This article was downloaded by: [University of Connecticut] On: 03 February 2015, At: 05:19 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Health Communication: International Perspectives Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/uhcm20
Understanding Public Resistance to Messages About Health Disparities a
Sarah E. Gollust & Joseph N. Cappella
b
a
University of Minnesota School of Public Health , Minneapolis , Minnesota , USA b
Annenberg School for Communication, University of Pennsylvania , Philadelphia , Pennsylvania , USA Published online: 13 Jan 2014.
Click for updates To cite this article: Sarah E. Gollust & Joseph N. Cappella (2014) Understanding Public Resistance to Messages About Health Disparities, Journal of Health Communication: International Perspectives, 19:4, 493-510, DOI: 10.1080/10810730.2013.821561 To link to this article: http://dx.doi.org/10.1080/10810730.2013.821561
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Downloaded by [University of Connecticut] at 05:19 03 February 2015
Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Journal of Health Communication, 19:493–510, 2014 Copyright # Taylor & Francis Group, LLC ISSN: 1081-0730 print=1087-0415 online DOI: 10.1080/10810730.2013.821561
Understanding Public Resistance to Messages About Health Disparities SARAH E. GOLLUST
Downloaded by [University of Connecticut] at 05:19 03 February 2015
University of Minnesota School of Public Health, Minneapolis, Minnesota, USA
JOSEPH N. CAPPELLA Annenberg School for Communication, University of Pennsylvania, Philadelphia, Pennsylvania, USA Advocates and policymakers strategically communicate about health disparities in an effort to raise public awareness, often by emphasizing the social and economic factors that influence these disparities. Previous research suggests that predisposing political orientation and values related to self-reliance and personal responsibility may produce resistance to such messages. In this study, the authors culled 4 messages about the causes of disparities in life expectancy from public discourse and randomly presented them to a nationally representative sample of 732 Americans. Three indicators of message resistance were measured: belief that messages are weak, elicitation of anger, and production of counterarguments. Expected political differences in message resistance were identified, with Republicans perceiving messages to be weaker, arousing less anger, and eliciting more counterarguing than for Democrats. Among 3 messages that described the social determinants of health disparities, a message that identified the role of personal choices (explicitly acknowledging personal responsibility) produced the least anger and counterarguing among Republicans. Political differences in anger arousal and counterarguing can be explained, in part, by predisposing values toward personal responsibility. These findings have relevance for policy advocates seeking to bridge public divides surrounding health disparities and for scholars advancing theories of reactance to policy-relevant health messaging.
In January 2011, the U.S. Centers for Disease Control and Prevention issued a landmark report about health disparities in the United States (Centers for Disease Control and Prevention, 2011). The report documented racial, ethnic, and socioeconomic disparities in insurance coverage, mortality, morbidity, and risky behaviors. In the introduction, Centers for Disease Control and Prevention Director Thomas Frieden alluded to one important goal of the report: ‘‘Differences in health based on race, ethnicity, or economics can be reduced, but will require public awareness and understanding of which groups are most vulnerable’’ [italics added] (p. 2). The conventional wisdom behind this and similar campaigns seems to be that enumerating health disparities will increase public awareness, which will lead to support for policy actions to ameliorate these disparities. Address correspondence to Sarah Gollust, University of Minnesota School of Public Health, 420 Delaware Street SE, MMC #729, Room 15-230 Phillips-Wangensteen Building, Minneapolis, MN 55455, USA. E-mail: [email protected]
493
Downloaded by [University of Connecticut] at 05:19 03 February 2015
494
S. E. Gollust and J. N. Cappella
Although this strategy has intuitive appeal, little is known about the public’s awareness of health disparities or their responses to different ways of framing the problem. One popular public health frame involves emphasizing the social and economic origins of health differences, which many consider more important than biological or behavioral factors (Braveman, Egerter, & Williams, 2010; Wilensky & Satcher, 2009). Yet, social psychological research shows that beliefs about the causes of group differences are polarized by ideology; liberals attribute status differences to social structural factors, whereas conservatives favor individual behavioral causes (Kluegel & Smith, 1986; Skitka, Mullen, Griffin, Hutchinson, & Chamberlin, 2002). As a result, claims about the social causes of health differences may meet resistance among those who do not share this causal attribution (Gollust, Lantz, & Ubel, 2009; Niederdeppe, Bu, Borah, Kindig, & Robert, 2008). Although U.S. public policy attention to health disparities has emerged and receded sporadically over the past couple decades, few actions to rectify disparities have been implemented (Stone, 2006). Bowen and Zwi (2005), in their review, argued that ‘‘the importance of values as a factor that influences the lack of action on health inequity has been poorly researched’’ (p. e166). Filling this gap, this study examines how political values influence public response to messages about health disparities.
