More Faith-Based Science

Sometimes you read things that you swear were published in The Onion, and it turns out to be a Christianist scandal sheet:

Soy is feminizing, and commonly leads to a decrease in the size of the penis, sexual confusion and homosexuality. That's why most of the medical (not socio-spiritual) blame for today's rise in homosexuality must fall upon the rise in soy formula and other soy products.

First off, there's no evidence that there is a "rise in homosexuality." We're just tired of hiding, that's all. However, there seem to be two mutually exclusive things going on here -- note that, in addition to leading to homosexuality, soy products also reduce penis size. However, there is this study (and we're talking real science here, not the faith-based stuff):

Abstract: The relation between sexual orientation and penile dimensions in a large sample of men was studied. Subjects were 5122 men interviewed by the Kinsey Institute for Research in Sex, Gender, and Reproduction from 1938 to 1963. They were dichotomously classified as either homosexual (n = 935) orheterosexual (n = 4187). Penile dimensions were assessed using five measures of penile length and circumference from Kinsey's original protocol. On all five measures, homosexual men reported larger penises than did heterosexual men. Explanations for these differences are discussed, including the possibility that these findings provide additional evidence that variations in prenatal hormonal levels (or other biological mechanisms affecting reproductive structures) affect sexual orientation development.

And before you ask, I'm not telling.

(OK, strictly speaking, penis size and incidence of homosexuality are not particularly related in the quote from WingNutDaily, since "decrease" is relative. [For all you size queens, think what it would be like without the vanilla soy latte every morning.] But I couldn't resist)

Hat tip to Andrew Sullivan, one of whose readers comes up with this observation:

Of course I am pleased by this result but the statistical part of my brain is more skeptical. One concern is that the sample sizes for homosexuals and heterosexuals are quite different. If the distributions are positively skewed (fewer observations on the positive end of the distribution, sigh) then it's possible that the higher mean in the smaller sample is an artifact of the distribution. It would be nice to know if this result holds when equating for sample size, or when calculating medians. I didn't see any note of this in the manuscript. Can your readers help?

Update:

Several readers at Andrew Sullivan say not to worry -- the likelihood of the results being skewed on that study is miniscule.