Comments on: Gender Humbug

By: eszter

eszter — Thu, 01 Jan 1970 00:00:00 +0000

Hey, great post, thanks! I’ve posted a follow-up (no trackback option for me) here:
http://www.esztersblog.com/archives/00000296.html

By: Invisible Adjunct

Invisible Adjunct — Thu, 01 Jan 1970 00:00:00 +0000

Great post! I love the bedsheets-threadcount question. To which I might add the following:

“When I bake a cake, I am careful to measure each ingredient accurately and precisely.”

By: Elusis

Elusis — Thu, 01 Jan 1970 00:00:00 +0000

THANK YOU. I took the tests yesterday out of curiosity and was overwhelmingly irritated with the questions myself.

Someone on Theresa Nielsen Hayden’s blog (from where I got the links) also asked a good question: cannot emotions be understood as a kind of system themselves? Why are they considered anti-systematic?

By: Kevin Drum

Kevin Drum — Thu, 01 Jan 1970 00:00:00 +0000

You PoMo blank slater, you.

OK, but a serious question. I’ve always been a little bugged by the emphasis on between group vs. within group differences. Even if there’s lots of within group variation, and I think there always would be in any study of human behavior, isn’t between group variation still of interest? Within group variation doesn’t invalidate between group variation, does it?

(Assuming, of course, that all the statistics are properly drawn etc. etc. I’m just asking as a hypothetical, nothing to do with the acutal quiz at hand.)

By: ArchPundit

ArchPundit — Thu, 01 Jan 1970 00:00:00 +0000

Good post—essentially much of this type of pop research might as well be based on second-rate comedians’ jokes about pulling over and asking for directions.

As for Kevin’s question—all data is good data and should be exploited. You want to use both and hopefully exploit it to answer what might commonly driving in-group and between group variation. In simple terms, are demographics of members of the different groups with similar scores similar or different? So a man who stays home with kids might consistently be lower on some scale as are women in similar situations. Quick example so don’t overanalyze it.

By: QrazyQat

QrazyQat — Thu, 01 Jan 1970 00:00:00 +0000

Suppose they asked questions like these instead:

A. When I buy a jock strap, I care deeply that the color be a beige, deep blue or black rather than pink, but care little about the details of its manufacture.

B. When I buy makeup, I study the ingredients, paying particular attention to details about its chemical composition.

By: Barry

Barry — Thu, 01 Jan 1970 00:00:00 +0000

Kevin, the obvious answer is that a large between/within group variation ratio would indicate that the group membership tag carries more information, compared to working with the groups pooled into a single group.

1) For example, let’s consider marketing/market research. If the male/female between-group differences were very large, compared to the within gender differences, then knowing if somebody was male or female would be a more useful marketing datum, than if the opposite were true.

2) If the the between-group differences are small, compared with the within-group variation, then the quality of the survey will be important – measurement error and bias could easily dominate.
As Kiernan, Invisible Adjunct and QrazyQat have pointed out, changing the questions could easily lead to flipping the results around.

3) If the the between-group differences are small, compared with the within-group variation, then the effects of environment will be important, and the data will reflect socialization, not inherent differences. To the extent that somebody is treated differently from birth depending on their genitals, true inherited genetic effects are swamped by socialization effects (i.e., measured heritibility is lowered).

In short, the conclusions resulting from bad science should be taken with a grain of salt.

By: Tom Scudder

Tom Scudder — Thu, 01 Jan 1970 00:00:00 +0000

To be fair, they do have the question,
“When I look at a painting, I’m interested in what technique was used” in there.

By: derrida derider

derrida derider — Thu, 01 Jan 1970 00:00:00 +0000

Barry, in an otherwise very good answer I think your third point has a non-sequitur. Surely what you mean is that effects which vary within the group will be more important than effects that vary between the groups. But it says nothing about whether such effects are environmental or hereditary.
I’ve never understood the allergy of social science people to arguments from heredity. Surely the real point to make is the one you made earlier – that group membership (whether membership is hereditarily or environmentally determined) tells us little where within-group variance is large relative to between-group variance.

By: gianna

gianna — Thu, 01 Jan 1970 00:00:00 +0000

Elusis asks “cannot emotions be understood as a kind of system themselves?” – yes: emotions are part of the limbic system.

i for one loathe psychometrics. :) the tests are easy to fake, people often understand the questions differently, there’s hardly ever any situational context…and so on.