- Email: [email protected]
Both bias against disconfirmatory evidence and political orientation partially explain the relationship between dogmatism and racial prejudice
Personality and Individual Differences 105 (2017) 89–94
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Both bias against disconfirmatory evidence and political orientation partially explain the relationship between dogmatism and racial prejudice Michael V. Bronstein ⁎, John F. Dovidio, Tyrone D. Cannon Department of Psychology, Yale University, 2 Hillhouse Ave, New Haven, CT, USA
a r t i c l e
i n f o
Article history: Received 16 June 2016 Received in revised form 17 September 2016 Accepted 20 September 2016 Available online xxxx Keywords: Dogmatism Prejudice BADE Evidence integration Belief revision
a b s t r a c t Background: Dogmatism and prejudice are strongly related. We hypothesized that bias against disconfirmatory evidence (BADE) might explain a portion of this relationship unique from that explained by constructs including Political Orientation. BADE (measured by the BADE task) comprises two facets, Evidence Integration Impairment and Positive Response Bias. Only Evidence Integration Impairment was expected to partially explain the prejudice-dogmatism relationship because in prior research it alone was associated with group differences in inflexible beliefs. Method: 254 MTurk participants completed measures of dogmatism, racial prejudice, BADE (Evidence Integration Impairment and Positive Response Bias), and Political Orientation. The hypothesized mediation effect was examined using a bootstrapping procedure. Results: Dogmatism predicted racial prejudice [b = 0.24, t(249) = 4.92, p b .001]; this relationship weakened in the presence of the above measures [b = 0.05, t(246) = 0.91, p = .363]. The 95% confidence interval for the size of the indirect effect of dogmatism on racial prejudice via Evidence Integration Impairment did not include zero [0.151, 0.331], confirming the hypothesized mediation effect. Conclusions: Evidence Integration Impairment accounts for a unique portion of the relationship between dogmatism and racial prejudice, suggesting that belief revision failures in ambiguous social situations may support prejudice in dogmatic individuals. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Dogmatism, which is defined as “relatively unchangeable, unjustified certainty” (Altemeyer, 2002a), has theoretical and empirical connections to prejudice (e.g., Corlett, 2003; Foley & Chamblin, 1982; Rokeach, 1960). Because dogmatism can account for almost 40% of the variance in prejudice toward Blacks (r = 0.63; Strickland & Weddell, 1972), elucidating the mechanisms underlying these connections has the potential to produce insights into the relationship between individual differences and the fundamental dynamics of prejudice. We therefore examined a novel, cognitively-based explanation for the relationship between dogmatism and prejudice. In his seminal work, The Open and Closed Mind, Rokeach (1960) suggested that dogmatism reflects a deficit in integrating information that threatens multiple prior beliefs into a new belief system. Recent research has demonstrated that more dogmatic individuals show greater persistence of a belief after the evidence that begot that belief has been discredited (Davies, 1993), and often ignore information that does not support their prior beliefs (Davies, 1998). Thus, dogmatic
⁎ Corresponding author. E-mail address: [email protected] (M.V. Bronstein).
http://dx.doi.org/10.1016/j.paid.2016.09.036 0191-8869/© 2016 Elsevier Ltd. All rights reserved.
individuals appear biased against revising their beliefs in response to disconfirmatory evidence. Prejudiced individuals may also exhibit bias against disconfirmatory evidence, as implied by the “rigid, inflexible belief system” that Allport (1954) attributes to prejudiced individuals. Consistent with this notion, in ambiguous situations prejudice is associated with the systematic discounting of evidence inconsistent with biased beliefs. For example, more prejudiced individuals more often perceive hostility in faces produced by blending those typical of happy and angry target group members (Hugenberg & Bodenhausen, 2003). Additionally, more prejudiced individuals judge Black job candidates less favorably than comparable White candidates when the adequacy of candidates' qualifications is ambiguous (Hodson, Dovidio, & Gaertner, 2002). The mutual association of dogmatism and prejudice with beliefrevision deficits is consistent with the notion that these deficits may partially account for the relationship between dogmatism and prejudice. Recent research suggests that bias against disconfirmatory evidence (BADE) consists of two facets, Evidence Integration Impairment and Positive Response Bias (Speechley, Moritz, Ngan, & Woodward, 2012). Evidence Integration Impairment reflects the degree to which people are willing to change their beliefs in the face of potentially valid disambiguating evidence. Positive Response Bias represents the degree to which people endorse beliefs that they perceive as justified;
90
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
Positive Response Bias is a form of general response bias reflecting variation in one's willingness to rate an explanation as highly plausible when justified (Speechley et al., 2012). Moreover, clinical populations with inflexible beliefs (delusions) exhibit greater Evidence Integration Impairment than non-clinical populations or clinical populations with other disorders, but they have comparable levels of Positive Response Bias (Sanford, Veckenstedt, Moritz, Balzan, & Woodward, 2014; Speechley et al., 2012). Given this evidence from clinical populations, we expected that Evidence Integration Impairment would be the primary facet of BADE relevant to the relationship between inflexible beliefs in the general population, including dogmatism and prejudice. This expectation was reinforced by the absence (to our knowledge) of research suggesting that more prejudiced individuals display a larger positive response bias in domains not relevant to prejudice. We therefore sought to test the hypothesis that Evidence Integration Impairment, but not Positive Response Bias, mediates the relationship between dogmatism and prejudice. To this end, we recruited 271 volunteers to complete a survey containing a reliable measure of Evidence Integration Impairment (the bias against disconfirmatory evidence [BADE] task; Woodward, Buchy, Moritz, & Liotti, 2007) along with measures of prejudice (the Attitudes Toward Blacks scale; Brigham, 1993) and Political Orientation (a modified Wilson-Patterson Inventory; Smith, Oxley, Hibbing, Alford, & Hibbing, 2011). We measured Political Orientation because it is a well-documented social factor that might also mediate the relationship between dogmatism and prejudice. Political Orientation (along the dimension of political conservatism-liberalism) could partially account for the relationship between dogmatism and prejudice because more politically conservative people tend to display greater prejudice toward Black individuals and demonstrate higher levels of dogmatism (Jost, Glaser, Kruglanski, & Sulloway, 2003a; Prezza, Zampatti, Pacilli, & Paoliello, 2008). Although we expected that Evidence Integration Impairment and Political Orientation would both mediate the relationship between dogmatism and prejudice, we hypothesized that Evidence Integration Impairment would mediate this relationship beyond any effects of Political Orientation. 2. Method 2.1. Participants Using a $1.00 incentive, we recruited 271 participants through Amazon's Mechanical Turk (MTurk) who were considered reliable respondents (N95% MTurk approval rating). MTurk data collected in this way are comparable to data collected in the laboratory (Johnson & Borden, 2012). We restricted participation to individuals who identified their race as White and lived in the United States because of research suggesting that the association between dogmatism and prejudice varies by race in the United States (Foley & Chamblin, 1982). Participants were excluded from all analyses if they completed this study faster than 95% of the other participants (i.e., faster than 27:48; average completion time = 59:00). Applying this exclusion criterion reduced the sample to 258 individuals. Potential outliers were detected in these individuals' data using studentized residuals with Bonferroni corrected p-values b .05. After list-wise exclusion of 4 individuals whose data may have included outliers, 254 individuals' data remained. Demographic information regarding this final sample can be seen in Table 1. 2.2. Measures Our central measure was an adapted version of the bias against disconfirmatory evidence (BADE) task (see Woodward et al., 2007; Appendix A of Speechley et al., 2012). In each trial of this general cognitive
