Torturing the Data

What they do... I can't help it. I've been listening to the recordings from the CRC's meeting, transcribing interesting sections. And here's the erudite Doctor Ruth, talking about whether there's a gay gene or not (as if that mattered a bit):

If you have an identical twin, and one twin is gay, if you're just born that way, the other twin ought to be gay, right? So even [unintelligible] gay papers, it's still fifty percent. But if you go to a twin registry, in Australia, it's eighty percent, so if one twin is gay, the other eighty percent are heterosexual.

OK, this is really strange. If she said what I think she said, she's saying that when a twin is gay in this Australian registry, eighty percent of the time the other twin is straight. Is that what you heard?

I have to assume this is limited to identical twins.

Now, Peter Sprigg was sitting at the table, and I have heard him say that two percent of the population is gay. So we'll say that's what "they" believe.

So listen, if there is no gay gene, then two percent of twins of a twin who is gay should be gay, too. Not twenty percent, two percent.

By Ruth Jacobs' own numbers, having a gay twin makes you ten times more likely to be gay yourself. That is a huge effect, statistically. If she's right.

This lady has a doctorate, she should know better. There may be many factors influencing sexual orientation reported in a dataset, including measurement biases, genetic interactions ... science would be simple if you rejected every hypothesis that had less than a hundred percent support.

As I was writing this, I decided to find the paper she's taking about. Surprisingly, it's online. I say surprisingly, because the APA doesn't usually let you put their published papers on the Internet, but these guys did. If you're interested, you can look HERE for the pdf.

I'm not going to go through the paper ... well, I will quote the line relevant to what she's talking about: "Men's same-sex concordance rate for MZ twins was significantly greater than that for same-sex DZ twins by a directional test, both for the strict criterion (Fisher's exact = .04) and the lenient criterion ( < .001)." [MZ=monozygotic or identical twins; DZ=dizygotic or fraternal twins] (Fisher's exact test is similar to a chi-square test for 2x2 contingency tables, but it gives only a p-value, no statistic to interpret.)

In other words, Dr. Jacobs' description of the data last night was the exact opposite of what the researchers themselves concluded: with identical twins, it was significantly more likely that the second twin was gay if the first one was, compared to fraternal twins. (I am leaving out the whole discussion about this particular researcher, whose work is considered "controversial" from any perspective. This is just a comment on how the data were reported to the assembled group last night.)

The CRC wants to argue against "innateness" by arguing against genetics, as if they were the same. Do they think they can convince reasonable people of their point of view, when they do this to the facts?