Wednesday, June 07, 2006

The Gay Marriage Ban

by B. Strong

Much of the flap over the same-sex marriage ban has been over (a) what, precisely, Bush & Co were intending to accomplish, and (b) the results of the ABS news poll vs. the media interpretion of these results.

In my never-ending effort to clutter up Tom's blog with GSS data, I thought I'd see what My Favorite Online Survey of Public Opinion has to say about the gay marriage issue.

The GSS has one question pertaining specifically to gay marriage: "Should homosexuals have the right to marry?" Responses range from "strongly agree" to "strongly disagree".

It's not an ideal question, because it's framed in terms of rights as opposed to prohibitions (i.e., a constitutional ban on gay marriage). But it does have three advantages over the ABC news poll numbers: (1) it's been asked twice in the survey's history, so you can get some sense of change; (2) it's based on face-to-face interviews rather than telephone surveys; and (3) it's easy to look at how attitudes toward gay marriage differ across groups.

Here are the raw percentages, by year (n=about 1200 per year):

Gays have the right to marry

Strongly agree3.411.4
Neither agree nor disagree15.014.6
Strongly disagree46.935.5

In 2004, the majority of GSS respondents (56%) didn't think gays should have the right to marry, whereas just under a third do think they should. Also, far more people were passionately against gay marriage than passionately for it, which presumably supports the "pandering to the rabid base" theory for GW & co's motivations.

Look, though, at the tremendous increase in the percentage of proponents of gay's rights between 1988 and 2004, and the correponding decrease in the percentage of detractors. It's dangerous to impute trends from two points, but still, if I were in charge of thinking of the long-term health of the Republican party, I wouldn't be willing to bet on gay marriage as the rallying issue.

So, who supports gays rights to marry, and who doesn't? I ran a quick little ordered logit, which is a fancified regression for ordinal dependent variables. The results don't show too many surprises:

  1. The more educated the respondent, the higher his or her (net) probability of supporting gays' right to marry

  2. The older the respondent, the lower the probability of supporting gay marriage

  3. As is true with most tolerance items, Jews are the most supportive of gay rights, followed close behind by respondents with no religion. Catholics and "other" religion tie for third and fourth, while Protestants are easily the most likely to oppose gays' right to marry. (Presumably, if I broke this down according to fundamentalism, you'd see the usual pattern, too.)

  4. Unmarried respondents are significantly more likely to oppose gays' rights to marry than either formerly married or married respondents.

  5. Women are far more likely to support gay marriage than men.

The only real curiousity in the results is the lack of a strong net race effect in support for gay marriage. Curious, that is, in light of the blog discussion about how the Republicans are trying to cater to the black vote.

Although there isn't a net race effect in the general sample, there is among people who voted for Gore in the 2000 election, with black Gore-voters less likely to support gay marriage than white Gore-voters. (The 2004 GSS was conducted before the Bush/Kerry election, so I can't look at Kerry-voters.) I have a hard time believing that blacks' average conservatism on the gay marriage issue is strong enough to overcome their liberalism on other social and economic issues -- after all, it certainly wasn't in the 2000 election, when the country was less tolerant of gay marriage.

But, I suppose this is precisely what the Republicans hope to accomplish.
Professor Kim-

OK, so have a general "liberal" trend between the two years. Did you run these models for both time periods? Older people are less likely to support gay marriage and the unmarried are unlikely as well, this trend seems more suprising. The population is aging and (I think) there are fewer people at any given time married that in the past (I'd have to confirm this last point). But, If so, demogrpahic change alone suggests a conservative movement rather than liberal. Are points 1 and 3 overwhleming points 2 and 4 (I assume the proportion of women in the population hasn't changed much, but its been a few years since I played demographer)?

I'll leave it to Kim to clarify how she defined "older," but I'd be surprised if individual aging caused opposition to gay marriage. It's important to remember (cf., e.g., Frontline's AIDS retrospective) that '88 is eons ago in perception-of-homosexuality years, and that only so much of the '88 public has died off since.

I'm not so surprised by the greater likelihood of support for gay marriage among the married. Purely anecdotally, I've seen some unmarried young conservative bloggers retype the party line with a vehemence that only a total lack of experience with marriage could bring. From my married perspective, viewing others in stable LTRs -- or who want to solemnize their LTRs -- as representing something that must be defended against seems absurd. (I also don't see defending bad marriages as defending the institution.)
Though not exactly the s-s marriage issues, attitudes toward homosexuality follow the pattern that Kim identifies: age, gender, education, religion. Also, region and rural vs. urban residence.

The age issue is particularly interesting (one I'm working on now). Though it is a trusim that people don't change too much as they age - most overall change is caused by the older generations dying - this issue is different. In both Canada and the US, people in all age cohorts are changing their minds as time goes by. Perhaps the result of rapidly changing social policy on the issue? On shifts in the culture?
I just used a linear specification of age. I tinkered around with squared and cubic terms, but the estimated coefficients were trivial.

There does seem to be some difference in the effect of age across the two periods, with age having a stronger conservatizing effect in 2004 than in 1988. I can't tell if this is truly an aging effect, or whether it's cohort or period effects.

As for region, respondents in the South are, not surprisingly, far less supportive of gay marriage than anyone else. The difference between the Midwest and the Northeast is pretty trivial, and not statistically significant in these moderately sized samples.

Didn't run urban/rural, because I didn't download the variable. Pathetic excuse, but, hey, it's a blog post.
Why wouldn't a decomposition of the change work? I don't know how to do this with more than one variable, but you could at least tell how much of the change is due to changing age structure versus changing attitudes at any given age. No? This seems wrong to me, but I don't know what I am missing.
Lars suggests a couple of fun exercises. 1. 'Reweight' the 1988 sample to '04 demographics. Prediction: Results close to '88 results. 2. Same, but applying '88 demographics to the '04 sample. Prediction: results close to '04.

Having T=2 limits the ability to fully discern the dynamics. But there are some simple descriptives that might show whether my underlying intuition is true, which is that older people in '04 are more 'liberal' than in '88 on average, but may show signs of having liberalized more slowly than younger groups. Of course, they aren't in the target audience for Will and Grace.
When I said "I can't tell if this is truly an aging effect, or whether it's cohort or period effects," I meant that I couldn't tell from the ordered logit. Yes, one could reweight, do additional descriptives, etc. If any of y'all want to play around with this, I'm happy to share my data set (it's in Stata), or of course it's readily download-able from the GSS site.

Me, I've got to put together a book project for one of Quinn's teacher's retirement, clean the house in advance of visiting relatives, repair water-damaged ceilings in two closets (thanks, leaky pipe, I needed that), finish painting the hall, spread 8 cubic yards of mulch, and review an AJS manuscript. Oh, and entertain a 4-year old.
If you add up the strongly agree, agree and the neither agree nor disagree rows for 2004, gays are doing better than Bush's approval ratings.

Sure, it means nothing, but it still brought a smile to my face.
Janelle Renee, it made me smile too. :]
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?