What's the difference?

As a teacher I've been particularly curious about the idea of having a biased test. I'm a bit of a numbers junkie, so I've analyzed the scores of basically every test I've ever given. Mostly I've broken things down by grade level and gender as I've yet to have a class with a sizable ethnic minority. In my physics class, males pretty much always outperform females on average on the exams. It made me wonder if there was something in my delivery of the material, in the wording of the questions, in some aspect of the classroom environment that would put females at a disadvantage.

Last summer I went to national conference on physics education and saw that there were numerous research groups who were investigating what they termed the "gender gap" which was apparently ubiquitous throughout the physics world. It put me at ease knowing that my results were the same as everyone else's.

So what is going on here? I see three possibilities for any widespread results on a test that are discrepant between certain groups:
1. Genetic differences - one groups brain actually processes information differently
2. Support differences - one group has more support/encouragement
3. Differences in traditions - one group is more familiar with certain types of analogies/analysis


For a gender difference within the same ethnicity (white males consistently score higher in math, but lower in english on standardized tests), it could be that a certain gender is influenced by society to be more interested in certain subjects. Or it could be that their brains actually function differently (a rather touchy subject, but quite substantiated by neurology), and that difference somehow affects the performance of certain activities.

For a discrepancy between ethnic groups within the same country, differences in stable, supportive homes is a major factor. I think that differences in ethnic traditions can play a role as well, but how large a role is a tough question. I know a professor that taught physics to exiled Buddhist monks and told me about how he had to drastically change the analogies and wording he would normally use. Yet at the same time the logic (there's that word again) required to solve the problem was the same, it was just couched in a different setting.

The New Haven firefighter case is an interesting situation. Here we have adults, largely of the same gender, taking an exam in a field within which they all had experience. Would it be possible to write an exam that favored one group over another in such a situation? As adults, home support seems like less of an issue (not that that would be a bias in the exam, just a factor in the results). I know of no evidence for a difference in brain function between african americans and caucasian americans. So the only possibility I see is if the test somehow favored the ethnic traditions of one group. While I guess this is possible, it seems highly unlikely.

Comments

Anonymous said…
There is significant data showing that blacks just do worse at school and on test and the like. That's why affirmative action helps them with entrance to schools-- There are not enough qualified blacks to get proportionate representation at institutions of higher education (like law school for example). Thus, to get proportionate representation, those institutions must accept black students that are generally less qualified than the other students.

So one explanation (other than the 2 you propose--brain function and biased exam) is that under-qualified or less-qualified black applicants were given jobs early on in the system and only had a problem when they got to this test.

The other explanation is simply cultural (which could be the same as your "support") which is simply that blacks think its "acting white" or being an "oreo" to do well in school, and that discourages them from studying for tests.

You seem to dissmiss support because they are further along in their carriers, but I'm not sure why just because they'd made it that far in the profession, they necessarily shed this cultural barrier. (Maybe you are thinking of support in terms of parent-child support?)
Brett said…
I guess that I was thinking primarily about parent-child support, but I was also considering the cultural support/barrier that you mention.

While a person certainly does not shed his culture when he enters his 20s or 30s (I'm assuming most of these firefighters fall in that range), the parent-child relationship is likely modified and I would think that someone is less swayed by negative cultural peer pressures than they were in their teens (when much of the standardized testing discrepancy manifests itself).
Anonymous said…
I remember in an education class reading studies that showing that the higher the proportion of boys in the same room taking a test, the lower the scores of the girls on math and science tests. Even the presence of one boy brought down the scores from an all girls testing room. The opposite was true for english language tests. The theory was that the presence of the boys reminded the females that they're in a group that's considered not as good at the testing subject and that was reflected in the results. You get the same results with ethnic minorities and the proportion of white people taking the test. I remember certain people didn't have their scores affected, those who don't care about their grade anyway just tend to do poorly no matter what, and those who are in a class because of some unique special talents that they are aware of. Unfortunately those with the biggest effects were the ones who generally do well in school.

It says something for single sex education. Maybe you should try having males and females take tests in separate rooms and see how the scores are affected? Other than that I think girls and minorities are stuck just waiting for stereotypes to go away until the don't influence them anymore.

-Theresa
Brett said…
Good to have you back Theresa. Thanks for the postcard; I've been to France several times but have never made it out to Mont Saint Michele.

Interesting point on the environmental influences on a test subject. I'm wondering if the effect is concentrated at the actual taking of the test or if it is more due to the overall classroom experience. I have a hard time believing that I could teach a coed class and drastically change my test results by simply separating boys and girls on the day of the test. Single sex education is an interesting option with a number of pros and cons, but I'm guessing that separating students along ethnic lines would not be a popular idea.

Popular posts from this blog

2019 Stats - Solar

Two Speeches

Books (2022 and 2023)