Thursday, November 5, 2015

Reliability of General Election Polls Continue Downward Slide In One Direction

For the most part, with some notable exceptions, the polls have been pretty accurate predictors of presidential and midterm elections.   That was until the recent midterm elections in the United States in 2014.  While they predicted the GOP would pick up some senate seats with an outside chance, if everything went right, of taking over the Senate, none of the polls predicted the tsunami wave by the GOP in not just winning the senate easily with room to spare, but also with big gains by the GOP in the House of Representatives and in pickups in the state houses and governorships.

The reliability of the polls was questioned by some, but not by many as you could say one election, as big as it was, do not the polls make.  But then you had a couple of big foreign elections that were missed by the pollsters and missed in a huge way. 

In March of 2015, Benjamin Netanyahu's Likud Party won in a landslide victory, and re-elected for an un-precedented 4th time Bibi Netanyahu as Prime Minister, that no poll showed would happen.  In fact, every single poll in Israel just days before the election had Netanyahu and Likud losing and losing in a big way. 

Then in May of 2015, you had polls showing that in Great Britain there was a very tight race for Prime Minister with a good chance the Conservative Party was about to be thrown out of power.  What happened, a huge win for the conservative party and re-election of David Cameron as Prime Minister.  

So, beginning with the 2014 Midterm elections, and continuing with major elections in Israel and Great Britain in 2015, the polls have been not just wrong, but not even close to predicting the correct results. 

You might conclude that it shows you can't rely on the polls in a major general election not just in the United States but around the world.  No, that would be the wrong conclusion.  That is one of the secondary conclusions you might surmise from this downward slide of accuracy by major pollsters but not the correct one.  

There is one common thread that those variances of what the election results showed with the polling results before hand.  In the United States midterms, and in Israel, and in Great Britain, the pollsters were wrong and wrong in a big way, all in one direction.  Their polls were all off in determining the strength of the conservatives [party's/individual candidates] in those countries and in overstating the strength of the liberals [party's/individual candidates].  Could that possibly be just an amazing coincidence or indicative of skewed polling results that we will be seeing in future elections?  Fool us with one bad polling result, shame on you.  Fool us with 3 bad polling results, and we still believe in you, shame on us. 
Conservative Republican Matt Bevin - Governor elect of KY in an upset that, surprise, surprise, the polls missed
Now it is November of 2015, getting very close to 2016, and we have yet another example of the pollsters being way off in an election.  In the Kentucky Gubernatorial race on Nov. 2, 2015, we have this from Fox News"Republican Matt Bevin, a businessman and Tea Party favorite, beat Democrat Jack Conway on Tuesday to win the race for Kentucky governor -- becoming only the second GOP governor in the state in four decades."   

Matt Bevin won in a landslide by over 9 points.  So, what did all the polls show would happen in this race? 

NoteThanks to NRO's Jim Geraghty's 'Morning Jolt' and Real Clear Politics for the following graph.

The polls were way off, none showing that Bevin would win in a landslide, in fact none showing that he would win at all. The Real Clear Politics average just before the election showed the Democrat Conway would win by 3 points. 

Once again, showing the downward slide in legitimacy of the polls, and once again all in one direction:  overstating the Democrat's strength, and underplaying by a large margin the conservative Republican's strength. 

Sorry, I just do not believe these are coincidental outliers that just happen to be wrong all in one direction.  I am not saying pollsters are purposely giving out bad polling results, but they are doing  something wrong in their methodology that is missing the strength of the conservative candidate or the Republican Party. 

But whatever the reason, the reliability of the general election polls continue their downward slide in one direction, and for that reason we should take with a grain of salt what they will show in the 2016 election.  In fact, I submit just add 3-5 points to the Republican candidate for president from what the polls will be showing and you probably will get a more accurate result.

No comments: