but are also believed to be more likely to engage in proxy behaviors that can fulfill this goal (J. W. Brehm, 1966; Quick et al., 2013). Future research should explore the behaviors triggered by reactance in response to graphic warning labels and images more specifically.

Summary and Conclusion

This study reexamines the conclusions of Erceg-Hurn and Steed (2011), correcting weaknesses and expanding the scope of their study in several important ways. First, we employed complete and validated measures of psychological reactance and explored our results using both tests of association (structural equation modeling) and tests of difference (MANOVA). Second, we increased external validity by using real cigarette packages with the actual FDA-approved images as our stimuli. Third, our study investigated how this tobacco control strategy and reactance may affect source appraisal. Finally, we examined the mean differences between high and low trait-reactance individuals across the measured variables.

Overall, our results suggest that using graphic imagery on cigarette packaging enhances freedom threat perceptions, reactance, and perceived source domineeringness. This finding was found among HTR and LTR persons, although to varying magnitudes. These results indicate that utilizing graphic images on tobacco packaging might not be as effective as some practitioners had initially hoped. In fact, the messages designed to deter smoking behaviors ignite freedom threat appraisal, which precedes reactance and, in turn, elevates source domineeringness. Each of these outcomes is counterproductive to this antitobacco strategy. In other words, the good intentions of this tobacco control measure may be for naught; in light of this study’s findings, using graphic warning images on tobacco packaging should be carefully considered before its implementation.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Nicole R. LaVoie (MA, University of Illinois at Urbana-Champaign, 2010) is a PhD candidate at the University of Illinois Urbana-Champaign. Her primary interests are in unintended consequences of health messages. Currently, she is working on the examination of message cues that affect the cognitive, affective, and behavioral processes of underlying uninteded effects such as stigma.

Brian L. Quick (PhD, Texas A & M University, 2005) is a Professor in the Department of Communication and in the College of Medicine at the University of Illinois at Urbana-Champaign. His primary research interests examine how individuals process persuasive messages across a host of health contexts.

Julius M. Riles (MA, University of Illinois at Urbana-Champaign, 2010) is a PhD candidate at the University of Illinois at Urbana-Champaign. His primary research interests pertain to the effects of media messages on stigma and other social perceptions.

Natalie J. Lambert (MA, University of Illinois at Urbana-Champaign, 2012) is a PhD student at the University of Illinois at Urbana-Champaign. She conducts organizational and health research using computational and traditional methods in order to develop communication theory that is informed by large-scale empirical data.

References

Arnett

J. J.

(2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55, 469-480. doi:10.1037/0003-066X.55.5.469

Arnett

J. J.

(2007). Emerging adulthood: What is it, and what is it good for? Childhood Development Perspectives, 1, 68-73. doi:10.1111/j.1750-8606.2007.00016.x

Bentler

P. M.

(2007). On tests and indices for evaluating structural models. Personality and Individual Differences, 42, 825-829. doi:10.1016/j.paid.2006.09.024

Berg

C. J.

Lust

Sanem

J. R.

Kirch

M. A.

Rudie

Ehlinger

. . . An

L. C.

(2009). Smoker self-identification versus recent smoking among college students. American Journal of Preventative Medicine, 36, 333-336. doi:10.1016/j.amepre.2008.11.010

Bessarabova

Fink

E. L.

Turner

E. M.

(2013). Reactance, restoration, and cognitive structure: Comparative statistics. Human Communication Research, 39, 339-364. doi:10.1111/hcre.12007

Brehm

J. W.

(1966). A theory of psychological reactance. New York, NY: Academic Press.

Brehm

S. S.

Brehm

J. W.

(1981). Psychological reactance: A theory of freedom and control. San Diego, CA: Academic Press.

Buller

D. B.

Borland

Burgoon

(1998). Impact of behavioral intention on effectiveness of message features: Evidence from the family sun safety project. Human Communication Research, 24, 433-453. doi:10.1111/j.1468-2958.1998.tb00424.x

Centers for Disease Control and Prevention. (2010). National Adult Tobacco Survey (NATS). Retrieved from http://www.cdc.gov/tobacco/data_statistics/surveys/nats/

10.

Centers for Disease Control and Prevention. (2013a). Economic facts about U.S. tobacco production and use. Retrieved from http://www.cdc.gov/tobacco/data_statistics/fact_sheets/economics/econ_facts/index.htm

11.

