Unconscious Bias Training: An assessment of the evidence for effectiveness

Introduction

Unconscious bias training: An assessment of the evidence for effectiveness by Doyin Atewologun, et al. is a literature review and meta-analysis of studies that examines the evidence for the effectiveness of unconscious bias  training programs in actually reducing unconscious bias in individuals and organizations. The authors first define unconscious bias as the automatic and unintentional stereotypes, attitudes, and beliefs that people hold about particular groups of people. They argue that these biases can have negative effects on people’s behaviors and decision-making, leading to discrimination and inequality. To address this issue, many organizations have started to implement unconscious bias training (UBT) programs, which aim to increase awareness of these biases and reduce their influence. 

UBT interventions target the automatic behavioral functions in the brain which employ heuristics, or mental short-cuts, to process the large amounts of information individuals receive quickly so that people can carry out tasks efficiently. These mental short cuts can reinforce negative social stereotypes, especially for women, ethnic minorities, differently abled individuals, neurodiverse individuals, and others with protected characteristics. Atewologun et al. reference a previous intervention carried out by Baroness McGregor-Smith, member of the UK House of Lords, entitled, “Race in the Workplace.” In the 2017 study, Mcgregor-Smith examined the deep structural and historical biases that prevented individuals with protected characteristics from progressing in their careers. Mcgregor-Smith recommended that the UK Government implement a free digital unconscious bias training resource, which became a widely utilized resource in both the UK public and private sectors. The authors reevaluate these interventions a year following their initial implementation.

Doyin Atewologun is the Dean of the Rhodes Scholarship at the University of Oxford and is an expert on diversity, leadership, organizational culture, and intersectionality.  Tinu Cornish is an Occupational Psychologist who focuses on psychological approaches to diversity and inclusion leadership. Fatima Tresh is a psychologist and expert in human cognition and behavior addressing diversity, equity, and inclusion barriers. 

Methods and Findings

Atewologun et al.’s assessment aims to achieve three objectives evaluating UBT program implementations across the UK: (1), to demonstrate evidence in favor of UBT’s effectiveness (2) to analyze the contextual conditions under which UBT is most effective; and (3), to highlight evidence gaps for further research. The authors underscore that “effectiveness” in this assessment is contingent on the aims of the training designer. A 3-part rapid evidence assessment methodology was utilized to assess the UBT’s effectiveness:

Identify evidence from online databases

• Published peer reviewed articles (N=57)
• Non-academic searches like reports (N=31)

Evaluate the quality of the evidence

• Utilize the Maryland Scale of Scientific Methods (ranking scale of 1-5: 1 exhibiting the lowest rigor (before-after comparison) 5 exhibiting the highest rigor (randomized control trial))
• Exclude sources with low scientific rigor (Level 1)

Analyze the evidence 

• Identify aims and design of the intervention, draw conclusions about outcomes, and review the gathered evidence (N=18)

Among the key findings in the article, the authors highlight four types of UBT interventions they will measure: (1) awareness raising, (2) implicit bias change, (3) explicit bias change, and (4) behavior change. 

Awareness Raising 

Among the selected studies, eleven studies explicitly aimed to raise awareness of implicit bias, and the authors conclude in their analyses that UBT interventions can substantially increase awareness of bias. Whatley’s 2018 Implicit Association Test highlighted a successful intervention. In the study, Whatley measured UBT’s effectiveness on a multidisciplinary team’s attitudes toward African American students in special education. Whatley’s pre- and post- evaluations of a bias literacy workshop indicated the UBT intervention substantially improved both staff vulnerability to bias and individual student expectations.

Implicit Bias Change

Of the eleven studies aiming to change implicit bias, Atewologun et al. found that there is mixed evidence for UBT’s effectiveness. Two studies suggested that UBT can reduce the strength of bias; yet, there was no evidence that UBT can reduce bias to the extent of “neutral” preference. In Girod et al.’s 2016 evaluation of an educational presentation on reducing gender bias, 281 faculty members across 13 clinical departments at Stanford University indicated in the pre-trial a slight preference for males in leadership positions (this was consistent across all racial groups), and this male preference reduced in the post-trial measure. Yet, the authors concluded that males and older participants held stronger racialized and gendered implicit biases in both the pre- and post- trial measures. 

Explicit Bias Change

Nine studies reviewed in the assessment indicated that UBT is effective in changing explicit bias, but less effective than awareness raising or implicit bias changing . Of the available research, it was unclear to the authors how to best measure explicit bias. In Moss-Racusin et al. ‘s 2016 “scientific diversity” workshop administered to 126 life sciences instructors, the aim of the study was to increase awareness of gender diversity, reduce gender bias, and increase diversity-promoting actions. In this particular study, all three aims were effectively achieved. 

Behavior Change

Of the ten studies aiming to change behavior, only two of these studies actually measured behavior change Because of this limitation, the authors concluded that there is insufficient evidence to indicate UBT’s effectiveness. Research examining behavior change is limited, and  methods evaluating behavior change have low validity because they do not measure actual observed behavior change. In Forscher’s 2017 UBT intervention administered to 292 United States university students, researchers found that the effects of unconscious bias awareness waned two weeks post-intervention. However, a follow-up study conducted two years later between the intervention group and a control group indicated possible long-term behavior change. Participants in the intervention group were more likely to publicly refute an essay endorsing racial stereotyping than the control group.

Conclusions

In the concluding remarks, the authors highlight a variety of high-level observations about the assessed literature. Among these observations, the most notable include:  

• Male participants hold stronger unconscious gender biases than female participants. This gap can be reduced with UBT, and UBT may be more effective for men with respect to gender biases. There is some evidence that online and face-to-face UBT are equally effective for awareness raising.
• Mandatory UBT is generally more effective for behavior change than voluntary UBT. This is not supported by more rigorous studies, however. 
• Bias reduction strategies are more effective for reducing implicit bias and ineffective in reducing explicit bias.
• Some mindfulness interventions can reduce implicit bias and possibly mitigate discriminatory actions. 
• Bias mitigation strategies may have back-firing effects if participants do not want to be influenced or do not agree with the proposed direction of influence. 

In the final remarks, the authors set forth a series of recommendations for practice and future research. The first recommendation is to have a more nuanced approach to UBT content: use an implicit association test to increase awareness of unconscious bias and measure changes in implicit bias, educate participants about unconscious bias theory, and integrate bias reduction strategies in the UBT to increase participant confidence in managing their biases. The second recommendation focuses on the UBT context, such as delivering training to those who work closely together in a team unit or otherwise. Finally, the third recommendation is to evaluate effectiveness: randomly assign matched participants to control and intervention groups and deliver training to control groups when effectiveness has been established.

The authors suggest multiple areas for further research: systematic comparisons of approaches and design characteristics, investigations of UBT’s effectiveness in reducing bias against all protected groups, uniform measurement outcomes of UBT, structural changes due to UBT interventions, additional cognitive or social processes integral to maintaining inequity, and the impact of mandatory versus voluntary attendance on UBT effectiveness.