Reducing defensive responding to implicit bias feedback: On the role of perceived moral threat and efficacy to change

https://doi.org/10.1016/j.jesp.2021.104165Get rights and content

Abstract

The last decade has seen a rush to address the causes and consequences of bias in applied contexts across the world. When and why might these initiatives promote attitudes and behavior that align with egalitarian goals? A common assumption is that increasing awareness of bias can motivate control over prejudiced responding. However, learning that one's actions are biased is threatening, and often motivates a range of self-protective responses to buffer that threat. In the current research, we tested a strategy for reducing such defensive responding and increasing the kind of awareness central to contemporary theories of prejudice regulation and egalitarian behavior. Four experiments (N > 2500) and a mini meta-analysis demonstrate that interventions that (a) decrease perceived moral blameworthiness for having bias and (b) increase the perceived ability to control bias, can reduce defensive responding and increase awareness both in the short-term and approximately 6 months later. Interventions that minimize threat and facilitate efficacy can motivate increased bias awareness and commitment to egalitarian values.

Section snippets

When is feedback motivating versus a cause for defensiveness?

All theories of goal pursuit allow that feedback energizes responses that serve the goal. It stimulates action aimed at addressing the negative feedback (e.g., Bandura, 1991; Carver & Scheier, 1998; Higgins, Strauman & Klein, 1986; Wicklund & Gollwitzer, 1982). Negative feedback triggers an uncomfortable state, a tension, which lead to stronger intentions and goal commitment. When a low prejudice person receives feedback about bias, research shows that it can motivate control over bias and

Addressing the conditions that cause bias feedback to be demotivating

Feedback about bias is presumed to play an important role in producing long-term change in attitudes and behavior relating to egalitarianism (a decrease in bias). This increased egalitarianism is hypothesized to occur through feedback triggering 1) an increased awareness of bias, 2) personal culpability for responding to the bias, 3) commitment to the goal of egalitarianism and an intention to be more egalitarian, and 4) increased egalitarian action (e.g., Forscher et al., 2019; Howell et al.,

The present research

The goal of the current research is to evaluate the effectiveness of the proposed intervention in reducing defensive responding and, as a result, indirectly increasing bias awareness. Across four samples (N = 1489), and while utilizing multiples measures of bias awareness and forms of providing bias feedback, we find clear and convergent evidence that the proposed intervention reduces defensive responding and, as a result, directly and indirectly increases bias awareness. The proposed

Design

Experiment 1 employed a single independent variable design (Feedback Only vs. Intervention) in which all participants received the same bias feedback. An additional experimental condition was run concomitantly with the other conditions described above but was designed to test hypotheses different from what is addressed here. Information about this condition can be found in the supplemental materials. We report all measures used in this analysis here, and provide the exact language used for all

Participants

Participants were 478 White U.S. citizens recruited from Amazon Mturk (57.5% females; mean age = 37.46, SD = 13.02). Most participants report a family income greater than 50 K (60.38%) and have earned at least a Bachelor's degree (73.22%). G*Power was used to determine the sample size needed to obtain adequate statistical power to detect mean level differences between the experimental and control group for medium effect sizes, and then Mturk participants were oversampled to adjust for the

Results and discussion

Study 1 tests hypothesis 2 and 3 (that the interventions will reduce defensive responding and indirectly increase bias awareness), which involves a comparison between the intervention and the Feedback Only conditions. Implicit racial attitudes were included as covariates for all analyses in Study 1. While the inclusion of control variables that covary with our constructs strengthens our confidence in the independence and robustness of our theorized effects, model estimates without covariates

Design

The results of Study 1 indicate that the intervention was successful in both reducing defensive responding and increasing awareness of one's personal bias. Given concerns about the replicability of established effects in psychology (e.g., Open Science Collaboration, 2015), it is critical to determine whether a similar pattern of results can be observed on an independent sample. Study 2 provides a replication of the hypothesis that the intervention will reduce defensiveness and, consequently,

Results and discussion

Study 2 tests Hypotheses 1–3. Only participants who completed the entire study were included in analyses. All other measures, manipulations, and exclusions are otherwise fully reported. Analyses were conducted in Study 2 the same way as for Study 1. For all analyses reported here, implicit and explicit racial attitudes, internal and external egalitarian motivations, SDO, and skepticism about social science are included as covariates. These measures were assessed to address research questions

