Tuesday, May 28, 2013

more on turnout

Just a parse to address some lingering questions on why turnout was so low in last week's election.  The 2009 and 2013 races had remarkably identical total ballots cast.  Generally speaking closer races generate bigger turnouts so it begs lots of questions. That aggregate similarity masks some big changes in turnout within the city of course.  So here is a scatterplot of all city precincts comparing percentage Black to the percentage change in ballots cast between 2009 and 2013.

What it means is open to lots of conjecture.  But this, admitedly simplistic, look implies that turnout declined by approximately 14-15% for the Black population, but increased by 7-8% for everyone else.  The divergence of those two estimates produces a pretty different mix of voters last week compared to past elections.  Since 2009 was in total a pretty low turnout election to begin with, for turnout within the Black population to drop significantly below that cycle says something.

Yes, it deserves a much fuller model and I'm pondering that.  I bet some variables on income, age and past voting patterns would all show up as pretty significant. Also for those who might ask, that result is unweighted on size of districts. Note that the vertical axis compares votes cast in each mayoral election, not total ballots cast technically.  Some folks who showed up last week did not vote at all in the mayoral election.


Anonymous MH said...

Can you assume that turnout for each race is independent of the racial distribution of the district? If not, will income and past voting patterns be enough to control for that?

Tuesday, May 28, 2013 11:58:00 PM  
Blogger C. Briem said...

Desperately seeking the perfect instrumental variable.......

Wednesday, May 29, 2013 6:46:00 PM  

Post a Comment

<< Home