The Associated Press released a story yesterday with this headline: “Polls gone wild: Political gripes in Internet age.” It is an interesting read about the role polls have played in the 2010 election season and I have a few interpretations regarding the story.
1. The griping of politicians about polls does not often seem to be based on the methodology of the poll. Rather, I think the politicians are trying to curry favor with supporters and voters who are also suspicious of polls. I would guess many Americans are suspicious of polls because they think they can be manipulated (which is true) and then throw out all poll results (when there are methods that make the polls better or worse). Some of this could be dealt with by dealing with innumeracy and educating citizens about how good polls are done.
2. There is a claim that earlier polls affect later polls and elections and that overall, polls help determine election outcomes. Are there studies that prove this? Or is this just more smoke and mirrors from politicians?
3. If there are charges to be made about manipulation, it sounds like the political campaigns are manipulating the figures more than the reputable polling firms which are aiming to be statistically sound.
4. Stories like this remind me of the genius of RealClearPolitics.com where multiple polls about the same races are put side by side. If one doesn’t trust polls as much, just look at how polls compare over time. The more reputable companies show generally similar results over time. Basing news stories and campaign literature on just one poll may look silly in a few years with all of these companies producing numerous polls on almost a daily basis.