Skip to content

Having daughters makes you more liberal. No, it makes you more conservative. No, it . . . ??

Tyler Cowen reports the following claim from sociologists Dalton Conley and Emily Rauscher:

Using nationally-representative data from the [1994] General Social Survey, we [Conley and Rauscher] find that female offspring induce more conservative political identification. We hypothesize that this results from the change in reproductive fitness strategy that daughters may evince.

This surprised me, because less than a year ago, we reported here on a study by economists Andrew Oswald and Nattavudh Powdthavee with the exact opposite finding:

We [Oswald and Powdthavee] document evidence that having daughters leads people to be more sympathetic to left-wing parties. Giving birth to sons, by contrast, seems to make people more likely to vote for a right-wing party. Our data, which are primarily from Great Britain, are longitudinal. We also report corroborative results for a German panel.

Understanding the results (possibly) in terms of “family values”

This is a fun problem: we have two different studies, both by reputable researchers, with opposite results! I took a look at both papers and can’t immediately see a resolution, but I will offer some speculations, followed by some scattered comments.

We already know that the Republicans are the party of families much more than Democrats are. Just for example, John McCain did 20 percentage points better among married than unmarried voters.

The way I’d interpret Conley and Rauscher’s finding is that having a child is likely to make you much more focused on family issues, and that having a daughter (as compared to a son) might make this effect even larger. To the extent that the Republicans were seen (as of 1994) as the family-values party, this could swing some votes (or, at least, some party identification). Commenter DN below also suggests a connection to views about crime. It should be possible to look into this by studying GSS responses to some different survey questions.

At least in the U.S. context, it makes much more sense to to me to see this sons/daughters thing as being a spinoff from the huge differences between married and unmarried voters than in terms of evolutionary strategies (more on this below).

How large are the effects?

– Oswald and Powdthavee estimate that having a daughter (as compared to a son) makes you 2 percentage points more likely to vote for a party on the left.
– Conley and Rauscher estimate that parents with daughters are 10% more likely than parents with sons to identify as Republican.

I complained earlier that Oswald and Powdthavee did some cool analyses but never got around to showing their raw data. Similarly, Conley and Rauscher don’t get around to showing the simple percentages. They have some data in their Figure 1, but (a) I can’t quite figure out what they’re actually plotting, and (b) that set of graphs omits one-child families.

So I went into GSS and made a quick crosstab myself, looking at kdsex1 (sex of first child) and partyid (political party affiliation). Here’s what turned up:

– Among the 582 respondents who answered the question and whose first child was a boy: 51% Democrat, 13% Independent/Other, 36% Republican.
– Among the 525 respondents who answered the question and whose first child was a girl: 44% Democrat, 14% Independent/Other, 42% Republican.

(I’ve followed the standard practice in political science of lumping the leaners with the partisans.)

I also looked at polviews (think of self as liberal or conservative):

– Among the 568 respondents who answered the question and whose first child was a boy: 26% liberal, 36% moderate, 38% conservative.
– Among the 512 respondents who answered the question and whose first child was a girl: 22% liberal, 37% moderate, 42% conservative.

Just for laffs, here’s kidsex2 (sex of second child):

– Among the 435 respondents who answered the question and whose second child was a boy: 51% Democrat, 13% Independent/Other, 36% Republican.
– Among the 429 respondents who answered the question and whose second child was a girl: 41% Democrat, 15% Independent/Other, 44% Republican.

Thufferin’ Thuccotash! That’s just amazing. I’d say this puts Conley and Rauscher in contention for the (hypothetical) award for finding the largest effect “in plain sight” that nobody has noticed before. (See here for a discussion of the previous contender.)

Other comments

1. The basic analyses makes sense: Oswald and Powdthavee characterize “the gender of a child arriving in the household as a kind of exogenous event.” As Conley and Rauscher put it, “In a society where antenatal sex-selective abortion is rare, the sex of a particular biological child is a random variable,” and so it’s basically ok to consider correlations as causally induced.

