Boteveg's Pub

This is late, and no one reads this blog, but on the off-chance someone does — don’t listen to the polls. They don’t know what’s going to happen. That’s the TLDR, click for why.


The first thing you should know about Polls and Surveys is that every result ONLY REFLECTS THE OPINIONS OF THE KINDS OF PEOPLE WHO ANSWER POLLS AND SURVEYS.

I put that in caps because it is important. In a social media age, we like to think that since we throw our opinions into the ether, they are reflected in polls. But no. Polls can only claim accuracy over the people they actually talk to. And if you didn’t answer the phone when someone you didn’t recognize called or if you didn’t talk to the person with the clipboard, your opinion was not counted.

Now. Think about the kind of person who actually decides they can take the time to AND ALSO WANTS TO talk to a pollster. To answer a series of dry questions that the person asking can’t explain. THEY CAN’T EXPLAIN what the question means.

Leading to the second problem with polls. An d I should say here that most of my thoughts here were learned from Charles Seife’s Proofiness, if you want to read about all the other ways math is manipulated to make things seem true.

Second problem — if a question in a survey/poll is in anyway not clear, you will get bad answers. There are literal classes about how to ask survey questions because how do *you* interpret the word “could” or “would” or “can”? Yes, there are some questions that are clear. “Are Republicans a danger to Democracy in the US?” Clearly a YES or no answer. But since polls and surveys are mostly looking for opinions, there will be problems with answers.

And I’m not a statistics person, but that Margin of Error of 3% is bullshit. And it’s a bigger problem the fewer people you include in your poll/survey. Most poll takers keep going until they can say the MoE is 3% or under, but that means they stop asking in most cases.

Again, TL/DR — almost never trust polls.