How Republican polling went wrong and why they should have known better
We watched it all go down; the supreme over confidence of the Romney camp, the attacks on “liberal” prediction models, and the embarrassment of Karl Rove. The Republican Party was absolutely shocked Tuesday night and they are now finally coming to grips with their mistakes. In an article by Jan Crawford from CBS News, she lists three major miscalculations by the Republican Party.
In short, she writes that Republican analysts misjudged Independents and undecided voters, and misread turnout. These problems are statistical factors involved in what scientists call the “experimental setup”.
The experimental setup is the framework in which you create your experiment.
Generally speaking, you have a problem you wish to solve and you model outcomes using different variables and weighting. You can modify your variables and weighting in order to see what possible outcomes may occur “in silico” before actually starting your experiment in real life.
In the case of modeling or predicting election outcomes, your weighting of the individual polls against one another is very critical, as well as the correct understanding of your individual variables. In the case of the Republican Party models, they terribly misjudged the independent voter variable as well as voter turnout. I believe that these misjudgments should have never happened and I will explain why.
Republicans felt that they had a big lead with Independent voters. State polls showed that Independents were breaking Republican. The problem was that these Independents were not true Independents.
They were simply Republican voters that were sick of the Republican Party, but would never vote for Democrat. The Republican analysts did not see this, but they should have. As most people know, Mitt Romney was caught on tape thrashing 47% of the American public as free loading bums that do not want to take responsibility of their own lives.
Shortly after the video went viral, a Wisconsin poll was released that asked the American public if they agreed with Romney’s statement.
Below are the two points on independent voters in the poll:
39% of independent voters say Romney’s comments made them less likely to vote for him
20% of independent voters consider them a positive
In one of my comments on TMV, I stated
What I like about this poll is that the comments are so ideologically divisive that you can see the real split between conservative, moderate, and liberal independents.
Only a hard core conservative could support the 47% comment. Anyone person saying that they support his comments is not a real independent…maybe just a conservative that dislikes the Republican party, but will vote R anyway.
First, I will toss out the first point. “39% are less likely to vote for him”. “less likely” is very subjective. It could mean Independent identifying liberals who could not possibly be further away from voting for Romney or just real Independents that were turned off by the statement…
The real information comes from the second point. As I wrote before, those 20% of Independents? …NOT Independents. These are conservatives that will always vote Republican. In my opinion, no moderate or Independent voter could ever identify with that statement.
Only very conservative voters could agree with Romney’s comment. This poll, and many like it, was not part of the Republican experimental setup and those Independents, that were really Republican, completely skewed a very important variable.
Instead of having a having close to 50-50 Independent split in their calculus, they had a very right-leaning split. This right-leaning split, in combination with the calculus that those Independents would also turn out as Republican voters, led to a double calculus of a large group of voters.
The miscalculation of Republican-“Independents” that were identifying as Independents hurt the Republican Party because they were not correctly weighted and added into the polling math, which led to an overconfidence that drove Romney to go after states like Pennsylvania and Wisconsin, when he should have been in Florida, Ohio, and Virginia.
Why did no one catch this problem? Most likely because of ideological bias. It quite possible that Republicans actually thought the non-conservative Independents agreed with the statement because they too believed it.
It is really hard to understand how they missed it.
The other two misses I will chalk up to ideology. It is unclear why they thought Hispanics or young voters were not going to show up at the polls. They have dismissed the two groups for decades. It is possible that there miscalculation was just an extension of their own prejudice.
The Republican Party has since turned the page on the Hispanic hate and even Sean Hannity wants to play nice. Whether that will translate in the trenches remains to be seen.
Granted, I have not personally seen the polling setup used by the Republican Party, but it is obviously very flawed and I think I have laid out the cause and why it should have been caught be someone, but wasn’t.
wrong graphic via shutterstock.com