Table 1 Participants after removal of outlier cases. Age 18–29 30–39 40–49 50+ Sex Male Female Education Some high school High school Some college 2 year college 4 year college Some graduate or professional school Graduate or professional school Total
N 71 89 33 61 87 167 2 41 61 38 76 2 34 254
task, participants viewed three statements about a different fictional individual. For each individual, participants evaluated the plausibility of four explanations for the first of these statements. They repeated this process with the same four explanations after each statement pertaining to that individual was revealed. These explanations could be grouped across trials into four categories (True, Absurd, Lure-A, Lure-B). Absurd explanations remained implausible throughout each trial. Lure-A/Lure-B explanations were the most plausible options initially but became less plausible when the latter two statements were presented to participants. True explanations became the most plausible option by the end of each trial. Each trial of this task begins with multiple equally plausible explanations (the Lure-A 1/Lure-B 1 explanations) for the initial statement but by its end a single explanation (the True explanation) becomes the most plausible. This pattern suggests that the information participants encounter during a trial disambiguates it. The most parsimonious explanation for this disambiguation is that participants revise beliefs about the plausibility of each explanation according to the final two statements they encounter. Thus, it appears likely that behavior changes during the BADE task must at least partially reflect the incorporation of disambiguating information into belief structures. According to the scoring procedure for this task recommended by Sanford et al. (2014), two BADE sub-scores, Evidence Integration Impairment and Positive Response Bias (also known as “Evidence Integration” and “Conservatism,” respectively), derive from the twelve average plausibility ratings given to explanations of each category after each statement was revealed. Evidence Integration Impairment scores depend upon a combination of one's ability to reject implausible response options (the Absurd items on the BADE task) and to change one's plausibility ratings (for True and Lure items) in response to new information. Higher Evidence Integration Impairment scores are indicative of poorer integration ability; these scores can range from negative 10, indicating optimal integration ability, to 50, representing the poorest possible integration ability. Positive Response Bias scores depend upon one's ratings of plausible explanations (Lure and True items) in the first two stages of each trial of the BADE task. Higher Positive Response Bias scores indicate greater willingness to give these explanations high plausibility ratings; these scores can range from zero, indicating an individual's tendency to display a negative response bias, to 60, reflecting the individual's tendency to display the largest possible positive response bias. Our survey also employed three measures not previously used in studies of BADE. The first of these measured dogmatism (the DOG scale; Altemeyer, 2002a). Participants respond to the 20 items in this measure by rating their agreement with a statement on a one to nine scale (1 = strongly disagree, 9 = strongly agree). Dogmatism scores are computed as the sum of the ratings on each item. This measure has high internal consistency (α ≈ 0.90 in Altemeyer, 2002a; in our
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
sample α = 0.93). Its validity has been supported across multiple studies (Altemeyer, 2002a; Crowson, DeBacker, & Davis, 2008). We selected this measure because it was designed to reflect dogmatism across a wide variety of topics and is not specific to a particular political philosophy (Altemeyer, 2002a). The second of these was a measure of racial prejudice, the Attitudes Toward Blacks (the ATB scale; Brigham, 1993). The ATB scale is more reliable than other measures of racial attitudes, including the Modern Racism scale, and is internally consistent (α = 0.88 in Brigham, 1993; in our sample α = 0.94). On each of the 20 items of the ATB scale, participants rated their agreement with a statement from one to seven (1 = strongly agree, 7 = strongly disagree). These ratings were then summed to yield a total score computed such that higher scores were indicative of greater racial prejudice. The last of these measures was a version of the Wilson-Patterson Inventory (WPI) modified to focus exclusively on political issues (Smith et al., 2011). The 21 items in the WPI ask participants whether they agree with, disagree with, or are uncertain regarding a politically charged policy (e.g., small government). Participants scored one point each time they agreed with a policy typically endorsed by political conservatives, zero points if they disagreed, and half a point if they were uncertain. Thus, higher scores reflected greater political conservatism. The WPI is internally consistent (α = 0.74 in Smith et al., 2011; in our sample α = 0.78). The tendency of individuals in the present study with higher scores on this modified inventory to also rate themselves as more conservative on a seven point scale (where 1 = liberal, 7 = conservative), r(252) = 0.78, p b .001, supports this modified inventory's validity as a measure of Political Orientation. 2.3. Procedure Participants were recruited using Amazon's MTurk. After giving informed consent in accordance with the Yale University Institutional Review Board, participants in this correlational study completed a survey containing the above measures (in random order) and were then debriefed. 2.4. Analysis We tested whether Political Orientation, Evidence Integration Impairment, and Positive Response Bias partially mediated the relationship between dogmatism and Attitudes Toward Blacks using a parallel mediation model (Model 4) in the PROCESS macro for SPSS (Preacher & Hays, 2008). In the course of this test, we bootstrapped our data 5000 times and produced a bias-corrected 95% confidence interval (CI) for the indirect effect of dogmatism on Attitudes Toward Blacks through Evidence Integration Impairment, Positive Response Bias, and Political Orientation (considered simultaneously). Confidence intervals that do not include zero indicate a statistically significant effect. Age, sex, and level of education were controlled for in all analyses because of their relationship with the variables of interest (e.g., Case, Greeley, & Fuchs, 1989; Ekehammar, Akrami, & Araya, 2003; Stewart, von Hippel, & Radvansky, 2009). 3. Results Descriptive statistics for measures of dogmatism, racial prejudice, Evidence Integration Impairment, and Positive Response Bias are presented in Table 2. Zero-order correlations between these measures are presented in Table 3. Employing a simultaneous mediation model involving these measures yielded support for a direct path between dogmatism and Attitudes Toward Blacks (path c), b = 0.24, t(249) = 4.92, p b .001. Support was also found for the possibility of indirect paths from dogmatism to Attitudes Toward Blacks via both Evidence Integration Impairment and Political Orientation. As hypothesized, dogmatism predicted Evidence Integration Impairment (path a1), b = 0.03, t(249) =
91
Table 2 Descriptive statistics regarding measures of interest.
