Marshall’s Law

Well it’s finally over. We don’t have to deal with this for another couple of years. No, I’m not talking about the Habs’s winning streak (sorry Jared!), I’m talking about provincial elections. And with this fun little exercise in democracy comes one of the press’s favourite pastimes: opinion polls.

Almost every day during the month leading up to the election, the polls on Montreal dailies’ front page saw the numbers tipping this way, then that way, and then in the third guy’s favour. The cycle would then merrily start over.

How I love to watch the second runner up squirm in response to questions about his third rank. “I never pay attention to polls,” he replies. Cue him one week later touting that it is the people’s will that he be elected, “why, just look at the polls!”

Opinion polls are not as reliable as many people believe. Take for example a poll on racism. If you were indeed discriminating against people based on their ethnicity would you be honest enough about it to admit it to a random stranger taking a poll?

A recent poll in the U.S., conducted by Gallup, asked a sample population what characteristics would make them not vote for a candidate. For example, if he was black, Mormon, gay, atheist, etc.

Fifty-three per cent of those surveyed said they would not vote for an atheist, 43 per cent would not vote for a homosexual, 24 per cent would not vote for a Mormon, 11 per cent would not vote for a woman and 5 per cent would not vote for a black candidate.

The results were received with mixed reaction. Many touted the fact that Americans were beyond the colour of Obama’s skin and that it didn’t matter he was black. Although the poll might show that racism isn’t a big factor in the American presidential race, it can actually reveal something different.

As one commentator pointed out, this poll seems not to be a reflection on racism in the U.S., but how socially acceptable some forms of discrimination still are.

According to the poll, it is more socially acceptable to be homophobic than racist or sexist.

Which brings us to the next problem related to polls. The sample population used.

According to a poll conducted, 66 per cent of those surveyed agreed that I should be named Mayor of Reykjavik. What you don’t know is that I surveyed a teddy bear, my dog and a glass of water. Damn that glass of water for disagreeing.

The preceding example might be outlandish, but it reflects the problem of how you can never be sure of how accurately representative your sample population is.

Take for example the Nielsen ratings, the famous way of rating how much viewership television shows attract.

When you hear about how a television series is tanking, it doesn’t mean that no one is watching. It just means that the Nielsen families aren’t. These are randomly selected individuals who agree to divulge what they watch on TV. This sample population determines whether a show has good ratings or not and then in turn whether it survives.

So you can blame these people for the demise of Arrested Development.

Next time you hear X number of people think so and so about this or that issue, think about why polls shouldn’t always be relied on.

Mark Twain was right when he wrote, “there are three kinds of lies: lies, damned lies and statistics.”

Related Posts