When Good Polls Go Bad

This political poll from the Associated Press and GFK is a prime example of why you should not rely on polls to "prove" that lots and lots of people support your pet project.  Check out the issues that people feel are "very important".  That would be . . . all of them.  The American public feels very strongly about every single issue you ask them about.  Forgive me if I think that most Americans are not wearing themselves to a frazzle over all of the following: healthcare, immigration, the economy, unemployment, the budget deficit, taxes, terrorism, the environment, energy, gas prices, our relations with other countries, Iraq, and Afghanistan.  I get exhausted just typing it.

Fr'instance, 67% of the people polled say that the situation in Afghanistan is "Extremely" or "Very" important to them.  Now, how many of those people do you think could name two major cities in this country with which they claim to be practically obsessed?

Most people have a couple of key issues they care about.  But few of them want to tell pollsters that no, they really don't much care what's happening in Afghanistan.  So you get nonsense like this.  The relative numbers matter--people clearly do care about health care and the budget deficit more than Afghanistan.  But the absolute numbers are nearly worthless.

This article available online at:

http://www.theatlantic.com/business/archive/2009/09/when-good-polls-go-bad/24727/