This probably isn't anything new, but for some reason I've found myself much more annoyed by it this time around.
News networks don't understand how to read their own exit polling data. These polls aren't made up of randomly selected people, so you can't extrapolate their findings tot he general public. For example:
McCain appears to have won NH, and MSNBC is using its exit polling to explain why. Romney, according to their polling, is considered by voters to have been most negative. McCain, by contest, is considered to be the best choice for Commander in Chief. But what does this tell us? McCain won the most votes, so why should it surprise anyone that he also gets the most favorable responses in the exit polling? As Matt Y said last week, people are reading the lines of causation backwards:
...voters in Iowa knew that "change" was Obama's message, and so people who showed up to vote for Obama also told pollsters they were primarily interested in change. Clinton voters, by contrast, are trained to talk about "experience." This kind of thing is, I think, a major failing of conventional polling methods which tends to fairly naively assume that respondents' reported candidate preferences are built out of their reported character trait and issue preferences. It's likely, however, to be the other way around -- people who like Candidate X come to embrace key parts of Candidate X's argument.
This is an important point to keep in mind with all forms of polling, but it is absolutely crucial to remember when looking at exit polls. Exit polls by definition include the responses of only those who turned out to vote, and as such, they only tell you something about those specific people.
Or to put it more simply: If McCain won, why should it surprise anyone that the voters who selected him believe he is most qualified to serve as president?
Subscribe to this blog's feed
