Sunday, August 23, 2009

Research Worth Reading (3)

There's a myth out there that gay men are self-obsessed and narcissistic, and I for one don't buy it.
David Nimmons, in Chapter 3 of his book Soul Beneath the Skin claims that on the contrary, gay men are actually more likely to get involved in volunteerism and altruistic behaviors, in part basing that on the enormous outpouring of volunteer time and energy devoted to AIDS service organizations.

Well, this is the first quantitative random sample analysis that really tries to get to the issue of whether gay men (and lesbians) are more or less altruistic than straight people. And that goal is what makes it 'research worth reading', in my opinion.
This is one of the few analyses that examines assets, rather than risks; one of a very small number of quantitative analyses to do so. So regardless of the results, regardless of the methodologic challenges, this is an important read in terms of thinking about the health of sexual minority populations.
    Self-reported altruistic and reciprocal behavior among homosexually and heterosexually experienced adults: implications for HIV/AIDS service organizations.

    Susan D. Cochran, Vickie Mays, Heather Corliss, Tom W. Smith, Joseph Turner

    AIDS Care 21(6):675-682. June, 2009

The authors used data already collected as part of the General Social Survey (GSS), a random-dialed telephone survey of the US population conducted annully since the 1960's, a staple dataset for the social sciences.
They looked at four potential measures of altruism across groups of people defined by whether they reported ever having had sex with someone of the same sex, or whether they reported only mixed-sex sexual relationships. The sample size of the GSS is fairly large (they combined 2 years to get 2,031 people), but the number of men and women reporting same sex partners in their lifetimes was pretty small (68 men and 51 women), so they elected not to divide that group farther, but lumped together men who were homosexually active and bisexually active together, and the same for the women.

The GSS asked a 7-item scale intended to measure empathic concern (other-oriented feelings), a 4-item altruistic feelings scale (similar idea, as far as I can tell), a set of 11 altruistic behaviors (things that one does for others that have no self-serving interest, like giving directions to a stranger, even at some risk to one's self, such as donating blood), and 4 reciprocal behaviors (things that one does for another, but do have self-serving aspects as well, like helping someone you know find a job, or lending a friend a considerable amount of money).

They found that gay/bi men (at least in terms of reported sexual behavior) were very similar on all of these measures, in terms of the average. Of course, the average is just the average, it doesn't really say much about the distribution of individuals in the population, meaning there might be a bunch of gay/bi men who are much more altruistic (for example: helped 7 people find jobs, not just one) but that wouldn't be reflected in the average.

The lesbian/bi women were also very similar across all four measures of empathy and altruism.

So, the results are pretty vanilla. Looks like we're about as caring and other-centered as everyone else.

Blood donations
One interesting side note is that they found lots of people in GSS reporting having given blood recently, over 20% of exclusively hetero men, and almost as many of the gay/bi men. That's similar to what I saw in the same dataset and reported on here.
I suspect that there's something wrong with that variable, not because it shows lots of men with same sex sexual behavior giving blood, but because it shows many many more people of all stripes giving blood than actually do.
I don't know what the problem in GSS is about blood donation. Maybe a lot of people are interpreting it to mean gave blood ever in their lives, rather than just in the last year. Maybe a lot of people are interpreting having given a tube of blood for medical tests as having given blood. I don't know.
But at any rate, there are more credible results on blood donation in the National Health and Nutrition Examination Survey, which I reported on here.

Study limitations
The General Social Survey (GSS) is an amazing dataset. They ask tons of questions, and repeat a lot of them year after year. But, it is also very general, and not necessarily designed for whatever analysis you or I might have in mind, so when re-purposing it, it is important to hold in mind that the results may not be what they seem for a variety of reasons.
As is typical for epidemiologists, I'll break them down into three large categories: errors in assessing sexual orientation; errors in assessing altruism; and other factors that may be associated with both sexual orientation and altruism.

errors in assessing sexual orientation
The GSS is a telephone survey, and it is a really long survey. And not everyone pays close attention the whole time they are on the phone. As a result, some people inadvertently give answers that don't actually represent their reality, and in very rare cases, the interviewer records something other than what the respondent said.
All that would be well and good if a little bit of error here and there gets swapped from one group to another, but it gets problematic when one group is much smaller than the other (like sexual minorities). In this sample, 5% of the women and 6.6% of the men reported at least one sexual partner of the same sex in their lifetimes (since age 18). It's possible that maybe 4% of the women and 5% of the men really did have same sex partners, and the other 1-1.5% represent people who were inattentive or miscoded for some reason (this is purely hypothetical, there's no way to know what the error rate really is). So, if that were the case, then gay/bi men and lesbian/bi women would look more like straight people than they should, because a bunch of the people we think are gay/bi/lesbian really aren't (Scout's Law of Fake Queers).
The GSS actually has more opportunity than most datasets to check on this kind of error, because at different points they ask about sexual partners ever in one's lifetime, in the last five years, and in the last year. And there's always people who say they have had no sex in the last five years, but they have in the last year. No way to tell which of those is correct, but they can't both be true, so you can get a sense of the error rate that way. I haven't done that analysis myself, but it could be done.
Another way to check is to see whether the queers identified in this study look like queers identified in other studies. Most demographic studies have found that people who describe themselves as LBG, or who report same sex partners, tend to be slightly younger, more highly educated, and especially less likely to be married. In this study, they were somewhat younger (much younger for the women), but not more highly educated. The LGB people identified in this study were less likely to report being married, but still about 30% reported being married, which is pretty high.
For those reasons, I'm a bit skeptical that the GSS sample has really accurately described the LGB population, I suspect that there's a fair amount of 'slop' from the heteros mixed in with us.

errors in assessing altruism
Another potential source of error is if altruism is not measured accurately. This is an area I'm much less familiar with, in part because I just don't trust scales. The known inaccuracy of the blood donation question gives me some cause to interpret this study cautiously, but that could be a problem mainly with that one question for whatever reason, and not reflect systemically on the other measures in the study.

other factors associated with sexual orientation and altruism
I don't know enough about how altruism is distributed in the population. They separated the population by sex, and that's probably the biggie. There weren't enough people to do stratified analyses across any other variable (like age, educational attainment), but they did do a mathematically smoothed model to try to partially adjust for these factors simultaneously, and found no big difference from the overall results.

Full disclosure: Susan Cochran, the lead author on this study, was also the external reader for my dissertation. She and I have never discussed this paper, though. I don't think seeing her name on the paper made me especially more likely to choose it, I was hooked by the word 'altruism' before I saw that she was associated with it. I do think that knowing that she and Vickie write good research is why I asked her to be my external reader.

No comments:

Post a Comment