Background on Values in Communication Political scientists and social psychologists have long demonstrated the power of political values in shaping opinion about policy matters (Feldman, 1988; Tetlock, 1986). A value is commonly defined as ‘‘an enduring belief that a specific mode of conduct or end-state of existence’’ is ‘‘personally or socially preferable’’ to another conduct or end-state (Rokeach, 1968). One important value underlying Americans’ beliefs about social policy issues is economic individualism (Conover & Feldman, 1984; Feldman, 1988; Markus, 2001), the notion that Americans can achieve success through hard work and self-reliance. The ethic of personal responsibility—that people, not the government, should be responsible for improving well-being— extends from this seminal value. In addition to serving as an independent predictor of policy opinion, values also condition responses to messages (Zaller, 1992). Experimental studies demonstrate that individuals’ social, cultural, or political values (e.g., egalitarianism, individualism, justice beliefs) moderate the effects of messages on attitudes and opinions (Brewer & Gross, 2005; Domke, Shah, & Wackman, 1998; Kahan, JenkinsSmith, & Braman, 2011; Lucas, Alexander, Firestone, & Lebreton, 2009; Nelson & Garst, 2005; Shen & Edwards, 2005). The literature on biased processing offers explanations for these moderators of message effects, showing that people are motivated to perceive the strength and credibility of messages in accordance with their predisposing beliefs and values (Edwards & Smith, 1996; Lord, Ross, & Lepper, 1979; Taber & Lodge, 2006). For example, people evaluate arguments supportive of their prior beliefs as stronger and more credible, spend more time counterarguing messages that conflict with their prior beliefs, and seek out confirming rather than disconfirming information (MacCoun & Paletz, 2009; Taber & Lodge, 2006). Biased processing can also lead to boomerang effects, whereby people hold a perception even more strongly after exposure to a countermessage (Hart & Nisbet, 2011; Nyhan & Reifler, 2010). A small but growing literature in health communication has advanced similar themes, coalescing into the resurgence of the idea of reactance (Brehm, 1966). This
Downloaded by [University of Connecticut] at 05:19 03 February 2015
Resistance to Health Disparities Messages
495
theory posits that any persuasive message may arouse a motivation, called reactance, to resist the advocacy (Dillard & Shen, 2005). This response is particularly likely when a message threatens an individuals’ sense of freedom, defined broadly to encompass behavioral actions (e.g., to smoke or not) but also emotions or attitudes (e.g., feelings of autonomy or responsibility; Brehm, 1966; Dillard & Shen, 2005). When individuals perceive their freedoms are threatened, they experience an aroused psychological state indicated by expressing negative thoughts, counterarguing, feeling anger, derogating the message source, and perceiving the message to be weak or not credible (Cameron, Jacks, & O’Brien, 2002; Fishbein & Cappella, 2006; Rains & Turner, 2007; Zuwerink & Devine, 1996). A research consensus is building that negative thoughts and anger are particularly promising measurable signals of reactance (Dillard & Shen, 2005; Quick & Stephenson, 2007). While most existing research has examined reactance as a response to individually oriented health promotion messages such as ‘‘don’t drink’’ or ‘‘do exercise’’ (Miller, Lane, Deatrick, Young, & Potts, 2007; Rains & Turner, 2007; Dillard & Shen, 2005), the theory should apply to policy-related messages perceived to threaten freedoms as well.
Values and Resistance in Health Disparities Communication The public has limited awareness of health disparities; for example, fewer than half of Americans believe that African Americans are worse off than Whites in terms of life expectancy or infant mortality (Benz, Espinosa, Welsh, & Fontes, 2011; Brodie, Connolly, & Deane, 2010). Public opinion studies show that conservatives are less likely to recognize health disparities than liberals (Booske, Robert, & Rohan, 2011), that the public is more aware of differences in health by socioeconomic status than by race (Booske et al., 2011), and that the public judges behavioral factors as more important determinants than social and economic factors (Robert & Booske, 2011). Research also demonstrates that journalists face challenges in covering the issue of health disparities (Wallington, Blake, Taylor-Clark, & Viswanath, 2010), but when they do, they tend to emphasize individual-level behavioral explanations for racial health differences (Kim, Kumanyika, Shive, Igweatu, & Kim, 2010). The emphasis on behavioral factors among the lay public and journalists alike makes sense, given the centrality of personal responsibility in American political thought in general and in health policy discourse in particular (Brownell et al., 2010; Leichter, 2003; Wikler, 2002). Thus, it follows that health communication messages that challenge or threaten notions of personal responsibility may activate reactance motivations, especially among subgroups that hold this value more highly. Recent studies demonstrate that political predispositions (which likely correlate with responsibility values) shape how Americans respond to a range of health messages (Baum, 2011; Gollust et al., 2009; Landau, 2009; Niederdeppe, Shapiro, & Porticella, 2011). For example, when study participants were exposed to a message about the social determinants of diabetes, Republicans were more likely to resist the message and have backlash against it (Gollust et al., 2009). Niederdeppe and colleagues (2011) also found that political ideology distinguished individuals’ responses to messages about the social-environmental determinants of obesity, with liberals more responsive to a narrative-based message. Consistent with these studies, the Robert Wood Johnson Foundation concluded that different strategies for communicating about the social determinants of health are required for Democrats and Republicans because of their divergent values and beliefs (Robert Wood Johnson Foundation, 2010).
496
S. E. Gollust and J. N. Cappella
Downloaded by [University of Connecticut] at 05:19 03 February 2015
Research Goals and Hypotheses While previous research hints at an important moderating role for political values in shaping response to communication about health disparities, this has not been tested directly, nor is it clear which of several common ways of framing health disparities will yield more or less resistance. Understanding these micro-level processes of resistance is critical for determining which methods of strategic communication may elicit more support—and less backlash—among subgroups of the public who may be predisposed to disagree, a necessary step for gaining broad public support to address health disparities. This study examined responses to four messages about socioeconomic health disparities, tests political differences in response, and evaluates the extent to which these differences can be explained by differences in two related values – economic individualism and personal responsibility. We posed the following research questions: 1. Which causal explanations for health disparities in public discourse do members of the public evaluate most favorably? 2. Are certain causal explanations evaluated differently by political partisans? On the basis of evidence that reactance is measured by cognitive (counterarguing) and affective (anger) responses (Dillard & Shen, 2005), we hypothesized: Hypothesis 1: A message that explicitly acknowledges the role of personal responsibility in health disparities will elicit less reactance for Republicans than does other messages. Hypothesis 2: Gaps between Republicans’ and Democrats’ responses to messages can be explained by differences in predisposing values of economic individualism and personal responsibility.
