The city is looking at modifying its voting precincts, Henry Graff reports for NBC-29. Right now there are eight precincts of wildly differing populations, which means that different precincts have to handle different turnouts and political parties that weight their nominations by precinct are making some people’s votes worth more than others’. Voter registrar Sheri Iachetta will be educating City Council on the topic, so that when census data comes out in February, they’ll be prepared to consider whether and how to modify those precinct boundaries based on that new data.
Out of curiosity, I took the results of the 2008 presidential elections by precinct, compared the percentage of the city’s vote that was cast in each precinct, and graphed how far off from the ideal percentage each precinct is. With eight precincts, each one should, ideally, be receiving 1/8 of voters, or 12.5%. Yet (for example) the Venable precinct received 18% of all votes, a 47% overrepresentation. The results look like this:
Obviously, turnout varies from election to election by a certain percentage, but even allowing for that, most of these precincts are wildly mis-sized. I’ve been told that the precincts haven’t really changed for many years. They were established back when the city was smaller, in terms of square miles, and as the city grew, precincts were simply expanded to reach the new borders. I don’t know if that’s true, but looking at the map, it makes sense.
Whatever changes Council winds up making, they’ll have to be signed off on by the Department of Justice, as is required under section 5 of the Voting Rights Act for most of the south, a result of historical discrimination against minorities.
So, since none of the elections held in the City are on a per-precinct basis, it seems like the voting rights act would be completely neutral. The votes for city council are all at-large and the city is entirely contained inside single districts for State offices and also for federal offices.
I suppose an argument could be made that votes in certain neighborhoods (say, surrounding Venable) are suppressed by having too many voters for too many machines, but even then, you’d likely have to show that there were significant disruptions and that people were unable to vote to turned away.
Although your chart presents the data in a fairly powerful form (% deviation from your “uniform distribution”), it’s not clear the disparities are that extreme. It’s equally possible that the inconvenience of a more crowded polling place is more than offset by being moved into a different precinct with an even less convenient polling location. The schools provide suitable buildings for polling locations, but not all areas have good central polling locations available.
Skimming the map, it certainly appears that voters from JPA and Venable could easily be shifted into the under-utilized Alumni Hall location – though those folks may feel such a move is not convenient.
I think you’re over-emphasizing the disparities.
Do you mean that the disparities aren’t as numerically great as I’m claiming that they are, or simply that they’re not important as one might conclude, based on my devoting a few thousand pixels to a graph? If the former, I always welcome corrections!
Shoot, I just noticed that, somehow, my X axis goes from -40% to 50%, despite my strong intent to display the same amount on both sides of the graph, white space be damned. A minor sin in graphing errors, but it’s a personal pet peeve.
simply that they’re not important as one might conclude, based on my devoting a few thousand pixels to a graph
That.
Basically, this is kind of like saying, “we see a 300% increase in the prevalence of foo when fee is present” – a handy trick employed by epidemiologists seeking to move policy – when the increase is from 1% of the population to 3% of the population – sure, that’s a 300% increase in the RATE.
This is not exactly what you are doing, but by presenting the discrepancy as a % variation, rather than the raw numbers (the voting rate or per-precinct % of total cast votes), then suddenly the graph isn’t so dramatic.
I have no idea if the numbers are correct – I’ll take your word for it – or Sheri Iachetta’s.
I don’t think the X-axis error is such a big deal.
Finally – in picking the vote totals for just one year, you leave your analysis vulnerable to year-to-year noise. It’s entirely possible the Venable neighborhood was highly motivated in the ’08 election (this would not surprise me, given the makeup of most of those neighborhoods). I’d expect the same bump in Carver, but then Carver has been drained of population over the years.
How’s about we do the analysis this way: get the voter rolls – registered voters – from Iachetta – and use that instead of turnout.
Go for it. :) You get the numbers, I’ll graph them however you like (or you can), and I’ll post them here!
Hmm…the SBE may not give me the rolls for C’ville, but I am going to try. I need to get the CD anyway. Perhaps after the election Iachetta will have time to give the totals.