Centers for Disease Control and Prevention. (2013b). Smoking and tobacco use. Retrieved from http://www.cdc.gov/tobacco/data_statistics/fact_sheets/adult_data/cig_smoking/index.htm

12.

Cho

Salmon

C. T.

(2006). Fear appeals for individuals in different stages of change: Intended and unintended effects and implications on public health campaigns. Health Communication, 20, 91-99. doi:10.1207/s15327027hc2001_9

13.

Dillard

J. P.

Palmer

M. T.

Kinney

T. A.

(1995). Relational judgments in an influence context. Human Communication Research, 21, 331-353. doi:10.1111/j.1468-2958. 1995.tb00350.x

14.

Dillard

J. P.

Shen

(2005). On the nature of reactance and its role in persuasive health communication. Communication Monographs, 72, 144-168. doi:10.1080/03637750500111815

15.

Erceg-Hurn

D. M.

Steed

L. G.

(2011). Does exposure to cigarette health warnings elicit psychological reactance in smokers? Journal of Applied Social Psychology, 41, 219-237. doi:10.1111/j.1559-1816.2010.00710.x

16.

Federal Trade Commission. (2013). Federal trade commission cigarette report for 2011. Retrieved from http://www.ftc.gov/reports/federal-trade-commission-cigarette-report-2011

17.

Grandpre

J. R.

Alvaro

E. M.

Burgoon

Miller

C. H.

Hall

J. R.

(2003). Adolescent reactance and antismoking campaigns: A theoretical approach. Health Communication, 15, 349-366. doi:10.1207/S15327027HC1503_6

18.

Hammond

Fong

G. T.

Borland

Cummings

K. M.

McNeill

Driezen

(2007). Text and graphic warnings on cigarette packages: Findings from the International Control Four Country Study. American Journal of Preventative Medicine, 32, 202-209. doi:10.1016/j.amepre.2006.11.011

19.

Hammond

Fong

G. T.

McDonald

P. W.

Brown

K. S.

Cameron

(2004). Graphic Canadian cigarette warning labels and adverse outcomes: Evidence from Canadian smokers. American Journal of Public Health, 94, 1142-1445. doi:10.2105/AJPH.94.8.1442

20.

Hammond

Fong

G. T.

McDonald

P. W.

Cameron

Brown

K. S.

(2003). Impact of the graphic Canadian warning labels on adult smoking behavior. Tobacco Control, 12, 391-395. doi:10.1136/tc.12.4.391

21.

Hammond

McDonald

P. W.

Fong

G. T.

Brown

K. S.

Cameron

(2004). The impact of cigarette warning labels and smoke-free bylaws on smoking cessation: Evidence from former smokers. Canadian Journal of Public Health, 95, 201-204.

22.

Henriksen

Dauphinee

A. L.

Wang

Fortmann

S. P.

(2006). Industry sponsored antismoking ads and adolescent reactance: Test of a boomerang effect. Tobacco Control, 15, 13-18. doi:10.1136/tc.2003.006361

23.

Hoft

N. R.

Corley

R. P.

McQueen

M. B.

Schlaepfer

I. R.

Huizinga

Ehringer

M. A.

(2009). Genetic association of the CHRNA6 and CHRNB3 genes with tobacco dependence in a nationally representative sample. Neuropsychopharmacology, 34, 698-706. doi:10.1038/npp.2008.122

24.

Holbert

R. L.

Stephenson

M. T.

(2002). Structural equation modeling in the communication sciences, 1995-2000. Human Communication Research, 28, 531-551. doi:10.1111/j.1468-2958.2002.tb00822.x

25.

Hong

Faedda

(1996). Refinement of the Hong psychological reactance scale. Educational and Psychology Measurement, 56, 173-182. doi:10.1177/0013164496056001014

26.

Huijding

de Jong

P. J.

Wiers

R. W.

Verkooijen

(2005). Implicit and explicit attitudes toward smoking in a smoking and a nonsmoking setting. Addictive Behaviors, 30, 949-961. doi:10.1016/j.addbeh.2004.09.014

27.

Kline

R. B.

(2005). Principles and practice of structural equation modeling (2nd ed.). New York, NY: Guilford Press.

