Exit Polling and the Herd Instinct

Much of the conservative Internet was ablaze earlier with rumors of exit polls showing big Kerry leads in various states. It is astonishing that these politically savvy people, after all that has been talked about and written about over the past months, could fall for it. It shows innumeracy: Kerry up 20 points in Pennsylvania simply could not be true given the history of polls in the state, the history of past election results, etc. -- it's simply too far outside the confidence interval to possibly be accurate. It also shows a lack of common sense: consider that exit polls are taken by human beings (often pretty young, maybe something like a graduate student or an AP stringer) standing outside polling places, likely not very well-trained, likely not very well-paid, likely not very happy to be standing in the rain with a clipboard, and likely semi-embarrassed at having to accost complete strangers who are hurrying off to work after voting. Are they really going to create a truly random sample? Are they going to get responses from the woman who is pushing a stroller and dragging a toddler out of the poll? Are they even going to bother to ask the grown up man with a real job who is wearing a suit and a scowl to stop and fill out his little form? Come on. He's going to ask the people who look like him and act like him because they are less threatening to him and have the time to answer. And, more likely than not, the type of person who wants to be an exit poll taker -- like the type of person who wants to go to Journalism School -- is going to be liberal. There is going to be bias in all of these things.

Perhaps most importantly, it pays to remember that everything costs money. Paying the bus fare or the mileage for an exit poller costs money. Paying for his travel time costs money. So where do you think the exit polls are going to be? Cities? Suburbs? Rural areas? Probably cities, which probably means they'll skew Democratic that way too.

Keep the faith. GOTV!