The Polls Weren't Far off in 2016. They Aren't Wrong Now, Either.

Over the last few weeks, my political conversations have started to change. They now often lead off with a question like “Can President Trump turn this election around?” or “What would it take for him to win a second term?” That’s quite a shift from just a few months ago, when Joe Biden’s lead over Trump was in the 4-to-7-point range and the incumbent’s approval ratings were in the mid-40s. Biden’s advantage today is between 8 and 14 points, while Trump’s job-approval ratings are now hovering around 40 percent, with his personal unfavorables far higher than his challenger’s.

The RealClearPolitics average of national polls pegs Biden’s lead at 8.7 points (49.6 to 40.9 percent), the poll-driven estimate at Nate Silver’s FiveThirtyEight is virtually the same at 8.8 points (51.1 to 41.5 percent), and The Economist’s intriguing model, unveiled last month, puts Biden up by 8.3 points (54 to 45.7 percent). The Economist model incorporates both national and state-level polls as well as economic data to produce estimates of the national popular vote, each candidate’s chances of winning individual states, and the probabilities each has of winning the Electoral College, based on 20,000 computer-simulated scenarios, updated daily as new data becomes available. The model, developed by Economist data journalist G. Elliott Morris in conjunction with Columbia University political scientists Andrew Gelman and Merlin Heidemanns, currently shows Biden ahead in states totaling 346 electoral votes, to 192 for Trump.

This is the point at which I often hear that the 2016 polls were all wrong, or that people lie to pollsters, or that Trump voters simply don’t talk to pollsters. These oft-repeated arguments ignore the fact that the national polls missed Hillary Clinton’s popular-vote margin by just one percentage point. The RealClearPolitics average of polls on the morning of the election showed her 3.2 points ahead of Trump as the race continued to narrow. When the votes were tallied, Clinton’s national popular-vote edge was 2.1 percentage points (equal to about 2.9 million votes). There couldn’t have been too many Trump voters who lied or wouldn’t respond to pollsters for them to be that close.

Still others tell me—not without reason—that given our Electoral College system, the national popular vote and five bucks will get you a cup of coffee at Starbucks. Indeed, given current state voting patterns, Democratic voters for president are spread much less efficiently around the country than Republican voters. Democrats run up big vote margins in California, New York, and Illinois, while Republicans win their most populous states with lower margins, “wasting” fewer votes. Thus, Biden will need to win the national popular vote with room to spare to hit 270 electoral votes—obviously 2.1 percentage points wasn’t enough for Clinton in 2016. Presumably a Biden margin above 4 points would more than cover a situation like last time, when Trump’s paper-thin margin of two-tenths of a percentage point in Michigan and seven-tenths of a point in Pennsylvania and Wisconsin (fewer than 78,000 votes total out of 137 million cast nationwide) tipped the balance and the outcome.

As for the state polls in 2016, 45 states and the District of Columbia went exactly where they were expected to go. Florida and North Carolina were considered toss-ups, along with Maine's and Nebraska’s 2nd Congressional Districts. Michigan, Pennsylvania, and Wisconsin were the only real surprises, as polls showed Clinton ahead in each. Put simply, the polls weren’t nearly as far off as many people continue to think they were.

Rewinding four years: During the first half of July 2016, Clinton’s lead over Trump in the RCP average ranged from 2.7 to 4.8 points. She led by as many 3.2 points in the third week, but in the last week of the month the race pulled even and Trump briefly pulled ahead by as many as 1.1 percentage points.

Also key is how undecided voters break at the end. Exit polls showed that 85 percent of voters four years ago made up their minds before the last week of the campaign, with Clinton winning that group by 2 points, 49 to 47 percent. But among the 13 percent who decided in the last week, Trump won by 5 points, 47 to 42 percent. As political scientist, demographer, and Democratic strategist Ruy Teixeira points out, it isn’t just the size of Biden’s margin over Trump in the polls that is impressive; it’s the absolute number, hovering at or above 50 percent. Should late-deciding voters break against him, he’s got some cushion.

So what could turn things around for Trump? Normally a foreign or domestic crisis might be the tonic an unpopular president needs to turn his fortunes around, but crises have not been Trump’s forte. Indeed, history may well show that, even while Trump never reached a 50 percent job-approval rating, the race was still quite competitive until a pair of crises came along: the coronavirus pandemic and the civil unrest following the killing of George Floyd by a Minneapolis police officer. Those factors, and the related economic downturn, caused a race in which he trailed by between 4 and 6 points to balloon out to a margin at least double that.

In an analysis of last month’s CNBC All America Economic Survey—conducted jointly by Hart Research and Public Opinion Strategies, the same bipartisan team of pollsters that conduct the NBC News/Wall Street Journal poll—Hart partner Jay Campbell points to several factors that changed the trajectory of this race so dramatically over the last three months. According to Campbell, “Biden’s margin is driven almost entirely by voters under age 35. Biden now leads by 33 points among this young cohort, compared to a 2-point lead among voters 35-64, and a tie among voters age 65+.” Campbell noted that while many have commented on Trump’s diminished numbers among seniors compared with 2016, “if there is a large Biden victory in November it will be because of young people.”

Campbell also spotlighted independents, noting that “many of them have shifted to undecided over the past few months, with all of the movement coming out of Trump’s column. He was up 6 points with them in April, but is down by 5 now,” with his approval ratings among independents at 29 percent. Just as with voters overall, Trump gets his best ratings from independents (46 percent) on his handling of the economy. The poll showed that “the economy and jobs” was the only issue on which Trump has a significant edge over Biden. That’s one of the ironies of this race: Despite the fact that the economy is in a horrible recession, it’s the best, perhaps even the only, card that Trump has to play

This story was originally published on nationaljournal.com on July 7, 2020