What the polls said, what the bookies said, and what happened

A narrow Biden victory is as good for the bookies as it’s bad for polling firms, says James Moore

Philadelphia protesters demand that every vote be counted (Rex)

When it comes to this year’s elections in the US, the polling industry served up what’s best described as a double fault. After predicting a close win for Hillary Clinton in 2016 when Donald Trump took the White House, they suggested Joe Biden would comfortably dethrone the incumbent in 2020.

Poll trackers consistently put Biden’s nationwide lead at about eight points, the biggest a candidate for the Oval Office has enjoyed since 1996 when Bill Clinton cruised to a second term. The final batch of pre-election surveys showed the Democratic Party leading the race for the presidency by between seven and 12 points, albeit with two notable exceptions.

An Investors Business Daily/TIPP poll had Biden just 2.8 points ahead, which now looks prescient. Rasmussen, meanwhile, put the lead at just a single percentage point.

IBD/TIPP boasts an A/B grade from the statistical gurus at Nate Silver’s FiveThirtyEight but Rasmussen, a favourite of the president’s (no doubt because it has consistently favoured his party) has a fairly miserable C+.

The polling data led FiveThirtyEight to give Biden an 88 per cent chance of victory in its final forecast. It should be noted that this means it still had 10 scenarios in which Trump won the Electoral College and one with a tie of 270 votes each. But it nonetheless underlines the scale of the pollsters’ miss.

The betting markets, by contrast, were much more sceptical. Paddy Power had Trump at 15/8 and Biden at 2/5 at 11pm (UK time) on election night. Biden’s implied probability of victory based on that was 71 per cent. Better than a toss up, but notably closer.

They suggested a far tighter race than the polls did from the off. Partly that’s because Trump has long been a favourite of political punters. The same firm said he accounted for 90 per cent of the bets taken by number and most of the money too. Bookies’ finance departments are probably decked out in Democrat blue and their employees could be forgiven for sporting Biden campaign apparel. A narrow Biden victory is as good for the bookies as it’s bad for polling firms.

The last US election put the latter industry into an existential crisis, with pollsters scrambling to update their methodologies. Get set for another mix of anguish and excuses. When it comes to the excuses, it should be noted that polls have always attracted a degree of scepticism (especially from those on the wrong side of them) and they always have to be taken with a pinch of salt.

They provide a snapshot at a given moment in time. Even the best, most conscientious firms are apt to produce outliers. Their findings are influenced by statistical eccentricities, sampling errors, the methodology they opt for, the way their questions are phrased.

Another factor worth remembering: sometimes people lie. Some people refuse to answer pollsters’ questions when contacted by phone, internet or face-to-face (all of which can influence poll results).

But even with all that, the scale of the miss this time around – Biden was only about 2.7 points ahead in terms of the popular vote at the time of writing – is jarring, especially after the events of 2016. Not to put too fine a point on it but the polling industry being so far out of touch is just bad for business. It doesn’t do you any favours with customers if your most high-profile product is viewed as the purveyor of a sophisticated (and expensive) fairytale.

In 2016, when the president lost the popular vote, Trump collected just under 63m ballots. This year he’s racked up more than 68.5m

Was the industry skewered by “shy Trumpers” – people unwilling to declare their support for a deeply divisive president when put on the spot?

That’s an interesting question. There was some scepticism expressed about the idea, talked up by Trump in the run up to the election, not least because “shy” isn’t a word one would generally use about the Maga hat-wearing hordes.

But it’s possible that the mixture of aggression and defiance shown by the people who attended Trump rallies, sported his bumper stickers and wore those baseball caps, might have helped create another group, embarrassed by their antics and unwilling to admit they too were in Trump’s camp as a result. 

How to find out whether that’s true… Um, more polling?

When it comes to the actual results, they’ve seen anguished Americans asking “is this who we really are” (a question often asked by anguished Britons too) because they’ve firmly squashed the notion that Trump’s presidency has somehow been an “aberration”, ushered in because Hillary Clinton was a uniquely bad candidate for the Democrats.

In 2016, when the president lost the popular vote, Trump collected just under 63m ballots. This year he’s racked up more than 69.6m (and counting) compared with Joe Biden’s record-setting 73.7m plus.

