The one-size-fits-all diversity training model is broken – here’s a better alternative

Diversity training is more effective^[1] when it’s personalized, according to my new research in the peer-reviewed journal Applied Psychology.

As a professor of management^[2], I partnered with Andrew Bryant^[3], who studies social marketing, to develop an algorithm that identifies people’s “personas,” or psychological profiles, as they participate in diversity training in real time. We embedded this algorithm into a training system that dynamically assigned participants to tailored versions of the training based on their personas.

We found that this personalized approach worked especially well for one particular group: the “skeptics.” When skeptics received training tailored to them, they responded more positively – and expressed a stronger desire to support their organizations’ diversity efforts – than those who received the same training as everyone else.

In the age of social media, where just about everything is customized and personalized, this sounds like a no-brainer. But with diversity training, where the one-size-fits-all approach still rules^[4], this is radical. In most diversity trainings, all participants hear the same message, regardless of their preexisting beliefs and attitudes toward diversity. Why would we assume that this would work?

Thankfully, the field is realizing the importance of a learner-centric approach. Researchers have theorized that several diversity trainee personas exist^[5]. These include the resistant trainee, who feels defensive; the overzealous trainee, who is hyper-engaged; and the anxious trainee, who is uncomfortable with diversity topics. Our algorithm, based on real-world data, identified two personas with empirical backing: skeptics and believers. This is proof of concept that trainee personas aren’t just theoretical – they’re real, and we can detect them in real time.

But identifying personas is just the beginning. What comes next is tailoring the message. To learn more about tailoring, we looked to the theory of jujitsu persuasion^[6]. In jujitsu, fighters don’t strike. They use their opponent’s energy to win. Similarly, in jujitsu persuasion, you yield to the audience, not challenge it. You use the audience’s beliefs, knowledge and values as leverage to make change.

In terms of diversity training, this doesn’t mean changing what the message is. It means changing how the message is framed. For example, the skeptics in our study still learned about the devastating harms of workplace bias. But they were more persuaded when the message was framed as a “business case” for diversity, rather than a “moral justice” message. The “business case” message is tailored to skeptics’ practical orientation. If diversity training researchers and practitioners embrace tailoring diversity training to different trainee personas, more creative approaches to tailoring will surely be designed.

Why it matters

The Trump administration is leading a backlash^[7] against diversity initiatives, and a backlash to that backlash^[8] is emerging. This isn’t entirely new: Diversity has long been a contentious issue.

Organizations like the Pew Research Center^[9], the United Nations^[10] and others have consistently reported a conservative-liberal split^[11], as well as a male-female split, around diversity. Diversity training has done little to bridge these gaps^[12].

For one, diversity training is often ineffective^[13] at reducing bias and improving diversity metrics in organizations. Many organizations treat diversity training efforts as a box-checking exercise^[14]. Worse, it’s not unusual for such efforts to backfire^[15].

Our research offers a solution: Identify the trainee personas represented in your audience and customize your training accordingly. This is what social media platforms like Facebook do: They learn about people in real time and then tailor the content they see.

To illustrate the importance of tailoring diversity training specifically, consider how differently skeptics and believers think. One skeptic in our study – which focused on gender diversity training – said: “The issue isn’t as great as feminists try to force us to believe. Women simply focus on other things in life; men focus on career first.” In contrast, a believer said: “In my own organization, all CEOs and managers are men. Women are not respected or promoted very often, if at all.”

Clearly, trainees are different. Tailoring the training to different personas, jujitsu style, may be how we change hearts and minds.

What still isn’t known

Algorithms are only as good as the data they rely on^[16]. Our algorithm identified personas based on information the trainees reported about themselves. More objective data, such as data culled from human resources systems, may identify personas more reliably.

Algorithms also improve as they learn over time. As artificial intelligence tools become more widely used in HR, persona-identifying algorithms will get smarter and faster. The training itself needs to get smarter. A one-time training session, even a tailored one, stands less of a chance at long-term change compared with periodic nudges^[17]. Nudges are bite-sized interventions that are unobtrusively delivered over time. Now, think about tailored nudges. They could be a game changer.

The Research Brief^[18] is a short take on interesting academic work.

References

1. ^{^} Diversity training is more effective (doi.org)
2. ^{^} professor of management (scholar.google.com)
3. ^{^} Andrew Bryant (csbapp.uncw.edu)
4. ^{^} still rules (journals.aom.org)
5. ^{^} several diversity trainee personas exist (doi.org)
6. ^{^} theory of jujitsu persuasion (doi.org)
7. ^{^} leading a backlash (www.whitehouse.gov)
8. ^{^} backlash to that backlash (theconversation.com)
9. ^{^} Pew Research Center (www.pewresearch.org)
10. ^{^} United Nations (www.un-ilibrary.org)
11. ^{^} conservative-liberal split (doi.org)
12. ^{^} little to bridge these gaps (doi.org)
13. ^{^} often ineffective (doi.org)
14. ^{^} box-checking exercise (doi.org)
15. ^{^} to backfire (www.hup.harvard.edu)
16. ^{^} only as good as the data they rely on (doi.org)
17. ^{^} periodic nudges (doi.org)
18. ^{^} Research Brief (theconversation.com)