Method Message Development The first stage in the research was to craft disparities messages that draw from common themes in public discourse. We conducted a qualitative content analysis of elite discourse and media coverage related to three events: the spring 2008 airing of a PBS documentary about health inequalities (Unnatural Causes), the August 2008 release of a World Health Organization report about social determinants of health (World Health Organization, 2008), and the activities and final report of the Robert Wood Johnson Foundation Commission to Build a Healthier America (Robert Wood Johnson Foundation, 2009). LexisNexis keyword searches1 identified news stories about these events appearing in major U.S. newspapers and national nightly TV news broadcasts from January 1, 2008, to April 15, 2009, and press releases were obtained from the websites of the three campaigns. These approaches yielded 25 documents (5 press releases, 20 news stories), which we examined qualitatively for common themes. On the basis of these themes, we constructed three messages describing the social determinants of health disparities and countered these with one message that described behavioral determinants. The messages were not attributed to a specific source; they indicated only that they were abridged from a recent U.S. health report. All messages described socioeconomic differences in life expectancy, with 1 Search terms: ((‘‘world health organization’’ and (gap OR disparity OR inequality OR determinants)) OR (‘‘unnatural causes’’ and PBS) OR (‘‘commission to build a healthier America’’)).
Downloaded by [University of Connecticut] at 05:19 03 February 2015
Resistance to Health Disparities Messages
497
no mention of racial=ethnic differences, to avoid triggering resistance based solely on an explicit mention of race (Mendelberg, 2001).2 Each message began with a two-sentence statement indicating that people ‘‘living in poverty’’ have ‘‘higher rates of disease, and lower life expectancy’’ than wealthier individuals, and specifying a 6-year gap in life expectancy between groups (see the Appendix for message text). The sentences that followed provided four (randomly assigned) causal explanations for the socioeconomic differences in health. The typical social determinants message concluded that social factors are the most important causes of those health differences. The acknowledge choices message similarly emphasized social factors as most important but acknowledged that personal responsibility plays a role. The universal message emphasized that social factors are the most important and affect everyone throughout the income distribution, including the middle class. In contrast with the three messages that emphasized social factors, the personal responsibility only message emphasized behavioral choices and personal responsibility as the most important determinants of health differences. Data and Measures We used data from the Annenberg National Health Communication Survey, through identical 10-min modules fielded after the core 25-min survey in June and July 2009. At this time, U.S. political attention to health reform legislation was building, but there was little national attention to health disparities, and the controversy over health reform had not escalated (as it would during the August 2009 congressional recess). The Annenberg National Health Communication Survey is a nationally representative monthly Internet-based survey of adult Americans’ health behaviors and attitudes, fielded by Knowledge Networks (now known as GfK), a firm that maintains a panel of research participants recruited to reflect the national population. For the June and July 2009 waves, 442 (55% completion rate) and 290 participants (59% completion rate) were surveyed respectively, for a total sample of 7323 (see Table 1 for sample characteristics). Upon completing the core survey, respondents answered questions about their values (including humanitarian, egalitarian, and individualistic values so as not to prime one orientation over another). Then they completed an unrelated task designed to be distracting (12 items asking about use of news sources). Next they were randomly assigned to read and evaluate one of the four messages. There were no significant differences across the four treatment groups in any demographic or political characteristics. Indicators of Resistance: Message Strength, Anger, and Counterarguing Perceived message strength, shown to predict effectiveness of health campaigns (Dillard, Weber, & Vail, 2007), was measured with four items (Zhao et al., 2011): ‘‘This paragraph about health is believable,’’ reported on a 5-point Likert scale ranging from 1 (not at all believable) to 5 (very believable); ‘‘This paragraph about health is convincing,’’ reported on a 5-point Likert scale ranging from 1 (not at all convincing) to 5 (very convincing); ‘‘How much do you agree overall with the 2 However, we recognize that social class and race are closely linked in the United States (Gilens, 1999). To assess whether racial attitudes were primed by messages, we included feeling thermometer measures at the conclusion of the study, asking participants to identify perceived warmth toward six groups: Whites, Blacks, Latinos, poor people, middle-class people, and rich people. Analyses of variance of these measures on the randomly assigned messages identified no significant (p < .05) differences. 3 Given the small sample size across cells and the distortion that can occur from application of weighting variables in small samples, the Knowledge Networks poststratification weights are not applied in these analyses.