Longitudinal study

The results of Study 1 and 2 demonstrate that the intervention can reduce defensive responding to implicit bias feedback, and, as a result, increase awareness of bias in the self and others. However, these effects were observed cross-sectionally, that is, immediately after experimental treatment—the extent to which these effects persist (or not) beyond the immediate context remains unknown. By reducing defensive responding, this intervention may help facilitate strategies that aid in both the

Results

This design allows for an analysis of the extent to which the indirect cross-sectional effects of the feedback intervention conditions on bias awareness persist 6-months later (H4). This test involves a comparison between the intervention group and the Feedback Only condition. Because these samples were recruited consecutively, and because the follow-up survey was administered concurrently for all eligible participants, a dummy-variable to represent the sample from Study 1 or 2 was not

Design

Study 3 manipulated perceived efficacy and moral threat as separate dimensions of the intervention, administered manipulated checks for the psychological constructs targeted by the intervention, utilized a validated measure of bias awareness (Perry et al., 2015), and provided feedback in a less exaggerated or condemnatory way. Accordingly, Study 3 employed a 1 (No Feedback Control) + 2 (Efficacy; High vs. No Information) x 2 (Moral Threat: Low vs. No Information) design. The condition with no

Participants

Participants were 754 White U.S. citizens recruited from Amazon Mturk (64.4% females; mean age = 35.15, SD = 11.63). Most participants report a family income greater than 50 K (50%) and have earned at least a Bachelor's degree (83.95%). G*Power was used to determine the sample size needed to obtain adequate statistical power to detect mean level differences between each experimental and control group for medium effect sizes, and then Mturk participants were oversampled to adjust for the

Results and discussion

In Study 3, we completed two sets of analysis. First, we seek to replicate the results of Study 1 and 2 by comparing the Intervention condition (i.e., High Efficacy, Low Moral Threat) to the No Feedback and Feedback Only conditions. This confirmatory test was conducted using a one-way between-subjects ANOVA to evaluate the effects of the intervention on defensive responding, compared to the control groups. Second, we also explore the independent effect of each feature of the intervention– the

Design

In Study 4, we seek to replicate the results of Studies 1, 2, and 3, and extend our investigations by manipulating levels of perceived moral threat (high vs. low vs. no information) and levels (high vs. low vs. no information) of two distinct conceptions of efficacy (self vs. response-efficacy). Efficacy beliefs may concern the perceived effectiveness of a behavioral response for attaining a specified goal (i.e., response-efficacy; e.g., “I can control the influence of implicit bias on my

Participants

Participants were 1005 White U.S. citizens recruited from Amazon Mturk (67.4% females; mean age = 37.27, SD = 12.20). Most participants report a family income greater than 50 K (52.34%) and have earned at least a Bachelor's degree (85.47%). Given the number of conditions included in this design, we targeted approximately 100 participants per cell, and then Mturk participants were oversampled to adjust for the inclusion of non-Whites in the sample. Because estimated sample size was determined

Results and discussion

In Experiment 4, we test Hypothesis 2 and 3, by comparing the Intervention (i.e., High Response-efficacy, High Self-efficacy, and Low Moral Threat) condition to the Feedback Only and No Feedback conditions. This confirmatory test was conducted using a one-way between-subjects ANOVA. Similarly, we explore the independent effect of each feature of the intervention by comparing those conditions to the Feedback Only condition using a one-way between-subjects ANOVA. Post-hoc pairwise comparisons for

Meta-analysis of studies 1–4

Next, we conducted a meta-analysis across all of our studies (N = 1489), in which we examine differences in defensive responding across the intervention (n = 571) and control (No Feedback n = 351; Feedback Only n = 567) conditions using two approaches. For both sets of analyses, we report the results of models that do not include any covariates. Table 7 reports the M(SD) for the key dependent variables used in this study, separated by and aggregated across study.