Dogmatism Attitudes Toward Blacks Evidence Integration Impairment Political Orientation Positive Response Bias
M
SD
76.72 52.41 −1.27 8.94 25.98
29.14 23.19 6.51 3.86 7.91
2.34, p = .020. Evidence Integration Impairment in turn predicted Attitudes Toward Blacks beyond any meditational effects of Political Orientation (path b1), b = 0.55, t(246) = 2.59, p = .010. Dogmatism also predicted Political Orientation (path a2), b = 0.07, t(249) = 10.39, p b .001. Political Orientation in turn predicted, beyond the effects of Evidence Integration Impairment, Attitudes Toward Blacks (path b2), b = 2.37, t(246) = 5.75, p b .001. These results suggest that the indirect paths from dogmatism to Attitudes Toward Blacks via both Evidence Integration Impairment and Political Orientation were potentially viable. In contrast, no support was found for a potential indirect path through Positive Response Bias. As expected, dogmatism did not predict Positive Response Bias (path a3), b = 0.02, t(249) = 1.10, p = .313, and Positive Response Bias did not predict Attitudes Toward Blacks (path b3), b = −0.08, t(246) = 0.47, p = .641. The pattern of these results is consistent with the possibility that Evidence Integration Impairment and Political Orientation, but not Positive Response Bias, account for unique variance in the relationship between dogmatism and Attitudes Toward Blacks. Accordingly, we tested whether direct path strength decreased in the presence of potential mediators. Critically, the direct path from dogmatism to Attitudes Toward Blacks was enervated when these potential mediators (and Positive Response Bias) were included in the prediction model (path c′; b = 0.05, t(246) = 0.91, p = .363) compared to when they were not included. The significance of this decrease was confirmed by a completely standardized 95% CI [0.151, 0.331] for the indirect effect of dogmatism on Attitudes Toward Blacks that did not include zero. This result suggests that the combination of candidate mediators tested completely accounted for the relationship between dogmatism and Attitudes Toward Blacks. Examining the effects of individual potential mediators revealed that, as hypothesized, the indirect path relating dogmatism to Attitudes Toward Blacks via Evidence Integration Impairment was significant, 95% CI [0.003 0.063]. Similarly, the indirect path through Political Orientation was significant [0.132 0.303]. In contrast, also as hypothesized, the indirect path relating dogmatism to Attitudes Toward Blacks via Positive Response Bias was not significant, [−0.020 0.004]. The significance of both Political Orientation and Evidence Integration Impairment as parallel mediators supports the hypothesis that Evidence Integration Impairment mediates (and accounts for a unique portion of) the relationship between dogmatism and Attitudes Toward Blacks. The results of this mediation model are depicted in Fig. 1. Statistics describing the overall model for each path, as well as those describing the variables controlled for, can be found in Table 4. We examined two additional models. Given that Positive Response Bias was not a significant mediator, we examined whether our results would change when excluding Positive Response Bias from the parallel mediation model. This alteration did not change these results (see Supplement, Section S1.0). We also examined a serial mediation model with an indirect path via Evidence Integration Impairment and Political Orientation (see Hodson & Busseri, 2012). This model (included in Supplementary material Section S2.0) revealed no serial mediation effect, 95% CI [−0.012 0.001]. 4. Discussion Our results support the hypothesis that a specific facet of bias against disconfirmatory evidence (BADE), Evidence Integration Impairment,
92
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
Table 3 Zero-order correlations between measures of interest. Dogmatism Dogmatism ATB Evidence Integration Impairment Political Orientation Positive Response Bias
ATB
Evidence Integration Impairment
Political Orientation
Positive Response Bias
0.29⁎⁎
0.14⁎ 0.18⁎⁎
0.55⁎⁎ 0.40⁎⁎ 0.01
0.06 0.02 0.24⁎⁎ 0.03
Note. ATB = Attitudes Toward Blacks. ⁎ p b .05. ⁎⁎ p b .01.
helps account for the relationship between dogmatism and racial prejudice among Whites. Consistent with previous research (see Jost, Glaser, Kruglanski, & Sulloway, 2003b for a review), participants who were more politically conservative displayed greater racial prejudice and higher levels of dogmatism. However, while Political Orientation also partially accounted for the relationship between dogmatism and prejudice, its effects in this regard were independent from that of Evidence Integration Impairment and appeared, unexpectedly, to be driven by more liberal individuals (see Supplement, Section S4.0). Empirically, Evidence Integration Impairment and Political Orientation were uncorrelated, and both significantly mediated the relationship between dogmatism and prejudice when considered simultaneously. Conceptually, whereas Evidence Integration Impairment represents a cognitive mechanism, Political Orientation involves a range of elements, including epistemic motivations related to resistance to change and the endorsement of inequality (Jost et al., 2003b). Evidence Integration Impairment is thus likely one of many factors that contribute to the relationship between dogmatism and prejudice, as the independent effect of Political Orientation and the modest correlation between dogmatism and Evidence Integration Impairment (r = 0.14, Table 2) that we observed suggest.
This caveat notwithstanding, these findings extend prior research linking dogmatism with failures of belief revision generally (Davies, 1993, 1998) by identifying a specific facet of belief revision ability– Evidence Integration–that may be impaired in dogmatic individuals. Given the aforementioned modest correlation between dogmatism and Evidence Integration Impairment and given that dogmatic individuals display deficits at stages of belief revision conceptually distinct from Evidence Integration Impairment (Davies, 1993, 1998), we conceptualize dogmatism and Evidence Integration Impairment as overlapping but unique constructs. Future research could further evaluate this conceptualization by examining whether factors (e.g., liberal acceptance and hypersalient evidence-hypothesis matches) theorized to produce Evidence Integration Impairment in other populations (see Sanford et al., 2014) do so in dogmatic individuals as well. Future research might also consider how person- and situationbased factors may jointly shape prejudice (Hodson & Dhont, 2015). Because Evidence Integration reflects the ability to incorporate disambiguating evidence into preexisting belief structures (Speechley et al., 2012), such research could examine whether impairments in Evidence Integration are particularly likely to lead to greater racial biases in situations in which more complex and potentially conflicting information must be considered (e.g., in decisions about hiring or promoting candidates for a position).
Table 4 Mediation models for the relationship between dogmatism and Attitudes Toward Blacks. Path
Subject variable
b
Standard error
t
a1: Evidence Integration Impairment Sex Educ. Age
−0.88 0.35 −0.07
0.85 0.30 0.03
1.03 1.16 2.28⁎
Sex Educ. Age
0.90 −0.29 0.02
0.42 0.15 0.02
2.13⁎ 1.92 1.14
Sex Educ. Age
0.80 0.41 0.01
1.05 0.38 0.04
0.77 1.07 0.23
Sex Educ. Age
−6.66 0.75 −0.13
2.78 0.99 0.11
2.39⁎ 0.76 1.24
Sex Educ. Age
−5.07 0.23 −0.13
2.94 1.05 0.11
1.72 0.22 1.16
a2: Political Orientation
a3: Positive Response Bias
b1.2.3 and c′
c
Fig. 1. Mediation model with paths labelled according to Baron and Kenny (1986). Dogmat. = dogmatism. Evidence Integrat. = Evidence Integration Impairment. Political Orient. = Political Orientation. P.R.B. = Positive Response Bias. This model depicts full mediation of the relationship between dogmatism and Attitudes Toward Blacks by the combination of Evidence Integration Impairment and Political Orientation.
F
R2
3.27⁎
0.05
30.39⁎⁎⁎
0.33
0.66
0.01
10.15⁎⁎⁎
0.22
7.15⁎⁎⁎
0.10
Note: Educ. = level of education achieved. Positive coefficients for sex indicate that being female predicts a higher score. F-values represent the significance of the path as a whole, including the main predictors of interest. Predictors in bold were significant. Paths were named according to Baron and Kenny (1986). ⁎ p b .05. ⁎⁎⁎ p b .001.
