Coolest debiasing study (I’ve) ever (read)

So this is another installment (only second; first here) in my series on cool studies that we read in my fall Law & Cognition seminar at HLS.

This one, Sommers, S.R. On Racial Diversity and Group Decision Making: Identifying Multiple Effects of Racial Composition on Jury Deliberations. Journal of Personality and Social Psychology 90, 597-612 (2006), looked at the impact of the racial composition of a (mock) jury panel on white jurors. Sommers found that white jurors on mixed African-American/white panels were more likely than ones on all-white to form pro-defendant fact perceptions and to support acquittal in a case involving an African-American defendant charged with sexual assault of a white victim.

That’s plenty interesting — but the really amazing part is that these effects were not a product of any exchange of views between the white and African-American jurors in deliberations. Rather they were a product of mental operations wholly internal to the white subjects.

There were two sorts of evidence for this conclusion. First, Sommers found that the pre-deliberation verdict preferences of the white subjects on the mixed juries was already more pro-defense than the preferences of those on the all-white juries. Second, during deliberations the white subjects on the mixed juries were more likely to mention pro-defendant evidence spontaneously (that is, on their own, without prompting the by African-American ones) and less likely to inject mistaken depictions of the evidence into discussion.

In sum, just knowing that they would be deliberating with African-American jurors influenced — indeed, demonstrably improved the quality of — the cognition of the white jurors.

How many cool things are going on here? Lots, but here are some that really register with me:

1. OCTUSW (“Of course–that’s unsurprising–so what”) response is DOA & No ITMBSWILESP, either!

OCTUSW is a predictable, lame response to a lot of cool social science studies. What makes it lame is that it is common to investigate phenomena for which there are plausible competing hypotheses; indeed, the clash of competing plausible hypotheses is often what motivates people to investigate. This is one of the key points in Duncan Watt’s great book, Everything Is Obvious Once You Know the Answer.

But here the result was a real surprise (to me and my students, at least) — so we can just skip the 10 mins it usually takes to shut the (inevitably pompous & self-important) OCTUSW guy up.

At the same time, the result isn’t insane-there-must-be-something-wrong-it’s-like-ESP (ITMBSWILESP), either. ITMBSWILESP results can take up 30-40-50 mins & leave everyone completely uncertain whether (if they decide the study is valid, reliable) they’ve been duped by the researcher or (if they dismiss it out of hand) they’ve been taken in by their own vulnerability to confirmation bias.

2. Super compelling evidence that unconscious bias is defeating moral commitments of those experiencing it.

The results in this study suggest that the white subjects on the all-white juries were displaying a lower quality of cognitive engagement with the evidence than the whites on the mixed-race juries. Why?

The most straightforward explanation (and the animating conjecture behind the study) was that the racial composition of the jury interacted with unconscious racial bias or “implicit social cognition.” Perhaps they were conforming their view of the evidence to priors founded on the correlation between race and criminality or were failing to experience a kind of investment in the interest of the defendant that would have focused their attention more effectively.

Knowing, in contrast, that they were on a jury with African Americans, and would be discussing the case with them after considering the evidence, jolted the whites on the mixed juries into paying greater attention, likely becuause of anxiety that mistakes would convey to the African-American subjects that they didn’t care very much about the possibility an African-American was being falsely accused of an interracial sexual assault. Because they paid more attention, they in fact formed a more accurate view of the facts.

But this “debiasing” effect would not have occurred unless the unconscious racial bias it dispelled was contrary to the white subjects’ conscious, higher-order commitment to deciding the case impartially.

Obviously, if the white subjects in the study were committed, conscious racists, then those who served on the mixed-race juries would have gotten just as much satisfaction from forming anti-defendant verdict preferences and inaccurate, anti-defendant fact perceptions as ones on the all-white juries.

Likewise, it is not very plausible to think the whites on the mixed-race juries would have been jolted into paying more attention unless they had a genuine commitment to racial impartiality. Otherwise, why would the prospect that they’d be perceived otherwise have been something that triggered an attention-focusing level of anxiety?

The conclusion I draw, then, is that the effect of unconscious bias on the jurors in the all-white juries is something that they themselves would likely have been disappointed by.  They and others in their position would thus concur in, and not resent, the use of procedures that reduce the likelihood that this cognitive dynamic will affect them as they perform that decisionmaking task.

That’s a concluison, too, that really heartens me.

My own research on “debiasing” cultural cognition rests on the premise that identity-protective cognition (a cousin of implicit social cognition) disappoints normative commitments that ordinary citizens have. If that’s not true–if in fact, individuals would rather be guided reliably to conclusions that fit the position of “their team” than be right when they are evaluating disputed evidence on issues like climate change and the effectiveness of the HPV vaccine — then what  I’m up to is either pointless or (worse) a self-deluded contribution to public manipulation.

So when I see a study like this, I feel a sense of relief as well as hope!

 3. The debiasing effect can’t be attributed to any sort of “demand effect.”

This is a related point. A “demand effect” describes a result that is attributable to the motives of the subjects to please the researcher rather than to the cognitive mechanism that the researcher is trying to test.

One common strategy that sometimes is held forth as counteracting motivated cognition — explicitly telling subjects to “consider the opposite” — is very vulnerable to this interpretation. (Indeed, studies that look at the effect of explicit “don’t be biased” instructions report highly variable results.)

But here there’s really no plausible worry about “demand effect.” The whites on the mixed-race juries couldn’t have been “trying harder” to make the researchers happy: they had no idea that their perceptions were being compared to subjects on all-white juries, much less that those jurors were failing to engage in the evidence in as careful a way as anyone might have wanted them to.

4. The effect in this study furnishes a highly suggestive model that can spawn hypotheses and study designs in related areas.

Precisely because it seems unlikely to me that simply admonishing individuals to be “impartial” or “objective” can do much real good, the project to identify devices that trigger effective unconscious counterwights to identity-protective cognition strikes me as of tremendous importance.

We have done a variety of studies of this sort. Mainly they have focused on devices — e.g., message framings, and source credibility — that neutralize the kinds of culturally threatening meanings that provoke defensive resistance to sound information.

The debiasing effect here involves a different dynamic. Again, as I understand it, the simple awareness that there were African-Americans on their jury activated white jurors’ own commitment to equality, thereby leading them to recruit cognitive resources that in fact promoted that commitment.

Generalizing, then, this is to me an example of how effective environmental cues (as it were) can activate unconscious processes that tie cognition more reliably to ends that individuals, at least in the decisionmaking context at hand, value more than partisan group allegiances.

Seeing the study this way, I now often find myself reflecting on what sorts of cues might have analogous effect in cultural cognition settings.

That’s something cool studies predictably do. They not only improve understanding of the phenomena they themselves investigated. They also supply curious people with vivid, generative models that help them to imagine how they might learn, and teach others something, too.

Leave a Comment