Tuesday, September 25, 2012

Reality has a well-known liberal bias

Wishful Thinking is a fallacy in which on argues that something is true because she wants it to be true. As should be clear, this is a fallacy because the universe is generally indifferent to what we want. Reality is objective and wishing things worked differently is not going to make them happen. In my daily reading of the news I cam across this story on Salon.com about a website, UnSkewed Polls which claims to correct for liberal bias in all national polls. According to the author of this site, all national polls are skewed towards the democrats due to liberal bias among the pollsters. The author then claims to be able to correct for this bias, and when he does so, it appears that Romney is leading Obama by 7-10 points, a significant margin.

The basic "methodology" if we can call it that, is to assume that the votes of Republicans are under-counted, so the author simply reweights the polls so that, in his opinion, Republicans are no longer under-counted. For example, the author assumes that 34% of voters are Republicans. So, if a poll shows that only 24% of the respondents are Republicans, then the data must be skewed. So, the author reweights the data as if those 24% of respondents actually accounted for 34% of respondents. Basically he just gives more weight to the polling data of people identified as Republicans. Unsurprisingly, this reweighting seems to always lead to Republican dominance in the polling. Basically he is counting each Republican vote multiple times, which naturally leads to Republican dominance in the polling. If you don't believe that this is what the author is doing, you can read the details of this "methodology" here.

As should be obvious, this is a classic example of wishful thinking. The author doesn't like how reality is structured, so he just wishes up a new reality and assumes that to be correct. The problem, of course, is that the people who conduct the polling are really quite good at it, and very good at correcting for various biases. As Nate Silver of the NY Times notes:
First, the polling by this time in the cycle has been reasonably good, especially when it comes to calling the winners and losers in the race. Of the 19 candidates who led in the polls at this stage since 1936, 18 won the popular vote (Thomas E. Dewey in 1948 is the exception), and 17 won the Electoral College (Al Gore lost it in 2000, along with Mr. Dewey).
If the author of UnSkewed Polls was correct that polls have a statistically significant liberal bias, then why are they accurate predictors of the winner of the popular vote 94% of the time? If such liberal bias was infecting the polls, we wouldn't see them as accurate predictors of the winner so often. Thus, the only way the author of Unskewed Polls can object to polling data is by making up his own methodology that tells him what he wants to hear. As Stephen Colbert famously remarked concerning George W. Bush's falling approval ratings:
Now, I know there are some polls out there saying this man has a 32 percent approval rating. But guys like us, we don't pay attention to the polls. We know that polls are just a collection of statistics that reflect what people are thinking in reality. And reality has a well-known liberal bias ...

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.