498
S. E. Gollust and J. N. Cappella Table 1. Descriptive characteristics of sample (N ¼ 732)
Downloaded by [University of Connecticut] at 05:19 03 February 2015
% Age (years) 18–29 30–49 50–64 65þ Female White, non-Hispanic Education Less than high school High school Some college College degree or higher Annual household income 0.6). The two coders divided up the responses, with 50% double-coded (n ¼ 276) for final interrater reliability determination. First, we determined whether the thought represented a refusal (i.e., ‘‘No’’; ‘‘I don’t want to’’; j ¼ 1.00, n ¼ 24). Next, we excluded responses that were irrelevant (j ¼ 0.855, n ¼ 9). For the remainder of thoughts (n ¼ 548), we coded the tone as neutral, supportive, mixed (some thoughts in favor of the message, some opposed), or opposed (j ¼ 0.706). Next, we coded for whether the respondent provided any causal argument about health disparities. Responses were coded as no causal argument, a causal argument consistent with the message they had received (e.g., mention of social factors for message 1), or a causal argument that was inconsistent with the message (e.g., mention of social factors for message 4, or any mention of biological factors) (j ¼ 0.755). Given the literature’s emphasis on negative tone and counterarguing as important antecedents of resistance, a variable that measured unfavorable thoughts and counterarguing best operationalized this concept. We constructed a variable where 0 was a neutral response, 1 indicated supportive thoughts and consistent causal argument, 2 indicated any combination of mixed support and=or counterargument, and 3 indicated opposition and=or counterarguing (j ¼ 0.689). For these analyses, we categorized this variable dichotomously as 1 for any unfavorable thoughts or any counterarguing and 0 for all neutral or consistent=supportive responses. We refer to this variable as counterarguing for simplicity. Regression analyses using the fourcategory variable (ordered logit models) were consistent with analyses presented here. Political Party Identification and Values We used the Knowledge Networks panel measure of political party identification, a 7-point scale ranging from 1 (strong Republican) to 7 (strong Democrat). We collapsed this variable into two groups, Republicans (including those leaning Republican) and Democrats (including those leaning Democrat), excluding pure Independents (n ¼ 8) and those for whom Knowledge Networks had missing data (n ¼ 54). The economic individualism measure was drawn from the American National Election Survey (Feldman, 1988). Factor analysis of the six-item instrument (with responses measured on a 5-point Likert scale where 1 ¼ strongly disagree and 5 ¼ strongly agree) yielded two factors, distinguishing the reverse-coded items from the others, as has been previously documented for other measures (Marsh, 1996). Given the two-factor structure, we used the three-item subscale with the higher reliability (Cronbach’s a ¼ 0.662): ‘‘Hard work offers little guarantee of success’’; ‘‘Even if people are ambitious, they often cannot succeed’’; ‘‘Getting ahead in the world is mostly a matter of getting the breaks’’ (M ¼ 3.17, SD ¼ 0.76). The personal responsibility value was measured with a single item used in the World Values Survey (http://www.worldvaluessurvey.org). Respondents were asked to place themselves on a scale in which 1 ¼ ‘‘Government should take more responsibility to ensure that everyone is provided for’’ and 10 ¼ ‘‘People should take more responsibility to provide for themselves’’ (M ¼ 6.35, SD ¼ 2.77). Analysis To evaluate Research Question 1 (which messages the full sample evaluated favorably), significant differences in response across the 4 messages were tested using an analysis of variance and chi-squared tests. To answer Research Question 2 (whether Democrats and Republicans evaluated the messages differently), we estimated linear and logistic regression models for each of the three outcomes on
Downloaded by [University of Connecticut] at 05:19 03 February 2015
500
S. E. Gollust and J. N. Cappella
indicator variables for each message, an indicator for Republican, and interactions between each message and the Republican variable. Because party identification is correlated with other social characteristics, we included income, education, gender, age, and race as controls (from Knowledge Networks panel data). To evaluate Hypothesis 1 (that a message that acknowledges personal responsibility will elicit less reactance than other social determinants messages), we calculated predicted means or probabilities from the aforementioned models to examine whether Republicans were less likely to become angered by and counterargue Message 2. To test Hypothesis 2, whether any partisan gap in response can be explained by confounding with political values, we reestimated the regression models, adding economic individualism and personal responsibility as covariates. To examine whether party identification and values both significantly moderate message response, we estimated regression models including party identification and values and all of their interactions with the messages.
Results Table 2 compares the three indicators of resistance for the four messages. In answering Research Question 1, we found that among the three social determinants messages employed in public discourse (Messages 1, 2, and 3), respondents evaluated the typical message as significantly stronger. The universal message aroused the most anger, significantly more than the message that acknowledged choices, which aroused the least anger among the three social determinants messages. The three social determinants messages aroused counterarguing equally, among about one fourth of the sample. In contrast, the message attributing disparities entirely to personal responsibility aroused counterarguing among more than half the sample, significantly more than the social determinants messages. Given expected biased processing, the results in Table 2 likely mask subgroup differences. Table 3 presents regression models that indicate whether message responses vary by political orientation (Model 1) and whether political orientation moderates message response (Model 2). Compared with Democrats, Republicans perceived health disparities messages to be weaker (b ¼ 0.24, p < .001), to elicit less anger (b ¼ 0.26, p < .01), and to arouse more counterarguing (b ¼ 0.59, p < .01; odds ratio ¼ 1.81, 95% CI [1.20, 2.73]). The interaction results (Model 2) show no significantly different political patterns by message for perceived strength. However, significant interaction terms Table 2. Indicators of resistance to messages about health inequalities
Message 1: Typical 2: Acknowledge choices 3: Universal 4: Personal responsibility Test of significance across messages p
Perceived strength (n ¼ 727) M (SD)
Elicitation of anger (n ¼ 723) M (SD)
Counterarguing (n ¼ 548) Proportion
3.59 (0.79)2,3,4 3.41 (0.79)1 3.44 (0.93)1 3.37 (0.87)1 F(3, 723) ¼ 2.63
2.64 (1.34)2 2.37 (1.21)1,3 2.71 (1.16)2 2.55 (1.28) F(3, 719) ¼ 2.42
0.214 0.254 0.264 0.531,2,3 2 v (3) ¼ 41.4
.049
.065
Journal of Health Communication: International Perspectives Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/uhcm20
Understanding Public Resistance to Messages About Health Disparities a
Sarah E. Gollust & Joseph N. Cappella
b
a
University of Minnesota School of Public Health , Minneapolis , Minnesota , USA b
Annenberg School for Communication, University of Pennsylvania , Philadelphia , Pennsylvania , USA Published online: 13 Jan 2014.
