The Polls Are Accurate But Don't Predict Anything — Except Biases

Pollsters will confidently and proudly remind us that their “opinion-gathering instruments” were dead-on regarding the 2016 presidential campaign.

They cite that on Tuesday, November 8, 2016, Election Day, Real Clear Politics reported that two polls (IBD/TIPP and LA Times/USC) had candidate Trump up by 2 and 3 points respectively over Clinton.

The only issue with this revisionist history is the number of polls referred to in the Real Clear Politics average between 10/26/16 through 11/7/16 — 65 percent to be exact (194 for Clinton out of 297 total polls) — which had Clinton ahead.

And to add to these the “allied perceptions” that those in the know (pollsters, consultants, politicians) thought these two polls were “outliers,” reinforcing the “group-think” that Hillary Clinton would certainly become the 45th President of the United States.

And there lies the problem — the fact that polls are accurate in the long run, but are not perfect regarding outcomes no matter how much we want them to be true. Within the industry, it’s referred to as predictive validity. And why polls are probably not usable in forecasting results especially when they are weeks away from Election Day.

Simply said, when making a forecast or wagering a bet in Vegas, it’s probably best not to use a poll for that particular human enterprise.

A possible way to better understand the business of polling is to look at three distinct types involved within the industry and their role in the process. Call these the metricists, the apologists, and the skepticists.

“Metricists” are operationally defined as professionals who are the methodologists within the polling industry (and not a person who writes in or analyses poetic metre). We see them hard at work making sure that every poll they construct is reliable. They understand that polls can do little to predict a future outcome, which they will say, is not their intent.

The “Apologists,” those who interpret polls with either an intentional or unintentional bias, are another important group within the industry. These folks include campaign managers, candidate/party spokespersons, or opinion columnists. They usually have an agenda and must be viewed with a high degree of skepticism.

“Skepticists” are the people who consider the poll’s methodology, the organization conducting the poll, and the different parties citing the poll before coming to an informed conclusion. This is where we all should strive to be because it allows for objective thinking rather than subjective feelings in our poll analysis.

An example of utilizing a “skepticist” approach in poll analysis is to consider the recent NBC-WSJ poll September 23, 2018, where Democrats have a 12-point generic ballot lead over Republicans. A number of issues come to mind.

First, what is the relationship between the generic ballot and individual congressional races throughout the country? In the past, there may have been a relationship, but in these historic times, may not apply. Remember we are in the Trump Era, where nothing is what it appears to be.

Secondly, what about that dirty little secret that there is a decline in Democratic Party affiliation? Again, oversampling Democrats may be the reason for the high generic ballot response and a potential Type I Error in our analysis — saying that there will be a “Blue Wave,” when in reality there will not be one at all.

Third, we are far enough from Election Day that things can change. Remember 2016 and how polling data began to change in the final weeks of the campaign, which many campaign professionals failed to see, because of “individual bias.”

And finally, there may not be enough angry activists “out there” to vote come November. In short, the vociferous, motivated, and determined-to-vote against Republicans may not be enough to “out-show” the silent majority.

Right now it’s the apologists shaping the polls interpretation, when skepticists should be keeping the process honest thereby reminding us that it’s always easier when you have marketing in mind.

Dr. John Tantillo is a marketing and branding expert, known as The Marketing Doctor. JT utilizes his doctoral skills in applied research psychology to analyze the issues and personalities of the day utilizing his marketing and branding lens. This provides his readers with additional insight needed to understand the “new normal” in politics, news, and culture. Dr. Tantillo is the OpEd writer for Political Vanguard. He is the author of "People Buy Brands, Not Companies,” and the Udemy course "Go Brand Yourself!" You can follow him on Twitter @marketingdoctor and at Facebook.com/dr.johntantillo. To read more of his reports — Click Here Now.