Queer @ Work

[differenceblog]

Gay men and women have many challenges in the workplace. Some are similar; some are different. Chung and Harmon (1994) suggest that gay men are more likely to be interested in traditionally feminine careers than straight men, although they found that masculinity/femininity was not a predictor (as determined by BSRI). Adams et al (2005) did not

Comments

[beowabbit]

Adams interviewed eight people; Nam Cam Trau and Härtel interviewed five.
Wow, I should seriously go into this field! You don’t have to have actual data to publish! I guess you get to footnote stuff “¹ These three guys we know, personal correspondence, 2002.” :-)

[differenceblog]

Oh, it's data -- it's "qualitative" data. You do "content analysis" on "interviews".

The research has to start somewhere. When I put together my research proposal, I can cite these guys in my literature review, and then I don't look as much like I'm pulling stuff out of thin air.

[astrogeek01]

Sometimes, if you've got three photons, you can manage to say something useful in astronomy.

Small numbers statistics does not mean no statistics... ;)

[randomguy3]

That's not really the same. In a deterministic system, you (in theory) only need a sample of one. People are not deterministic, though, except as part of a large population.

Statistics tell you something about the population, not about an individual. Sure, you can get a probability of something being true of a particular person, but that assumes that you fix a certain number of attributes and randomly select a person from everyone having those attributes.

There are things called confidence intervals. Usually you want a 95% confidence interval. If a statistic is true with a 95% confidence interval, that basically says that in 95% of the populations the sample could have been take from (remember - you know nothing about the population you haven't sampled), the statistic will be true. [Caveat: this may be slightly incorrect if I am misremembering my A-level stats]. You need a large sample (both in real terms and in comparison to your population) to get that

[differenceblog]

A big difficulty in studying LGBT populations in large numbers is how to sample them. One option is to go to LGBT organizations and recruit for your studies there. (This was how both these studies found their volunteers). If you try to do it through more random (and hopefully more representative) populations, you run into a problem with how few people actually identify as LGBT. For an example see Ellis et al (2005):Based on a sample of nearly 8,000 college students, this study was undertaken to determine the prevalence of non-heterosexuality using four different measures of sexual orientation: a self-identity measure, an attraction measure, a fantasy measure, and an experiential measure. Over 97% of both males and females labeled themselves as heterosexual, with the proportion of homosexuals and bisexuals combined constituting nearly 3% of the male sample and about 2% of the female sample. Roughly 80-85% of both sexes would be classified as exclusively heterosexual, with about 10% of both sexes having at least half of their sexual