28.

Lindsey

L. L. M.

(2005). Anticipated guilt as behavioral motivation: An examination of appeals to help unknown others through bone marrow donation. Human Communication Research, 31, 453-481. doi:10.1111/j.1468-2958.2005.tb00879.x

29.

MacKinnon

D. P.

Fritz

M. S.

Williams

Lockwood

C. M.

(2007). Distribution of the product confidence limits for the indirect effect: Program PRODCLIN. Behavior Research Methods, 39, 384-389. doi:10.3758/BF03193007

30.

Marcus

M. B.

(2011). Genes may make quitting tougher for smokers. Retrieved from http://consumer.healthday.com/public-health-information-30/centers-for-disease-control-news-120/genes-may-make-quitting-tougher-for-smokers-659214.html

31.

McCroskey

J. C.

(1966). Scales for the measurement of ethos. Speech Communication, 33, 65-72. doi:10.1080/03637756609375482

32.

Miller

C. H.

Burgoon

Grandpre

J. R.

Alvaro

E. M.

(2006). Identifying principal risk factors for the initiation of adolescent smoking behaviors: The significance of psychological reactance. Health Communication, 19, 241-252. doi:10.1207/s15327027hc1903_6

33.

Miller

C. H.

Ivanov

Sims

Compton

Harrison

K. J.

Parker

K. A.

. . . Averbeck

J. M.

(2013). Boosting the potency of resistance: Combining the motivational forces of inoculation and psychological reactance. Human Communication Research, 39, 127-155. doi:10.1111/j.1468-2958.2012.01438.x

34.

Miller

C. H.

Lane

L. T.

Deatrick

L. M.

Young

A. M.

Potts

K. A.

(2007). Psychological reactance and promotional health messages: The effects of controlling language, lexical concreteness, and the restoration of freedom. Human Communication Research, 33, 219-240. doi:10.1111/j.1468-2958.2007.00297.x

35.

Miller

C. H.

Quick

B. L.

(2010). Sensation seeking and psychological reactance as health risk predictors for an emerging adult population. Health Communication, 26, 266-275. doi:10.1080/10410231003698945

36.

National Survey on Drug Use and Health. (2012). Substance Abuse and Mental Health Services Administration. Available from http://www.samhsa.gov

37.

O’Hegarty

Pederson

L. L.

Nelson

D. E.

Mowery

Gable

J. M.

Wortley

(2006). Reactions of young adult smokers to warning labels on cigarette packages. American Journal of Preventative Medicine, 30, 467-473. doi:10.1016/j.amepre.2006.01.018

38.

Pelofsky

Yukhananov

(2011, November 29). Obama admin appeals cigarette warning ruling. Reuters. Retrieved from http://www.reuters.com/article/2011/11/29/us-fda-tobacco-idUSTRE7AS2GT20111129

39.

Peters

Romer

Slovic

Hall Jamieson

Wharfield

Mertz

C. K.

Carpenter

S. M.

(2007). The impact and acceptability of Canadian-style cigarette warnings among U.S. smokers and nonsmokers. Nicotine & Tobacco Research, 9, 473-481. doi:10.1080/14622200701239639

40.

Quick

B. L.

(2012). What is the best measure of psychological reactance? An empirical test of two measures. Health Communication, 27, 1-9. doi:10.1080/10410236.2011.567446

41.

Quick

B. L.

Bates

B. R.

(2010). The use of gain- or loss-frame messages and efficacy appeals to dissuade excessive alcohol consumption among college students: A test of psychological reactance theory. Journal of Health Communication, 15, 603-628. doi:10.1080/10810730.2010.499593

42.

Quick

B. L.

Bates

B. R.

Quinlan

M. R.

(2009). The utility of anger in promoting clean indoor air policies. Health Communication, 24, 548-561. doi:10.1080/10410230903104939

43.

Quick

B. L.

Considine

J. R.

(2008). Examining the use of forceful language when designing exercise persuasive messages for adults: A test of conceptualizing reactance arousal as a two-step process. Health Communication, 23, 483-491. doi:10.1080/10410230802342150

44.

Quick

B. L.

Shen

L. J.

Dillard

J. P.

(2013). Reactance theory and persuasion. In Dillard

J. P.

Shen

L. J.

(Eds.), The persuasion handbook: Developments in theory and practice (2nd ed., pp. 167-183). Los Angeles, CA: Sage.

