Thursday, November 08, 2012

How Do You View Pollsters and Polls




Not everyone likes pollsters. I get that. But I think that they serve a helpful purpose in our culture. In part this is because I AM a pollster, albeit not one that makes it on Real Clear Politics. As a pollster and an evangelical Christian, several people have asked me to weigh in on polls and this election.
I've commented several times about the polls on Twitter and Facebook. (I am a researcher and, not surprisingly, I comment on polls.) On a couple of occasions, I pointed out that it was statistically unlikely for Governor Romney to pull off an upset. Also, when I simply listed the poll numbers indicating President Obama "won" the second and third debates, some screamed "no way" (among other things).
Each time, some folks went crazy, explaining how the stats are all biased, particularly the ones from "those bad people at CNN." People questioned MY judgment and called me naive-- some said I was GLAD President Obama was winning (though I did not support the President's reelection). Yet, now that everything is over, it appears that my judgment was not the issue-- but there are issues of judgment to consider here.

Pollsters work very hard to get things right. They did so in this election. Contrary to the angry conspiracy theorists, pollsters do not generally seek to influence the election (at least the credible ones, and those are the ones you see on the news). We even have a code of ethics we all follow-- you can see what they asked, the sample methods, etc. And, to be honest, they all know they will be graded the day after the election.
Yet, some people were quite convinced that all the polls were lying. Furthermore, some made accusations about pollsters, including me for quoting them. That's not helpful and makes you look like you don't care about facts.
The most blatant display of this was when a number of people targeted one individual, Nate Silver. Nate is not really a pollster, per se; he doesn't ask any questions. He is a statistician who analyzes information. He first started out in baseball and was quite successful, but in the last several years has turned his attention to politics, and analyzing election polls throughout the campaign season. You can see more about the model he uses here.
When Nate shared a prediction at his blog, a number of people who were hoping for a different outcome were outraged. They said there was no way he could predict what would happen. And while it is true that trends can change quickly, it is also true that statistics are a helpful tool in looking at the big picture. And in case you were wondering, his projections were right.
So, what can we learn about polls from this situation?
  1. Just because everyone YOU know thinks one way does not mean it is the majority. When everyone you know is voting for the same candidate you are, the polls can seem unbelievable. But as we saw in the final results, they were accurate.
  2. Making accusations can hurt your credibility. Facts are our friends. And, everyone had some similar facts here. The truth is that while everyone acknowledged the momentum gained after the first debate, Governor Romney never had a sustained lead in the swing states.
  3. People have selective memory--we can think, "But I saw when he was up in the battleground states!" Yes-- in a few polls he was ahead, and those were given prominence on The Drudge Report. But, that was never consistent in the majority of polls.
So, the reality is that we WANT to believe the polls that agree with our own positions--and in a similar way, we tend to think that all others are biased, have the wrong sample, or are driven by an agenda. After the barrage of tweets I received after quoting two polls saying that President Obama won a debate, I tweeted:
"Only trust polls supporting what you believe! All others are biased, the wrong sample, or deeply flawed." -Mr Human Nature

Featured Post

Did Jesus Die For All Men

Did Christ Die for all Men or Only His elect?   The following is a written response to a brother with the following question about l...