Managing the Threat of Impending Implicit Attitude Feedback

Abstract

People facing potentially threatening feedback sometimes opt to avoid it in an attempt to preserve a cherished self-view. In three studies, we examined whether people would adopt such a strategy in the context of the Black–White Implicit Association Test (IAT), which has the potential to reveal implicit prejudice. Study 1 demonstrated that people expect their IAT results to indicate less implicit prejudice than the results actually do, and perceive feedback from the Black–White IAT as potentially threatening. In addition, people who would rather avoid learning their results regretted receiving their feedback. In Studies 2 and 3, more participants declined to learn their IAT results when cued to expect unfavorable, rather than favorable, IAT results. Importantly, participants who received no expectation cue generally opted to receive their IAT feedback, suggesting that participants likely expect favorable IAT feedback.

Keywords

information avoidance expectations threat management implicit attitudes racism

People are generally adept at dealing with threatening information and may derogate or dismiss information that challenges a desired self-view (Pyszczynski & Greenberg, 1987; Shepperd, Malone, & Sweeny, 2008). Responding defensively to threats to identity is consistent with both self-enhancement and consistency theories (Sedikides & Gregg, 2008; Swann, 1990).Yet, people do not merely respond to information after the fact. They sometimes address challenging information proactively (e.g., Carroll, Sweeny, & Shepperd, 2006; Sweeny, Melnyk, Miller, & Shepperd, 2010).

One proactive strategy for responding to threat is information avoidance. Information avoidance entails preventing or delaying “the acquisition of available but potentially unwanted information” (Sweeny et al., 2010, p. 4). Although people may avoid information for many reasons, one potential reason that has not received experimental test is that the information threatens a cherished belief (Sweeny et al., 2010) such as a desired self-view. We explored in three studies whether people would adopt this strategy in anticipation of feedback that may reveal that they are implicitly prejudiced.

IAT Feedback as a Threat to Self

In American society where racial favoritism is reviled and an egalitarian outlook is valued (Crandall, Eshleman, & O’Brien, 2002), accusations or evidence that one is prejudiced can be quite threatening. Indeed, concerns about appearing egalitarian can lead people to report less prejudice than they actually harbor (Greenwald, Poehlman, Uhlmann, & Banaji, 2009; Nier, 2005). As a consequence, researchers have developed implicit measures to bypass the problem of self-presentation in self-report measures of prejudice (Lane, Banaji, Nosek, & Greenwald, 2007). Chief among implicit measures is the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998), which researchers have used extensively to assess automatic attitudes and preferences (Greenwald et al., 2009). For instance, as of 2010 the Project Implicit website (https://implicit.harvard.edu), where people can complete various versions of the IAT, reported over 10 million participants (Banaji & Heiphetz, 2010).

Perhaps the best-known IAT is the Black–White IAT, which assesses the relative preference for Black versus White faces. Evidence reveals a low correspondence between self-reported explicit attitudes toward Blacks versus Whites and responses to the Black–White IAT—people consistently report less bias than their IAT results indicate (Lane et al., 2007). Importantly, the correspondence between implicit and explicit attitudes increases when participants complete explicit measures under time pressure, are subjected to a bogus lie-detection procedure, or receive instructions to give their “gut” feelings about a target group (Nier, 2005; Ranganath, Smith, & Nosek, 2008). These findings suggest that people may be aware of their implicit attitudes at some level, but opt not to express them. Indeed, people may perceive or present themselves as more egalitarian than their IAT scores suggest and be unwilling to acknowledge that they are racially prejudiced. For these people, IAT results that suggest they are biased may be unexpected and threatening.

Two studies offer preliminary evidence that the Black–White IAT provides potentially threatening feedback. First is the finding that 70–90% of Whites learn that they favor Whites more than Blacks (Nosek, Banaji, & Greenwald, 2002). Given that an egalitarian self-perception is desirable to most Americans, the IAT likely produces feedback that is appreciably worse than desired for many people. Second is the finding that merely completing the Black–White IAT produces negative affect especially when participants link their performance to racial issues (Monteith, Voils, & Ashburn-Nardo, 2001). We suspect that the negative affect occurs when people glean from the difficulty of the task and the pattern of errors they receive that their IAT feedback will suggest that they are prejudiced.