Discussion

Do interventions and workshops regulate individual-level bias? Are people receptive to learning about bias? Does education about the many subtle forms of bias having unintended effects on how one thinks and acts (for a review, see Banaji & Greenwald, 2016; Bargh, 2017) lead to the desired effect? The path from negative feedback to bias reduction is fraught. Critically, there are times it has the opposite of the desired effect, triggering resentment, anger, denial, polarization, motivated

Data availability

All of the data and data syntax are available at: https://osf.io/pbrdn/?view_only=b2070c097d7143f3917cb72215aa5c37

Author contributions

J.A. Vitriol is responsible for developing the initial project concept. Both authors were responsible for the theoretical framework, drafting the final manuscript, and the study design and data collection, analysis, and interpretation. Study 1, Study 2, and the Pilot Study reported in this paper was submitted by J.A. Vitriol to the Psychology Department at the University of Minnesota in partial fulfillments of the requirements for a doctoral degree.

Declaration of Competing Interest

The authors declare no competing interests.

Acknowledgements

The authors would like to thank Eugene Borgida, Mark Snyder, Christopher Federico, and Howard Lavine for their help in developing the study concept; Mahzarin R. Banaji and members of her lab, Dominic Packer, Adam Magerman, Naomi Rothman, Alexandera Sackett, and Jose Causadias for critical comments and feedback on early drafts and iterations of this project; Andrew Sell and Alicia Hofelich Mohr at the University of Minnesota for help with software design; Center for the Study of Political

References (90)

  • J.A. Vitriol et al.

    Ideological uncertainty and investment of the self in politics

    Journal of Experimental Social Psychology

    (2019)
  • I. Ajzcn

    The theory of planned behavior

    Organizational Behavior and Human Decision Processes

    (1991)
  • D.M. Amodio et al.

    Individual differences in the regulation of intergroup bias: The role of conflict monitoring and neural signals for control

    Journal of Personality and Social Psychology

    (2008)
  • D.M. Amodio et al.

    Intergroup anxiety effects on implicit racial evaluation and stereotyping

    Emotion

    (2012)
  • J.R. Axt et al.

    Reducing social judgment biases may require identifying the potential source of bias

    Personality and Social Psychology Bulletin

    (2018)
  • M.R. Banaji et al.

    Blindspot: Hidden biases of good people

    (2016)
  • A. Bandura

    Self-efficacy: Towards a unifying theory and the organization

    Psychological Review

    (1977)
  • A. Bandura

    Self-efficacy mechanism in human agency

    American Psychologist

    (1982)
  • A. Bandura

    Self-regulation of motivation and action through internal standards and goal systems

  • A. Bandura

    Self-regulation of motivation through anticipatory and self-reactive mechanisms

  • J. Bargh

    Before you know it: The unconscious reasons we do what we do

    (2017)
  • R.F. Baumeister et al.

    The intrinsic appeal of evil: Sadism, sensational thrills, and threatened egotism

    Personality and Social Psychology Review

    (1999)
  • A.J. Berinsky et al.

    Evaluating online labor markets for experimental research: Amazon.com’s mechanical Turk

    Political Analysis

    (2012)
  • M. Carnes et al.

    The effect of an intervention to break the gender bias habit for faculty at one institution: A cluster randomized, controlled trial

    Academic Medicine

    (2015)
  • E.C. Carter et al.

    What do we know about (implicit) bias and what does it mean for bias reduction training?

    Behavioral Science & Policy

    (2021)
  • C.S. Carver et al.

    On the self-regulation of behavior

    (1998)
  • C.S. Crandall et al.

    Social norms and the expression and suppression of prejudice: The struggle for internalization

    Journal of Personality and Social Psychology

    (2002)
  • A.M. Czopp et al.

    Standing up for change. Reducing bias through interpersonal confrontation

    Journal of Personality and Social Psychology

    (2006)
  • N.M. Daumeyer et al.

    Consequences of attributing discrimination to implicit vs. explicit bias

    Journal of Experimental Social Psychology

    (2019)
  • P.G. Devine et al.

    Prejudice with and without compunction

    Journal of Personality and Social Psychology

    (1991)
  • P.H. Ditto et al.

    Motivated skepticism: Use of differential decision criteria for preferred and nonpreferred conclusions

    Journal of Personality and Social Psychology

    (1992)
  • J.F. Dovidio et al.