45.

Quick

B. L.

Stephenson

M. T.

(2007). Further evidence that psychological reactance can be modeled as a combination of anger and negative cognitions. Communication Research, 34, 255-276. doi:10.1177/0093650207300427

46.

Quick

B. L.

Stephenson

M. T.

(2008). Examining the role of trait reactance and sensation seeking on perceived threat, state reactance, and reactance restoration. Human Communication Research, 34, 448-476. doi:10.1111/j.1468-2958.2008.00328.x

47.

Rains

S. A.

(2013). The nature of psychological reactance revisited: A meta-analytic review. Human Communication Research, 39, 47-73. doi:10.1111/j.1468-2958.2012.01443.x

48.

Reinhart

A. M.

Marshall

H. M.

Feeley

T. H.

Tutzauer

(2007). The persuasive effects of message framing in organ donation: The mediating role of psychological reactance. Communication Monographs, 74, 229-255. doi:10.1080/03637750701397098

49.

R. J. Reynolds Tobacco Company, Lorillard Tobacco Company, Commonwealth Brands, Inc., Liggett Group, LLC, and Santa Fe Natural Tobacco Company, Inc. v. U.S. Food and Drug Administration, Hamburg, and Sebelius. 11-1482 RJL 19. (US District Court 2012). Retrieved from https://ecf.dcd.uscourts.gov/cgi-bin/show_public_doc?2011cv1482-58

50.

Ruiter

R. A. C.

Kok

(2005). Saying is not (always) doing: Cigarette warning labels are useless. European Journal of Public Health, 15, 329-330. doi:10.1093/eurpub/cki095

51.

Salahi

(2010, November 10). FDA wants cigarette packs to include images of corpses, diseased lungs. Retrieved from http://abcnews.go.com/Health/Wellness/cigarette-packaging-graphic/story?id=12109439#.Ttwmv4PT-fE.email

52.

Shen

(2010). Mitigating psychological reactance: The role of message-induced empathy in persuasion. Human Communication Research, 36, 297-422. doi:10.1111/j.1468-2958.2010.01381.x

53.

Svanson

J. E.

Rudman

L. A.

Greenwald

A. G.

(2001). Using the Implicit Association Test to investigate attitude-behavior consistency for stigmatised behaviour. Cognition and Emotion, 15, 207-230.

54.

van den Putte

Yzer

Willemsen

M. C.

de Bruijn

(2009). The effects of smoking self-identity and quitting self-identity on attempts to quit smoking. Health Psychology, 28, 535-544. doi:10.1037/a0015199

55.

Webb

T. L.

Sheeran

(2006). Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin, 132, 249-268.

56.

Weiss

R. B.

Baker

T. B.

Cannon

D. S.

von Niederhausern

Dunn

D. M.

Matsunami

. . . Leppert

M. F.

(2008). A candidate gene approach identifies the CHRNA5-A3-B4 region as a risk factor for age-dependent nicotine addiction. PLoS Genetics, 4(7): e1000125. doi:10.1371/journal.pgen.1000125

57.

Wicklund

R. A.

(1974). Freedom and reactance. Potomac, MD: Lawrence Erlbaum.

58.

Willemsen

M. C.

(2005). The new EU cigarette health warnings benefit smokers who want to quit the habit: Results from the Dutch continuous survey of smoking habits. European Journal of Public Health, 15, 389-392. doi:10.1093/eurpub/cki061

59.

Wohlburg

J. M.

(2009). College students’ response to antismoking messages: Denial, defiance, and other boomerang effects. Journal of Consumer Affairs, 40, 294-323. doi:10.1111/j.1745-6606.2006.00059.x

60.

Wood

Quinn

J. M.

Kashy

(2002). Habits in everyday life: The thought and feel of action. Journal of Personality and Social Psychology, 83, 1281-1297. doi:10.1037//0022-3514.83.6.1281

61.

World Health Organization. (2013a). Full list of signatories and parties to the WHO framework convention on tobacco control. Retrieved from http://www.who.int/fctc/signatories_parties/en/index.html

62.

World Health Organization. (2013b). WHO report on the global tobacco epidemic. Retrieved from http://www.who.int/tobacco/global_report/2011/en/