Bryan Caplan  

How Selfish Are Our Views About Abortion?

Reasoning from multiple price ... Uber, taxis and the issue of s...
[Update: Since I wrote this post, Weeden re-did some of his results - including his abortion results - controlling for ideology.  I leave judgment to the reader.]

[Further Update: I didn't notice that the abortion opinion measure I was using coded "not asked" as "no."  Corrected regressions here.]

In The Hidden Agenda of the Political Mind, Weeden and Kurzban argue that individuals' "inclusive interests" have a large effect on their position on abortion.  On the one hand, this seems implausible.  On the other hand, Jason Weeden wrote his dissertation on abortion attitudes, so there must be something to his story.  What's really going on?  Like Weeden and Kurzban, I turn to the General Social Study to find out.  My evidence is only exploratory; I don't pursue every statistical permutation.  However, I did precommit to blog the topic before I looked at the data.

Recall Weeden and Kurzban's novel claim:
People who party and sleep around have an interest in other people not bringing legal or moral costs to bear on them for doing so.  People who want to delay having children while partying and sleeping around have an interest in the availability of family planning, including the backstop of legal abortion.  Their mental Boards of Directors will prefer moral and political policies that help them live the lives they want to live.
The GSS has many measures of abortion opinion, but the best is probably ABORTION, which =1 if the respondent thinks abortion should be legal for any reason, and =2 otherwise.  Given the starkness of the former position, it's not too surprising that the split is 23/77.

The GSS also has many measures of sexual activity.*  Taking both relevance and missing data into account, the best measures are probably:

1. Number of sexual partners in the last year (PARTNERS), binned into 8 categories:  0=no partners, 1=1 partner, 2=2 partners, 3=3 partners, 4=4 partners, 5=5-10 partners, 6=11-20 partners, 7=21-100 partners, 8=100+ partners.

2. Number of sexual partners in the last five years (PARTNRS5), binned into the same 8 categories as PARTNERS.

3. Lifetime number of male sexual partners (NUMMEN) and female sexual partners (NUMWOMEN).  These measures are continuous, but I recode responses into the same 8 bins as the two previous variables.  The recoded variables are NUMMENBIN and NUMWOMENBIN.

So what happens if we regression ABORTION on a constant, all four measures of sexual activity, age, age squared, year, and SEX (=1 if male, =2 if female)?


Weeden and Kurzban are seemingly on to something.  Every step up the ladder of lifetime partners makes respondents about 2 percentage-points more pro-choice.  Strangely, though, number of partners in the past year and past five years make very little difference; their coefficients are statistically insignificant even though there are over 15,000 valid observations.  In terms of self-interest, of course, you'd expect recent sexual behavior to matter more.

But what happens if we control self-reported ideology, which runs from 1 (most liberal) to 7 (most conservative)?  This.

Left-right ideology is far more important than number of sexual partners.  A single step on the 1-7 ideology ladder matters more than three steps on the lifetime sexual partners ladder.  A liberal virgin is more pro-choice than a conservative with over 100 lifetime partners.

What about religion?  Weeden and Kurzban want to count church attendance as a measure of interests.  The GSS variable is ATTEND, which ranges from 0 (never attends religious services) to 8 (attends more than once per week).  But once you're controlling for all four measures of sexual partners, it's hard to see why attendance continues to be a credible measure of sexual risk-taking.  In any case, Weeden and Kurzban neglect a more philosophical GSS measure of religiosity: your views on the Bible.  Response options:
1. The Bible is the actual word of God and is to be taken literally, word for word.

2. The Bible is the inspired word of God but not everything in it should be taken literally, word for word.

3. The Bible is an ancient book of fables, legends, history, and moral precepts recorded by men.
What happens if we add ATTEND and BIBLE to the list of explanatory variables?


Left-right ideology remains the strongest predictor, but both religious measures are comparably important.  The coefficients on sexual activity are miniscule by comparison.  Moving from a moderate Christian to a fundamentalist Christian view of the Bible has a bigger effect than moving from 0 to 100+ lifetime sexual partners.

Since Weeden wrote his dissertation on this topic, he surely knows this data better than I do.  Still, I'll be very surprised if he can add anything to my final regression that makes interests look more important than ideas.  If he posts any such regressions, you'll be the first to know.

P.S. As far as I can tell, GenCon attendees are liberal and irreligious yet have few sexual partners.  Does anyone who's attended seriously expect these gamer nerds to be pro-life?

* Weeden and Kurzban have a bunch of other measures of the "Freewheeling" lifestyle, but it's hard to see how they remain relevant to the interest-abortion nexus after controlling for number of sexual partners.  For example, one of their explanatory variables is how often respondents visit bars.  But after controlling for number of sexual partners, it's hard to see how abortion serves the interests of barflies.

COMMENTS (7 to date)
Emily writes:

I don't think controlling for ideology is getting you where you think it is here. How conservative or liberal you are is just a summary of your political beliefs, so of course it's correlated with individual political beliefs. People call themselves liberal or conservative because of, among other things, their beliefs about abortion. Their beliefs about abortion are not caused by calling themselves liberal or conservative. With religious attendance/beliefs, it's a little less straightforward, but it's still not clear to me whether you're seeing causes, or just mechanisms by which self-interest is working.

Nathan Ashby writes:

I think one mistake you're making is conflating the number of sex partners someone HAS with the number that they WANT. I'm willing to bet that a lot of those pro choice virgins at GenCon would WANT more sexual partners than a representative pro lifer - even if the latter was in reality more sexually active. Especially when someone is young and has most of their sexual experience ahead of them, this distinction seems important.

CC writes:

What Emily said. I'm sure if you added PROCHOICE as one of the explanatory variables, it would show a very high t-stat as well. :)

Lupis42 writes:


That's possible, but it seems just as plausible to me that, with some exceptions, people who self-identify as conservative will default to the opinions of other conservatives whenever there isn't a pressing reason to do otherwise.

For example, I know a few liberals who are pro gun rights, but almost all of them either have a firearms related hobby, or friends and family who are closely involved with firearms in some way. Most people who don't think about the issue much nevertheless seem to take a default position that corresponds with their ideology.

If I were trying to unify the theories, I would suggest that people tend to deviate from self-identified ideology for reasons of self-interest, especially if there's a strong personal or immediate social circle reason to do so.

Ian writes:

Just a technical question. Are these coefficients from ordinary linear regressions or logistic regressions? Logistic regressions would be more appropriate given the binary response variable (ABORTION).

Larks writes:

[Comment removed for supplying false email address. Email the to request restoring your comment privileges. We'd be happy to publish your comment. A valid email address is nevertheless required to post comments on EconLog and EconTalk.--Econlib Ed.]

Brian writes:


So just to be clear, your first two tables show that accounting for ideology has no meaningful impact on the number-of-partners effect. Wouldn't your theory of ideology-plus-noise say that the correlation should be greatly reduced, if not eliminated?

I'd say the data clearly support Weeden's "just so" story.

I'd also add that I think it's obvious, even before looking at any data, that ideology is mostly a correlate and not a cause political views.

Comments for this entry have been closed
Return to top