David Moore, author of The Opinion Makers: An Insider Reveals the Truth Behind the Polls, is a former Vice President of the Gallup Organization and Managing Editor of the Gallup Poll. He’ll be guest blogging here all week.
THE MYTH OF WAR SUPPORT….Numerous excellent comments followed my post yesterday, in which I argued that polls don’t give us an accurate picture of public opinion — and that current poll results about oil drilling were just the most recent examples of this inaccuracy. Among many comments worth discussing in more detail is the one that suggested the problem appeared not to be with the polls themselves, but with the reporting of poll results.
I agree that news reports are often to blame for the misrepresentation of public opinion, but I can’t let us pollsters off the hook. Media pollsters themselves rarely attempt to measure other than superficial responses, in part because pollsters are integrated into the news process itself. Either the polls are owned and run by the news organizations, or the hired pollsters work hand in glove with the news editors and reporters to produce results deemed useful for news stories.
A prime example of the problem occurred in 2003, just before the United States launched the invasion of Iraq. All the major media polls at the time found widespread support for the war, typically by margins of two-to-one or greater. The questions included on those polls asked some version of “Do you favor or oppose sending American ground troops to the Persian Gulf in an attempt to remove Saddam Hussein from power in Iraq?”
In a February 2003 poll, Gallup asked a standard version of the question that all the other pollsters asked, and like the other polls, found a substantial majority in favor of the war — 59 percent to 38 percent, a 21-point margin. Only 3 percent said they did not have an opinion. However, as part of a special experiment which I helped design (as a senior editor of the Gallup Poll), the standard question was followed up with another, which essentially asked if people really cared that their opinions might prevail. And the results here revealed a very different public from the one that has come to dominate conventional wisdom.
To people who said they favored the war, we asked if they would be upset if the government did not send troops to Iraq. And to people who opposed the war, we asked if they would be upset if the government did send troops. Just over half of the supposed supporters and a fifth of the opponents said they would not be upset if their opinions were ignored.
The net result: Only 29 percent of Americans supported the war and said they would be upset if it didn’t come about, while 30 percent were opposed to the war and said they would be upset if it did occur. Another 38 percent, who had just expressed an opinion either for or against the proposed invasion, said they would not be upset if the government did the opposite of what they had just opined. Add to this number the 3 percent who initially expressed no opinion, and that makes 41 percent who didn’t care one way or the other.
What this experiment revealed was that instead of a war-hungry public, Americans were evenly divided over whether to go to war — three in ten in favor, three in ten opposed, with a plurality willing to do whatever the political leaders thought best.
These results from the experimental follow-up question reveal the absurdity of much public opinion polling. A democracy is supposed to represent, or at least take into account, the “will” of the people, not the uncaring, unreflective, top-of-mind responses many people give to pollsters.
If people don’t care that the views they tell pollsters are ignored by their political leaders, then it hardly makes sense that pollsters should treat such top-of-mind responses as a Holy Grail. Yet, typically pollsters and the media do treat those superficial results as though they represent what Americans are really thinking, with pollsters making no distinction between those who express deeply held views and those who have hardly, if at all, thought about an issue.