- While Democratic nominee Hillary Clinton was heavily favored to win the presidential election in 2016, election results surprised many pollsters when now-President Donald Trump emerged victorious.
- Numerous circumstances led to erroneous state polling in 2016, including undervaluing non-college educated voters and vast amounts of undecided voters.
- A report from the New York Times examines what the outcome of the electoral college will be this November if state polling proves to be as incorrect as it was in 2016. The report shows Joe Biden defeating President Trump, but just barely.
- Visit Business Insider’s homepage for more stories.
The results of the 2016 presidential election were a shock to pollsters and experts around the country: How could the polls and predictions be so wrong? In fact, the New York Times famously had Hillary Clinton’s chances of defeating now-President Donald Trump at 91% just three weeks before the election.
As of Sept. 29 with 35 days until the election, the New York Times released a followup of what the electoral college could look like in 2020 if the same mistakes were made by newspapers and pollsters alike.
—Nathan Tankus (@NathanTankus) October 1, 2020
The mistakes from 2016 include:
- Surveys in 2016 did not properly weight the value of non-college educated voters.
- Large amounts of undecided voters, which experts had difficulty predicting, who ultimately sided with Trump.
- Many polls and predictions came before the release of the Comey letter and after the infamous “Access Hollywood” tapes where Trump bragged about grabbing women “by the p—y.”
According to the New York Times, if all polling translates perfectly into votes (which the Times notes realistically will not happen), then Biden will receive 359 electoral votes — much more than the 270 necessary to win the presidential election.
However, if the same mistakes are made in 2020 as were made in 2016, Biden would only receive 280 electoral votes, or 22% less than current polling would suggest; 270 electoral votes are needed to win the election.
Have pollsters and experts learned from 2016? It’s unclear.
FiveThirtyEight recently said that presidential primary polls in 2020 had a weighted average error of 10.2 percentage points — about the same as the 2016 presidential primaries. The same FiveThirtyEight report also noted that in the final 21 days of the election in all races since 1998 that ended in 6-10 point margins — the approximate lead that Biden currently holds with less than 35 days to go — polls have been right 84% of the time.
Pollsters and experts alike have adjusted their models since 2016, but as FiveThirtyEight points out, the closer and the tighter the race is the higher the likelihood of an upset or results which differ from polls.
As the New York Times suggests, “there’s no reason the polls couldn’t be off by even more than they were four years ago.”