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
Additional work could examine how dogmatism and related impairments in Evidence Integration might moderate the effectiveness of interventions on prejudice. For example, prejudice can be reduced through positive contact with members of target groups (Pettigrew & Tropp, 2006). However, more dogmatic individuals may benefit less from this intergroup contact for two reasons. First, these individuals may be less likely to engage in intergroup contact because they maintain greater social distance from other groups (Kirtley & Harkless, 1969; Orpen & Rookledge, 1972). Second, even when more dogmatic individuals engage in positive intergroup contact, they might be less able to integrate this positive experience into their belief structures and thereby revise the way they view members of other groups. Interventions that address Evidence Integration deficits (or BADE generally) may therefore make subsequent intergroup contact more effective in highly dogmatic individuals. Metacognitive Training (MCT), a recently developed therapy, contains a module to reduce BADE that might be used to realize this putative benefit (Moritz & Woodward, 2007). Targeting BADE, rather than dogmatism, might more efficiently reduce prejudice in dogmatic individuals because dogmatism is both a more general personality trait and is characterized by resistance to change. Future research might test whether an intervention using this module of MCT would enhance the impact of positive intergroup contact for reducing prejudice, especially for more dogmatic individuals. Despite these important implications, this study has several limitations. First, we examined the prejudice of Whites toward Blacks. Reversing the races of our target group and perpetrators of bias may have changed our results (see Foley & Chamblin, 1982). It therefore remains to be seen whether our results generalize to prejudices held by other groups. It also remains to be seen whether our results generalize to other prejudices held by Whites. Given the particular importance of individual differences for predicting generalized prejudice (Hodson & Dhont, 2015), we expect our results to generalize in this manner (e.g., to prejudice against gay/lesbian individuals; see Altemeyer, 2002b; Altemeyer & Hunsberger, 1992). Second, our results derive from a single study. Mitigating this limitation, a preliminary study that did not include Political Orientation replicated the pattern of zero-order correlations and provided some support for the indirect path through Evidence Integration Impairment reported above. Results of this replication study can be seen in Supplementary material, Section S3.0. Future research should strive to address these limitations by replicating our findings using other populations or while measuring prejudice toward other groups. 5. Conclusion Through a novel application of Woodward, Moritz, Cuttler, and Whitman's (2006) BADE task, we procured a pattern of results suggesting that Evidence Integration Impairment accounts for a unique portion of the relationship between dogmatism and prejudice. This pattern suggests that failures of belief revision in ambiguous social situations may contribute to the maintenance of prejudice in dogmatic individuals. This suggestion lays a theoretical foundation for experimental research on whether or not aspects of belief revision failure might support individual differences in response to interventions on prejudice. Such research should contribute to our understanding of the bulwarks of prejudice, and thus holds out the promise of leading toward the development of more effective interventions on prejudiced beliefs. Acknowledgement The authors declare no potential conflicts of interest regarding this article. The authors thank Dr. Arielle Baskin-Sommers for contributing data processing scripts, and Dr. Todd Woodward and colleagues for providing us with their BADE task. We thank two anonymous reviewers for
93
providing us with useful feedback on an earlier version of this manuscript. The authors received no financial support for the production of this article. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.paid.2016.09.036. References Allport, G. W. (1954). The nature of prejudice. Cambridge, MA: Addison-Wesley. Altemeyer, B. (2002a). Dogmatic behavior among students: Testing a new measure of dogmatism. The Journal of Social Psychology, 142(6), 713–721. Altemeyer, B. (2002b). Changes in attitudes toward homosexuals. Journal of Homosexuality, 42(2), 63–75. Altemeyer, B., & Hunsberger, B. (1992). Authoritarianism, religious fundamentalism, quest, and prejudice. The International Journal for the Psychology of Religion, 2(2), 113–133. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182. Brigham, J. C. (1993). College students' racial attitudes. Journal of Applied Social Psychology, 23(23), 1933–1967. Case, C. E., Greeley, A. M., & Fuchs, S. (1989). Social determinants of racial prejudice. Sociological Perspectives, 32(4), 469–483. Corlett, J. A. (2003). Race, racism, and reparations. Cornell University Press. Crowson, H. M., DeBacker, T. K., & Davis, K. A. (2008). The DOG Scale: A valid measure of dogmatism? Journal of Individual Differences, 29(1), 17–24. Davies, M. E. (1993). Dogmatism and the persistence of discredited beliefs. Personality and Social Psychology Bulletin, 19, 692–699. Davies, M. (1998). Dogmatism and belief formation: Output interference in the processing of supporting and contradictory cognitions. Journal of Personality and Social Psychology, 75, 456–466. Ekehammar, B., Akrami, N., & Araya, T. (2003). Gender differences in implicit prejudice. Personality and Individual Differences, 34(8), 1509–1523. Foley, L. A., & Chamblin, M. H. (1982). The effect of race and personality on mock jurors' decisions. The Journal of Psychology, 112(1), 47–51. Hodson, G., & Busseri, M. (2012). Bright minds and dark attitudes: Lower cognitive ability predicts greater prejudice through right-wing ideology and low intergroup contact. Psychological Science, 23, 187–195. Hodson, G., & Dhont, K. (2015). The person-based nature of prejudice: Individual difference predictors of intergroup negativity. European Review of Social Psychology, 26(1), 1–42. Hodson, G., Dovidio, J. F., & Gaertner, S. L. (2002). Processes in racial discrimination: Differential weighting of conflicting information. Personality and Social Psychology Bulletin, 28(4), 460–471. Hugenberg, K., & Bodenhausen, G. V. (2003). Facing prejudice implicit prejudice and the perception of facial threat. Psychological Science, 14(6), 640–643. Johnson, D. R., & Borden, L. A. (2012). Participants at your fingertips using Amazon's Mechanical Turk to increase student–faculty collaborative research. Teaching of Psychology, 39(4), 245–251. Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003a). Exceptions that prove the rule—Using a theory of motivated social cognition to account for ideological incongruities and political anomalies: Reply to Greenberg and Jonas (2003). Psychological Bulletin, 129, 383–393. Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003b). Political conservatism as motivated social cognition. Psychological Bulletin, 129(3), 339–375. Kirtley, D., & Harkless, R. (1969). Some personality and attitudinal correlates of dogmatism. Psychological Reports, 24(3), 851–854. Moritz, S., & Woodward, T. S. (2007). Metacognitive training in schizophrenia: From basic research to knowledge translation and intervention. Current Opinion in Psychiatry, 20, 619–625. Orpen, C., & Rookledge, Q. (1972). Dogmatism and prejudice in White South Africa. The Journal of Social Psychology, 86(1), 151–153. Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90(5), 751–783. Preacher, K. J., & Hays, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891. Prezza, M., Zampatti, E., Pacilli, M. G., & Paoliello, A. (2008). Territorial sense of community, ethnic prejudice and political orientation. Journal of Community and Applied Social Psychology, 18(4), 315–332. Rokeach, M. (1960). The open and closed mind. New York, NY: Basic Books. Sanford, N., Veckenstedt, R., Moritz, S., Balzan, R. P., & Woodward, T. S. (2014). Impaired integration of disambiguating evidence in delusional schizophrenia patients. Psychological Medicine, 44(13), 2729–2738. Smith, K. B., Oxley, D. R., Hibbing, M. V., Alford, J. R., & Hibbing, J. R. (2011). Linking genetics and political attitudes: Reconceptualizing political ideology. Political Psychology, 32(3), 369–397. Speechley, W. J., Moritz, S., Ngan, E. T. C., & Woodward, T. S. (2012). Impaired evidence integration and delusions in schizophrenia. Journal of Experimental Psychology, 3, 688–701.
94
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
Stewart, B. D., von Hippel, W., & Radvansky, G. A. (2009). Age, race, and implicit prejudice using process dissociation to separate the underlying components. Psychological Science, 20(2), 164–168. Strickland, B. R., & Weddell, S. C. (1972). Religious orientation, racial prejudice, and dogmatism: A study of Baptists and Unitarians. Journal for the Scientific Study of Religion, 11, 395–399.
Woodward, T. S., Moritz, S., Cuttler, C., & Whitman, J. C. (2006). The contribution of a cognitive bias against disconfirmatory evidence (BADE) to delusions in schizophrenia. Journal of Clinical and Experimental Neuropsychology, 28(4), 605–617. Woodward, T. S., Buchy, L., Moritz, S., & Liotti, M. (2007). A bias against disconfirmatory evidence is associated with delusion proneness in a nonclinical sample. Schizophrenia Bulletin, 33(4), 1023–1028.
