While current polling appears to suggest that Joe Biden would beat Donald Trump if the election was held today, there is widespread sentiment that these polls could be wrong. Trump has tweeted recently that, “Final RCP Polling Averages Had Hillary Clinton Winning MI, WI, and PA [...] The Polls are Fake just like much of the reported news. I won it all against Crooked Hillary!” Meanwhile, Biden’s campaign manager recently wrote a memo saying that in certain swing states, the “race is far closer than some of the punditry we’re seeing on Twitter and on TV would suggest.” So, with the election only weeks away, I thought it would be good to compare Biden’s polling lead with 2016 polling and the 2016 election results.
Nationally, FiveThirtyEight’s polling aggregate currently has Biden up 52.2% to 41.9% (+10.3), while RealClearPolitics’s aggregate has him leading 51.1% to 42.5% (+8.6). FiveThirtyEight gives Biden an 87% chance of winning the election (data accurate as of 23:59 CDT 10/20). Meanwhile, on election day in 2016, 538 had 45.7% Clinton to 41.8% (+3.9) Trump, and gave her a 71.4% chance of winning. Other models were much more bullish on Clinton---the New York Times gave her an 85% chance of winning, and Huffington Post gave her a whopping 98% chance of victory.
Now, how were HuffPo and NYT so wrong last time? Well, one major reason is that most statistical models underrated the extent to which state-level polling errors would be likely to be similar in different states. In order for Clinton to lose the election, she had to lose multiple states where she was projected to win. So, if these events were entirely independent of each other, it would have been statistically highly unlikely for Clinton to lose most battleground states. However, state outcomes are highly correlated with each other empirically, especially when those states are demographically similar. As FiveThirtyEight wrote in October 2016: “In 2012, Obama beat his polling by 2 or 3 percentage points in almost every swing state.” Because of this, Trump was just one normal polling error away from victory on election day.
Differences between the Biden and Clinton leads
The most obvious distinction between Biden and Clinton’s respective leads is the sheer magnitude of Biden’s lead over Trump over time, which is best shown in graph form. Note: the y-axes on these graphs are different. Graphs from RealClearPolitics.
While RCP had Trump down by 11 points near the end of March 2016, the race significantly tightened in the final months of the campaign, with Trump trailing by just 1.3 points on November 2nd in a race where Clinton ultimately won the popular vote by 2.1 points. Meanwhile, in recent months this year, Trump has consistently trailed by no less than 6 points, though there is still time for polls to tighten.
A second crucial distinction can be seen in the relative popularity of the two Democratic candidates. In 2016, both candidates were unpopular with 18% of voters having a negative opinion of both Clinton and Trump according to exit polling. This meant that many voters did not decide whom they wanted to vote for until the end of the campaign. On 10/24/16, FiveThirtyEight wrote, “In national polls, Clinton and Trump together have approximately 85 percent of the vote, while Mitt Romney and Barack Obama had about 95 percent of the vote at this time four years ago.” So, negative news cycles for Clinton late in the campaign, like the Comey letter cycle, likely swung a critical amount of support in Trump’s favor. Meanwhile, this year, Biden’s net favorability rating is at +7.4 while Trump’s is at -11.3. Additionally, using the RCP numbers, Biden and Trump already command 93.6% of the vote in the polls, a much higher percentage than this time in 2016, likely meaning fewer late-breaking voters.
Thirdly, the demographics of Clinton and Biden’s support are different. Biden is faring much better among senior citizens. In 2016, Trump won senior citizens by 7 points, which was 9 points better than his overall performance. By contrast, a recent NYT/Siena College poll had Biden up 10 points with seniors compared to 9 points overall. It is unclear just how well Biden is doing with seniors---two polls conducted after the first presidential debate had Biden up 21 and 27 points, when he was leading those polls by 16 and 14 points overall. Meanwhile, a recent YouGov poll found that Biden was losing seniors by 3 points while winning overall by 9 points, though this is perhaps an outlier. In any case, Biden is doing much better than Clinton, and this influx of senior support seems to, at least partially, predate coronavirus, as a January poll from CNN said that Biden was up 9 points overall and also 9 points with seniors.
Concurrently, Democratic support among Hispanic voters has fallen over the past four years. While 2016 exit polls say that Clinton won Latino voters 66% to 28%, Biden is only winning Latino Americans by a tally of 50% to 30% according to NYT/Siena, despite having a significantly bigger lead overall compared Clinton’s popular vote win margin. (Note: Hispanic and Latino are not exact synonyms.) Contemporaneously, Morning Consult gives Biden a 62%-34% lead with people who are Hispanic. Crucially, this could lead to a Trump victory in Florida, where he currently trails by under four points. Latino voters comprise 17% of Florida’s electorate this year. This swing towards Trump seems to be a part of a larger trend, as Trump probably did better in 2016 than Mitt Romney did in 2012 with Latino voters.
Welcome to Political Union's blog! All opinions expressed are those of our writers, and not NU Political Union.