Trump makes Nate SAD!

First, I would just like to state that I see what Nate Silver is doing and I don’t like it. ‘Our polls did less bad than most polls’, alright Nate, I know you have to save face but this article’s a little sad.

He's actually on the statistician and analyst side of polling but ignore that.

          It’s true that most polls and forecasts led the public to believe Clinton would be the leader of this country right now. While still hanging on to the popular vote, she did so by a smaller margin than anticipated, and with her unpredicted performance in the Midwest, she was unable to gather the much-needed votes to win the Electoral College.

Naturally, pollsters, and much of the American public was shocked. After all, most polls predicted a Hillary Clinton win by a pretty immodest margin.

Even though Nate’s article has some real arrogant vibes I think he’s right about a lot of things. For example, he explains that this was an election of high uncertainty, and we knew that from the polls. The margin of error for many of them was great enough that any of the dozens of polls that gave Clinton an edge over Trump could all be wrong within the parameters of a pretty expected polling error.

 

Honestly, I would have a latte with this man.

We know polls were wrong about the winner, but I think there’s room to give credit for what they were right about. The polls were right about Hillary winning the popular vote, which, in my opinion is the most we can really ask polls of such scale and uncertainty to get right. The polls were also right about many of the states that Trump would win, just not the ones he would lose. It’s hard to say exactly what went wrong, especially in the final days, when polls said Hillary was pulling further ahead.

A few theories:

-There’s a stigma around supporting Trump, (warranted or unwarranted) which may have cost pollsters nonresponse bias, or untruthful answers.

-There’s a certain demographic of Trump supporters pollsters just couldn’t reach, perhaps people in rural areas or those not easily reached through email or phone.

-Undecided or apathetic voters turned out for Trump at the final hour.

All of these are feasible explanations for why even the most prominent polls did not predict a likely Trump victory.

If I were to conduct my own poll for this election, I would have wanted qualitative information, since, well, the quantitative information didn’t do us much good to begin with. I’d ask questions like, “What issues are most important to you this election?” and “Does Trump’s stance on immigration policy make you more or less likely to vote for him?”. I find detailed and nuanced responses to be far more substantive in giving meaning to the results of the election than polling of voter preference.

I think the thing that makes polling so interesting and such a big part of the election cycle is the very fact that it is often wrong. I found Silver’s comments on the disbelief and shock of the nation intriguing. We know that polls are only estimated probability, but so many of us buried our heads in our hands on election night asking where the data went wrong. Polls are comfort for those they favor, and motivation for those they don’t, demonstrating an important part in the political process, given their apparent ability to influence elections.

But like anything, there are certainly limitations. Silver mentioned polls missing in the wrong direction, and I think he makes a fair point. In some cases there’s a group of people that can’t be reached and the only solution is extrapolating what we already know about those groups of people to predict their vote, which is fairly dubious way to go about election forecasting given the unpredictability of humans.

 It’s almost funny to me how much time and effort is invested in the practice of polling when we will eventually find out the real, accurate results soon enough. But this day in age is all about immediacy, and polling gives us a fast and effective way of getting a handle on public opinion, which can influence candidates and constituents for better, or for worse.