Nothing better illustrates the problem with polls than the current chaos of the Republican Party. Is Donald Trump the current GOP favorite for the presidency? Or is he merely the best-known name? Or is he merely the one jokers name when they are asked questions in a poll? Are the respondents teasing the pollster? Have Palin and Gingrich really fallen so far?
We don't know.
Obviously, a long time before an election polls are especially inaccurate. But polls have built-in inaccuracies.
The most accurate poll is the exit poll, where those who just voted are asked how they voted. Exit polls are far superior because, most of all, only those who actually voted are polled. When most voters don't vote, political polls have to be taken with one huge dose of salt.
I think this can partly be corrected by polls asking respondents "how strongly do you feel about the upcoming election?" and weighting the response accordingly, on the assumption that those who feel most strongly about an election are most likely to vote. Which is not always true, especially a long tie before the election.
Or only those who always vote can be identified -- I nearly always vote--and their opinions given extra weight.
There are other problems with polling. There is the fact that every response is voluntary. Can one assume that those who are too busy to waste time answering a bunch of questions do not show a distinctive voting pattern? Then there are the
No comments:
Post a Comment