Although my job title is not “statistician”, I do work with statistics fairly often. I’m a professional geographer and work with both social data (census data, demographics, etc.) and physical data (water samples, soil samples, etc.). Much though people think “all we do” is make maps and have fun with Google Earth, it is much more complex than that. We regularly perform “spatial analysis”: exploring data over space and identifying how different elements correlate, and trying to determine why. So when I saw a the description of the following article, I was intrigued.
First, before I even start, I want to identify one thing. There is an age-old adage in statistics: correlation does not imply causation (mathbits.com). This is often forgotten in simple analyses. Just because it is observed that two measurements are correlated (when one goes up, the other goes up proportionally; when one goes down, the other goes down proportionally), one does not necessarily cause the other. There might be a common link to both that remains undetected, or there may be other activities that have not been measured affecting both. Since this post has to do with suicide of LGBT youth, that is a good example.
It has been known for a long time that suicide rates among LGBT people are higher than average. Some make the “causation” error; they believe that homosexuality is a depressing “lifestyle” (and I suppose that for a heterosexual, it is) and that we therefore commit suicide more often. Similarly, others have said that “LGBT students commit suicide because they know they are sinful”. A quote from the linked video of pastor Damon Thompson’s preaching: “the boy already hated himself because he was dealing with the demonic force of homosexuality… being gay is not normal.” No, Mr. Thompson: it is only in a statistical sense that being gay is not “normal”… as in “not average”. Otherwise, for the gay person: it can comfortable, and right, and completely healthy. It is not “demonic”. However, this does bring up the affect of environmental influences: trusted social links like parents and pastors and teachers and peers. They can choose to treat homosexuality as acceptable (or at least neutrally): or, they treat it as though it is “sinful”; “unacceptable”; “bad”. As with pastor Thompson. In that case, those trusted social links are directly pushing youth to suicide. Youth do not commit suicide “because they know they are sinful.” (By that argument, and standard Christian theology, we are all sinful and therefore everyone in a church should all be jumping off cliffs.) They commit suicide because they’ve been told they are especially sinful, and by someone they trust. The environment (the effect of anti-gay teaching) is the element that remains unmeasured and affects the correlation.
So anyway, that’s the wrong way to look at statistics. And for a long time, in this case, I thought it was going to be impossible to try to measure that environmental variable. But it’s been tried and it works.
According to a study from last year, LGBT youth in progressive areas are “about 25 percent less likely to attempt suicide than their counterparts in politically conservative areas that lack school programs supporting gay rights.” (US News, 2011) “Researchers, led by Mark Hatzenbuehler, studied surveys from 34 counties of Oregon. The state was chosen because it is the only one that records sexual orientation and suicide attempts in detail.” (CBS News, 2011) There were a number of factors combined to determine a rating for how supportive an area was for LGBT youth: “The researchers… scored 34 of Oregon’s 36 counties on how supportive of gays and lesbians the environment was based on the proportion of same-sex couples in the community; the proportion of registered Democrats in the community; whether schools had gay-straight alliances; and whether schools had anti-bullying and antidiscrimination policies specifically protecting lesbian, gay and bisexual students.” (LA Times, 2011)
I find it deeply distressing that the researchers would choose to include party affiliation as an indicator of “LGBT supportiveness”; but even worse that it worked. If there had been no correlation, the affect of party affiliation would have been random (and I wish it had been). I know some very compassionate Republicans. Apparently, however, they are sufficiently in the minority in this regard that they don’t affect the overall outcome.
It should be noted that this effect is not limited to LGBT youth, although the effect is not as noticeable in general. “What’s more, the rate of suicide attempts among straight teens in conservative communities was also higher — by 9% — than in areas that were more politically and socially liberal. The finding suggests that widespread acceptance and support contribute to the well-being of all community members, not just those who identify with minority groups.” (Time, 2011) Apparently the concept of “acceptance” is a generalized commodity that affects everyone… as Marin Luther King theorized many times, fifty years ago. It’s good to see it as an evident byproduct of the data.
I note, however, that there are some who are critical of these so-called “obviously skewed” results. (Doctor Bulldog and Ronin, 2011… not exactly a peer reviewed piece, but the best I could find.) His reasoning? “Liberal researchers… promoting the gay agenda (what did I say about that agenda?)… no such thing as an [sic] accuracy in the methodology.” Apparently Mr. Bulldog’s skills lie neither in the areas of objectivity nor grammatical constructs. (Nor does his link to an original article work very well.)
What is interesting is that every single one of these critical posts made reference to a 2006 paper written in the Netherlands which also found higher suicide rates among LBGT people. The assumption is fascinating: “the Netherlands, which has one of the most tolerant and permissive societies on planet Earth…” Well… no. The culture in the Netherlands is more “tolerant” than the U.S., perhaps, but that’s not difficult. Canada has many of the same perspectives as the Netherlands regarding LGBT people, yet discrimination is far from eliminated. Even in the quoted paper, from the Archives of Sexual Behaviour and which has a strong anti-gay perspective from the outset, admits in the synopsis, “Among homosexual men, perceived discrimination was associated with suicidality.” (de Graaf et al, 2005, 253) So apparently the Netherlands is not quite as “tolerant” as it was assumed.
So apparently there is some slight link between party affiliation and the rate of gay teen suicides (and to a lesser extent, the rate of straight suicides). Any comments?