Click for updates To cite this article: Sarah E. Gollust & Joseph N. Cappella (2014) Understanding Public Resistance to Messages About Health Disparities, Journal of Health Communication: International Perspectives, 19:4, 493-510, DOI: 10.1080/10810730.2013.821561 To link to this article: http://dx.doi.org/10.1080/10810730.2013.821561
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Downloaded by [University of Connecticut] at 05:19 03 February 2015
Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Journal of Health Communication, 19:493–510, 2014 Copyright # Taylor & Francis Group, LLC ISSN: 1081-0730 print=1087-0415 online DOI: 10.1080/10810730.2013.821561
Understanding Public Resistance to Messages About Health Disparities SARAH E. GOLLUST
Downloaded by [University of Connecticut] at 05:19 03 February 2015
University of Minnesota School of Public Health, Minneapolis, Minnesota, USA
JOSEPH N. CAPPELLA Annenberg School for Communication, University of Pennsylvania, Philadelphia, Pennsylvania, USA Advocates and policymakers strategically communicate about health disparities in an effort to raise public awareness, often by emphasizing the social and economic factors that influence these disparities. Previous research suggests that predisposing political orientation and values related to self-reliance and personal responsibility may produce resistance to such messages. In this study, the authors culled 4 messages about the causes of disparities in life expectancy from public discourse and randomly presented them to a nationally representative sample of 732 Americans. Three indicators of message resistance were measured: belief that messages are weak, elicitation of anger, and production of counterarguments. Expected political differences in message resistance were identified, with Republicans perceiving messages to be weaker, arousing less anger, and eliciting more counterarguing than for Democrats. Among 3 messages that described the social determinants of health disparities, a message that identified the role of personal choices (explicitly acknowledging personal responsibility) produced the least anger and counterarguing among Republicans. Political differences in anger arousal and counterarguing can be explained, in part, by predisposing values toward personal responsibility. These findings have relevance for policy advocates seeking to bridge public divides surrounding health disparities and for scholars advancing theories of reactance to policy-relevant health messaging.
In January 2011, the U.S. Centers for Disease Control and Prevention issued a landmark report about health disparities in the United States (Centers for Disease Control and Prevention, 2011). The report documented racial, ethnic, and socioeconomic disparities in insurance coverage, mortality, morbidity, and risky behaviors. In the introduction, Centers for Disease Control and Prevention Director Thomas Frieden alluded to one important goal of the report: ‘‘Differences in health based on race, ethnicity, or economics can be reduced, but will require public awareness and understanding of which groups are most vulnerable’’ [italics added] (p. 2). The conventional wisdom behind this and similar campaigns seems to be that enumerating health disparities will increase public awareness, which will lead to support for policy actions to ameliorate these disparities. Address correspondence to Sarah Gollust, University of Minnesota School of Public Health, 420 Delaware Street SE, MMC #729, Room 15-230 Phillips-Wangensteen Building, Minneapolis, MN 55455, USA. E-mail: [email protected]
493
Downloaded by [University of Connecticut] at 05:19 03 February 2015
494
S. E. Gollust and J. N. Cappella
Although this strategy has intuitive appeal, little is known about the public’s awareness of health disparities or their responses to different ways of framing the problem. One popular public health frame involves emphasizing the social and economic origins of health differences, which many consider more important than biological or behavioral factors (Braveman, Egerter, & Williams, 2010; Wilensky & Satcher, 2009). Yet, social psychological research shows that beliefs about the causes of group differences are polarized by ideology; liberals attribute status differences to social structural factors, whereas conservatives favor individual behavioral causes (Kluegel & Smith, 1986; Skitka, Mullen, Griffin, Hutchinson, & Chamberlin, 2002). As a result, claims about the social causes of health differences may meet resistance among those who do not share this causal attribution (Gollust, Lantz, & Ubel, 2009; Niederdeppe, Bu, Borah, Kindig, & Robert, 2008). Although U.S. public policy attention to health disparities has emerged and receded sporadically over the past couple decades, few actions to rectify disparities have been implemented (Stone, 2006). Bowen and Zwi (2005), in their review, argued that ‘‘the importance of values as a factor that influences the lack of action on health inequity has been poorly researched’’ (p. e166). Filling this gap, this study examines how political values influence public response to messages about health disparities.