We believe that people do not welcome feedback that threatens a desired self-view, and if given a choice, they would decline or avoid such feedback. With respect to the Black–White IAT, we tested directly what feedback people expected to receive on the IAT and how people believed they would feel if they received unfavorable feedback. We predicted that most people in our sample (White Americans) would underestimate how biased they are in favor of White Americans, and would find news that they are biased to be distressing. In addition, we expected that our participants would decline to receive their feedback when cued to believe that their feedback might reveal prejudice toward Blacks.

Theoretical Contribution

The present research makes two novel contributions to the literature. First, our research is the first to examine the IAT results people expect to receive and how people respond to receiving typical IAT feedback. Millions of people have completed the IAT (Greenwald et al., 2009); yet, we know little about how the 70–90% of White individuals who received feedback indicating they are implicitly prejudiced (Nosek et al., 2002) reacted to this feedback. Second, no study has examined whether the IAT evokes proactive threat management strategies like information avoidance. That is, we do not know whether people would avoid learning their IAT results if given the chance. Our studies address these gaps in the literature.

Overview

We conducted three studies to investigate information avoidance in anticipation of IAT feedback. In Study 1, we investigated expectations for and reactions to IAT feedback. In Studies 2 and 3, we assessed whether people opt to avoid their IAT feedback when they receive suggestive evidence that the feedback will be unfavorable. We used two approaches to providing unfavorable feedback. In Study 2, we directly manipulated expectations for IAT feedback. Specifically, we told participants either that their IAT results would likely reveal that they were implicitly prejudiced, implicitly egalitarian, or we provided no expectation information. In Study 3, we created similar expectations about IAT results indirectly. Specifically, we asked participants to list recent examples of times they engaged in interracially egalitarian behavior; half of the participants listed two examples (an easy task) and the other half listed eight examples (a difficult task). We informed participants in both conditions that egalitarian individuals found the task to be easy. We reasoned that participants would infer that their feedback would reveal them to be prejudiced in the difficult task condition but not in the easy task condition. In both studies, we hypothesized greater information avoidance when we cued participants to expect unfavorable feedback (indicating they were implicitly prejudiced) than when we cued participants to expect favorable feedback (indicating they were implicitly egalitarian) or when we provided no cue.

Study 1

In Study 1, we examined people’s expectations for and reactions to their IAT feedback.

Participants

Participants were 64 (37 women) White native-English-speaking residents of the United States aged 18 to 64 (M = 33.0, SD = 10.0) recruited and paid $0.51 on Amazon.com’s Mechanical Turk (see Buhrmester, Kwang, & Gosling, 2011 for more information).

Procedure and Measures

After consenting to participate, participants completed demographic measures, read a description of the BIAT, and then completed the Black–White version of the Brief IAT (Sriram & Greenwald, 2009). The Brief IAT (BIAT) is a computerized task that assesses the relative strength of association between target groups (i.e., Black vs. White) and positivity by pairing one set of stimuli at a time (e.g., White and Good) and using response latency to operationalize attitude strength.

After completing the BIAT, participants indicated what feedback they expected to receive using a scale ranging from 1 = strong automatic preference for Black individuals to 7 = strong automatic preference for White individuals, with a midpoint of 4 = favor Black and White individuals equally. Next participants indicated whether they agreed that learning that they were more implicitly biased than they thought would be distressing, and whether learning that they were less implicitly biased than expected would make them happy (1 = strongly disagree to 9 = strongly agree).

We also assessed the extent to which participants desired to avoid their feedback using an 8-item information avoidance index (α = .87). The index included items like “If I am unconsciously prejudiced, I do not want to know,” and “Even if it will upset me, I want to know if I am unconsciously prejudiced,” which was reverse coded. Next, participants received their BIAT feedback. Specifically, they read that they either “strongly,” “moderately,” or “slightly” favored either White individuals over Black individuals, Black individuals over White individuals, or read that they had no automatic preference. After receiving their feedback, participants indicated on a scale ranging from not at all (1) to very much (7) the extent to which they wished they had not learned their BIAT results and regretted learning these results. We combined these two items to form an index assessing regret (α = .86). Participants also completed 10 items from the Positive and Negative Affect Schedule-Expanded Form (Watson & Clark, 1994). Nine of the items assessed the extent to which participants were experiencing negative emotions (e.g., guilty, sad, upset; α = .95) and 1 item assessed the extent to which participants were “surprised.”

