« Irony? | Main | It's Only One Game »
October 11, 2004
Polling
Interesting post from Mark Blumenthal on the peculiarities of "likely voter" polling. One of the questions used by pollsters to determine this catogory is to ask who the respondent voted for in 2000. When you total up those responses, it comes out that about 60% of people asked claim to have voted, even though the actual turnout was closer to 51%. It begs the question, if pollsters can't get an accurate read on an election that happened 4 years ago, how in the world can you trust them to predict the future?
--------
Posted by stan at October 11, 2004 11:23 AM