why don't minorities vote?

As we learn who is running for office in the primary coming up, there are at least 9 Republicans running for PA's 5th Congressional District.. Nine?! What do you think the winning candidate will need percentage-wise to pull that off? (update: The Centre Daily Times sums up the situation best with an editorial: Who's on First and even the CQPolitics leads off with the situation in PA5).

Here is one more followup from the fall election... but it could apply to any election really. With the biggest news coming out of the national primaries thus far being the surge in voter participation, you have to figure at least some of that is due to above average interest in the African American community. If true, then maybe we can get past the recurrent question that comes up every election cycle about why minorities don't vote. Questions like that have an implied premise that nobody ever seems to question.

Yet, even though the pundits will perpetually try to answer the question, they never really ask whether they need to. Is it true that non-minorities are more likely to vote than minorities? and even if true elsewhere, does that necessarily mean it's true here?

OK. the first question is what to measure? On election day, everyone is concerned with voter turnout: votes cast divided by registered voters. If you are a political operative turnout is the important metric because by election day, you are either registered or you are not. If you are not registered, you are not part of the equation in the short term. The more important metric in the big picture is voter participation: votes cast divided by voting age population.

OK. here is my scatter plot by voting district of % Black among the voting age population compared to voter participation rates (votes cast as percentage of total population age 18 and over):

Voter Participation vs. % Black
City of Pittsburgh Voting Districts, November 2007 General Election

Yes, lots of background to making that which I am sure you dont want to read about here.. but the methodology is similar to what I used here: http://www.briem.com/votingpatterns.pdf. If you have some other question let me know.

I added a linear regression line to that. Does this show lower voter participation for Blacks compared to everyone else? It does actually, but just a little. If that was a steep line down it would show a big disparity between the two, but it really isn't much. It comes out to a few percentage point difference between the two groups. All done? Not even close. Voting districts differ in size, so here is a bubble chart of the same data with the size of each point scaled by the voting age population in the district. What does this tell me? I bet even that minimal correlation (negative correlation that is) between % Black and voter participation fades away if you weight for size of districts. You see that a lot of big mostly White districts have the lowest participation rates. That would pull down that linear regression line on the left and I bet make the disparity in participation rates fade away. Done now? Of course not.

Now we have to remember that younger people don't vote. OK OK before I get hate mail from PC let's rephrase that: young people vote just marginally less than those above the median age. Nonetheless, all election demographics are impacted by the disproportionate rate elderly vote. Now remember that there is this little issue of health disparities that impact the size of elderly cohorts by race. Black mortality rates are higher than for others. That means that on average the White population has proportionately more of its population in that supervoting age range. So even if there was some minimal difference in voter participation in Black and White districts, you can bet that is being caused by the different age patterns across districts. For example, in the City of Pittsburgh in 2000, 27% of the population was Black. For the population age 65 and over in the city of Pittsburgh, it's just over 20% Black. Think that difference does not skew the voter turnout numbers? Given the scale of the health disparities here, I would not be surprised if I could not engineer a positive correlation locally between %Black and voter participation once you control for some of these factors.

So even if you see some statistic that says Blacks are not turning out to vote, it just does not necessarily mean that if you pulled a random Black person of a given age from the community and compared their voting record to a random White person of the same age, you would expect to see much different. Think about that the next time you hear some long discussion of minority voting patterns.

and with all of that.... it will be interesting to compare those recent results with what happens in the Pennsylvania primary if indeed things are still unsettled on the D side at least.