Prior to data analysis, we removed data from two participants who did not meet standard IAT-score inclusion criteria (Greenwald, Nosek, & Banaji, 2003). That is, they either completed over 10% of trials faster than 300 ms, made more than 30% errors overall, or they made more than 40% errors in any one-trial block.

Results and Discussion

Expectations for BIAT Feedback

As expected, the majority of participants underestimated their implicit preference for White individuals. Table 1 shows the distribution of expected and actual scores on the BIAT. Table 1 also shows the percentage of people in each expectation category (i.e., slight, moderate, and strong preference for White or Black individuals) whose actual BIAT scores indicated they were more prejudiced than they expected. For instance, 62% of people who indicated they expected to learn they had a “Slight Automatic Preference for White Individuals” actually had a moderate or strong automatic preference for White targets on the BIAT. A within-subjects analysis of these scores revealed that participants generally predicted that they would score lower on the BIAT scale (M = 4.2, SD = .90) than they actually scored (M = 5.1, SD = 1.3), t(61) = 4.97, p < .001, within-subjects d = .27. Moreover, participants’ estimated results were uncorrelated with their actual BIAT results, r(62) = .12, p = .32.

Table 1.

Study 1 Distribution of Predicted and Actual IAT Results.

	Strong Automatic Preference for Black Individuals	Moderate Automatic Preference for Black Individuals	Slight Automatic Preference for Black Individuals	Favor Black and White Individuals Equally	Slight Automatic Preference for White Individuals	Moderate Automatic Preference for White Individuals	Strong Automatic Preference for White Individuals
D score^a	< −.65	−.64 to −.35	−.34 to −.15	−.14 to .15	.16 to .34	.35 to .64	> .65
Actual feedback (%)	0	3.2	6.5	22.6	24.2	24.2	19.4
Predicted feedback (%)	3.2	0	7.9	61.9	20.6	6.3	0
% of participants receiving IAT scores indicating greater bias than they predicted	100	—	100	55.3	61.5	25.0	—

Note. IAT = implicit association test.

^aIAT D-scores are the average of the D scores for the BIAT. Each score is constructed by subtracting the average latency from trials when responding to White + good from the average latency when responding to Black + good and dividing that number by the pooled standard deviation of these latencies. Positive scores thus indicate a stronger association between “good” + “White” relative to “good” + “Black.”

Participants indicated that learning that they were more implicitly biased than they expected would be distressing (M = 6.4, SD = 2.3) but that learning they were less implicitly biased than they expected would make them happy (M = 7.1, SD = 1.8), both ts (61) > 4.7, ps < .001, compared with the midpoint of 5.0. Participants’ desire to avoid was uncorrelated with their expected, r(62) = −.005, p = .97, and actual BIAT results, r(62) = −.12, p = .34.

Reactions to BIAT Feedback

All variables measuring reactions to feedback were positively skewed. Therefore, we used nonparametric tests to assess all relationships. Among participants who received feedback indicating they were more biased than expected, participants reporting a greater desire to avoid learning their BIAT feedback also reported greater regret, ρ(37) = .72, p < .001. This relationship did not hold among participants who learned they were less or equally biased, ρ(25) = −.03, ns; interaction term: β = .25, p = .008. These findings indicate that people regretted learning their BIAT scores most when they did not want to learn them and their feedback indicated they were more biased than they realized.

The greater the discrepancy between predicted and actual BIAT scores, the more surprised participants were by their results, ρ(62) = .37, p < .001. That is, participants were more surprised if they learned they mispredicted their BIAT results (M = 3.3, SD = 1.8) than if they accurately predicted their BIAT results (M = 1.8), K(1, N = 62) = 8.62, p = .003. Moreover, learning that one was more biased than expected corresponded with greater negative affect, ρ(62) = .55, p < .001, and more regret, ρ(62) = .45, p < .001. Specifically, participants who were more (n = 37), equally (n = 16), and less (n = 9) biased than they predicted differed in regret and negative affect, Ks (2, N = 62) > 13.14, p < .001. Participants who underestimated their bias experienced greater regret (M = 1.9, SD = 1.2) and greater negative affect (M = 2.8, SD = 1.4) than did participants who were accurate (M _regret = 1.0, SD = 0.1; M _affect = 1.5, SD = 0.8), Ks (1, n = 53) > 8.32, p < .005, and participants who overestimated their bias (M _regret = 1.0, SD = 0; M _affect = 1.2, SD = 0.4), Ks (1, n = 46) > 6.00, p < .02. These latter two groups did not differ in negative affect and regret, Ks (1, n = 53) < 1, ns.

