I've been looking at Project Implicit, which says it "blends basic research and educational outreach in a virtual laboratory at which visitors can examine their own hidden biases. Project Implicit is the product of research by three scientists whose work produced a new approach to understanding of attitudes, biases, and stereotypes." There's a demonstration site for the Implicit Association Test, where you can complete their Three Countries or U.S. Election tasks. I tried them both:
- For Three Countries, you complete "short questionnaires and tasks in which you will identify words associated with China, India, and Japan and group them into categories as quickly as possible." First they asked if I preferred one country over another. Then they presented a series of words -- if I saw a "good" word, or a word associated with the particular country being tested, I was supposed to press "P" on my keyboard; for all others, I pressed "Q." So for India, I pressed "P" for things like "friend" and "curry". And so on.
- During the U.S. Election timed task, they showed photos of Obama and McCain, and I quickly clicked either "E" or "I" on my keyboard to indicate which I saw. Then they showed "good" and "bad" words and I did the same thing -- finally, they associated those things together.
What does the evidence say? Here are some of the conclusions they've drawn: "Findings observed in seven years of operation of the Project Implicit web site:
- Implicit biases are pervasive. They appear as statistically "large" effects that are often shown by majorities of samples of Americans. Over 80% of web respondents show implicit negativity toward the elderly compared to the young...
- People are often unaware of their implicit biases. Ordinary people, including the researchers who direct this project, are found to harbor negative associations in relation to various social groups (i.e., implicit biases) even while honestly (the researchers believe) reporting that they regard themselves as lacking these biases.
- Implicit biases predict behavior. From simple acts of friendliness and inclusion to more consequential acts such as the evaluation of work quality, those who are higher in implicit bias have been shown to display greater discrimination."
Gaining ground? The IAT folks say "The published scientific evidence is rapidly accumulating. Over 200 published scientific investigations have made use of one or another version of the IAT." On the Media's Brooke Gladstone interviewed one of the IAT creators, Tony Greenwald, recently (listen to Hidden Persuaders) -- but some commenters were unimpressed: Matt in Arlington, VA says "The test does not reveal anything at all! Matching words and images has absolutely no external validity. OTM should have done their due diligence before elevating a parlor game to the prominence of coverage on NPR. That due diligence would include familiarizing yourself with the statistical and methodological theories underlying experimental designs that Tony Greenwald purports to operate within. He does not operate within the methodological mainstream because he ignores the external validity of IAT."
Transparent evidence? Sort of. Related research articles are easily accessible from Project Implicit. MEGO alert! But my eyes glazed over, so proceed with caution: These are typical academic research articles, filled with dense language and an apparent inability to just get to the point. (See example* at the end of this post.)
Is there a practical application? I'm skeptical. Is this a big "so what," or can this approach teach us something meaningful about ourselves? Greenwald et al.* have said that "The IAT’s property of producing a palpable and possibly unsettling reaction during performance may be its central asset." Project Implicit offers corporate services, workshops, etc. that might have some value: "We provide a compelling face to face demonstration of the pervasive role of unconscious mental processes in everyday behavior. We can develop segments drawn from organizational situations that elicit unintended biases." Now if they could develop a way to test for implicit biases against evidence -- and the pursuit of evidence-based management -- we'd really have something.
*Excerpt: "Any new psychological measure is under challenge to establish its validity. Validity may include theoretical value (construct validity), empirical value (predictive validity), or applied value (ecological and consequential validity) [Ed. note: So far, so good.] Blanton and Jaccard... offered a new validity label, arbitrariness: 'We define a metric as arbitrary when it is not known where a given score locates an individual on the underlying psychological dimension or how a one-unit change on the observed score reflects the magnitude of change on the underlying dimension'.... Like Blanton and Jaccard, we use meaningfulness as the opposite of arbitrariness, and we therefore take metric meaningfulness to be the appropriate label for their preferred alternative. Blanton and Jaccard’s (2006) conception of metric meaningfulness is similar to and is largely contained within the concept of consequential validity that was described in this journal by Messick (1995, also cited by Blanton & Jaccard). Messick defined consequential validity as the aspect of construct validity that 'appraises the value implications of score interpretation as a basis for action as well as the actual and potential consequences of test use, especially in regard to sources of invalidity related to issues of bias, fairness, and distributive justice'"
From Greenwald, A. G., Nosek, B. A., & Sriram, N. (2006). Consequential validity of the Implicit Association Test: Comment on the article by Blanton and Jaccard. American Psychologist, 61(1), 56-61.
Comments