A Trump-supporter Facebook friend asserted over this weekend that polls are fake. I argued they’re not, in the sense that polling firms aren’t sitting at their computers simply pulling numbers out of their navels. I also conceded that all the polls showing Biden winning could be wrong. And they could be! Let’s look at it.
The polls in 2016 kept giving Clinton the lead and indeed she did win the popular vote by about 2.9 million (2%). The polls failed, however, to discern a relatively tiny number of disaffected urban working-class voters in three states — Wisconsin, Michigan and Pennsylvania — that tipped the Electoral College to Trump. Literally only about 80,000 people split between these three states trumped all the other millions. Forgive the pun.
My friend is wrong to say the polls are flat-out fake. That’s too cynical, too easy a rationalization. They are, however, a mix of art and science. Like adding too much yeast to flour, a little error here or an oversight there can wreck everything.
Among other things, pollsters have concluded they failed to account for education level after postmortem analyses of 2016 detected strong differences between those who attended college (voted Clinton) and those who didn’t (voted Trump). I think they also failed to look at the fundamental politics: Clinton’s arrogance and entitlement, calling workers a “Basket of Deplorables,” versus Trump’s attention and appeal to the working class. Whether his appeal was genuine is immaterial; it worked.
Is Biden Really Going to Win?
I’ll leave the politics for another discussion. Today I’m focusing on the polls and their methodology.
- Are the 2020 polls this year accurate in showing Biden winning?
- Or is another Trump surprise coming?
Unlike their counterparts, in 2016 the polling firm Trafalgar Group accurately predicted Trump’s Electoral College victory and predicts the same will happen again this year. They’ve achieved an impressive track record in many different elections. The Trafalgar Group polls much larger samples than other firms, reducing its margin of error, and it’s sensitive to the existence of so-called “shy Trump voters” who are reluctant to answer pollsters’ questions honestly.
Michael Moore, filmmaker and political prognosticator, adamantly warned Democrats in 2016 that Trump was not the joke everyone thought but instead a real threat. He saw Clinton’s miscalculations in Michigan and predicted the vote in Wisconsin, Michigan and Pennsylvania. He appeared last week on The Rising with his new warnings for 2020.
Now to the true science of election prediction. On Facebook this weekend my attention was called to the “Lochel’s Bakery presidential cookie poll” in Hatboro, Pennsylvania, near Philadelphia. I’ve never heard of it, so I promised to check it out. Here’s what I found. They’ve predicted the last three presidential elections correctly. SPOILER ALERT for those who choose to watch.
A Look Back at 2016
I’ll close with a video by FiveThirtyEight, a firm that aggregates and analyzes all the different national and state polls. FiveThirtyEight database journalist Dhrumil Mehta explains polling in general — a good overview if you’re not familiar — what happened in 2016 and what we might expect in 2020.
A Request to my Facebook Friends: If you have a comment I encourage you to enter it below instead of on Facebook. This way everyone can participate in the conversation!
A Request to Everyone: All opinions are welcome. I only ask that we remain civil and respectful of one another.