In sum, the majority of our participants learned that they were more implicitly biased than they expected. In addition, BIAT results indicating greater bias than expected corresponded with greater negative affect and with greater regret about receiving one’s scores. Among participants who learned they were more biased than expected, the more they desired to avoid learning their BIAT results, the greater regret they reported in response to receiving their BIAT feedback. Although participants may have reported more explicit prejudice (and less discrepancy) had we asked about their expectations differently (e.g., asking for gut reactions; Ranganath et al., 2008), the discrepancy between reported expectations and actual BIAT feedback is nevertheless striking. Study 1 thus confirms that many people receive BIAT feedback that is more unfavorable than expected, and that this unfavorable, unexpected BIAT feedback produces negative affect and regret about receiving BIAT feedback, particularly among people who do not want their feedback.

Given the hypotheses of the present article, it may seem surprising that participants’ expectations for their BIAT results were unrelated to their tendency to avoid. However, we do not think that avoidance stems from expectations for prejudiced feedback generally, but rather from the expectation of bad news—the worry that the IAT will reveal results that indicate greater prejudice than expected. In Studies 2 and 3, we aimed to create this type of expectation.

Study 2

In Study 2, we evaluated whether warning people that their IAT results might be unfavorable would prompt greater information avoidance. We led participants to expect unfavorable IAT feedback, favorable IAT feedback, or we did not manipulate IAT expectations. We hypothesized that more participants would opt to avoid learning their IAT results in the unfavorable expectation condition than in the other conditions.

Participants

Participants were 102 (64 women) White native-English-speaking residents of the United States aged 18 to 68 (M = 32.8, SD = 17.0) recruited via e-mail, announcements on social networking websites, volunteer requests on message boards, and a post on Amazon.com’s Mechanical Turk (Buhrmester et al., 2011).

Procedure and Measures

As in Study 1, participants consented, completed demographic measures, and took an abbreviated version of the Brief IAT. Unlike Study 1, participants in Study 2 did not complete the full set of counterbalanced BIAT trials, so their results were not recorded. Specifically, all participants completed only one congruent (i.e., White-Positive) and one incongruent (i.e., Black-Positive) trial block. We randomly assigned the block (incongruent or congruent) that participants completed first. We chose not to have participants complete the full set of trials to reduce participant burden. Because IAT results were uncorrelated with avoidance in Study 1, it is unlikely that we would gain additional information from requiring them to complete a full set of trials, which would more than double the length of our survey. Rather, the task served to expose participants to the conditions that typically surround IAT feedback.

We then assigned participants to one of the three conditions. In the Unfavorable Expectation condition, participants read that “Based on information you gave earlier in the study, you are more likely to have a score indicating you are unconsciously prejudiced on our unconscious racism measure when compared to the population in general.” The message included a histogram that placed them in the top 25% in likelihood of receiving such feedback. In the Favorable Expectation condition, the message was identical except that “prejudiced” was replaced with “egalitarian.” In the No Expectation condition, participants received no message. We did not specify which information the computer used to determine their likelihood. The only items participants completed prior to the IAT were demographic measures (i.e., age, sex, ethnicity, race, political orientation, religion, and education). We also did not assess what participants believed we meant by earlier information. Next, participants indicated whether they wished to see their IAT results (yes or no). This decision represented our primary dependent measure.

To pilot test our manipulation, a separate group of participants (N = 128) completed a study identical to the one just described. However, instead of choosing whether to receive IAT feedback, they reported whether they expected to receive IAT feedback indicating they were biased using a 1 = extremely unconsciously prejudiced (favor White individuals) to 7 = not at all unconsciously prejudiced (favor Black and White individuals equally) scale. After reversing responses so that higher numbers reflect greater bias, a one-way analysis of variance revealed that participants’ expectations differed by condition, F(2, 122) = 4.19, p = .02, η = .28. An examination of the simple main effects revealed that participants were more inclined to expect feedback to indicate they were biased in the Unfavorable Expectation condition (M = 3.8, SE = .24) than in the Favorable Expectation condition (M = 2.7, SE = .19), F(1, 88) = 4.77, p = .03, η = .23, and the no expectation condition (M = 2.5, SE = .22), F(1, 75) = 7.73, p = .007, η = .31. Participants in the favorable expectation and no expectation conditions did not differ in their anticipated implicit attitude, F(1, 90) < 1, ns. As such, we expected that participants would be more likely to avoid learning their IAT results in the Unfavorable Expectation condition than in No Expectation or Favorable Expectation conditions, but that participants in these latter two conditions would not differ in avoidance.

