As we get very close to Election Day, I wonder more and more about the value of polls. How many are accurate and how many are constructed so as to give a preordained result? For election races, the polls should all converge as we get close to the election. Even for those pollsters who inject bias into their numbers, there is a desire to appear to be correct when the actual numbers are counted. On the other hand, for questions like the job approval of the President which are never tested with actual voting, the impetus to move towards accuracy is never there. This often leads to strange results.
A good example of this phenomenon is the recent Newsweek poll of the public about Obama's job performance. During October 10 separate major polling organizations asked the public about they rating of obama's job performance. Of these polls, only two found more people approving of Obama than disapproving. One of the two was in Newsweek, the new self-proclaimed journal of liberal opinion. The Newsweek poll, taken at the end of last week found the public approves of Obama's performance by a margin of 54% to 40%. The other poll showing approval was from Pew and it found a one percent margin.
Is it mere coincidence that liberal Newsweek found a 14% margin of approval for Obama when no one else did? Is it possible that Newsweek had its poll tailored to give a result that its readers wanted to see? The statistical chance that the Newsweek poll is just the result of chance is very small. Newsweek supposedly polled over 1000 people to get its result while the other polls questioned over 8000 people.
The only reasonable conclusion that one can draw is that Newsweek rigged its poll to produce the result it wanted.
No comments:
Post a Comment