It’s Election Day across the nation, and they’re spitting out polls right and left. For every election and every issue there are more polls and stats than you can find in major league baseball. What’s the reason behind this abundance of “data?” Well, I like to call it Poll-itics (a portmanteau of poll and lunatic). Now more than ever there are tens (perhaps hundreds) of thousands of poll-iticers working around the clock making calls and robocalls, sending emails, and doing in person questioning to determine the prevailing winds in the USA on any issue. Certainly, this is a useful tool, but it starts to turn ugly when people treat these things as the gospel truth or stretch them beyond what they are. While polls are generally accurate in evaluating broad trends and feelings, all they really tell us is how the people answered the questions asked. So, as you watch the eledction returns flow in tonight, keep these things in mind as you are bombarded by a myriad of poll-itics:
1) It’s all about the method behind the madness. Not surprisingly, a poll is only as good as the method used to take it. Lots of polls purport to take a “random” sampling of adults from across the country and take their views on any issue. But, a sneaky polltaker can skew the results by skeing the sample. For example, if you are a GOP pollster and you want to reflect that Obama’s policies are unpopular, where are you going to select random people? Probably from areas that voted pro-McCain and GOP in the last election. It also all depends on how the question is framed. If you call a bunch of people and ask if they would support a health plan that would require the government to determine whether old people live or die, it isn’t exactly fair to use those numbers to say that the “overwhelming majority of Americans don’t support Obama’s health plan.”
2. Polls aren’t perfect. If you go to the doctor with the flu, what does he or she do? Well, you get your temperature and vitals taken, you describe your symptoms, and you get examined. Then, the doctor diagnoses and determines the course of treatment. What if, instead of the exam, your doctor called 100 people in the community and asked them how they were feeling to determine what was wrong with you and how to treat it? Sure, he or she might get an idea of the ailments in the local area, but if everyone else was suffering from the chicken pox, you’d be in a bind. It’s the same with polls. While they might give an idea of how people generally feel, it doesn’t do much to figure out what you will do when standing in the voting booth.
3. Correlation isn’t causation. What do polls really tell us? They tell us how each of the people asked the questions responded, and that’s it. Just because 55% of people responding said “x” doesn’t mean that 55% of Americans believe the same. They also don’t give us the why. If 55% of people support “x,” we don’t know if they do it for personal reasons, religious ones, political ones, or because they flipped a coin. For the same reason, trying to tie two different polls together to explain one another is crap. What’s my point? Well, that means that when people try to to explain why people in polls feel the way that they do, they are probably wrong.
Let’s look at an example. The big race to watch today is the New Jersey Governor’s race. It pits the incumbent (and unpopular) Democrat against a Republican challenger. Both sides have their polls about who is winning and whether people support the President and his policies. If the Repub wins, expect tomorrow’s press bites to say that it was because NJ wanted to send a message to Obama (and they will have polls to back up their belief). If the Dem wins, expect the soundbites to say that it was a message about how NJ wants hope and change (And they will have their polls too). The truth is that all it will tell us is who NJ wants for their governor. The other interpretation is like saying that people hate Michael Vick because they disagree with the policies of the NFL commissioner. Guess what, some people might hate Michael Vick because of the dog-fighting thing or for reasons entirely different.
In the end, people should use polls for what they are–general indicators. Trying to extrapolate the why is about as useful as trying to guess why John and Kate’s marriage didn’t work out. So, when you are watching the election returns, look at the actual votes (not the polls) and realize that all they are really telling you is who won or lost.