Study 2: Results and Discussion

As Figure 1 shows, avoidance differed significantly between conditions, χ²(2, N = 88) = 13.48, p = .001, Φ > .39. An examination of the simple main effects revealed that more participants avoided learning their IAT results in the Unfavorable Expectation condition (57%), than in the No Expectation (18%), χ²(1, n = 62) = 9.85, p = .002, Φ = .40, and Favorable Expectation (15%), χ²(1, n = 49) = 9.12, p = .003, Φ = .43, conditions. Participants in the latter two conditions did not differ in avoidance, χ² < 1, ns. These finding suggest people are more inclined to avoid their IAT results if they receive preliminary evidence suggesting that their results will reveal them to be prejudiced than if they receive preliminary evidence suggesting that their results will reveal them to be egalitarian or if they receive no preliminary evidence.

Figure 1.

Study 2: Avoidance by expectation condition.

Study 3

People often receive no overt warning that forthcoming feedback might be unfavorable. Rather, they form expectations from surrounding circumstances or environmental cues. As noted earlier, merely completing the IAT can cue people that their results might be worse than expected (Monteith et al., 2001). Study 3 examined whether the effects of Study 2 would replicate when participants received an indirect hint that their forthcoming IAT feedback might be unfavorable. Specifically, we assigned participants to an easy or difficult task that was supposedly easy for implicitly egalitarian individuals. We expected that participants completing the difficult version of the task would use task difficulty as a cue about their implicit attitudes and therefore be would more inclined to avoid learning their feedback.

Participants

Participants were 46 (27 women) White native English-speaking residents of United States aged 18 to 88 (M = 34.3, SD = 14.1) recruited online as in Study 2.

Procedures

Procedures were identical to Study 2 with two exceptions. First, we omitted the No Expectation Condition. Second, rather than explicitly stating what their IAT was likely to reveal, we offered participants a subtle, indirect hint based on their experience completing a listing task (Schwarz et al., 1991). We instructed participants to generate either two or eight recent examples of their own interracially egalitarian behavior. Earlier work with similar tasks shows that participants use ease of retrieval as a cue for their own attitudes. For instance, when asked to generate either 2 or 12 examples of recent assertive behavior, participants asked to generate 12 examples (a difficult task) rated themselves as less assertive than did participants asked to generate two examples (an easy task; Schwarz et al., 1991).

To determine the point at which listing recent examples of interracially egalitarian behavior task was difficult, we asked 20 pilot participants to list as many examples as they could and to indicate the point where listing became difficult. Pilot participants generated between 7 and 16 examples (M = 11.1, SD = 2.52), and found the task difficult after listing, on average, 5.5 examples (SD = 1.57). These two values were highly correlated such that participants who could list fewer examples found the task difficult sooner than did participants who could list more examples (r = .76, p < .001). An examination of the lists revealed that none of the participants found it difficult to generate two examples of recent egalitarian behavior, but that all found it difficult to generate eight examples. Thus, we presumed that participants would expect favorable IAT feedback when they listed two examples (an easy task) and unfavorable IAT feedback when they listed eight examples (a difficult task). Data from pilot participants confirmed this suspicion. Specifically, 85 pilot participants indicated on a 1 = extremely unconsciously prejudiced (favor White individuals) to 7 = not at all unconsciously prejudiced (favor Black and White individuals equally) scale their expectations about the IAT feedback they would receive following the listing task. After reverse coding the scale so that higher scores reflected greater bias, participants were more inclined to expect feedback indicating they were implicitly biased when instructed to list eight examples (M = 3.2, SE = .20) than when instructed to list two examples of personal egalitarianism (M = 2.1, SE = .22), F(1, 83) = 13.07, p < .001, η= .33.

In sum, we assigned participants to list either eight examples or two examples of recent interracially egalitarian behavior. Because pilot participants expected more egalitarian feedback when they listed two examples, we predicted that participants would avoid their IAT feedback more in the eight examples condition than in the two examples condition.