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Both bias against disconfirmatory evidence and political orientation partially explain the relationship between dogmatism and racial prejudice Michael V. Bronstein ⁎, John F. Dovidio, Tyrone D. Cannon Department of Psychology, Yale University, 2 Hillhouse Ave, New Haven, CT, USA
a r t i c l e
i n f o
Article history: Received 16 June 2016 Received in revised form 17 September 2016 Accepted 20 September 2016 Available online xxxx Keywords: Dogmatism Prejudice BADE Evidence integration Belief revision
a b s t r a c t Background: Dogmatism and prejudice are strongly related. We hypothesized that bias against disconfirmatory evidence (BADE) might explain a portion of this relationship unique from that explained by constructs including Political Orientation. BADE (measured by the BADE task) comprises two facets, Evidence Integration Impairment and Positive Response Bias. Only Evidence Integration Impairment was expected to partially explain the prejudice-dogmatism relationship because in prior research it alone was associated with group differences in inflexible beliefs. Method: 254 MTurk participants completed measures of dogmatism, racial prejudice, BADE (Evidence Integration Impairment and Positive Response Bias), and Political Orientation. The hypothesized mediation effect was examined using a bootstrapping procedure. Results: Dogmatism predicted racial prejudice [b = 0.24, t(249) = 4.92, p b .001]; this relationship weakened in the presence of the above measures [b = 0.05, t(246) = 0.91, p = .363]. The 95% confidence interval for the size of the indirect effect of dogmatism on racial prejudice via Evidence Integration Impairment did not include zero [0.151, 0.331], confirming the hypothesized mediation effect. Conclusions: Evidence Integration Impairment accounts for a unique portion of the relationship between dogmatism and racial prejudice, suggesting that belief revision failures in ambiguous social situations may support prejudice in dogmatic individuals. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Dogmatism, which is defined as “relatively unchangeable, unjustified certainty” (Altemeyer, 2002a), has theoretical and empirical connections to prejudice (e.g., Corlett, 2003; Foley & Chamblin, 1982; Rokeach, 1960). Because dogmatism can account for almost 40% of the variance in prejudice toward Blacks (r = 0.63; Strickland & Weddell, 1972), elucidating the mechanisms underlying these connections has the potential to produce insights into the relationship between individual differences and the fundamental dynamics of prejudice. We therefore examined a novel, cognitively-based explanation for the relationship between dogmatism and prejudice. In his seminal work, The Open and Closed Mind, Rokeach (1960) suggested that dogmatism reflects a deficit in integrating information that threatens multiple prior beliefs into a new belief system. Recent research has demonstrated that more dogmatic individuals show greater persistence of a belief after the evidence that begot that belief has been discredited (Davies, 1993), and often ignore information that does not support their prior beliefs (Davies, 1998). Thus, dogmatic
⁎ Corresponding author. E-mail address: [email protected] (M.V. Bronstein).
http://dx.doi.org/10.1016/j.paid.2016.09.036 0191-8869/© 2016 Elsevier Ltd. All rights reserved.
individuals appear biased against revising their beliefs in response to disconfirmatory evidence. Prejudiced individuals may also exhibit bias against disconfirmatory evidence, as implied by the “rigid, inflexible belief system” that Allport (1954) attributes to prejudiced individuals. Consistent with this notion, in ambiguous situations prejudice is associated with the systematic discounting of evidence inconsistent with biased beliefs. For example, more prejudiced individuals more often perceive hostility in faces produced by blending those typical of happy and angry target group members (Hugenberg & Bodenhausen, 2003). Additionally, more prejudiced individuals judge Black job candidates less favorably than comparable White candidates when the adequacy of candidates' qualifications is ambiguous (Hodson, Dovidio, & Gaertner, 2002). The mutual association of dogmatism and prejudice with beliefrevision deficits is consistent with the notion that these deficits may partially account for the relationship between dogmatism and prejudice. Recent research suggests that bias against disconfirmatory evidence (BADE) consists of two facets, Evidence Integration Impairment and Positive Response Bias (Speechley, Moritz, Ngan, & Woodward, 2012). Evidence Integration Impairment reflects the degree to which people are willing to change their beliefs in the face of potentially valid disambiguating evidence. Positive Response Bias represents the degree to which people endorse beliefs that they perceive as justified;
90
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
Positive Response Bias is a form of general response bias reflecting variation in one's willingness to rate an explanation as highly plausible when justified (Speechley et al., 2012). Moreover, clinical populations with inflexible beliefs (delusions) exhibit greater Evidence Integration Impairment than non-clinical populations or clinical populations with other disorders, but they have comparable levels of Positive Response Bias (Sanford, Veckenstedt, Moritz, Balzan, & Woodward, 2014; Speechley et al., 2012). Given this evidence from clinical populations, we expected that Evidence Integration Impairment would be the primary facet of BADE relevant to the relationship between inflexible beliefs in the general population, including dogmatism and prejudice. This expectation was reinforced by the absence (to our knowledge) of research suggesting that more prejudiced individuals display a larger positive response bias in domains not relevant to prejudice. We therefore sought to test the hypothesis that Evidence Integration Impairment, but not Positive Response Bias, mediates the relationship between dogmatism and prejudice. To this end, we recruited 271 volunteers to complete a survey containing a reliable measure of Evidence Integration Impairment (the bias against disconfirmatory evidence [BADE] task; Woodward, Buchy, Moritz, & Liotti, 2007) along with measures of prejudice (the Attitudes Toward Blacks scale; Brigham, 1993) and Political Orientation (a modified Wilson-Patterson Inventory; Smith, Oxley, Hibbing, Alford, & Hibbing, 2011). We measured Political Orientation because it is a well-documented social factor that might also mediate the relationship between dogmatism and prejudice. Political Orientation (along the dimension of political conservatism-liberalism) could partially account for the relationship between dogmatism and prejudice because more politically conservative people tend to display greater prejudice toward Black individuals and demonstrate higher levels of dogmatism (Jost, Glaser, Kruglanski, & Sulloway, 2003a; Prezza, Zampatti, Pacilli, & Paoliello, 2008). Although we expected that Evidence Integration Impairment and Political Orientation would both mediate the relationship between dogmatism and prejudice, we hypothesized that Evidence Integration Impairment would mediate this relationship beyond any effects of Political Orientation. 2. Method 2.1. Participants Using a $1.00 incentive, we recruited 271 participants through Amazon's Mechanical Turk (MTurk) who were considered reliable respondents (N95% MTurk approval rating). MTurk data collected in this way are comparable to data collected in the laboratory (Johnson & Borden, 2012). We restricted participation to individuals who identified their race as White and lived in the United States because of research suggesting that the association between dogmatism and prejudice varies by race in the United States (Foley & Chamblin, 1982). Participants were excluded from all analyses if they completed this study faster than 95% of the other participants (i.e., faster than 27:48; average completion time = 59:00). Applying this exclusion criterion reduced the sample to 258 individuals. Potential outliers were detected in these individuals' data using studentized residuals with Bonferroni corrected p-values b .05. After list-wise exclusion of 4 individuals whose data may have included outliers, 254 individuals' data remained. Demographic information regarding this final sample can be seen in Table 1. 2.2. Measures Our central measure was an adapted version of the bias against disconfirmatory evidence (BADE) task (see Woodward et al., 2007; Appendix A of Speechley et al., 2012). In each trial of this general cognitive
