If pollsters and researchers continue sticking to old, traditional methods, they will not survive in a political environment where innovation is key to success.
In Michigan, four prominent research groups and news outlets had Hillary Clinton ahead by an average of five percentage points. Donald Trump won the state of Michigan by 0.3 percent. In Wisconsin, the final RCP average had Clinton ahead by 6.5 percentage points. Trump won the state by 0.7 percent. In Pennsylvania, pollsters had Clinton beating Trump by an average of 2.1 percent. Trump won by 0.7 percent. This trend goes on and on, as state poll after state poll was wrong each time, and by wide margins.
As we all know, Trump narrowly defeated Clinton in the 2016 presidential election. Had the Clinton campaign known how close the race really was in some of the swing states, we would likely have a different president in the White House today. Something seemingly so small, like accurate polling, has had huge ramifications on the country, from president on down.
This begs the question: why has polling become so inaccurate? One decade ago, polling had a great degree of precision. Often times, the only varying factor between polls was estimated voter turnout. Pollsters were able to predict elections like weathermen predicting the weather. They were usually right, but not always. Now they seem to usually be wrong. While weathermen have gotten better at predicting the weather, pollsters have gotten worse at predicting elections. Why is this?
The answer is simple. To continue with the analogy, weather forecasting has improved because of innovation and new technologies, whereas polling has largely stuck to traditional methods and thus fallen behind.
Some of the key factors that put traditional pollsters out of touch during the last election were polarization, the decline in telephone usage, and the lack of measuring voter enthusiasm. People were lying to pollsters, choosing the more socially acceptable choice; People were not answering their telephones because of advances as simple as contact lists; and people who said they were going to vote chose to stay home because they were more conflicted or less enthusiastic than pollsters predicted.
Society has changed tremendously in short time. Most outlets that were authoritative on this issue ten years ago have failed to evolve. Even today, many of the same researchers and outlets who were blatantly wrong about the last election still refuse to change their methods. Many have blamed the uniqueness of the Trump vs. Clinton election.
What new technologies could be implemented into the polling sphere in the near future? One possible proposed solution is blockchain technology. Blockchain technology allows for data to be decentralized, incorruptible, and protected using cryptography. It eliminates the possibility of manipulating data, as many have accused mainstream outlets of doing in 2016.
Platforms poised to try this idea in 2018, like OpenPoll, claim that blockchain technology “is the future of unbiased market research and polling.” When I reached out to OpenPoll over the phone inquiring how they will conduct their research, they said:
“We aren’t trying to predict elections. We are simply collecting accurate data on individuals and using statistics on voter demographics from the US Census to portray a more accurate picture of the electorate than traditional methods of polling.”
Blockchain technology is just one example of possibly disruptive innovation, but it certainly has market potential.
Another innovation, and one you are more likely to have heard about, is online polling. Online polling has existed for some time now, but has often been criticized as unreliable and inaccurate; however, new, more scientific methodology implemented by online pollsters has flipped the scales. There is evidence to suggest that online polls were more accurate predictors than “live” polls in 2016. This article from FiveThirtyEight, published less than three months before the election, shows just that.
Polling is a critical component of politics in the modern era. Ask Hillary Clinton who might have won the election had pollsters been more accurate. Or Gary Johnson, whose presence in the presidential debates was determined by his success in mainstream polls. Or Mike Huckabee, Chris Christie, or other presidential candidates who missed primary debates because of their poor performance in the mainstream polls.
Regardless of your thoughts on the last election cycle, or your views on current political issues, one thing we can all agree on is that the way we conduct polling is due for an upgrade.