Tuesday, November 29, 2011

Silly And Not-So-Silly Things For Today

None of these are silly in an innocent way unless you take a very, very long view and regard yourself as belonging to the ethnic group of pixies, orcs or elves. And then to the male gender.

1. These pieces of advice to "single girls" of the late 1930s look silly to us now, right? But wait a moment! It's not that different from 2011, after all! And I hear that McMillan's ideas might become a television series, too.

2. Ten things the iPhone Siri will help you get instead of an abortion. Self-explanatory silliness.

3. Andrew Sullivan's lazy post on how "politically correct egalitarianism" about racial differences in IQ scores has strangled the study of intelligence. What is silly about that is Sullivan's assumption that he can scribble down a few lines on a vast, vast topic which comes with a long history of ghoulies and ghosts, having to do with the misuse of such scores and other similar measures in the past, without any further responsibility.

Ta-Nehisis Coates reminds us of those ghoulies and ghosts, and Amanda at Pandagon has a good post on the problems of identifying IQ scores with some nebulous concept of general intelligence. Here is what I wrote on the topic in the autism post comments below:
The general problems:
1. The IQ tests obviously are affected by what used to be called "nurture." You have to be able to read and write and understand the language the test is in (including the language in the sense of the types of words it uses, such as long ones, perhaps). Because of this, the test results are correlated with a) parental income, b) the quality of school a person attends or attended and c) social class, when defined widely. If a particular ethnic group has a lower average income, worse schools and perhaps a different culture at home (such as immigrants might), then -- surprise! --- they are going to score lower, on average, EVEN IF whatever the actual innate part the IQ test measured were the same as that of some other ethnic group.

2. This means that you have to standardize, to compare like with like, if you wish to draw conclusions from IQ tests. That links to a second problem, which is that any attempt to compare test results across groups should control for the way the samples are selected. Some countries, for instance, might use IQ tests mostly to find gifted children, some others might use them to find children who need special schooling because they can't manage ordinary classrooms. You can't compare two countries which differ in this way and expect to find something general about the intelligence of the two populations. One country would test low IQ scores, the other one high IQ scores. This is a relevant concern also when we compare groups of any kind, because UNLESS the people are randomly selected and UNLESS there is no particular reason WHY the test is taken, we are quite likely to get different samples from different populations.

3. One can do better on an IQ test by practicing similar type of problems. This works both against the idea that the test measures something fixed and innate, as Amanda points out, but it also means that those people who get more practice in their everyday lives will do better than those people who do not get that practice. Both home environment and schools matter here, too. This is not quite the same as the first argument, because I'm talking about practicing the specific kinds of things here, rather than general environmental factors.

4. The IQ test was not initially created as a measure of general intelligence, but of a child's ability to benefit from ordinary school classes. That modest goal has been forgotten altogether, when it turned into a possible measure of general intelligence. BUT the test doesn't even pretend to measure aspects which clearly ARE important in the latter concept: Memory and creativlty, for example, are not measured at all. Neither are any talents which don't translate into paper-and-pen answers or keyboard answers.

5. The Flynn effect: The average scores have crept up over time. Because it's very unlikely that evolution could change an innate characteristic that quickly, this also serves to point out the non-innate aspect of the tests.

6. The stereotype threat: This refers to people doing worse on a test if they are told that their group usually does worse on it, as compared to a case where they are not told that. This is important, because EVEN FALSE findings about racial or gender differences in IQ scores CAN then become self-fulfilling prophesies. It is the stereotype threat which makes Sullivan's lazy few lines on the topic a crime in my view. If you are going to write on a topic of this type, you owe your audience some hours of proper research first.