The increase was fuelled by a record turnout, particularly in some of the decisive midwest battleground states. But it’s still striking. The last four years have seen the norms of US politics and government trashed, a presidential impeachment, the publication of a number of lurid tell-alls penned by insiders, the type of scandal that might even have moved the late Richard Nixon (he of Watergate fame) to say “steady on”.

And yet millions more Americans still reciprocated Trump’s beloved thumbs up gesture by backing him in the polling booth than did the same thing four years ago. The aftermath of the election has seen commentators pointing to the absence of the “blue wave” the Democrats had anticipated. I’ve seen it described as “a ripple at best”.

But take a look at the results. They say something different. The vote count tells you the wave was clearly there on polling day. It was a veritable tsunami. It was just matched by a Republican red wave of nearly equal force and intensity.

With the GOP seeming set to retain control of the US senate, a Joe Biden presidency (now all but nailed on) is going to find getting a cabinet confirmed tough, let alone enacting his agenda.

If the Democrats are going to have to spend more time, energy, and money in those states for the foreseeable future, the Republicans have work to do in some of their former citadels too

The markets were quite pleased. The Dow Jones Industrial Average finished Wednesday up by an impressive 367.63 points or 1.34 per cent, the technology-laden Nasdaq did even better, posting a rise of 430.21, or 3.85 per cent.

Pay attention to that: Silicon Valley has been taking heat from both sides of the US political divide. But divided government inevitably means reforms are going to be much harder to enact. Big tech looks set to be a bigger winner than even the betting industry. It may offer up a banana or two to placate critics, monitor hate speech and misinformation a little more closely, but that’s about it. Otherwise normal service looks set to be resumed.

Just like in 2016, the midwest region played a pivotal role in the outcome. Assuming the results hold, Joe Biden took Wisconsin by a wafer thin margin and Michigan by a relatively slim 2.6 points, just as they were only narrowly won by Trump in 2016. Pennsylvania hadn’t been called at the time of writing but was similarly close (with Biden in the lead). These three states were once considered reliable parts of the Democrats’ rock-solid blue wall, not quite on a level with some of Britain’s safe parliamentary seats, nowhere near as deep blue as, say, a California but solid all the same.

The last Republican favoured by Wisconsin before it fell to Donald Trump in 2016 was Ronald Reagan in the 1984 Republican landslide when he carried every state with the exception of Minnesota (the District of Columbia also backed his opponent Walter Mondale). Michigan was among the 40 states that backed Reagan’s successor and former vice-president George HW Bush in 1988, but after that it was held by the Democrats until 2016. Ditto Pennsylvania.

State-wide polls in Wisconsin and Michigan had suggested they were poised to revert to type with high quality pollsters finding Biden leads of 7 points or greater in Wisconsin. It looked better still for the Democrats in Michigan. Surveys suggested a tighter race in Pennsylvania, but not as tight as it has proved. Rust-belt residents who dislike politics had best hide. Those states are going to get a lot of attention from here on out.

As for Ohio? This big state, with an enticing 18 electoral college votes, used to be a genuine swing state, backing the winning candidate in every race since 1962. Today it looks like an increasingly solid red state. Trump carried it by eight points. Its most important elected officials are all Republicans, as is the state legislature and a big majority of the congressmen and women it sends to the House of Representatives in Washington. Senator Sherrod Brown, who was re-elected in 2018 and doesn’t face voters again until 2024, is a noteworthy exception.

If the Democrats are going to have to spend more time, energy, and money in those states for the foreseeable future, the Republicans have work to do in some of their former citadels too. Polling suggested tight races favouring the Democrats in North Carolina in the south and Arizona in the southwest. Georgia was predicted to be a Republican-leaning toss up. It’s recounting but Biden is favoured. Regardless, demographic changes have turned these states purple.

A word on those demographics. Much of the data analysing who voted which way so far comes from exit polling (so, yes, it’s polling again). But this election really ought to see the end of treating “Hispanic voters” as a single monolithic bloc as some in the media still do. While it’s true that they are an important part of the Democrats’ coalition, they are also a highly diverse group. The term “Hispanic” in Florida, for example, encompasses a sizeable community of Cuban Americans, who were cock-a-hoop at their state staying red to an even greater degree than in 2016 as many non-Cuban Latinos voted Republican.

Florida is another place in which the Democrats have work to do if they’re to repeat the winning performances of Barack Obama and Bill Clinton any time soon. A lot of work.