2. Despite their repeated use of the term “conservative,” Conley and Rauscher actually study party identification, not political ideology. My quick tables above suggest that their results will still hold when looking at respondents’ self-placement on a liberal/conservative scale, but really they should check this before too quickly using the term “conservative” when they are really saying “Republican.” (See here for some exploration of the differences.)

3. The next step is to look at lots of outcome variables: not just political party and ideology, but also attitudes on individual issues such as abortion (see below) and composite indexes on social and economic attitudes.

4. Oswald and Powdthavee analyze panel data (people who are surveyed more than once) and find, “when compared to the year before the birth, men and women alter their political opinions.” Conley and Rauscher analyze cross-sectional data, which makes their findings harder to interpret; still, they do see clear differences between parents of sons and parents of daughters.

5. How did the two studies end up with opposite findings. My guess is that it’s differences between different countries. Conley and Rauscher suggest that it’s data issues (in particular, the inclusion or exclusion of adopted children or children who are no longer living at home), but both of the articles here consider various alternative specifications and the results don’t change, so I doubt this is what’s going on.

6. As an aside, I would very strongly doubt any claims that people with different political attitudes differ much in their probabilities of having boys or girls. To their credit, neither of the papers being discussed here make this argument.

Economics and politics

Perhaps unsurprisingly, Oswald and Powdthavee, being economists, focus on economic issues:

The interesting recent work of Campbell (2004) documents systematic gender differences in modern British political attitudes. The author tabulates answers given in the British Election Survey of 2001. She shows that the single most important concern to males is that of low taxes. For females, by contrast, it is the quality of the National Health Service.

Similarly, left-wing parties tend to be more supportive of women-friendly policies such as family leave; and right-wing parties tend to favor increased spending on the military, which is of course mostly staffed by men.

Conley and Rauscher, being sociologists, center their story on social issues>

Males’ optimal reproductive strategy is to sire many offspring with a range of mates and push the parenting requirements onto the mothers. Meanwhile, the mother seeks to maximize not only the genetic fitness of the sire, but also to induce more post-conception investment in rearing the offspring from the father. Seen in this light, more conservative policies that increase the cost of promiscuity–particularly for males–will enhance the reproductive bargaining power of women. . . . The conservative emphasis on family, traditional values and gender roles, and prolife/anti-abortion sentiments all stress investment in children – for both men and women. Conservative policies mirror the genetic interests of women, writ large. They attempt to promote paternal investment in offspring.

To me, this seems to be a lot to hang on a fragile thread. When it comes to economic policy, Democrats are pushing higher taxes and public services while Republicans want lower taxes and public services–and I don’t see how these map at all on to Conley and Rauscher’s categories. Even when you come to the particular issue of abortion, I don’t see how how an anti-abortion policy is an “attempt to promote paternal investment in offspring.” There are some deadbeat-dad laws out there, but I didn’t think of them as being associated more with one party than the other.

On the other hand, Conley and Rauscher seem to have found a real pattern, so I don’t want to dismiss their explanations too quickly. But I see their discussions of “reproductive bargaining power” and the like to be a bit of a distraction. To me it’s simpler and clearer to think of liberalism and conservatism, or Democratism and Republicanism, to be linked to general attitudes about the family.

P.S. See here for more.


  1. EmilyKennedy says:

    That's an interesting finding, and I remember your post about the Oswald and Powdthavee study from last year. I agree that the two different findings are probably due to the studies using data from different countries.

    I'm kind of surprised that as sociologists Conley and Rauscher start off their explanation with biological reasoning. That doesn't get at the sociology of gender much at all.

  2. John says:

    Although the cross-tabs from the GSS support their case, they use the proportion of females in a family as their predictor—but is this really as good as randomly assigned?

    Here is an example that I think shows that within-family proportion is not exogenous.

