Implicit bias training has little effect on police officers' behaviour

Police departments across the United States are implementing training programmes aimed at reducing racial inequality in policing. Many focus on reducing unconscious biases — a bias towards automatically perceiving a Black person as posing more of a threat than a White person, for example. But do these programmes actually change how officers behave?

This has been hard to answer, as most training programmes have not been evaluated empirically, write Calvin K. Lai at Washington University and Jaclyn A. Lisnek at the University of Virginia in Psychological Science. However, the pair has now evaluated a popular one-day programme. Their findings "suggest that diversity trainings as they are currently practiced are unlikely to change police behaviour."

Lai and Lisnek evaluated the 'Managing Bias' programme, which was devised by the Anti-Defamation League, a non-governmental organisation for combating discrimination, and which is run by pairs of external educators. The interactive workshop starts with activities designed to improve the participants' awareness of implicit (and explicit) biases, and to help them to appreciate how these biases might shape their behaviour. Officers are then taught practical skills aimed at reducing the potential impact of these biases. They learn five strategies, including the use of 'stereotype substitution' (replacing thoughts of negative stereotypes with positive mental images) and taking another person's perspective.

In total, Lai and Lisnek evaluated 251 Managing Bias training sessions conducted in police departments in which officers had shown unequal treatment of Black people, stopping them more regularly or using force more often than with White individuals. Some of these sessions were run pre-Covid, and entirely in person, and the participants completed surveys immediately before and immediately after the training.

Some of the sessions were run during the pandemic, however. In these cases, the training was run in two stages, the first via self-paced modules, the second by video-conferencing. These participants completed the surveys as before but also a third survey, one month later.

When Lai and Lisnek analysed the responses to the surveys conducted immediately before and after the programme, they found that the training was effective at increasing officers' knowledge about bias and their concerns about bias. After the training, the participants also reported being motivated to use the highlighted strategies to manage bias. Some of the comments made by the officers seemed to support this finding: "One officer said that they 'will definitely use [these tools] in future', and another said it was 'great information that I will utilize for the rest of my career."

However, when Lai and Lisnek looked at the follow-up surveys conducted one month later, they found that participants did not follow through on these intentions. While the follow-up surveys showed that participants remembered what they'd learned about bias, and still felt the strategies were feasible, they were actually using these strategies less often than before the training even started.  Their level of concern about bias had also dropped back down. The training programme "was ineffective at durably increasing concerns or strategy use," the pair concludes.

What might explain this? Perhaps officers had tried the strategies, but ended up feeling that they weren't helpful in their everyday work, the team suggests. Alternatively, perhaps they had initially felt they had been effectively using the strategies to manage bias, but the training had made them realise that they hadn't been doing so properly. This might account for why they subsequently rated themselves as using these strategies less often.

This is far from the only study to conclude that anti-bias diversity training has only limited impacts on behaviour, or even fails to change it. Lai and Lisnek suggest that perhaps a one-day, one-off programme run by outsiders isn't sufficient — and that, to make a real difference, anti-bias, diversity training should be integral part of the activity of a police department. "Future research may find that greater investment and integration of training programs into organization-wide initiatives could increase their impact," they write.

This may turn out to be the case. However, a serious point of caution has to be made about the sample sizes in this study. Lai and Linsek report gathering "usable data" on a total of 3,764 officers who had completed both the survey immediately before and immediately after the training programme. However, only 53 of these officers also completed the third survey one month later.  In fact, the researchers gathered a total of just 300 usable third surveys (mostly from officers who had not completed both of the initial surveys).

This means that their main conclusion — that current programmes are unlikely to change police behaviour — is based on a very small sample. More research on anti-bias programmes aimed at police officers is certainly needed — but not least to better evaluate the current programmes.