Monday, May 12, 2008

 

The Dangers of Polling and Statistical Analysis

1/20 a polling result will fall outside the margin of error.... 1/20. I value this number greatly. Many elections will fall inside the margin of error, thus allowing voters to ultimately decide the election. But when the polls predict a result outside the margin of error, I worry that many voters see this result and turnout is either depressed or "the bandwagon" effect takes hold. Let's face it, if you know that a candidate is favored to win by 10 points and you have to take the kids to hockey, you probably will think "the election is already decided". Yet, I can always point to that one out of twenty times (although it is probably greater than this because all polls have to be wrong at once) when the polls are dead.... and I mean dead wrong. Take New Hampshire Primary 2008... Barack Obama was going to cruise over Clinton by 10+ points... but it was not to be... Clinton won by 3. But with the new fivethirtyeight.com, which I have praised, I honestly wonder if we might have reached the point where we ask "why even vote?"

If primaries are so predictable, why even hold them? If the matchup in the fall is so predictable, why hold the campaign season...

Now, I know elections aren't that predictable and never will be... But I always do fear the day.

Comments: Post a Comment

Subscribe to Post Comments [Atom]





<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]