    Aversive racism and selection decisions: 1989 and 1999

    Psychological Science

    (2000)
  • P. Forscher et al.

    A meta-analysis of change in implicit bias

    Psychological Bulletin

    (2019)
  • C.M. Frantz et al.

    A threat in the computer: The race implicit association test as a stereotype threat experience

    Personality and Social Psychology Bulletin

    (2004)
  • D.C. Funder et al.

    Evaluating effect size in psychological research: Send and nonsense

    Advances in Methods and Practices in Psychological Science

    (2019)
  • J.X. Goh et al.

    Mini meta-analysis of your own studies: Some arguments on why and a primer on how

    Social and Personality Psychology Compass

    (2016)
  • A.G. Greenwald et al.

    Understanding and using the implicit association test: I. An improved scoring algorithm

    Journal of Personality and Social Psychology

    (2003)
  • A.G. Greenwald et al.

    Understanding and using the implicit association test: III. Meta-analysis of predictive validity

    Journal of Personality and Social Psychology

    (2009)
  • F. Hansen

    Diversity’s business case: doesn’t add up

    Workforce

    (2003)
  • E.T. Higgins et al.

    Standards and the process of selfevaluation: Multiple affects from multiple stages

  • A.L. Hillard et al.

    Reactions to the implicit association test as an educational tool: A mixed methods study

    Social Psychology of Education

    (2013)
  • J.L. Howell et al.

    Managing the threat of impending implicit attitude feedback

    Social Psychological and Personality Science

    (2013)
  • J.L. Howell et al.

    Caught in the middle: Defensive responses to IAT feedback among whites, blacks, and biracial black/whites

    Social Psychological and Personality Science

    (2015)
  • J.L. Howell et al.

    Not your average bigot: The better than average effect and defensive responding to implicit association test feedback

    British Journal of Social Psychology

    (2016)
  • J.L. Howell et al.

    Responding defensively to IAT feedback

    Social Cognition

    (2017)
  • Cited by (24)

    • Equating silence with violence: When White Americans feel threatened by anti-racist messages

      2022, Journal of Experimental Social Psychology
      Citation Excerpt :

      Still, useful connections can be drawn between our research, and research focusing on the role of moral threat in White Americans' willingness to acknowledge individual (implicit) biases (Daumeyer, Onyeador, Brown, & Richeson, 2019; Vitriol & Moskowitz, 2021). Research suggests that White individuals might resist acknowledging that they hold implicit racial biases if they feel they will be morally condemned as racist for holding those implicit biases in the same way that they would be for expressing explicit prejudice (Vitriol & Moskowitz, 2021). Based on our research findings, it is likely that equalizing interpretations might contribute to these feelings of threat.

    • Biases left unattended: People are surprised at racial bias feedback until they pay attention to their biased reactions

      2022, Journal of Experimental Social Psychology
      Citation Excerpt :

      Given the remaining ambiguity of these results, we find it important to note that how IAT feedback is communicated may still be an important factor that influences how people react to IATs. For instance Vitriol and Moskowitz (2021) show in their studies that if IAT results are communicated such that participants feel less blamed and perceive more control over their biases, it reduces their defensiveness and increases bias awareness. In the present studies, we only wanted to test the effect of the feedback qualifiers (“slight”, “moderate”, and “strong”) specifically, and found only mixed to negative results.

    • Confronting intergroup biases: Validity and impugnment as determinants of other-confrontation consequences

      2022, Advances in Experimental Social Psychology
      Citation Excerpt :

      For example, after taking the White/Black Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998), people may be informed that their performance indicated racial bias because they responded faster when White people were paired with GOOD and Black people were paired with BAD than the reverse. When this feedback is presented without further explanatory information, people may react by denying that the test had anything to do with racial biases (Howell, Redford, Pogge, & Ratliff, 2017; Vitriol & Moskowitz, 2021). By recategorizing the confronted behavior as not biased, any discrepancy that otherwise would exist with one's personal standards can be resolved.

    • Responding to feedback about implicit bias

      2024, Social and Personality Psychology Compass
    View all citing articles on Scopus

    This paper has been recommended for acceptance by Dr Vanessa Bohns.

    View full text