Alaska isn't the first place you'd expect to see a woman elected to higher office. With its harsh climate and reliance on traditionally male-dominated industries like fishing, mining, and oil extraction, it has the most male population in the country: 106 men for every 100 women. Things are a bit worse still for the guys on Alaska's single scene -- the ratio of unmarried men (15 years or older) to unmarried women is 114:100. Throughout the rest of the United States, the men have it a bit easier, as the ratio is 86:100 nationally.
And yet, Alaska is one of just five states to have elected a female governor -- the irrepressible Sarah Palin. One of its two Senators, Lisa Murkowski, is also a woman.
But Alaska is a quirky state, and presumably this is highly irregular behavior. Except that -- it really isn't. Although women are still having a relatively tough time getting elected in general -- they represent just 17 percent of the members of the U.S. Congress -- Congresswomen, as opposed to Congressmen, are more plentiful in areas where the male-to-female ratio is higher.
I have a database containing the names of 535 members of the Congress -- 435 Representatives plus 100 Senators -- as well as a bunch of demographic information on their states and districts. Of these 535 geographies, 91, or about 17 percent, elected a woman at their last opportunity. (A couple of methodological notes: state-level observations are deliberately double-counted, to represent the two senators that each state has. I also look at the identity of the person who was last elected in each geography, so if someone has since resigned their seat in the Congress, I'm still counting as holding their seat unless there has already been a special election held to name their replacement.)
The chart below lists the 25 most male "districts" in the country (from here forward, I will refer to state-level observations on the Senate side as "districts" in addition to actual congressional districts from the House side), along with the 25 most female districts. The female office-holders are highlighted in red.
Nine of the 25 most male-dominated districts (36%) most recently elected a woman to office, as compared with 4 of the 25 most female-dominated districts (16%). This alone is somewhat interesting -- however, it actually conceals the strength of the relationship because female-dominated districts are more likely to vote Democratic, and Democratic-leaning districts are more likely to elect women to office regardless of their sex ratios. Let's look, for instance, at what's happening only in strongly Democratic districts, which I define as those with a PVI of D+10 or higher:
The most male-dominated from among these strongly Democratic districts elected women in 10 out of 15 instances. The 15 most female districts elected just 3 women.
Next, moving to moderately Democratic districts with a PVI of between D+3 and D+9.
A less impressive result here: 3 of the 15 most male districts elected women, as compared with 4 of the top 17 most female districts. (Note that we include 17 districts rather than 15 on the gals' side because there's a three-way tie between Delaware's two senate seats and its one U.S. House seat in 15th place).
Next up are the swing districts:
Just two of the 15 most male swing districts -- IL-8 and NV-3 -- elected women. But women were shut out in the 15 most female-dominated swing districts.
Moving on to lean Republican districts:
Once again, it's the more male districts that are more inclined to elect women to office.
Lastly, the strongly Republican districts:
Same story here. Among the 15 most female districts, only OH-2, which reelected the inimitable Jean Schmidt, sent a woman to Congress. All told, after controlling for the district's partisan affiliation, male-dominated districts were more than twice as likely to elect a Congresswoman as were female-dominated districts.
We can generalize this result by means of a logistic regression analysis. This is the estimated probability of electing a woman to Congress based on the sex ratio in three types of districts: a strongly Democratic one, a strongly Republican one, and a neutral one.
Note that, although the pattern manifests itself regardless of the partisan affiliation of the district, it is strongest in Democratic-leaning districts and weaker in Republican ones.
What I don't have for you is a ready explanation for this seeming paradox. What's especially perplexing is that although the sex ratios differ some from district to district, they don't differ all that much. In the most male district in the country, CA-20, about 54 of every 100 people you'd encounter on the street will be a man, as compared with 45 of 100 in the most female district, Philadelphia's PA-2. Unless perhaps you were specifically looking for a marriage partner, you might not notice those discrepancies. For that matter, I recently moved from the fifth-most male district in the country, IL-4, to the second-most female, NY-11. I couldn't in good conscience tell you I had any idea about that until I looked up these statistics.
Nevertheless, I cross-checked at least a dozen different demographic variables in addition to the sex ratios, such as race, income and educational status; none of them were statistically significant once the sex ratio of the district was accounted for. There is also no doubt as to the statistical significance of the effect; we are more than 99.9 percent certain that it isn't the result of chance alone.
It's possible, and maybe even somewhat likely, that there is some sort of latent variable affecting both the sex ratios and elections to the Congress that I haven't accounted for. If this really is being driven by the sex ratios, however, and it's being driven in this extremely counterintuitive way, it's one of the more fascinating things that I've come across. Perhaps in male-dominated areas, women are more likely to violate traditional sex roles including something like running for political office, which has traditionally been a male-dominated occupation -- the Sarah Palin frontierswoman caricature works well here. It would be interesting to know whether there more women actually running for office in male-dominated areas, or rather, whether they are winning more often when they do run. Or perhaps this is a phenomenon that goes beyond politics, and career growth is retarded for the dominant gender when there is an insufficient number of the opposite one. Or perhaps there is even something more Freudian: a lack of female companionship (or vice versa) triggers a yearning for it that is manifested in the way we vote.