Study 3: Results and Discussion

Consistent with Study 2, more participants declined to learn their IAT results when asked to list eight examples of personal egalitarian behavior (56%) than when asked to list two examples of personal egalitarian behavior (10%), χ²(1, N = 46) = 10.18, p < .01, Φ = .50. These finding suggest that people are more inclined to avoid their IAT results if they receive preliminary implicit evidence suggesting that their results will reveal them to be prejudiced than if they receive preliminary implicit evidence suggesting that their results will reveal them to be egalitarian.

General Discussion

Results from three studies reveal that the IAT may produce unexpected and undesired feedback and that cuing people to expect such feedback increases avoidance of that feedback. Study 1 revealed that our sample of White Americans expected to receive more egalitarian IAT feedback than they actually received and reported that they would be upset if their results indicated they were implicitly prejudiced. Moreover, the more they wanted to avoid learning their IAT results, the greater regret they reported in response to receiving their IAT results. Studies 2 and 3 showed that participants who were led to anticipate that their IAT feedback might be unfavorable were more inclined to avoid learning their IAT feedback than were participants who were led to expect favorable feedback or who were not prompted to expect particular feedback.

Our findings offer important contributions to both the IAT and information avoidance literatures. Expanding on the IAT literature, we found that people who receive IAT feedback indicating that they are more biased than they expected report negative affect and regret about learning their IAT scores. Moreover, we found that people are not always willing to receive IAT feedback, especially if cued to expect that the feedback might be negative. Our work also expands the literature on information avoidance by showing that manipulating feedback expectations influenced information avoidance. We found that people show greater information avoidance when led to expect unfavorable news than when led to expect favorable news.

Although our findings are theoretically informative, it remains unknown whether they generalize to non-White participants and to other versions of the IAT (e.g., the Weight or Age IAT). Presumably non-White participants and participants completing other IAT measures would respond similarly to the way they responded in the present studies, provided that unfavorable results would threaten an important self-view. The role of prior expectations also remains unknown. In our studies, we chose not to measure participants’ expectations prior to our manipulations, because we feared that doing so would produce demand characteristics. However, it is possible that people who expect at the outset that their IAT results will reveal prejudice are less threatened by such feedback. Alternatively, they might be more threatened by the possibility of having their expectations confirmed. We thus think it important for future studies to investigate the role of initial expectations on avoidance of implicit feedback.

Of course, the prospect of unfavorable IAT feedback might elicit other threat management strategies. For instance, when IAT feedback cannot be avoided, people may brace for possible bad news by shifting their expectations downward (Carroll et al., 2006). And people who receive undesired IAT feedback may dismiss or derogate it (Shepperd, 1993) or deny it altogether (McQueen, Vernon, & Swank, 2012). People use a variety of strategies to maintain their view of self (Tesser, 2000). However, the prevalence and interchangeability of such strategies remain untested in the context of IAT feedback.

Finally, little is known about how threat management strategies, like information avoidance, affect people’s perceptions of their biases. It is possible that, when people respond defensively (e.g., by avoiding or derogating the feedback) they unintentionally impede some of the positive personal change that may result from learning that they are implicitly biased. As such, future work should investigate how defensive processing of IAT feedback affects people’s perceptions of their own bias.

The last decade has witnessed an explosion of opportunities for people to acquire important information about the self from online disease-risk calculators to direct-to-consumer genetic testing. The IAT is but one example of the growing array of tools that people can use to acquire self-relevant information. Yet, we know little about how people respond to the results they receive or whether they are even prepared to learn their results. Our findings suggest that the feedback many people receive is sometimes unexpected. When people suspect that their results might be undesired, they may opt to decline receiving those results.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This article supported by National Science Foundation Graduate Research Fellowships awarded to Jennifer L. Howell and Steve M. Newell, under Grant No. DGE-0802270, by an Intergovernmental Personnel Act Fellowship awarded to James A. Shepperd by the National Cancer Institute, and a developmental grant to James A. Shepperd funded through the Southeast Center for Research to Reduce Disparities in Oral Health (U54DE019261-02) by the National Institute of Dental and Craniofacial Research.

References

Banaji

M. R.

Heiphetz

(2010). Attitudes. In Fiske

S. T.

Gilbert

Lindzey

(Eds.), Handbook of social psychology (5th ed., Vol. 1, pp. 353–393). New York, NY: McGraw-Hill.

