241 on polling

From the Wall Street Journa, October 13, 2008:

Good Polling Means Not Influencing the Outcome

I appreciate Thomas Riehle's review of my book, "The Opinion Makers: An Insider Exposes the Truth Behind the Polls" ("The Art of Yes, No and Maybe," Sept. 26), because I think it reflects more generally the industry's response to my concerns about the way most media polls distort public opinion. My concern is on the substantive opinions reported by media polls and why we can't trust them.

Measuring public opinion should include measuring how many people don't have strongly held views, as well as those who do. The failure of most polls to differentiate between the lightly and strongly held views of the public is one reason we can't trust the polls to reflect the will of the people.

My view is that providing information to respondents in a sample results in a distortion of what the public at large is really thinking. Different pollsters give different information, and then present the results, which often conflict with each other, as though they represent the American people.

Just three days before his review was published, three major media polls, all conducted in the same time period, illustrated the problems I discuss in the book. They came up with three wildly different pictures of public opinion on the bailout of the financial industry: The Los Angeles Times/Bloomberg Poll found a 24-point majority of Americans against the bailout, while the Pew Research Center found a 27-point majority in favor. In the middle was the ABC/Washington Post, which found the public evenly divided. Those are hardly useful "insights." Instead, they make a mockery of polling and of the claim that polls can measure the will of the public.

Until pollsters address the consequences of their manipulation of respondents into giving meaningless answers to poll questions, public-opinion polls will continue to confuse more than enlighten.

David W. Moore
Durham, N.H.