I was again impressed by the way Levitt was able to find data that would answer his questions. However, most data out there can't be used the way Levitt uses it. Just because the data is related to a question doesn’t mean it can be used to answer it. I was a teacher, and the perennial question is how to evaluate students. For example, how would you prove that an objective test measures insight and understanding? What scores would show mastery of content? Are essays a better instrument? Is it meaningful to compare students to each other? Etc.
Some of those questions are matters of principle. I don’t think comparative grading tells you much, but that’s what teachers do. Claiming that a student’s performance is measured against some expected standard just interposes a layer of obfuscation, which may soothe the teacher’s conscience, and certainly reassures parents. But grades merely quantify two features: the student’s stage of development; and their family’s socio-economic status.
Other questions are worth asking, but answers require data sets that are hard to come by. For example, it’s fairly obvious that the test questions must relate to the insights we wish to measure. It’s not obvious how such questions should be framed. Nor is it obvious how to determine whether the results tell us anything useful about anything else, such as the student’s future performance. What data we have show that test results correlate most strongly with postal codes, which in turn identify neighbourhoods, which in turn correlate with socio-economic status.
As you see, Freakonomics prompts musings and questions. That alone makes it worth reading. It’s well written, entertaining, and mind-stretching. ****
Footnote: Many years ago, I administered a series of vocabulary tests to my classes. I found that my students consistently picked the same "incorrect" answers for some words. The reason? Subtle differences between the regional dialects spoken by my students and the test makers.
No comments:
Post a Comment