Table 1 Participants after removal of outlier cases. Age 18–29 30–39 40–49 50+ Sex Male Female Education Some high school High school Some college 2 year college 4 year college Some graduate or professional school Graduate or professional school Total
N 71 89 33 61 87 167 2 41 61 38 76 2 34 254
task, participants viewed three statements about a different fictional individual. For each individual, participants evaluated the plausibility of four explanations for the first of these statements. They repeated this process with the same four explanations after each statement pertaining to that individual was revealed. These explanations could be grouped across trials into four categories (True, Absurd, Lure-A, Lure-B). Absurd explanations remained implausible throughout each trial. Lure-A/Lure-B explanations were the most plausible options initially but became less plausible when the latter two statements were presented to participants. True explanations became the most plausible option by the end of each trial. Each trial of this task begins with multiple equally plausible explanations (the Lure-A 1/Lure-B 1 explanations) for the initial statement but by its end a single explanation (the True explanation) becomes the most plausible. This pattern suggests that the information participants encounter during a trial disambiguates it. The most parsimonious explanation for this disambiguation is that participants revise beliefs about the plausibility of each explanation according to the final two statements they encounter. Thus, it appears likely that behavior changes during the BADE task must at least partially reflect the incorporation of disambiguating information into belief structures. According to the scoring procedure for this task recommended by Sanford et al. (2014), two BADE sub-scores, Evidence Integration Impairment and Positive Response Bias (also known as “Evidence Integration” and “Conservatism,” respectively), derive from the twelve average plausibility ratings given to explanations of each category after each statement was revealed. Evidence Integration Impairment scores depend upon a combination of one's ability to reject implausible response options (the Absurd items on the BADE task) and to change one's plausibility ratings (for True and Lure items) in response to new information. Higher Evidence Integration Impairment scores are indicative of poorer integration ability; these scores can range from negative 10, indicating optimal integration ability, to 50, representing the poorest possible integration ability. Positive Response Bias scores depend upon one's ratings of plausible explanations (Lure and True items) in the first two stages of each trial of the BADE task. Higher Positive Response Bias scores indicate greater willingness to give these explanations high plausibility ratings; these scores can range from zero, indicating an individual's tendency to display a negative response bias, to 60, reflecting the individual's tendency to display the largest possible positive response bias. Our survey also employed three measures not previously used in studies of BADE. The first of these measured dogmatism (the DOG scale; Altemeyer, 2002a). Participants respond to the 20 items in this measure by rating their agreement with a statement on a one to nine scale (1 = strongly disagree, 9 = strongly agree). Dogmatism scores are computed as the sum of the ratings on each item. This measure has high internal consistency (α ≈ 0.90 in Altemeyer, 2002a; in our
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
sample α = 0.93). Its validity has been supported across multiple studies (Altemeyer, 2002a; Crowson, DeBacker, & Davis, 2008). We selected this measure because it was designed to reflect dogmatism across a wide variety of topics and is not specific to a particular political philosophy (Altemeyer, 2002a). The second of these was a measure of racial prejudice, the Attitudes Toward Blacks (the ATB scale; Brigham, 1993). The ATB scale is more reliable than other measures of racial attitudes, including the Modern Racism scale, and is internally consistent (α = 0.88 in Brigham, 1993; in our sample α = 0.94). On each of the 20 items of the ATB scale, participants rated their agreement with a statement from one to seven (1 = strongly agree, 7 = strongly disagree). These ratings were then summed to yield a total score computed such that higher scores were indicative of greater racial prejudice. The last of these measures was a version of the Wilson-Patterson Inventory (WPI) modified to focus exclusively on political issues (Smith et al., 2011). The 21 items in the WPI ask participants whether they agree with, disagree with, or are uncertain regarding a politically charged policy (e.g., small government). Participants scored one point each time they agreed with a policy typically endorsed by political conservatives, zero points if they disagreed, and half a point if they were uncertain. Thus, higher scores reflected greater political conservatism. The WPI is internally consistent (α = 0.74 in Smith et al., 2011; in our sample α = 0.78). The tendency of individuals in the present study with higher scores on this modified inventory to also rate themselves as more conservative on a seven point scale (where 1 = liberal, 7 = conservative), r(252) = 0.78, p b .001, supports this modified inventory's validity as a measure of Political Orientation. 2.3. Procedure Participants were recruited using Amazon's MTurk. After giving informed consent in accordance with the Yale University Institutional Review Board, participants in this correlational study completed a survey containing the above measures (in random order) and were then debriefed. 2.4. Analysis We tested whether Political Orientation, Evidence Integration Impairment, and Positive Response Bias partially mediated the relationship between dogmatism and Attitudes Toward Blacks using a parallel mediation model (Model 4) in the PROCESS macro for SPSS (Preacher & Hays, 2008). In the course of this test, we bootstrapped our data 5000 times and produced a bias-corrected 95% confidence interval (CI) for the indirect effect of dogmatism on Attitudes Toward Blacks through Evidence Integration Impairment, Positive Response Bias, and Political Orientation (considered simultaneously). Confidence intervals that do not include zero indicate a statistically significant effect. Age, sex, and level of education were controlled for in all analyses because of their relationship with the variables of interest (e.g., Case, Greeley, & Fuchs, 1989; Ekehammar, Akrami, & Araya, 2003; Stewart, von Hippel, & Radvansky, 2009). 3. Results Descriptive statistics for measures of dogmatism, racial prejudice, Evidence Integration Impairment, and Positive Response Bias are presented in Table 2. Zero-order correlations between these measures are presented in Table 3. Employing a simultaneous mediation model involving these measures yielded support for a direct path between dogmatism and Attitudes Toward Blacks (path c), b = 0.24, t(249) = 4.92, p b .001. Support was also found for the possibility of indirect paths from dogmatism to Attitudes Toward Blacks via both Evidence Integration Impairment and Political Orientation. As hypothesized, dogmatism predicted Evidence Integration Impairment (path a1), b = 0.03, t(249) =
91
Table 2 Descriptive statistics regarding measures of interest.
