Pluto downgrade shows why polls can't always be trusted

Pluto / AP

As the presidential campaign heats up, we should look closely at a faraway object for a lesson about polls. Billions of miles away from us, Pluto spins happily around the sun, ignoring all polls and surveys.

We would be wise to adopt its attitude.

In 2006, it was a poll of pocket-protector-wearing, hung-over, and disco-bobulated astronomers that determined Pluto wasn’t a planet after all (officially, we now have only eight planets in the solar system - Pluto was sent to the kiddies’ table).

Now, news outlets are going to be pushing polls on us for the next few months, as the November 2016 election approaches. There will be polls pitting Hillary Clinton against Donald Trump in a variety of scenerios. There will be polls to gauge the public’s mood on everything from the economy to national security.

So here’s a quick guide to reading polls.

First, size matters. The smaller the poll, the less representative it likely is.

That 2006 vote on Pluto is a great example. In August of that year, the International Astronomical Union (IAU) met in Prague in the Czech Republic for a 10-day conference. Like any professional conference, there were plenty of “social activities.” On the eve of the last day, for example, a free “disco party” was thrown for the delegates. Wouldn’t video from that event be interesting?

The number of astronomers attending each morning session dwindled, according to news reports. Only 2,700 of the IAU’s 9,000 members attended the conference in the first place, and by the final morning, only 800 showed up. That may have had something to do with the disco party the night before.

And of those 800, only 424 cast ballots on the planet question.

Is that poll representative of how most astronomers feel about Pluto? Within days, hundreds of IAU members who had either not attended the conference or left before the last day circulated a petition opposing the decision. The majority of astronomers, it turns out, are sane, and were quite willing to leave poor Pluto with its dignity.

This clearly demonstrates that polls can be skewed by any number of factors, including size. And Czech beer.

A second important thing to consider is just whom the poll is polling. There are polls that sample the public, or “registered voters,” or “likely voters.” In general, the polls surveying “likely voters” are more accurate.

And all polls will list a “margin of error.” This is basically just an expression of uncertainty. It’s a mathematical calculation based on sample size and the “confidence interval.”

So a poll that lists a three-point advantage for Candidate X, and acknowledges a margin of error of five points, means less than nothing.

Polls do have their uses. They can help candidates focus in on issues that voters want to hear more about, for example.

But for the most part, they’re merely fodder for political types to talk about (at least, when they agree with the results).

Just ask those Disco Astronomers. But not too loudly - they don’t feel well.

Recommended for you