Thursday, September 11, 2008

Stereotypes and Generalization

I guess my thinking was partially triggered by a conversation in which a writer told me that a book on dissertations said that a dissertation would take something like 1047 hours of work. I forget the exact number, and I'm not going to mention the title of the book lest my comments be construed as a criticism of the book, which I've not read. But to suggest such an exact number? It seems to me that it might make sense to discuss the rough order of magnitude of the time that goes into writing a dissertation, but not to try to claim such precision. It seems an undue, or inappropriate generalization.

Generalization is a very useful tool. But we want to be careful of it. Stereotypes are a form of generalization. There's nothing wrong with stereotypes, in themselves, but due caution is necessary in their use.

One writer I was talking with suggested that quantitative studies take less time than qualitative studies, and that therefore it would be easier to write a quantitative study. It is in reasoning like this that the use of generalizations goes wrong. Just because one can make a general statement about a group, doesn't mean that statement holds for all members of the group. It's like saying that because men are taller than women (a safe, and easily verifiable generalization), all men are taller than all women, which is, of course, absolute hooey. I meet plenty of women taller than I. I meet more women who are shorter than I, but that doesn't make the opposite true.

One reason to be careful of generalization is that if you use it poorly in academic writing, you open yourself to ridicule. I worked with a writer once who was writing a dissertation that explicitly argued against "essentialism" in reasoning, and against racism, yet the writer consistently made generalizations about race--in direct contradiction to the anti-essentialist positions she claimed to hold. It was a frustrating experience because much of the writing and thinking in the work would have been sound, but for the use of essentialist reasoning and the careless use of generalization--"all black people experience X," she would write, "and all white people experience Y."

But more importantly than looking like a fool for expressing oneself in crude generalizations, one can benefit by thinking things through properly.

I heard the "quantitative studies are easier" from one writer who had previously almost completely planned a good deal of her study. She had planned a qualitative study, and she had worked out many details related to structure of the study and recruitment. Then, on hearing from a friend that quantitative studies are shorter, she abandoned all the work she had previously done and spent several weeks just trying to orient herself to a new project. I don't know how much shorter the quantitative study was compared to the qualitative study, but I do know that a significant amount of past effort was scrapped and due to that decision, a significant amount of new effort was necessary. Now, if we know that a dissertation takes 1047 hours of work ;-), on average, what's the difference in the amount of work necessary to do the quantitative vs. the qualitative? 10% (or 100 hours)? or is it 5%? or 25%? The crude generalization doesn't really help us. And again, we might also think about the dissertation not in hours of work, but in weeks or months elapsed from start to finish. 1000 hours is 50 20-hour weeks. If the writer lost four or more weeks (which she did) switching topics, then her time cost was about 10% of a 50-week project, or 5% of a 100-week project. So for that writer, the generalization would only pay off in the case of her study being more than 5% shorter.

Anyway, this isn't really so much about writing, I suppose, as it is about logic. My basic point is that, while generalization can be useful in many ways, one must always check whether the generalization is apt to the specific case.

To put this in slightly different terms, two forms of logical inference depend on generalization. These two forms are very useful, each in their own way, but they are not logically certain in the same way that logical deduction is. The two logical forms: induction and abduction.

No comments: