Built-in occupational survey gender bias?

December 31, 2019 6:32 AM

It is often stated - usually for the purpose of saying that we don't need more women in STEM careers - that the most reliable difference between men and women is that "men are more interested in things and women are more interested in people." Cordelia Fine and Virginia Valian have criticized this conclusion based on the limitations of the surveys used to determine it. I'm having trouble accessing the surveys, as they all seem to be products behind paywalls. Can anyone who does have access expand on the accuracy and relevance of their criticisms?

Men and Things, Women and People: A Meta-Analysis of Sex Differences in Interests is the most commonly quoted summary of the results from a number of personality studies. There are some obvious problems with their approach, like calling artistic endeavours (which often produce a thing) a "people" interest, but Fine and Valian say that the problems go deeper than that, down to gender assumptions built into the specific questions:

For example, the three subscales of [Prediger's take on Holland's] inventory that make up the "thing" dimension require "thing" to be interpreted so broadly - including "the global economy, string theory, mental representations, or tennis" - that the term becomes "vacuous." Valian also suggests that preconceptions about which sex does stuff with things have influenced the creation of the items. Why, for instance, don't activities like "Take apart and try to reassemble a dress" or "Try to recreate a dish tasted in a restaurant" appear on such scales?

I can imagine how that would produce skew. (I'd add that to count something like getting together with your buddies to add a turbo to your car and paint on some racing stripes is pretty questionable as a "thing" activity, especially if you're counting painting a picture as a "people" activity.) I don't know anything specific about what's on the surveys, though. Is Fine's criticism a fair criticism of the contents of the surveys used?

Does the criticism she raises apply more strongly to the personality surveys which produce the most extreme gender differences?

From the meta-analysis linked above, the most extreme gender differences on the Things-People scale:

Self-Directed Search (Form R) [SDS-R]: 1.65
Armed Services Vocational Aptitude Battery [ASVAB]: 1.58
Occupational Aptitude Survey and Interest Schedule, 3rd edition [OASIS-3 IS]: 1.20
Harrington-O'Shea Career Decision-Making System, Revised [CDM-R]: 1.17
Kuder General Interest Survey [KGIS (Form E)]: 1.16

Least extreme gender differences on the Things-People scale:

Career Occupational Preference System Interest Inventory, Revised [COPS-R]: 0.14
Vocational Interest Inventory, Revised [VII-R]: 0.36
Career Assessment Inventory, Enhanced Version [CAI-E]: 0.41
Vocational Research Interest Inventory [VRII]: 0.30
O*NET Interest Profiler [O*NET IP]: 0.55

Of course, even if there is a difference in interests it is quite probably the result of socialization into gender roles rather than - as is typically assumed by people who don't want more women in STEM - something biological, and Fine does a fine job of covering that point. I'm specifically interested in the narrower criticism she makes of the survey questions themselves.

posted by clawsoon to Society & Culture