This was the first election where the new voter ID requirement was in place in the state of Texas.  Democrats have claimed that it would suppress voter turnout while Republicans have claimed that it would not.  After election day, CNN featured an article by conservative blogger Bryan Preston claiming that voter turnout was not suppressed.

Preston’s methodology was comparing 2013 voter turnout to the most recent comparable election.  2013 was a off-year featuring proposed consitutional amendments, so Preston used 2011 as the basis for the comparison.  The most voted-upon amendment in 2011 has 690,052 votes, with an average of 672,874 voters for each of the ten amendments.  In 2013, the most popular one garnered 1,144,844 voters, with an average of 1,099,679 across nine proposed amendments.  Preston shows this increase as proof that voter turnout was not suppressed and then gets into demographical breakdowns in an attempt to prove the point more convincingly.  Not only was voter turnout not suppressed, but there was a huge surge!  Case closed, right?

Unfortunately, the entire argument is built upon the foundation of 2011 and 2013 being “comparable” elections.  In a nutshell, that we should expect roughly the same number of voters in those elections.  Let’s look back over a few cycles of these “comparable” constitutional amendment elections.  Let’s look at the amendment that garnered the most voters in each year, since that is the floor for the number of distinct voters.  Let’s trot over to the Texas secretary of state’s site.

  • 2013 – 1,144,844
  • 2011 – 690,052
  • 2009 – 1,055,330
  • 2007 – 1,096,410
  • 2005 – 2,260,695
  • 2003 – 1,470,443

Obviously, Preston is focusing on the wrong problem.  Instead of wondering if voter turnout was suppressed in 2013, he should be focusing on the voter suppression that occurred in 2007, when there was a more than 50% decline from the previous “comparable” election.

Or perhaps these off-year elections, which get very little turnout (the 2013 numbers touted by Preston was an 8% turnout) are driven by other issues, making it impossible to predict one election’s turnout from the previous election.

This is pretty obvious cherry picking from Preston, comparing 2013 results to an extreme statistical outlier.  The demographic breakdown he does is unfortunately worthless, because he’s working with bad data from square one.  You can’t make a silk purse from a sow’s ear.

How volatile are these elections, compared to presidential elections?  Let’s look at the past six presidential election totals.

  • 2012 – 7,993,851
  • 2008 – 8,077,795
  • 2004 – 7,410,765
  • 2000 – 6,407,637
  • 1996 – 5,611,644
  • 1992 – 6,154,018

There’s even quite a bit of volatility here (albeit over a timeframe that is twice as long), but the the high water year has just 1.44 times the number of voters of the lowest, compared to the high/low ratio of 3.28 for the constitutional amendment elections.

The key point is that voter suppression is not measured by whether vote total went up or down from a previous election, but by whether vote totals are lower than they would have been if the action that is alleged to suppress had not been in place.  That’s not easy – or perhaps even possible – to measure.  However, that’s no excuse to take a lazy shortcut and pass off the results as being the answer to the more difficult question.

Does the Texas voter ID law suppress voter turnout?  I honestly have no idea.

 

 

Enhanced by Zemanta