Public opinion polls of any sort typically carry a margin of error, a point especially important to remember when polls try to measure an extraordinarily split electorate like that in the United States. And despite ever-more-sophisticated polling techniques, mistakes still happen.
Take the recent midterm elections, in which polls mostly didn't catch the strong Republican tilt across the nation. In the Georgia governor's race, for instance, a poll taken just a week before the election wasn't off by a few percentage points in predicting a winner; it was flat-out wrong.
What's behind the mistakes? One big issue, as pollsters will acknowledge, is less than accurate methods for predicting turnout in the face of the nation's growing ethnic and racial diversity.
And reaching potential voters by telephone just isn't as reliable, even when pollsters do the needed callback legwork. More people use cellphones, and screen calls at home with technology like Caller ID and other telemarketing avoidance tools.
Further, polls can easily be as subject to abuse as they are to good use.
Politicians and ideologues often use them as statistics - to further their own objectives. What about a political party's own polls, for example, that indicate that negative ads actually work, even though they are so disliked by so many? Or what about polls regarding going to war with Iraq?
Citizens, of course, should not rely solely on polls as a lens for viewing world events.
Using larger sample sizes, conducting more focus groups, and asking more neutral, open-ended questions are among the remedies. But that takes time and money, and pollsters are in business to make money. Polls should always be taken with several grains of salt.