Wednesday, November 13, 2013

Election Shenanigans - Waiter There's a Fly In My Soup!


As I previously indicated, I decided to take a look at the numbers that came out of the election to try to see what happened. My supposition is that by comparing the detailed election results for my county from this year’s gubernatorial election against the results from the 2009 election, I should be able to learn something about what is happening to GOP voters or find some indicators about any nefarious goings on. The following is an accounting of my journey through these numbers. If you are so inclined, grab a cup of coffee, and join me in what will be for most a boring adventure in numberland.

Sources

There are few sources that I pulled from for this analysis. For election results from the 2013 election I got the spreadsheet export from the Board of elections site here.

Similarly, for the numbers for the 2009 election, I pulled from the Prince William County site, which had the results in a much less helpful format of a PDF file. This meant that I had to painstakingly hand enter the numbers from the PDF file into my own spreadsheet. I will freely admit that though I was careful, there may be one or two errors that resulted in that process. Here is the source PDF.

For population numbers I pulled data from the census sites, here and here.

Methodology

In preparing the data I had to make some decision. As a rapidly growing county in the suburbs of Washington, DC, Prince William County has had its voting districts big and small redrawn a bit during this time period between 2009 and 2013. This makes it sometimes exceedingly difficult to do a precinct by precinct comparison. To account for that, I compared each precinct and matched up names. I discarded results from precincts that either was eliminated or was created for the 2013 election. I know that every vote counts, but this is a statistical analysis and I had to base it on something comparable. I ended up with 63 precincts. For this analysis I also did not analyze the absentee ballots as they are indeterminate from a geographical viewpoint. If I have time I will come back to them.

Population vs. Voting
The only definitive numbers from the census that I could find indicated that there were 402,002 persons living in the county in 2009 and 430,289 in 2012. That’s a 6.57% change in two years, or 3.29% annual change. Applying this percentage across the years you get the following:


With that in mind, let’s look at some overall voting numbers. In 2009 there were 75123 votes cast for Governor. That means that only 19.32% of the county population participated in the election. In 2012 there were 97060 votes cast for Governor. Applying the same formulas, that means that population participation jumped to 21.98%, or a 2.65% increase in participation. This all looks perfectly reasonable right? Given the fascination with the first black president, a modest increase in participation of 2.65% seems reasonable, right. Well, don’t eat that cookie quite yet. Let’s look at it just a little differently. The population increased at an annual rate of 3.29% right? Thus, from 2009 to 2013 that would be a total increase of 13.16%. So if voter rates were to follow population trends, then you would expect a similar increase in total number of voters. Given the number of votes for both years above, the actual number is 13.74%. Since that is less than a 1% variance, it appears reasonable. No Flies in my soup yet, though I may see a very tiny hair in there.

Of course just looking at total number of voters is not enough because elections are about people making choices and a winner being chosen. An important, nay the important question is how they voted and how did that change since the last election. Taking overall numbers from the fore mentioned 63 precincts, in 2009, you have:
  • McDonald (R) 58.38%
  •  Deeds        (D) 41.62%

 In 2013 we have:
  • Cuccinelli (R) 43.43% 
  • Macaliffe (D) 52.37%

That’s a decrease of 14.33% in republican votes and an increase of democrats by 10.12%, or a net change in votes for democrats by a whopping 25.70%. Now let’s consider those population numbers again. Even if every single new participating voter that moved into the county was a democrat, this swing is twice as much as you would expect. There is something very odd going on here that warrants a deeper look at the specific numbers. Ah, I think I see a tip of the a fly wing down in there hiding behind a meatball...

Here’s a more detailed plot of all the precincts net change:


Now admittedly that really isn’t unexpected, given that you already know that the overall net loss for Republicans was 25%.

Let’s take a look at a ‘typical’ precinct. Here are the 2009 and 2013 results for the Chinn precinct:

2009
2013
Rep
Dem
Tot
Rep %
Dem %
Rep
Dem
Ind
Tot
Rep %
Dem %
463
351
814
56.88%
43.12%
377
582
42
1001
37.66%
58.14%
Note that total for that precinct only increased by 187 votes, or 22%. While the Republicans had a decrease in 86 voters, the Democrats saw an increase of 231 voters. Now let’s assume for a minute that half of the missing Republicans were disaffected voters that decided to cast a ‘pox on both your houses’ vote and account for all of the Independent voters, and the other half got sick, had to work, etc. That still doesn’t explain how in a precinct where only 1001 votes were cast, the Democrats managed to generate 231 new votes or an increase of 28%. Now if you are going to say these are Obamabots, remember that the 2009 election was just one year after Obama first won the presidency. Those voters were already there! The question then is where did all these new voters come from?

Let’s take a closer look at this precinct. Here’s the official map of that precinct:


I am very familiar with it as I go to that library all the time. All those roads at the top of the region are a townhouse community where the majority of the population for the precinct lives. The bottom 2/3rds of the precinct are older established homes. So if there were a turnover in the community, I would expect most of it would come from the townhouse community.
  
I did a quick search of ‘homes recently sold’ atRealtor.com and came up with 42 homes sold in that area within the last 6 months, 30 of which appear to be in that precinct. So if that held true for the last 4 years, that would be about 240 sales, and assuming that all of the new residents had at least one voter, and each of those voters were Democrats, then that would account for the additional votes. Let’s just see how improbable that is. First, consider where I noted above that about 20% of the population in the county voted. Assuming a two person household for the above 240 houses, that’s:
480 * 20% = 96 voters
Now it really isn’t reasonable to assume all of those new voters are Democrats, so lest use the election results to divvy them up and we come up with:
96 * 58% = 55 Democrat voters 
That seems a really far cry from 231 voters needed to satisfy what is being reported. The other assumption is that rate of sales has remained constant throughout the last four years. Anyone who has thought about selling a house in this area can tell you that is just not true. Along with the rest of the country, sales (and prices) have been very depressed over the last four years and have only picked up in the last year. Yep! I can see it now... A big old nasty horse fly, swimming in my soup; and I think he's doing a backstroke!

I am not a statistician, but from what I can see here, these numbers look extremely suspicious. It appears that voters have appeared out of thin air to pull a lever for the Democrats. I welcome constructive analysis, based on actual numbers and logic to explain where these voters came from, because I don’t see it. If I get time, I intend to winnow down further into this, perhaps taking a specific precinct, voter list, and knocking on doors to see if the people who are recorded as voting actually did. In the meantime, I welcome all constructive, logical, analysis of what I have presented here.

No comments:

Post a Comment