Background on Values in Communication Political scientists and social psychologists have long demonstrated the power of political values in shaping opinion about policy matters (Feldman, 1988; Tetlock, 1986). A value is commonly defined as ‘‘an enduring belief that a specific mode of conduct or end-state of existence’’ is ‘‘personally or socially preferable’’ to another conduct or end-state (Rokeach, 1968). One important value underlying Americans’ beliefs about social policy issues is economic individualism (Conover & Feldman, 1984; Feldman, 1988; Markus, 2001), the notion that Americans can achieve success through hard work and self-reliance. The ethic of personal responsibility—that people, not the government, should be responsible for improving well-being— extends from this seminal value. In addition to serving as an independent predictor of policy opinion, values also condition responses to messages (Zaller, 1992). Experimental studies demonstrate that individuals’ social, cultural, or political values (e.g., egalitarianism, individualism, justice beliefs) moderate the effects of messages on attitudes and opinions (Brewer & Gross, 2005; Domke, Shah, & Wackman, 1998; Kahan, JenkinsSmith, & Braman, 2011; Lucas, Alexander, Firestone, & Lebreton, 2009; Nelson & Garst, 2005; Shen & Edwards, 2005). The literature on biased processing offers explanations for these moderators of message effects, showing that people are motivated to perceive the strength and credibility of messages in accordance with their predisposing beliefs and values (Edwards & Smith, 1996; Lord, Ross, & Lepper, 1979; Taber & Lodge, 2006). For example, people evaluate arguments supportive of their prior beliefs as stronger and more credible, spend more time counterarguing messages that conflict with their prior beliefs, and seek out confirming rather than disconfirming information (MacCoun & Paletz, 2009; Taber & Lodge, 2006). Biased processing can also lead to boomerang effects, whereby people hold a perception even more strongly after exposure to a countermessage (Hart & Nisbet, 2011; Nyhan & Reifler, 2010). A small but growing literature in health communication has advanced similar themes, coalescing into the resurgence of the idea of reactance (Brehm, 1966). This
Downloaded by [University of Connecticut] at 05:19 03 February 2015
Resistance to Health Disparities Messages
495
theory posits that any persuasive message may arouse a motivation, called reactance, to resist the advocacy (Dillard & Shen, 2005). This response is particularly likely when a message threatens an individuals’ sense of freedom, defined broadly to encompass behavioral actions (e.g., to smoke or not) but also emotions or attitudes (e.g., feelings of autonomy or responsibility; Brehm, 1966; Dillard & Shen, 2005). When individuals perceive their freedoms are threatened, they experience an aroused psychological state indicated by expressing negative thoughts, counterarguing, feeling anger, derogating the message source, and perceiving the message to be weak or not credible (Cameron, Jacks, & O’Brien, 2002; Fishbein & Cappella, 2006; Rains & Turner, 2007; Zuwerink & Devine, 1996). A research consensus is building that negative thoughts and anger are particularly promising measurable signals of reactance (Dillard & Shen, 2005; Quick & Stephenson, 2007). While most existing research has examined reactance as a response to individually oriented health promotion messages such as ‘‘don’t drink’’ or ‘‘do exercise’’ (Miller, Lane, Deatrick, Young, & Potts, 2007; Rains & Turner, 2007; Dillard & Shen, 2005), the theory should apply to policy-related messages perceived to threaten freedoms as well.
Values and Resistance in Health Disparities Communication The public has limited awareness of health disparities; for example, fewer than half of Americans believe that African Americans are worse off than Whites in terms of life expectancy or infant mortality (Benz, Espinosa, Welsh, & Fontes, 2011; Brodie, Connolly, & Deane, 2010). Public opinion studies show that conservatives are less likely to recognize health disparities than liberals (Booske, Robert, & Rohan, 2011), that the public is more aware of differences in health by socioeconomic status than by race (Booske et al., 2011), and that the public judges behavioral factors as more important determinants than social and economic factors (Robert & Booske, 2011). Research also demonstrates that journalists face challenges in covering the issue of health disparities (Wallington, Blake, Taylor-Clark, & Viswanath, 2010), but when they do, they tend to emphasize individual-level behavioral explanations for racial health differences (Kim, Kumanyika, Shive, Igweatu, & Kim, 2010). The emphasis on behavioral factors among the lay public and journalists alike makes sense, given the centrality of personal responsibility in American political thought in general and in health policy discourse in particular (Brownell et al., 2010; Leichter, 2003; Wikler, 2002). Thus, it follows that health communication messages that challenge or threaten notions of personal responsibility may activate reactance motivations, especially among subgroups that hold this value more highly. Recent studies demonstrate that political predispositions (which likely correlate with responsibility values) shape how Americans respond to a range of health messages (Baum, 2011; Gollust et al., 2009; Landau, 2009; Niederdeppe, Shapiro, & Porticella, 2011). For example, when study participants were exposed to a message about the social determinants of diabetes, Republicans were more likely to resist the message and have backlash against it (Gollust et al., 2009). Niederdeppe and colleagues (2011) also found that political ideology distinguished individuals’ responses to messages about the social-environmental determinants of obesity, with liberals more responsive to a narrative-based message. Consistent with these studies, the Robert Wood Johnson Foundation concluded that different strategies for communicating about the social determinants of health are required for Democrats and Republicans because of their divergent values and beliefs (Robert Wood Johnson Foundation, 2010).
496
S. E. Gollust and J. N. Cappella
Downloaded by [University of Connecticut] at 05:19 03 February 2015
Research Goals and Hypotheses While previous research hints at an important moderating role for political values in shaping response to communication about health disparities, this has not been tested directly, nor is it clear which of several common ways of framing health disparities will yield more or less resistance. Understanding these micro-level processes of resistance is critical for determining which methods of strategic communication may elicit more support—and less backlash—among subgroups of the public who may be predisposed to disagree, a necessary step for gaining broad public support to address health disparities. This study examined responses to four messages about socioeconomic health disparities, tests political differences in response, and evaluates the extent to which these differences can be explained by differences in two related values – economic individualism and personal responsibility. We posed the following research questions: 1. Which causal explanations for health disparities in public discourse do members of the public evaluate most favorably? 2. Are certain causal explanations evaluated differently by political partisans? On the basis of evidence that reactance is measured by cognitive (counterarguing) and affective (anger) responses (Dillard & Shen, 2005), we hypothesized: Hypothesis 1: A message that explicitly acknowledges the role of personal responsibility in health disparities will elicit less reactance for Republicans than does other messages. Hypothesis 2: Gaps between Republicans’ and Democrats’ responses to messages can be explained by differences in predisposing values of economic individualism and personal responsibility.
