Wednesday, October 29, 2014

The 2014 Midterm Elections

The 2014 midterm elections are now a week away, and it’s time for a prediction on how the election will turn out.  (My 2012 attempt at prognostication indicates this may be an exercise in futility, but here goes.)

In Colorado, we have a new mail voting system, so results will be different from prior years.  Even so, I’ll provide two charts for comparison.

The first is for 2012, and has Colorado turnout results from October 31, just under a week prior to the 2012 General Election.  The second is for this year, and has Colorado turnout results from October 27, just over a week before our midterm election.  Because of the timing difference and the fact that turnout for midterm elections is generally lower than for general elections, the overall voting for Colorado so far this year is about one-half (660,113) of what it was in the 2012 report (1,150,698).

For comparison, let’s assume that if 2012 behavior occurred in 2014, we would have about twice the number of votes in place for 2014.  I’ll compare three counties: Douglas, Denver, and Boulder.  Douglas has a majority of Republican voters; Denver and Boulder are Democratic Party strongholds.

Here’s a chart of numbers for 2012 and 2014, with 2014 numbers doubled (approximately) to facilitate a comparison with 2012:

County      2012D    2014D (actual)   2012R    2014R (actual)

Douglas       18,158      20,000 (9,864)      44,487     55,000 (27,672)

Denver         68,204     74,000 (37,296)    21,483     30,000 (15,048)

Boulder        38,672      37,000 (18,512)    16,440     20,000 (10,239)

TOTAL       125,034     131,000 (65,672)   82,410    105,000 (52,959)

Comparing the three counties, we see that 2014 Democrats are voting in slightly higher numbers than they did in 2012 (about 5 to 10% greater turnout) while 2014 Republicans are voting with greater intensity (about 25 to 30% more).  If turnout is a guide, Republicans have a good chance of making Colorado “less purple” in 2014.

But what about other state elections?  Two polling services worth a look are and

Real Clear Politics gives a listing of public polling results, with an average (RCP Average) calculated for the last five polls.  Five Thirty Eight uses roughly the same polling results, but “handicaps” the data based on the voting fundamentals of a given state and the past performance of the various polling agencies.  Here is a listing of the ten “battleground” states that will determine control of the U. S. Senate next year:

State   538          RCP Average

AK      R+1.8  R+2.2 (R=Sullivan; D=Begich)

AR      R+3.7  R+5.0 (R=Cotton; D=Pryor)

CO      R+2.1  R+3.3 (R=Gardner; D=Udall)

GA      R+0.6  R+0.5 (R=Perdue; D=Nunn)

IA        R+0.9  R+2.1 (R=Ernst; D=Braley)

KY      R+3.5  R+4.4 (R=McConnell; D=Grimes)

KS       I+0.4   I+0.9 (R=Roberts; The “Independent” is Orman)

LA       R+4.7  R+4.5 (R=Cassidy; D=Landrieu)

NC      D+1.6  D+1.0 (R=Tillis; D=Hagan)

NH      D+2.6  D+2.2 (R=Brown; D=Shaheen)

With one week until the election, here are two things we know:

--Real Clear Politics seems to have a slight polling bias in favor of Republicans.
--Four  races (GA, IA, KS, NC) are within the “Margin of Fraud.”

That “Margin of Fraud” terminology might need some clarification.  It is a term promoted by Glenn Reynolds to describe the fact that close races are routinely decided in favor of the Democratic Party.  Republicans, unfortunately, have to win elections by at least a 2% margin.

Here in Colorado, we see the Margin of Fraud potential in places like Boulder.  With our new mail voting system, the only check on widespread fraudulent voting is Signature Verification.  Without a dual-party control mechanism, those counties dominated by one political party can easily manipulate voting.  And keep in mind you don’t have to be a citizen to vote in Colorado.

We will know the results of the 2014 election next week (assuming no run-off in Georgia or Louisiana), and my prediction is that Republicans will take control of the Senate by a 51 to 49 majority.

That means Republicans must win six of the ten races shown above.  Maybe Colorado’s Cory Gardner will be one of the six!

UPDATE 10/31/2014:
Here is the Colorado turnout report for today.  The data are similar to the 2012 report referenced above.  It shows unaffiliated turnout about the same as 2012, but 33,680 fewer Democratic Party voters and 36,408 more Republican Party voters.  In Douglas County, Republicans are voting in the same numbers as 2012, but Democrats and unaffiliated voters are turning out in fewer numbers.  If that trend holds, it will be good for Republicans in Colorado.

UPDATE 11/5/2014:
Republicans currently have 52 states in the Senate and the Democrats 46.  Alaska and Louisiana are still in play.  My prediction of 51 states for Republicans was too conservative.  More than likely, the Republican pickup will be 54 states.  We'll know on December 7, 2014.

UPDATE 11/11/2014:
Nate Silver analyzes the polling bias in the 2014 Midterms, showing there was a 4% bias in favor of Democrats.  As a result, few pollsters were predicting the extent of the Republican "wave."

It appears that about half the time there is a greater than 2% bias in polling data.  (About half the time, there is a less than 1% polling bias.)  What determines that effect?  Perhaps it's the American media.

In my next prediction, I'll see if I can determine if there is a 2% bias going on, and adjust for it.

Return to Bottom