By Brad Torch
There are 16 months until the 2020 election. The Democratic field of 24 candidates begin debating in a week and President Trump has kicked off his re-election campaign. It’s incredibly early to take a guess who is going to win the Democratic nomination, let alone trying to figure out if Trump is going to continue his time in office or if a new President will take over The White House. Yet here we are already, hearing about early polling. Polling that likely means nothing in terms of who will be leading our country through 2024. I have so many questions surrounding this phenomenon, I’ve decided to delve into it head first. Over the last week multiple states have been polled, including Michigan, Iowa, North Carolina, and Texas. Most of these polls from different pollsters show Trump trailing up to 6 potential Democratic contenders, by up to 12 points. National polls also show potential Democratic nominees leading Trump as much as 13 points. How much stock can we possibly put into these polls this early in the election cycle? How much can we trust the polls after 2016 polling was so inaccurate? Why do we even care so much about these polls?
Presidential polling dates back to 1824 when John Quincy Adams, Andrew Jackson, and William H. Crawford all took a considerable share of the electoral college and the U.S. House of Representatives had to perform a contingent election to select a President. Needless to say, the polling which took place in taverns, militia offices, and public meetings, was hardly scientific, however even two centuries ago, there was a desire to forecast the results of the election. Our desire to predict the future has always existed in the same way we try to predict outcomes of sporting events, the stock market, and many other future events. Obviously, polling has evolved in many ways. Between 1984 and 2000, polling skyrocketed with access to further means of communication. After 2000, polling continued to grow with the development of interactive-voice-response (IVR) and internet polling opportunities. With more access to more voters, is it possible we have taken a step backwards? Nonprobability polls such as opt-in and internet polls are gaining popularity and do not provide accurate results due to sampling issues. Probability polls are also becoming more difficult to obtain due to non response and the increasing number of mobile phone only households. Furthermore, increased politicization in media creates motivation to systematically skew a polls results by weighting the sample one direction or another. This makes today’s polls extremely difficult to trust.
As most people who had any interest in the 2016 Presidential election know, polls showed Hillary Clinton leading substantially throughout the election cycle, but the final results were not what the polls told us. Analysts still haven’t been able to pinpoint exactly what went wrong with the polling, but the issues above likely played a role in the inaccuracy of the results. What I question is, whether or not adjustments can and have been made to the polling process to elicit more accurate results for the 2020 election? Many analysts predicted that the 2018 midterm election polling would again be inaccurate, however nonpartisan polls were more accurate than the average polls since 1998. This could make reading polls in this election cycle even more confusing than ever before. Another interesting consideration is the potential effect of polling from 2016 on polling, as well as voting behavior for the 2020 election. Polling can have an effect on how we vote, as it did in 2016. Many voters did not vote due to their belief that Hillary Clinton was going to “run away" with the election, based on polling numbers. I suspect that the lack of trust in polling may create the opposite effect, driving more people to the voting booth for the candidate who is leading in the polls.
Plain and simple, my advice is to take any and all polling with a grain of salt. We have a long and daunting election cycle to get through, and at the end of the election, accurate or inaccurate, polling should have no bearing on the results. As Americans, we should be voting not for who the polls indicate is the frontrunner or the underdog, but for whom each of us individually believe is the right person for the job. Polls can be an indicator, but are not an exact science, as was proven in 2016. Instead of trying to predict the future, it is our job to create it.