Biden’s poll wobble is mostly froth – Trump is in deep trouble

Today is another hot day in the UK but the summer has been pretty miserly when it comes to sunshine. It’s a timely reminder that sampling data too discretely can indeed lead to conclusions that run counter to the evidence of the real world.

That should certainly be borne in mind when reading headlines today that Joe Biden’s odds have lengthened slightly in the US election. CNN has analysed the Presidential Approval rating and concluded that “Trump’s position has stabilised and perhaps improved a few points”.

That’s probably true, but we’re also well within margins of error, themselves rendered fairly meaningless because we’re averaging across multiple polls. According to FiveThirtyEight, Biden now has a 7.6% lead across national polls. Biden is on 50% of the vote, Trump 42.5%. Not that national polls are particularly helpful (battleground states are still the numbers to watch), but it does tell us something that we already knew about the Trump bedrock. It would need tectonic levels of movement even to penetrate below 40%.

Look at the long-term graphs, however, and the reality of Trump’s position is more evident. This wobble around his vote share is very characteristic of this presidency, where one man largely controls the news cycle and uses it to speak to loyalists even as he riles and motivates his opponents. He experienced a terrible June and early July as deaths to COVID-19 rose. August has seen a slight decline in cases, and Trump’s numbers have seen a very slight improvement. It is not, however, anything extraordinary – certainly nothing we haven’t seen before, and nothing to suggest that he’s winning over moderate and independent voters currently trending towards Biden.

The other thing the latest numbers underscore is how no two polls are the same. If Biden looks to have dipped especially in recent days it probably comes down to timing, with two polls coming in on August 4 that pegged Biden at a surprisingly low 3% lead. Those two polls were conducted by Harris and Rasmussen. Again, it’s worth bearing in mind that when it comes to polls, we’re often comparing apple to oranges. Trump particularly loves Rasmussen, calling it (with some justification) “one of the most accurate in predicting the 2016 Election”. Rasmussen indeed called the previous election well, but the number crunchers at FiveThirtyEight consider them mid-table, less reliable than Ipsos and YouGov.

To highlight the predictive fallibility of it all Rasmussen this week had Trump’s approval rating at 46%, with 52% of the country disapproving. Looking better for Trump? Well, their last poll, for July, had him at 48% and 50%, respectively. That’s a four-point decline in a single month. Reuters/IPSOS’s most recent poll suggests the difference is now 19 points, with Trump’s approval sitting at 40% (disapproval at 59%).

Confused?

It’s natural if you are. Polling tells only one half of the story and the predictive quality of polls is often overstated. It’s more important to consider them in the context of the political narrative – such as the fact that they completely overlooked the weak enthusiasm for Clinton in 2016 which was widely discussed but barely factored in.

Forget “Thighland”, “Yo, Semite”, and a terrible Axios interview that probably mattered more to political junkies than it did to the American public – Trump has not had a bad couple of weeks by his standards. He’s doing his daily press conferences again, reassuring his base that he’s clamping down on the “violent left” in Portland, while Democrats in Congress didn’t lay a glove on Attorney General Barr. Daily Covid cases are now declining and Trump is promising a vaccine by election day. A mild uptick in his grassroots vote makes sense.

What doesn’t make sense is any narrative that speaks of radical changes to our expectations. Unless we see significant shifts – a big Trump success or a sizable Biden gaffe (not improbable) – these variations are no more than background flutter. What remains critical are the key battleground states: Arizona, Florida, Michigan, North Carolina, Pennsylvania, and Wisconsin. As it stands, Biden holds moderate leads in all six. Here are the latest polls from the swing states:

Arizona: Biden +3.7 (down from around 5 in late June)

Florida: Biden +4.0 (down from +8.4)

Michigan: Biden +6.3 (down from a high of 8.2 in July)

North Carolina: Biden +4.6 (holding steady)

Pennsylvania: Biden +4.7 (down from a high of 8.5 in late July)

Wisconsin: Biden +5.0 (down from a high of 6.5 in July)

Again, it’s worth emphasising trends. “Down from June/July” means that Trump has strengthened a little from a particularly low point in the early summer as coronavirus spread across the US. The trend over time is much more telling, indicative of people’s long-term opinions of the President. In Arizona, for example, only between February and March, the period when Trump boasted of his victory over Covid-19, did Trump enjoy a lead (less than one point) in the polls. Since then, Biden’s lead has been as high 5.3 and as low as 2.8. Realistically, Biden is probably trending back towards the two-point lead he had before Covid struck.

Pennsylvania’s dip might also look alarming for Biden, a native of the state, but in 2020 Biden’s lead has varied from a low point of 2.3% in April – a notable feature across all polls where Trump seems to have enjoyed a surge of support – and a high of 8.5% in July.

What does it all mean?

It means what we already know: that events are reflected in polls, but trends are more indicative of people’s opinions of the President. It also means that the election hasn’t even started. Neither candidate has yet officially been nominated by their parties. The campaign has yet to kick into gear, so we haven’t begun to see the toing-and-froing that comes from the usual election mechanics.

Biden remains an uninspiring choice against a hugely unpopular President, and one which a big portion of the country is determined to kick out of office. A bit of a wobble in the numbers shouldn’t fundamentally alter our sense of that reality, which remains good for Biden – and awful for Trump.