Buhrmester

Kwang

Gosling

S. D.

(2011). Amazon’s mechanical turk: A new source of inexpensive, yet high-quality, data? Perspective on Psychological Science, 6, 3–5.

Carroll

Sweeny

Shepperd

J. A.

(2006). Forsaking optimism. Review of General Psychology, 10, 56–73.

Crandall

C. S.

Eshleman

O’Brien

L. T.

(2002). Social norms and the expression of prejudice: The struggle for internalization. Journal of Personality and Social Psychology, 82 , 359–378.

Greenwald

A. G.

McGhee

D. E.

Schwartz

J. L. K.

(1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464–1480.

Greenwald

A. G.

Nosek

B. A.

Banaji

M. R.

(2003). Understanding and using the Implicit Association Test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85, 197–216.

Greenwald

A. G.

Poehlman

Uhlmann

Banaji

M. R.

(2009). Understanding and using the Implicit Association Test: III. Meta-analysis of predictive validity. Journal of Personality and Social Psychology, 97, 17–41. doi:10.1037/a0015575

Lane

K. A.

Banaji

M. R.

Nosek

B. A.

Greenwald

A. G.

(2007). Understanding and using the Implicit Association Test: IV. What we know (so far). In Wittenbrink

Schwarz

N. S.

(Eds.), Implicit measures of attitudes: Procedures and controversies (pp. 59–102). New York, NY: Guilford.

McQueen

Vernon

S. W.

Swank

P. R.

(2012). Construct definition and scale development for defensive information processing: An application to colorectal cancer screening. Health Psychology, 32, 190–202. doi:10.1037/a0027311

10.

Monteith

M. J.

Voils

C. I.

Ashburn-Nardo

(2001). Taking a look underground: Detecting, interpreting, and reacting to implicit racial biases. Social Cognition, 19, 395–417. doi:10.1521/soco.19.4.395.20759

11.

Nier

J. A.

(2005). How dissociated are implicit and explicit racial attitudes? A bogus pipeline approach. Group Processes & Intergroup Relations, 8, 39–52. doi:10.1177/1368430205048615

12.

Nosek

B. A.

Banaji

M. R.

Greenwald

A. G.

(2002). Harvesting implicit group attitudes and beliefs from a demonstration web site. Group Dynamics: Theory, Research, and Practice, 6, 101–115.

13.

Pyszczynski

Greenberg

(1987). Toward an integration of cognitive and motivational perspectives on social inference: A biased hypothesis-testing model. Advances in Experimental Social Psychology, 20, 297–340.

14.

Ranganath

K. A.

Smith

C. T.

Nosek

B. A.

(2008). Distinguishing automatic and controlled components of attitudes from direct and indirect measurement methods. Journal of Experimental Social Psychology, 44, 386–396.

15.

Schwarz

Bless

Strack

Klumpp

Rittenauer-Schatka

Simons

(1991). Ease of retrieval as information: Another look at the availability heuristic. Journal of Personality and Social Psychology, 61, 195–202. doi:10.1037/0022-3514.61.2.195

16.

Sedikides

Gregg

A. P.

(2008). Self-enhancement: Food for thought. Perspectives on Psychological Science, 3, 102–116.

17.

Shepperd

J. A.

(1993). Student derogation of the SAT: Biases in perceptions and presentations of college board scores. Basic & Applied Social Psychology, 14, 455–473.

18.

Shepperd

J. A.

Malone

Sweeny

(2008). Exploring causes of the self-serving bias. Social and Personality Psychology Compass, 2, 895–908.

19.

Sriram

Greenwald

A. G.

(2009). The brief Implicit Association Test. Experimental Psychology, 56, 283–294.

20.

Swann

W. B.

Jr (1990). To be adored or to be known? The interplay of self-enhancement and self-verification. In Higgins

E. T.

Sorrentino

R. M.

(Eds.), Motivation and cognition: Foundations of social behavior (Vol. 2, pp. 408–448). New York, NY: Guilford.

21.

Sweeny

Melnyk

Miller

Shepperd

J. A.

(2010). Information avoidance: Who, what, when & why. Review of General Psychology, 14, 340–353.

22.

Tesser

(2000). On the confluence of self-esteem maintenance mechanisms. Personality and Social Psychology Review, 4, 290–299.

23.

Watson

Clark

L. A.

(1994). The PANAS-X: Manual for the positive and negative affect schedule-expanded form. Ames: University of Iowa.