How Election Polling Works On 'Stuff You Should Know'


On this episode of Stuff You Should Know, Chuck and Josh are talking about election polling. When Donald Trump won the presidential election in 2016, it came as a surprise to most of the country, who had been hearing the news media confidently predict Hillary Clinton’s landslide win for months. People decried the science of polling and said all the polls had been completely incorrect, but when statisticians went back to figure out what happened, they discovered that the polls had actually been more accurate than in the past twelve presidential elections. It wasn’t the polls that were wrong – it was how they were being reported on, and how we understand the numbers. Not a math person? Josh and Chuck aren’t, either, but they make it manageable.

Polls are taken by creating a random population sampling of American voters and asking them questions to determine who they’ll vote for or how they feel about certain issues. But there are a lot of different types of Americans in our huge country, so it’s important to make sure the poll has something of an accurate representation of the populace; that’s why questions about college education, age, race, gender, voting habits, marital status, even whether or not you frequently volunteer, are included. Once at least 800 responses are gathered (and quite a lot goes into that process), the pollsters have to “weigh” the results, which involves adjusting the sample population against the actual population. For example, if the responses end up being from 80% white voters, they’ll adjust the numbers to reflect the 66% of voters they actually represent. 

There’s always going to be a margin of error, because it’s not possible to interview every single American voter and compile their data perfectly. Usually this falls within three or four points, but that can be quite a big difference: If someone is leading in a poll, 51% to 49%, that means they could win by as much as 55% or lose by as much as 46% and still be within the margin of error. But the bigger problem is that news media and campaign messaging like to convert these numbers to “horse race numbers,” to claim that someone has some percent “chance of winning.” Poll data isn’t meant to be converted this way: A 51% to 49% race is a close race, but it converts into the leader having an 82% chance of winning, which sounds like a foregone conclusion. Hear more about how polling is actually conducted, how to know that you’re looking at a legitimate poll, and why polling is so important for democracy, on this episode of Stuff You Should Know.

If you want to be sure you're listening to the podcasts everyone else is checking out, iHeartRadio has you covered. Every Monday, iHeartRadio releases a chart showing the most popular podcasts of the week. Stay up to date on what's trending by checking out the chart here. There's even a chart just for radio podcasts here, featuring all your favorite iHeartRadio personalities like Bobby Bones, Elvis Duran, Steve Harvey and dozens of others.

Photo: Getty Images