Why do black communities elect white politicians?

8/15/2014 04:26:00 PM

About what it feels like when a member of a minority goes to vote.
Vox tells me that 67 percent of the residents of Ferguson, Missouri are black, yet the police chief, mayor, and 5 out of 6 city council members are white. From the protests and racial tensions, it seems unlikely that this is because the black community thought these were capable leaders that can be trusted to look after their community's interests. So how does this disparity happen in a democracy?

This is not at all unique to Ferguson. Black representation has gotten a bit better in the US House of Representatives but is still awful in the Senate and most state legislatures. In the US House 9 percent of House members are black, compared to 13.6 percent of the population, which is a huge improvement over just a few years ago--the gains mostly due to direct federal intervention into the state redistricting process. The Senate, by contrast, is only 2 percent black. Nor is this disparity limited to the black community: women comprise half the population but hardly 17 percent of US Representatives.

There are lots of reason for this relative disenfranchisement of disadvantaged groups: gerrymandering; voter ID laws that make it harder for poor communities to vote; tampering with precinct locations, absentee ballots, and early voting dates in ways that make it disproportionately hard for these communities to vote; racist felony laws that have the effect of making large segments of minorities permanently ineligible to vote. Moreover, many local governments are not actually democracies, and city councils can effectively appoint their own members. How that works is retiring council members agree to retire mid-term, allowing the city to appoint their favorite candidate to fill out the term and breeze through a sham general election as a virtually unbeatable incumbent.

But even without all that, we'd still most likely have legislatures disproportionately dominated by white men. Democracy is strongly biased towards discrimination. To see why, consider a thought experiment: Consider a district comprised of half men and half women. Suppose women are equal-opportunity voters who are equally likely to vote for a man as for a woman, but the men are all sexist pigs who won't vote for any woman. Thus, in an election between a male and a female candidate, the male candidate will automatically get the 50 percent of voters who are men, plus an additional 25 percent who are women, for a landslide victory of 75-25. That is, democratic elections inherently favor the more discriminatory of two voting blocks, regardless of the actual percentages of votes each voting block commands.

Obviously, that example--where zero men are willing to vote for women--was a bit extreme. To see exactly how powerful this effect is though, I've done some simulations for you. In the US, there are an average of 710,767 voters in each House district. Let's assume that in each district, exactly half of voters are women, and the other half are men. Women are equally likely to vote for a male candidate as a female one, but men are only slightly biased towards men--specifically, we will assume that each man is has a 51 percent likelihood of voting for the male candidate, and 49 percent likelihood of voting for the female candidate. There are 435 House districts, so my simulation consists of 355383 women randomly casting votes according to a Bernoulli distribution with probability of voting male p=0.5, and 355383 men randomly casting votes according to a Bernoulli distribution with probability p=0.51, repeated 435 times to simulate all the House districts. I've then repeated this simulation 1,000 times to bootstrap estimates of the average number of men and women elected to congress in this system. The results are summarized in the table below:
Average Number of Women:0
Average Number of Men:435
Holy cow! Remember, men were barely, barely more likely to vote for men than women, yet in none of the 1000 simulations did a Woman win a single one of the 435 House seats! Not one! If you think I'm lying, try the simulation yourself. [It is, in principle, possible to analytically compute the probability of a woman winning a House seat under these conditions, and I included code for that in my R file, but this turned out to be more computationally intensive than the simulation. The probability is so small that the computer--the computer!--had trouble with the rounding errors. The simulation indicates a probability of less than 0.0000023.]

The lesson here is that discrimination--the bias of straight white men to vote for straight white men most of the time--causes massive inequality in electoral outcomes in a democracy. This may explain why there are hardly any women in congress, or why overwhelmingly black municipalities consistently elect nearly-all-white city councils.

This is why I think Democrats are largely on the right track in focusing on candidates' views on minority rights. In a world where a certain percentage of straight white men will vote only for other straight white men, it is important for the rest of us react by ruling out all candidates who do not care deeply about fixing the racial, gender, and other disparities in our society. If we choose indifference--if we pretend to be impartial--then the racists win. The nature of democracy demands that we be partial.
Harald Korneliussen 8/16/2014 08:07:00 AM
You seem to have gone awfully quickly from a model where straight white men - and only straight white men - are biased towards their own, to assuming this is actually the case, and this is actually the problem. But as I recall, minorities and women are also biased towards white men.

What your model shows is that in a winner takes all system, even a tiny bias turns into big inequality. But this ought to make you question the system, rather than attempting to fix voters - after all, how realistic is it to get unfair biases down to the absurdly small levels where they wouldn't make a difference?

Identity groups may be the most obviously disadvantaged. But there is plenty of unfairness connected to non-identity groups. Poor people (who have less sense of shared identity) and non-lawyers (who have virtually none) are also ridiculously unfavored by the electoral system, but there will never be identitarian justice groups fighting for equality for them. In the unlikely event of success, groups with "class consciousness" would still have power over groups without it.

Ultimately. the way to get representative government is the same way you get a representative sample: by random selection. Originally, that's what democracy meant. Aristotle said that voting was as natural for oligarchy as lottery (sortition) was for democracy.