8 thoughts on “Defund Implicit Bias Research

  1. “The z-curve plot for traditional articles in this literature looks even worse”: I didn’t get the difference for the 214 vs 154 studies analysis – could you explain how the studies are different? thanks

    1. The analysis excluded the two articles by Lai et al. (2014, 2016) that are discussed in more detail below. These articles also differ from the typical articles in that they follow Open Science practices and used large samples.

  2. ah ok I see now – two papers with this many empirical results is very unusual so it threw me – but looking them up they are quite unusual.

    It does seem to be something of a pattern across psychology – a “striking” result in a sexy journal comes out with an implausibly large effect size (that no-one questions) with a tiny sample, which hints towards exciting real world applications, grants get written and funded, a deluge of n = tiny studies get run (“assuming Farkwardinson et al.’s effect size of 1.8 means our n = 10 has 98% power…”) which have to be p-hacked into publications to have something to show from it, till eventually its realised that the effect sizes are pretty small and close to null in a lot of cases and actually very hard to reproduce even in the lab with 20 year old psychology students, and so really no hope to ever see any real world applications. But it was fun while it lasted.

    1. I think you describe the way experimental social psychology works pretty well. I hope you agree that it is pointless to fund this kind of silly research and to use funding for projects that can actually make a difference.

  3. Excellent article. I do not think Greenwald’s original work was subjected to the same level of scrutiny as others might have been. Worst is the crazy logic surrounding the relationship between explicit and implicit measures. Any size of correlation can be made to fit the conclusion “The IAT measures something different and valuable” (typically made when the relationship is small/non-significant). I’d love to know how we might falsify the claim that the IAT is a bona-fide measure of implicit attitudes. To me, evidence is still lacking (or rather I have concerns over validity) after over 20 years.

    1. The main point is that implicit bias is a poorly defined construct without a good measure. That makes it impossible to say something meaningful about it. The most plausible form of implicit forms of bias are those where individuals fail to attribute their behavior to race (e.g., call the police on somebody without thinking it was because of race, when race was a factor). In my opinion, systemic and explicit racism by a powerful minority are bigger problem (see http://dx.doi.org/10.1037/xge0000983)

Leave a Reply