« Proving Discrimination: Why Armstrong's Gotta Go. | Main | Springtime for Bichons »
April 8, 2006
Think There's Such A Thing As Colorblindness or Gender-Blindness? Think Again.
Shih studies, among other things, the impact of stereotype and social identity on performance. Her work explodes the notion that anyone really is "colorblind" or "gender-blind," even if they think they are.
In one study, for example, Shih and her assistants gave a group of female Asian-American undergrads a math test. Before taking the test, some of them had to fill out a questionnaire that highlighted their Asian identity. Some of them had to fill out a questionnaire that highlighted their status as female. And some (a control group) had to fill out a questionnaire about their telephone company. The results: the women whose Asian identity had been highlighted scored significantly better than the control group, and the women whose gender had been highlighted scored significantly worse!
In another experiment, subjects had an email exchange with somebody they thought was a Harvard student, but was actually the researcher. At some point in the exchange, the researcher mentioned her SAT scores. Some of the subjects were corresponding with a person whose email address was "chen@harvard.edu," some were corresponding with a person whose email address was "Amy@harvard.edu," and some were corresponding with a person whose email address was "ac@harvard.edu." After the exchange, subjects were asked to recall their correspondent's SAT scores. Those who were corresponding with the email address that implied Asian ancestry ("chen@harvard.edu") remembered that their correspondent scored higher on the math test than the verbal test. And those who were corresponding with the email address that implied their correspondent was female ("Amy@harvard.edu") remembered that their correspondent scored higher in verbal than in math.
Fascinating and troubling.
UPDATE: This piece from last week's Newsweek is not entirely unrelated.
Posted by Eric at April 8, 2006 7:16 AM
Comments
According to the brief overview, the first study was based on a 12-question math test and 46 students broken down into three study arms (or perhaps about 15 each for the ethnicity, gender, and neutral arms).
With this brief test and small groups, the differences would have to be very large for her to make those claims.
As you attended this presentation, did you get anything on statistical significance testing or even data on mean number of answers correct per arm? There is also mention in the abstract of using "controls" for other factors. Do you know what she controlled for in this tiny population?
[ELM: No, I didn't get any such info. She was presenting to law profs, not scientists, so I imagine that her presentation was tailored to go light on the design study/statistical measures stuff.]
Posted by: F. McBride at April 8, 2006 11:31 AM
These results are not new. There are a fair number of social pyschologiss working in the area who have obtained similar results over a wide range of tests, although most of the other studies I have come across focused on blacks and whites. So this should not be seen as controversial (which is more in the way of a response to the comment above).
It is not clear to me that the inferences we should draw have that much to do with stereotyping, as that strikes me as about perceptions of others. It seems to me to be more about expectations.
Posted by: T. Gracchus at April 8, 2006 7:03 PM
Sounds like a study that would have gone very well with a book I recently read called Blink, about how we make "snap" judgements. I reviewed it on my blog and you can see any number of very interesting and sometimes troubling things:
http://jiggle.anaze.us/archives/2006/03/march_book_reading.html
Posted by: Jonathan Arnold at April 8, 2006 7:13 PM
Is there an actual debate about whether discrimination exists? I don't think so. Discrimination is real and that is why we must be as objective as possible about grading, judging, sentencing, etc.
Let's look at the actual behavior of the subjects. Do you know which group scored the best on the math test? Did the "identified" asians actually get more correct answers or was there some subjectivitiy to the grading? I guess my main question is HOW the discrimination affected the results? I didn't get that from the pages that were connected.
Proof that we do not live in a colorblind society is not an excuse to not insitute colorblind policies.
Posted by: Greg at April 9, 2006 8:08 AM