Dogmatism Attitudes Toward Blacks Evidence Integration Impairment Political Orientation Positive Response Bias
M
SD
76.72 52.41 −1.27 8.94 25.98
29.14 23.19 6.51 3.86 7.91
2.34, p = .020. Evidence Integration Impairment in turn predicted Attitudes Toward Blacks beyond any meditational effects of Political Orientation (path b1), b = 0.55, t(246) = 2.59, p = .010. Dogmatism also predicted Political Orientation (path a2), b = 0.07, t(249) = 10.39, p b .001. Political Orientation in turn predicted, beyond the effects of Evidence Integration Impairment, Attitudes Toward Blacks (path b2), b = 2.37, t(246) = 5.75, p b .001. These results suggest that the indirect paths from dogmatism to Attitudes Toward Blacks via both Evidence Integration Impairment and Political Orientation were potentially viable. In contrast, no support was found for a potential indirect path through Positive Response Bias. As expected, dogmatism did not predict Positive Response Bias (path a3), b = 0.02, t(249) = 1.10, p = .313, and Positive Response Bias did not predict Attitudes Toward Blacks (path b3), b = −0.08, t(246) = 0.47, p = .641. The pattern of these results is consistent with the possibility that Evidence Integration Impairment and Political Orientation, but not Positive Response Bias, account for unique variance in the relationship between dogmatism and Attitudes Toward Blacks. Accordingly, we tested whether direct path strength decreased in the presence of potential mediators. Critically, the direct path from dogmatism to Attitudes Toward Blacks was enervated when these potential mediators (and Positive Response Bias) were included in the prediction model (path c′; b = 0.05, t(246) = 0.91, p = .363) compared to when they were not included. The significance of this decrease was confirmed by a completely standardized 95% CI [0.151, 0.331] for the indirect effect of dogmatism on Attitudes Toward Blacks that did not include zero. This result suggests that the combination of candidate mediators tested completely accounted for the relationship between dogmatism and Attitudes Toward Blacks. Examining the effects of individual potential mediators revealed that, as hypothesized, the indirect path relating dogmatism to Attitudes Toward Blacks via Evidence Integration Impairment was significant, 95% CI [0.003 0.063]. Similarly, the indirect path through Political Orientation was significant [0.132 0.303]. In contrast, also as hypothesized, the indirect path relating dogmatism to Attitudes Toward Blacks via Positive Response Bias was not significant, [−0.020 0.004]. The significance of both Political Orientation and Evidence Integration Impairment as parallel mediators supports the hypothesis that Evidence Integration Impairment mediates (and accounts for a unique portion of) the relationship between dogmatism and Attitudes Toward Blacks. The results of this mediation model are depicted in Fig. 1. Statistics describing the overall model for each path, as well as those describing the variables controlled for, can be found in Table 4. We examined two additional models. Given that Positive Response Bias was not a significant mediator, we examined whether our results would change when excluding Positive Response Bias from the parallel mediation model. This alteration did not change these results (see Supplement, Section S1.0). We also examined a serial mediation model with an indirect path via Evidence Integration Impairment and Political Orientation (see Hodson & Busseri, 2012). This model (included in Supplementary material Section S2.0) revealed no serial mediation effect, 95% CI [−0.012 0.001]. 4. Discussion Our results support the hypothesis that a specific facet of bias against disconfirmatory evidence (BADE), Evidence Integration Impairment,
92
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
Table 3 Zero-order correlations between measures of interest. Dogmatism Dogmatism ATB Evidence Integration Impairment Political Orientation Positive Response Bias
ATB
Evidence Integration Impairment
Political Orientation
Positive Response Bias
0.29⁎⁎
0.14⁎ 0.18⁎⁎
0.55⁎⁎ 0.40⁎⁎ 0.01
0.06 0.02 0.24⁎⁎ 0.03
Note. ATB = Attitudes Toward Blacks. ⁎ p b .05. ⁎⁎ p b .01.
helps account for the relationship between dogmatism and racial prejudice among Whites. Consistent with previous research (see Jost, Glaser, Kruglanski, & Sulloway, 2003b for a review), participants who were more politically conservative displayed greater racial prejudice and higher levels of dogmatism. However, while Political Orientation also partially accounted for the relationship between dogmatism and prejudice, its effects in this regard were independent from that of Evidence Integration Impairment and appeared, unexpectedly, to be driven by more liberal individuals (see Supplement, Section S4.0). Empirically, Evidence Integration Impairment and Political Orientation were uncorrelated, and both significantly mediated the relationship between dogmatism and prejudice when considered simultaneously. Conceptually, whereas Evidence Integration Impairment represents a cognitive mechanism, Political Orientation involves a range of elements, including epistemic motivations related to resistance to change and the endorsement of inequality (Jost et al., 2003b). Evidence Integration Impairment is thus likely one of many factors that contribute to the relationship between dogmatism and prejudice, as the independent effect of Political Orientation and the modest correlation between dogmatism and Evidence Integration Impairment (r = 0.14, Table 2) that we observed suggest.
This caveat notwithstanding, these findings extend prior research linking dogmatism with failures of belief revision generally (Davies, 1993, 1998) by identifying a specific facet of belief revision ability– Evidence Integration–that may be impaired in dogmatic individuals. Given the aforementioned modest correlation between dogmatism and Evidence Integration Impairment and given that dogmatic individuals display deficits at stages of belief revision conceptually distinct from Evidence Integration Impairment (Davies, 1993, 1998), we conceptualize dogmatism and Evidence Integration Impairment as overlapping but unique constructs. Future research could further evaluate this conceptualization by examining whether factors (e.g., liberal acceptance and hypersalient evidence-hypothesis matches) theorized to produce Evidence Integration Impairment in other populations (see Sanford et al., 2014) do so in dogmatic individuals as well. Future research might also consider how person- and situationbased factors may jointly shape prejudice (Hodson & Dhont, 2015). Because Evidence Integration reflects the ability to incorporate disambiguating evidence into preexisting belief structures (Speechley et al., 2012), such research could examine whether impairments in Evidence Integration are particularly likely to lead to greater racial biases in situations in which more complex and potentially conflicting information must be considered (e.g., in decisions about hiring or promoting candidates for a position).
Table 4 Mediation models for the relationship between dogmatism and Attitudes Toward Blacks. Path
Subject variable
b
Standard error
t
a1: Evidence Integration Impairment Sex Educ. Age
−0.88 0.35 −0.07
0.85 0.30 0.03
1.03 1.16 2.28⁎
Sex Educ. Age
0.90 −0.29 0.02
0.42 0.15 0.02
2.13⁎ 1.92 1.14
Sex Educ. Age
0.80 0.41 0.01
1.05 0.38 0.04
0.77 1.07 0.23
Sex Educ. Age
−6.66 0.75 −0.13
2.78 0.99 0.11
2.39⁎ 0.76 1.24
Sex Educ. Age
−5.07 0.23 −0.13
2.94 1.05 0.11
1.72 0.22 1.16
a2: Political Orientation
a3: Positive Response Bias
b1.2.3 and c′
c
Fig. 1. Mediation model with paths labelled according to Baron and Kenny (1986). Dogmat. = dogmatism. Evidence Integrat. = Evidence Integration Impairment. Political Orient. = Political Orientation. P.R.B. = Positive Response Bias. This model depicts full mediation of the relationship between dogmatism and Attitudes Toward Blacks by the combination of Evidence Integration Impairment and Political Orientation.
F
R2
3.27⁎
0.05
30.39⁎⁎⁎
0.33
0.66
0.01
10.15⁎⁎⁎
0.22
7.15⁎⁎⁎
0.10
Note: Educ. = level of education achieved. Positive coefficients for sex indicate that being female predicts a higher score. F-values represent the significance of the path as a whole, including the main predictors of interest. Predictors in bold were significant. Paths were named according to Baron and Kenny (1986). ⁎ p b .05. ⁎⁎⁎ p b .001.
