What is the purpose of exit polls? I’m not talking about questions that try to determine which groups of voters did what and why. I’m talking about horse-race exit polls, released just after (often just before) the close of voting–polls that project who will win, who will lose, and by how much. What is the point of releasing these horse-race poll results? As far as I can tell, the purpose is to satisfy curiosity, thereby allowing voters to quickly switch back to their favorite entertainment programming, or even (heaven forbid) turn the television off. But at what cost?
If exit polls are accurate, they accomplish these ends. But what if the polls are wrong? In that case, they potentially distort the race. Exit polling in New Hampshire wrongly called Obama the victor. That mistake magnified the significance of Hillary’s victory. Yes, to a degree that would have happened anyway, since the exit polls were only confirming what pre-election polling had already said. Even so, the release of faulty exit polls favorable to Obama artificially magnified the impact of Hillary Clinton’s New Hampshire victory. In other words, the presidential race was significantly influenced by an entirely avoidable error–the decision to release possibly mistaken exit poll results before actual results were known.
Tonight’s exit polls show McCain with a five point lead over Huckabee. Now it’s true that the networks are not actually calling the race for McCain. They’re saying the race is too close to call, and that the exit polling may therefore be off by more than five percent. But if Huckabee turns out to be the victor, his win will have been artificially magnified by the “come from behind” image created by faulty exit polling. This may or may not happen tonight (apparently not–the nets seem to have just called it for McCain), but it happened in New Hampshire, and could easily happen again in other states. So why are we letting our presidential election process be distorted by exit-poll results that can clearly be mistaken?
I’m no expert on exit polling. Maybe someone can explain why quickly released horse-race exit polls are actually very helpful, and why the good they do outweighs any distorting effect caused by errors. But in a year when primaries are everything, and where faulty exit polling has already had a distorting effect on the race, it seems like time to at least raise the question.
The practice of releasing “horse-race” exit polls may be easier to change than we think. The networks cooperate (right?), and so carry the same “horse-race” exit poll results. No competitive advantage there. If we chuck the polls and go strictly with real results instead, doesn’t that just bring more viewers to the coverage? That gives the networks a motive to change. So it would seem to be in everybody’s interest to get rid of horse-race exit polling (but not the more detailed sort of exit polling). Or am I missing something?