KCK Man’s Tenuous Grasp Of Statistics Results In Embarrassing Letter To Star

The last campaign season saw a new obsession with the science of polling, with RealClearPolitics and 538 seeing spikes in popularity they’d never expected (though Nate Silver probably expects everything). A common question, though, is just how these polls work. Can you really evaluate the national mood based on a survey of 1,500 people? How about 600? Turns out, given our advances in statistical technology, that: yeah, you totally can. But you know who doesn’t buy a word of that mumbo-jumbo? Al Rolls of KCK, who reveals little more than a basic misunderstanding of math in his letter to the editor today.

Polls, polls, polls. The media spout poll results as if they are a true reflection of the opinions and thoughts of the majority of Americans.

The reason some pollsters do not tell us how many people are polled is because it is a very small number. Most of us do not get to express our opinion except on Election Day. I have voted every year for more than 45 years, and I have never been polled about anything.

Think about it. How does anyone know what TV show you watched last night? The Star could enlighten the public by doing its own poll asking how many people in the Kansas City area have ever been polled. The low number compared with our population would open eyes and shed light on this ruse.

Dear Al,
Without getting too technical, here’s how it works. To achieve statistical accuracy, you basically need a margin of error under 3 percent. You find that by dividing one by the square root of the number of people in the survey. As luck would have it, polling a relatively small number of people actually does extrapolate accurately to larger populations — and that’s why we’re able to poll 1,600 people and get a startlingly accurate snapshot of the American mood.

As for your question re: knowing what you watched last night, the short answer is “same thing.” And we’d merely point out, on the subject of you never being polled, that it’s wrong to mistake correlation for causation. Of course, given your Fermatian math skills, you probably already knew that.