Method Message Development The first stage in the research was to craft disparities messages that draw from common themes in public discourse. We conducted a qualitative content analysis of elite discourse and media coverage related to three events: the spring 2008 airing of a PBS documentary about health inequalities (Unnatural Causes), the August 2008 release of a World Health Organization report about social determinants of health (World Health Organization, 2008), and the activities and final report of the Robert Wood Johnson Foundation Commission to Build a Healthier America (Robert Wood Johnson Foundation, 2009). LexisNexis keyword searches1 identified news stories about these events appearing in major U.S. newspapers and national nightly TV news broadcasts from January 1, 2008, to April 15, 2009, and press releases were obtained from the websites of the three campaigns. These approaches yielded 25 documents (5 press releases, 20 news stories), which we examined qualitatively for common themes. On the basis of these themes, we constructed three messages describing the social determinants of health disparities and countered these with one message that described behavioral determinants. The messages were not attributed to a specific source; they indicated only that they were abridged from a recent U.S. health report. All messages described socioeconomic differences in life expectancy, with 1 Search terms: ((‘‘world health organization’’ and (gap OR disparity OR inequality OR determinants)) OR (‘‘unnatural causes’’ and PBS) OR (‘‘commission to build a healthier America’’)).
Downloaded by [University of Connecticut] at 05:19 03 February 2015
Resistance to Health Disparities Messages
497
no mention of racial=ethnic differences, to avoid triggering resistance based solely on an explicit mention of race (Mendelberg, 2001).2 Each message began with a two-sentence statement indicating that people ‘‘living in poverty’’ have ‘‘higher rates of disease, and lower life expectancy’’ than wealthier individuals, and specifying a 6-year gap in life expectancy between groups (see the Appendix for message text). The sentences that followed provided four (randomly assigned) causal explanations for the socioeconomic differences in health. The typical social determinants message concluded that social factors are the most important causes of those health differences. The acknowledge choices message similarly emphasized social factors as most important but acknowledged that personal responsibility plays a role. The universal message emphasized that social factors are the most important and affect everyone throughout the income distribution, including the middle class. In contrast with the three messages that emphasized social factors, the personal responsibility only message emphasized behavioral choices and personal responsibility as the most important determinants of health differences. Data and Measures We used data from the Annenberg National Health Communication Survey, through identical 10-min modules fielded after the core 25-min survey in June and July 2009. At this time, U.S. political attention to health reform legislation was building, but there was little national attention to health disparities, and the controversy over health reform had not escalated (as it would during the August 2009 congressional recess). The Annenberg National Health Communication Survey is a nationally representative monthly Internet-based survey of adult Americans’ health behaviors and attitudes, fielded by Knowledge Networks (now known as GfK), a firm that maintains a panel of research participants recruited to reflect the national population. For the June and July 2009 waves, 442 (55% completion rate) and 290 participants (59% completion rate) were surveyed respectively, for a total sample of 7323 (see Table 1 for sample characteristics). Upon completing the core survey, respondents answered questions about their values (including humanitarian, egalitarian, and individualistic values so as not to prime one orientation over another). Then they completed an unrelated task designed to be distracting (12 items asking about use of news sources). Next they were randomly assigned to read and evaluate one of the four messages. There were no significant differences across the four treatment groups in any demographic or political characteristics. Indicators of Resistance: Message Strength, Anger, and Counterarguing Perceived message strength, shown to predict effectiveness of health campaigns (Dillard, Weber, & Vail, 2007), was measured with four items (Zhao et al., 2011): ‘‘This paragraph about health is believable,’’ reported on a 5-point Likert scale ranging from 1 (not at all believable) to 5 (very believable); ‘‘This paragraph about health is convincing,’’ reported on a 5-point Likert scale ranging from 1 (not at all convincing) to 5 (very convincing); ‘‘How much do you agree overall with the 2 However, we recognize that social class and race are closely linked in the United States (Gilens, 1999). To assess whether racial attitudes were primed by messages, we included feeling thermometer measures at the conclusion of the study, asking participants to identify perceived warmth toward six groups: Whites, Blacks, Latinos, poor people, middle-class people, and rich people. Analyses of variance of these measures on the randomly assigned messages identified no significant (p < .05) differences. 3 Given the small sample size across cells and the distortion that can occur from application of weighting variables in small samples, the Knowledge Networks poststratification weights are not applied in these analyses.
