The pink pound

Via Tim comes this report, claiming that gay men earn almost £10,000 a year - 38% - more than straight ones.
I find this implausibly large. I suspect the gap is biased upwards because men who are confident enough to tell a market researcher that they are gay are disproportionately likely to be high earners; the causality between confidence and earnings probably runs both ways.
There is more rigorous academic evidence on gays' pay here. The researchers have found that gays do indeed earn more than straights - just under 10% more.
However, this is because gays are more likely to: work in London; have degrees; and work in large firms. On these counts, they should earn even more than they do - about 10% more, the research estimates.
However, all this "discrimination" is against gay men. Lesbians, they claim, have higher pay than straight women with similar characteristics.
Maybe this suggests Tim's fundamental point is right. The "discrimination" women face is a discrimination against those who have (or  are expected to have) children. 

Comments

[Matthew]

There's two points here I would make.

1. The survey is of the readers of the 'Gay Times', so I don't think your point holds (though obviously this raises other questions about whether their readers are representative)
2. The gap between gay men's and lesbian women's pay is as large as other surveys find between all men and women. As gays and lesbians are less likely (presumably) to have children than non gays and lesbians doesn't this provide some (if not particularly good given the survey's base) evidence of a non-child related pay gap?

[skuds]

I came to the same conclusion (that it was rubbish) but for a different reason.

I figured that the readership of Diva and Gay Times only represented the more affluent section of the gay population and therefore not a valid group to compare with the general population.

A better comparision would have been with that group and the readership of a similarly upmarket publication with a predominantly straight readership - but I couldn't think of one. Or if the poll had found a sample which was more representative it could have made a valid comparision with national averages.

Someone pointed out to me that the Guardian's corrections column today contains a very similar point.