The Mixed Effects of Online Diversity Training

Introduction

One-time diversity training is a common tool deployed by more than half of mid-size and large organizations seeking to promote equality in the workplace. However, the existing body of research on the effectiveness of diversity training is limited by a lack of field experiments and the difficulty of identifying and measuring objective behavioral outcomes. Because of these limitations, it is still somewhat unclear whether those trainings significantly impact attitudes and behaviors toward women and other historically marginalized and underserved groups. 

With collaboration from a large international organization, the authors designed and executed a diversity training module to gauge the effects of training on employee attitudes and workplace behaviors. The results of this study suggest that diversity training has varying effects on attitudinal and behavioral change based on prior individual attitudes and belonging to historically disadvantaged groups. 

Edward H. Chang is an Assistant Professor of Business Administration at Harvard Business School. Katherine L. Milkman is a Professor of Operations, Information and Decisions and the James G. Dinan Endowed Chair at the University of Pennsylvania’s Wharton School. Other contributors included Dena M. Gromet, the Executive Director of the Behavior Change for Good Initiative, Robert W. Rebele, a Senior Research Fellow at The Wharton People Analytics Initiative, and Cade Massey, who is a Practice Professor in the Wharton School’s Operations, Information and Decisions Department. Lastly, Angela L. Duckworth, the Founder and CEO of Character Lab, and Adam M. Grant, the Saul P. Steinberg Professor of Management also contributed to the paper. 

Methods and Findings

The authors used an online training module to gauge the effects of diversity training on inclusive attitudes and behaviors toward women with a secondary analysis of its impact on attitudes and behaviors toward racial minorities. Employees of a large organization were asked to participate by voluntarily completing the training module as part of a broader strategy to promote inclusion and inclusive leadership. 

Participants were randomly assigned to one of three experimental conditions, including a gender-bias group, a general-bias group, and a control group. In the treatment groups, participants received instruction in how stereotypes reinforce bias and adversely impact equity and inclusion in the workplace. Control group participants received structurally similar training on psychological safety and active listening with no stereotyping content.

The study found significantly increased attitudinal support for women among treatment groups relative to the control group, driven by non-U.S. participants. Contrastingly, U.S. employees in the treatment group showed no significant change in attitudinal support for women. Participants in the treatment group also demonstrated increases in willingness to acknowledge their own gender and racial biases and gender-inclusive intentions, although the latter effect was again driven by non-U.S. employees. Behaviorally, U.S. employees in the treatment group nominated more women for an office-wide mentorship program and for office-wide excellence recognition by a three-week follow-up date. 

Conclusions

In a body of research that has provided inconsistent evidence on the impact of diversity training, this study found that the effectiveness of training may be dependent on the modules’ audience and their preexisting attitudes toward women and other social groups. Specifically, the authors suggest that the presence of attitudinal effects exclusively for non-U.S. employees and the presence of behavioral effects exclusively for U.S. employees may be related to the fact that non-U.S. employees on average reported initial lower attitudinal support toward women, indicating more room to grow attitudinally but lower susceptibility to behavioral change.

The study also highlights unexpected drivers of behavioral change, having found in additional explanatory analyses that increased mentorship of U.S. women was driven by those employees seeking out mentorship proactively following the treatment. This lack of behavioral change among dominant group members indicates a need for additional strategies to create a truly equitable, inclusive, and representational work environment for women and racial minorities. 

The authors acknowledge that this study is limited by its analysis of employees from a single organization, and on a narrow set of attitudinal and behavioral outcomes. To overcome these limitations, future research may explore the behavioral and attitudinal effects of diversity training in organizations with varying levels of pre-existing attitudinal support for women and racial minorities. Future research may also study the mechanisms through which diversity training motivates women and other underrepresented groups to seek mentorship opportunities with senior officials in their organization.