498
S. E. Gollust and J. N. Cappella Table 1. Descriptive characteristics of sample (N ¼ 732)
Downloaded by [University of Connecticut] at 05:19 03 February 2015
% Age (years) 18–29 30–49 50–64 65þ Female White, non-Hispanic Education Less than high school High school Some college College degree or higher Annual household income 0.6). The two coders divided up the responses, with 50% double-coded (n ¼ 276) for final interrater reliability determination. First, we determined whether the thought represented a refusal (i.e., ‘‘No’’; ‘‘I don’t want to’’; j ¼ 1.00, n ¼ 24). Next, we excluded responses that were irrelevant (j ¼ 0.855, n ¼ 9). For the remainder of thoughts (n ¼ 548), we coded the tone as neutral, supportive, mixed (some thoughts in favor of the message, some opposed), or opposed (j ¼ 0.706). Next, we coded for whether the respondent provided any causal argument about health disparities. Responses were coded as no causal argument, a causal argument consistent with the message they had received (e.g., mention of social factors for message 1), or a causal argument that was inconsistent with the message (e.g., mention of social factors for message 4, or any mention of biological factors) (j ¼ 0.755). Given the literature’s emphasis on negative tone and counterarguing as important antecedents of resistance, a variable that measured unfavorable thoughts and counterarguing best operationalized this concept. We constructed a variable where 0 was a neutral response, 1 indicated supportive thoughts and consistent causal argument, 2 indicated any combination of mixed support and=or counterargument, and 3 indicated opposition and=or counterarguing (j ¼ 0.689). For these analyses, we categorized this variable dichotomously as 1 for any unfavorable thoughts or any counterarguing and 0 for all neutral or consistent=supportive responses. We refer to this variable as counterarguing for simplicity. Regression analyses using the fourcategory variable (ordered logit models) were consistent with analyses presented here. Political Party Identification and Values We used the Knowledge Networks panel measure of political party identification, a 7-point scale ranging from 1 (strong Republican) to 7 (strong Democrat). We collapsed this variable into two groups, Republicans (including those leaning Republican) and Democrats (including those leaning Democrat), excluding pure Independents (n ¼ 8) and those for whom Knowledge Networks had missing data (n ¼ 54). The economic individualism measure was drawn from the American National Election Survey (Feldman, 1988). Factor analysis of the six-item instrument (with responses measured on a 5-point Likert scale where 1 ¼ strongly disagree and 5 ¼ strongly agree) yielded two factors, distinguishing the reverse-coded items from the others, as has been previously documented for other measures (Marsh, 1996). Given the two-factor structure, we used the three-item subscale with the higher reliability (Cronbach’s a ¼ 0.662): ‘‘Hard work offers little guarantee of success’’; ‘‘Even if people are ambitious, they often cannot succeed’’; ‘‘Getting ahead in the world is mostly a matter of getting the breaks’’ (M ¼ 3.17, SD ¼ 0.76). The personal responsibility value was measured with a single item used in the World Values Survey (http://www.worldvaluessurvey.org). Respondents were asked to place themselves on a scale in which 1 ¼ ‘‘Government should take more responsibility to ensure that everyone is provided for’’ and 10 ¼ ‘‘People should take more responsibility to provide for themselves’’ (M ¼ 6.35, SD ¼ 2.77). Analysis To evaluate Research Question 1 (which messages the full sample evaluated favorably), significant differences in response across the 4 messages were tested using an analysis of variance and chi-squared tests. To answer Research Question 2 (whether Democrats and Republicans evaluated the messages differently), we estimated linear and logistic regression models for each of the three outcomes on
Downloaded by [University of Connecticut] at 05:19 03 February 2015
500
S. E. Gollust and J. N. Cappella
indicator variables for each message, an indicator for Republican, and interactions between each message and the Republican variable. Because party identification is correlated with other social characteristics, we included income, education, gender, age, and race as controls (from Knowledge Networks panel data). To evaluate Hypothesis 1 (that a message that acknowledges personal responsibility will elicit less reactance than other social determinants messages), we calculated predicted means or probabilities from the aforementioned models to examine whether Republicans were less likely to become angered by and counterargue Message 2. To test Hypothesis 2, whether any partisan gap in response can be explained by confounding with political values, we reestimated the regression models, adding economic individualism and personal responsibility as covariates. To examine whether party identification and values both significantly moderate message response, we estimated regression models including party identification and values and all of their interactions with the messages.
Results Table 2 compares the three indicators of resistance for the four messages. In answering Research Question 1, we found that among the three social determinants messages employed in public discourse (Messages 1, 2, and 3), respondents evaluated the typical message as significantly stronger. The universal message aroused the most anger, significantly more than the message that acknowledged choices, which aroused the least anger among the three social determinants messages. The three social determinants messages aroused counterarguing equally, among about one fourth of the sample. In contrast, the message attributing disparities entirely to personal responsibility aroused counterarguing among more than half the sample, significantly more than the social determinants messages. Given expected biased processing, the results in Table 2 likely mask subgroup differences. Table 3 presents regression models that indicate whether message responses vary by political orientation (Model 1) and whether political orientation moderates message response (Model 2). Compared with Democrats, Republicans perceived health disparities messages to be weaker (b ¼ 0.24, p < .001), to elicit less anger (b ¼ 0.26, p < .01), and to arouse more counterarguing (b ¼ 0.59, p < .01; odds ratio ¼ 1.81, 95% CI [1.20, 2.73]). The interaction results (Model 2) show no significantly different political patterns by message for perceived strength. However, significant interaction terms Table 2. Indicators of resistance to messages about health inequalities
Message 1: Typical 2: Acknowledge choices 3: Universal 4: Personal responsibility Test of significance across messages p
Perceived strength (n ¼ 727) M (SD)
Elicitation of anger (n ¼ 723) M (SD)
Counterarguing (n ¼ 548) Proportion
3.59 (0.79)2,3,4 3.41 (0.79)1 3.44 (0.93)1 3.37 (0.87)1 F(3, 723) ¼ 2.63
2.64 (1.34)2 2.37 (1.21)1,3 2.71 (1.16)2 2.55 (1.28) F(3, 719) ¼ 2.42
0.214 0.254 0.264 0.531,2,3 2 v (3) ¼ 41.4
.049
.065