    Assume that different groups have different "stopping rules."

    Republican rule: Have a child. If girl, have another, otherwise stop at 1. Stop after 2 children.

    Democratic rule: Have one child. With probability 1/2, have another child. Stop after 2 children.

    Republican Child Distribution:
    (Outcome, Proportion Boys, Probability of Outcome, Expected Proportion)
    (GB, 1/2,1/4,1/8)
    (GG, 0, 1/4, 0)
    For republicans, the expected proportion of males in a family is 5/8. For democrats, the expected proportion is 1/2. Note that controlling for family size doesn't change anything – both groups have the same family size.

    Is this right?

  3. Andrew Gelman says:

    John: Because of this sort of concern, they also run their analysis not controlling for family size.

  4. Michael says:

    A small quibble on point #6, worth mention only because I use your blog in class: it is never to an author's credit that he fails to discuss a reason to doubt the exogeneity of an explanatory variable.

  5. DN says:

    Seems to me that this is more likely a "law and order" effect in the US. Families with daughters are more likely to be concerned about 'protecting' their daughters against violent crime. These impacts are either mitigated in Germany and the UK (because of perceptions of the parties) or swamped by other peceived advantages of left-wing parties.

  6. Sarang says:

    I find DN's interpretation pretty appealing. And I suspect the salience and esp. the political dimension of violent crime is probably a lot lower in Europe. Incidentally I wonder if this correlation can be used in a twin-studies-ish way to separate out selection effects (e.g., conservatives are likelier to get married and have kids) from effects due to people's views actually changing because they had kids.

  7. Derrell Durrett says:

    The married/unmarried dichotomy (in the linked Gallup poll data for 2008) doesn't appear to distinguish between married w/children v. married/childless. Is there similar data that you know of for recent US elections that explores this?

    FWIW, party id in the US seems far more complicated than either of these studies seem prepared to explore. Votes in specific elections are perhaps less complicated because of the perceptions of the individual candidate (correctly perceived or not).

  8. William Ockham says:

    This is an interesting idea, but I would quibble with one of your side points:

    We already know that the Republicans are the party of families much more than Democrats are. Just for example, John McCain did 20 percentage points better among married than unmarried voters.

    If you adjust for age (obviously marriage rates vary significantly by age; only 9% of 18-24 year-olds are married while 68% of 40-49 year-olds are), then McCain only does 4 points better among the married than the unmarried. The Republicans are much more the party of older people than they are the party of families.

  9. Can the data distinguish social and economic liberalism/conservativism?

  10. Andrew Gelman says:

    Yup. The GSS has a lot of questions.

  11. William Ockham says:

    Oops, I made a stupid math error. Age only accounts for less than half of the difference (about 8 of 20% difference). So while the difference is less extreme, McCain is doing better among married people than would be expected.

  12. Jim says:

    Maybe families with a daughter (1st, 2nd, or otherwise) are more likely to be bigger families, and maybe family size is endogenously correlated with unobservable individual characteristics. For example, if my preference is to have a small family, but I also have a weak preference for a son, I might be content stopping after one child, given that the first one is a son. Treating the event of a girl baby as purely exogenous might be a slight stretch.

  13. The male/female wage premium has declined faster during Republican administrations than during Democratic ones. Maybe the parents of daughters just have their relative welfare at heart.

  14. Caravelle says:

    Wasn't there a study of US congresspeople that showed that having daughters led to a more feminist voting record ? Given the Republican party is more and more extreme (hence presumably less appealing to feminists) these days how does that square with this study ?

  15. J. Cross says:

    Just thinking out loud, parents of daughters are somewhat more likely to be divorced and daughters with divorced parents are more likely to live with their mom than sons are. Fathers also spend more time with sons (I'm taking this from Shelly Lundberg 2005). Not sure any of these factors are nearly big enough to explain such a large party affiliation differences or even how they would relate.