Why the Polls Got it Soooo Wrong

Photo: Hillary Clinton in Philadelphia on July 8, 2016. (AP/Matt Rourke)

The US Election is over and the result was shocking for many, and surprising to most. Secretary Hillary Clinton was a clear favorite, and several tracking polls had her in the lead for several months. So how could so many experienced pollsters get it wrong?

According to RealClearPolitics, polls that consistently had Clinton in the lead prior to the election included Bloomberg Politics, CBS News, Fox News, Reuters/Ipsos, USA TODAY/Suffolk, Quinnipiac, Monmouth, Economist/YouGov and NBC News/SM. Even the quintessential number cruncher, Nate Silver, on his FiveThirtyEight.com website showed Clinton with a high probability of winning the White House. Failing to accurately predict swing states like Florida, Ohio and North Carolina is understandable since polling numbers consistently fell within the margin of error and it is difficult to predict what undecided voters will do on Election Day. But Democrat strong holds like Pennsylvania, Michigan and Wisconsin flipped in favor of Donald Trump and these states, according to most pollsters, were virtually “in the bag” for Clinton.

It is generally difficult to predict elections, especially when you have two candidates in a close race where the margin of error will be at its peak. An added problematic factor in US elections is the large number of abstentions. Estimated at around 56% (votes still being counted), voter turnout in 2016 was lower than expected and significantly lower than the elections that catapulted Barack Obama into the presidency. This means pollsters have to estimate the likelihood a respondent will actually vote. Voter fatigue and institutional factors could have come into play and may have affected some Clinton voters. Federal Bureau of Investigation director James Comey’s “disclosure” in a letter just 11 days before the election is another issue which may have caused a late swing with some undecided voters. Based on exit polling, it appears the Trump campaign played heavily to the fears and anxieties of white voters, further adding to the uncertainty of previously non-participatory voters deciding to exercise their franchise.

Most polls are done by phone to ensure they reach a wider demographic while producing results relatively quickly. Limitations such as secret numbers and the fact that around half the US electorate only use mobile phones can be overcome with random-digit-dialing methods. However, given their intrusive nature, phone surveys have notoriously low response rates, sometimes below 10%. And not reaching certain demographics can introduce bias. But the potentially biggest factor may have been all the negative ads and media coverage surrounding Trump’s troublesome, if not genuinely irreverent, behavior, causing his voters to be less likely to admit they intended to vote for him. US citizens are constantly bombarded by phone sales representatives, making it increasingly difficult for pollsters to cut through. And when they do, the degree of trust they receive may be quite low, especially considering the sensitivity of being asked which candidate they will vote for.

Interestingly, a limited number of online polls did predict a Trump victory. The Los Angeles Times/University of Southern California tracking poll was one of them. This poll was based on an online survey with a recruited panel of voters. The criticism of online panels is that panel members are not selected randomly. Instead the structure of the panel is matched to the demographics of the population. Online surveys, however, have been growing steadily over the years as they offer the two things respondents value most, convenience and privacy. If some voters were sheepish about admitting to a human pollster that they would vote for Trump this would not have been an issue online.

Could it be that randomness in the face of convenience and privacy is over-rated? It really depends. Yet pollsters will no doubt look into this matter to improve accuracy and predictability. Pundits will also ponder the claim that Trump awakened some sort of “silent majority” that somehow polls failed to capture. What we do know is that the 2016 US election result proved to be a total surprise and the way the race was won was completely unprecedented. Are pollsters facing a shake up as well? It is possible, because after this election it seems nothing is impossible anymore!

If you found this article useful, please remember to ‘Like’, share on social media and/or hit the ‘Follow’ button to never miss an article. You may also be interested in the following article on How to Engage Cambodian Youth – Politically.

About the Author: Daniel Lindgren is the Founder of Rapid Asia Co., Ltd., a management consultancy firm based in Bangkok that specializes in evaluations for programs, projects, social marketing campaigns and other social development initiatives. Learn more about our work on: www.rapid-asia.com.

Connect With Rapid Asia l Facebook l LinkedIn l Twitter l