I spoke too soon

OK, thanks to everyone who sent me Numbers Guy links. I'll need to be more astute in the future.

Carl Bialik talks about the "social effect" of some poll interviews and how it can skew results.

For instance, President Clinton said during the Super Bowl that a third of American households had contributed to tsunami aid. Where'd that number come from? Polls.

But Americans likely haven't been quite as charitable as Mr. Clinton suggested. His figure appears to have been based on telephone surveys, and that's no way to measure charitable donations.

That's not to say Americans haven't been very generous in giving to victims of the deadly Asian tsunami. But it turns out to be very tough to get a precise measure of how many Americans have donated. Asking the major aid groups won't work, because you'd inevitably miss a few, and anyway you'd double-count for households that have donated to multiple groups. But a poll has a serious problem of its own: People tend to fudge when they feel social pressure to answer questions a certain way -- in this case, by saying they've given to a good cause.

So why were these polls even done? Well, it's polling organizations' job to do polls, and if people are curious about the number of U.S. contributors, I guess the orgs can do a poll on it.

But where are the media types questioning the validity of such polls? More and more, these are making my head spin. I don't discount them all. But it's been awhile since I've believed everything I read in them.