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
Additional work could examine how dogmatism and related impairments in Evidence Integration might moderate the effectiveness of interventions on prejudice. For example, prejudice can be reduced through positive contact with members of target groups (Pettigrew & Tropp, 2006). However, more dogmatic individuals may benefit less from this intergroup contact for two reasons. First, these individuals may be less likely to engage in intergroup contact because they maintain greater social distance from other groups (Kirtley & Harkless, 1969; Orpen & Rookledge, 1972). Second, even when more dogmatic individuals engage in positive intergroup contact, they might be less able to integrate this positive experience into their belief structures and thereby revise the way they view members of other groups. Interventions that address Evidence Integration deficits (or BADE generally) may therefore make subsequent intergroup contact more effective in highly dogmatic individuals. Metacognitive Training (MCT), a recently developed therapy, contains a module to reduce BADE that might be used to realize this putative benefit (Moritz & Woodward, 2007). Targeting BADE, rather than dogmatism, might more efficiently reduce prejudice in dogmatic individuals because dogmatism is both a more general personality trait and is characterized by resistance to change. Future research might test whether an intervention using this module of MCT would enhance the impact of positive intergroup contact for reducing prejudice, especially for more dogmatic individuals. Despite these important implications, this study has several limitations. First, we examined the prejudice of Whites toward Blacks. Reversing the races of our target group and perpetrators of bias may have changed our results (see Foley & Chamblin, 1982). It therefore remains to be seen whether our results generalize to prejudices held by other groups. It also remains to be seen whether our results generalize to other prejudices held by Whites. Given the particular importance of individual differences for predicting generalized prejudice (Hodson & Dhont, 2015), we expect our results to generalize in this manner (e.g., to prejudice against gay/lesbian individuals; see Altemeyer, 2002b; Altemeyer & Hunsberger, 1992). Second, our results derive from a single study. Mitigating this limitation, a preliminary study that did not include Political Orientation replicated the pattern of zero-order correlations and provided some support for the indirect path through Evidence Integration Impairment reported above. Results of this replication study can be seen in Supplementary material, Section S3.0. Future research should strive to address these limitations by replicating our findings using other populations or while measuring prejudice toward other groups. 5. Conclusion Through a novel application of Woodward, Moritz, Cuttler, and Whitman's (2006) BADE task, we procured a pattern of results suggesting that Evidence Integration Impairment accounts for a unique portion of the relationship between dogmatism and prejudice. This pattern suggests that failures of belief revision in ambiguous social situations may contribute to the maintenance of prejudice in dogmatic individuals. This suggestion lays a theoretical foundation for experimental research on whether or not aspects of belief revision failure might support individual differences in response to interventions on prejudice. Such research should contribute to our understanding of the bulwarks of prejudice, and thus holds out the promise of leading toward the development of more effective interventions on prejudiced beliefs. Acknowledgement The authors declare no potential conflicts of interest regarding this article. The authors thank Dr. Arielle Baskin-Sommers for contributing data processing scripts, and Dr. Todd Woodward and colleagues for providing us with their BADE task. We thank two anonymous reviewers for
93
providing us with useful feedback on an earlier version of this manuscript. The authors received no financial support for the production of this article. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.paid.2016.09.036. References Allport, G. W. (1954). The nature of prejudice. Cambridge, MA: Addison-Wesley. Altemeyer, B. (2002a). Dogmatic behavior among students: Testing a new measure of dogmatism. The Journal of Social Psychology, 142(6), 713–721. Altemeyer, B. (2002b). Changes in attitudes toward homosexuals. Journal of Homosexuality, 42(2), 63–75. Altemeyer, B., & Hunsberger, B. (1992). Authoritarianism, religious fundamentalism, quest, and prejudice. The International Journal for the Psychology of Religion, 2(2), 113–133. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182. Brigham, J. C. (1993). College students' racial attitudes. Journal of Applied Social Psychology, 23(23), 1933–1967. Case, C. E., Greeley, A. M., & Fuchs, S. (1989). Social determinants of racial prejudice. Sociological Perspectives, 32(4), 469–483. Corlett, J. A. (2003). Race, racism, and reparations. Cornell University Press. Crowson, H. M., DeBacker, T. K., & Davis, K. A. (2008). The DOG Scale: A valid measure of dogmatism? Journal of Individual Differences, 29(1), 17–24. Davies, M. E. (1993). Dogmatism and the persistence of discredited beliefs. Personality and Social Psychology Bulletin, 19, 692–699. Davies, M. (1998). Dogmatism and belief formation: Output interference in the processing of supporting and contradictory cognitions. Journal of Personality and Social Psychology, 75, 456–466. Ekehammar, B., Akrami, N., & Araya, T. (2003). Gender differences in implicit prejudice. Personality and Individual Differences, 34(8), 1509–1523. Foley, L. A., & Chamblin, M. H. (1982). The effect of race and personality on mock jurors' decisions. The Journal of Psychology, 112(1), 47–51. Hodson, G., & Busseri, M. (2012). Bright minds and dark attitudes: Lower cognitive ability predicts greater prejudice through right-wing ideology and low intergroup contact. Psychological Science, 23, 187–195. Hodson, G., & Dhont, K. (2015). The person-based nature of prejudice: Individual difference predictors of intergroup negativity. European Review of Social Psychology, 26(1), 1–42. Hodson, G., Dovidio, J. F., & Gaertner, S. L. (2002). Processes in racial discrimination: Differential weighting of conflicting information. Personality and Social Psychology Bulletin, 28(4), 460–471. Hugenberg, K., & Bodenhausen, G. V. (2003). Facing prejudice implicit prejudice and the perception of facial threat. Psychological Science, 14(6), 640–643. Johnson, D. R., & Borden, L. A. (2012). Participants at your fingertips using Amazon's Mechanical Turk to increase student–faculty collaborative research. Teaching of Psychology, 39(4), 245–251. Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003a). Exceptions that prove the rule—Using a theory of motivated social cognition to account for ideological incongruities and political anomalies: Reply to Greenberg and Jonas (2003). Psychological Bulletin, 129, 383–393. Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003b). Political conservatism as motivated social cognition. Psychological Bulletin, 129(3), 339–375. Kirtley, D., & Harkless, R. (1969). Some personality and attitudinal correlates of dogmatism. Psychological Reports, 24(3), 851–854. Moritz, S., & Woodward, T. S. (2007). Metacognitive training in schizophrenia: From basic research to knowledge translation and intervention. Current Opinion in Psychiatry, 20, 619–625. Orpen, C., & Rookledge, Q. (1972). Dogmatism and prejudice in White South Africa. The Journal of Social Psychology, 86(1), 151–153. Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90(5), 751–783. Preacher, K. J., & Hays, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891. Prezza, M., Zampatti, E., Pacilli, M. G., & Paoliello, A. (2008). Territorial sense of community, ethnic prejudice and political orientation. Journal of Community and Applied Social Psychology, 18(4), 315–332. Rokeach, M. (1960). The open and closed mind. New York, NY: Basic Books. Sanford, N., Veckenstedt, R., Moritz, S., Balzan, R. P., & Woodward, T. S. (2014). Impaired integration of disambiguating evidence in delusional schizophrenia patients. Psychological Medicine, 44(13), 2729–2738. Smith, K. B., Oxley, D. R., Hibbing, M. V., Alford, J. R., & Hibbing, J. R. (2011). Linking genetics and political attitudes: Reconceptualizing political ideology. Political Psychology, 32(3), 369–397. Speechley, W. J., Moritz, S., Ngan, E. T. C., & Woodward, T. S. (2012). Impaired evidence integration and delusions in schizophrenia. Journal of Experimental Psychology, 3, 688–701.
94
M.V. Bronstein et al. / Personality and Individual Differences 105 (2017) 89–94
Stewart, B. D., von Hippel, W., & Radvansky, G. A. (2009). Age, race, and implicit prejudice using process dissociation to separate the underlying components. Psychological Science, 20(2), 164–168. Strickland, B. R., & Weddell, S. C. (1972). Religious orientation, racial prejudice, and dogmatism: A study of Baptists and Unitarians. Journal for the Scientific Study of Religion, 11, 395–399.
Woodward, T. S., Moritz, S., Cuttler, C., & Whitman, J. C. (2006). The contribution of a cognitive bias against disconfirmatory evidence (BADE) to delusions in schizophrenia. Journal of Clinical and Experimental Neuropsychology, 28(4), 605–617. Woodward, T. S., Buchy, L., Moritz, S., & Liotti, M. (2007). A bias against disconfirmatory evidence is associated with delusion proneness in a nonclinical sample. Schizophrenia Bulletin, 33(4), 1023–1028.