Donald Trump’s self-worth has long been defined by numbers. His net worth (which he would often inflate). The ratings for “The Apprentice.” And once he entered politics, polling. He would watching polling closely, loudly celebrating his leads in the 2016 Republican primary and angrily deriding his deficits in most polls thereafter. He’d make polls up, like his incessantly exaggerated (and ever increasing) claims about his grip on Republicans. He’d hype outlier polls that showed him doing well on Twitter, claiming that they came from greatly respected polling firms; if those polls turned against him, the president turned against them just as quickly.
In June 2020, Trump offered one of his more unusual responses to a bad poll. CNN and its polling partner SSRS found Trump trailing Joe Biden in the presidential election by 14 points, a wider margin than any other poll in the same period. At the time, Biden’s lead in the FiveThirtyEight polling average was about half that size. It seemed likely to be an outlier (as I wrote at the time), but was worth keeping an eye on as a possible preview of a bad period for the incumbent president.
But then Trump tweeted out a poll from his preferred pollster, John McLaughlin. In it, McLaughlin disparaged not the poll’s methodology but the pollsters’ integrity. He criticized not the pattern of polls but the way in which the poll was conducted. He criticized CNN and SSRS as making the same mistake that pollsters had made in 2016 when state polls underestimated Trump’s support. It was a blistering, personal, aggressive memo seemingly aimed less at assuaging the concerns of Republicans than at assuaging the feelings of Trump.
Within a few weeks, Biden’s lead over Trump in national polling in fact widened, though not to 14 points. In the end, polling did again underestimate Trump’s support, though it accurately estimated Biden’s.
So what happened? The CNN poll does appear to have been an outlier, though, with a margin of sampling error of 3.4 points, its estimate of Trump’s support was not statistically different than the polling average. McLaughlin’s complaints about intent and Trump’s odd suggestion that pollsters were trying to suppress turnout in an election that wouldn’t see votes cast for another three months were just shouting. But, you know. Very Trump.
So it was unexpected when, on Thursday, the Biden White House deployed a similar tactic, though with a much more Biden-y approach.
The prompt was a poll from Quinnipiac University released on Wednesday. In it, approval for President Biden’s performance was at a grim 33 percent, nearly 10 points lower than the FiveThirtyEight average. The result no doubt triggered alarm bells in campaign offices around the country, but, again, it’s not clear if the survey is an outlier or the beginning of a dire new trend.
For the Biden administration, there’s no space to have people think it might be the latter. While Trump’s June 2020 complaints were clearly spurred by personal defensiveness, the Biden team is in a very fragile moment from a policy standpoint. It needs every vote it can get on its twin priorities of social spending and bolstering federal voting laws. If Democratic legislators think that only a third of Americans approve of how Biden’s doing, there will be a broad instinct to instead side with the majority that doesn’t.
So, a memo from deputy chief of staff Jen O’Malley Dillon, as first obtained by Axios. Dillon eschews the vitriol that McLaughlin used (quite possibly with Trump standing over his shoulder) instead making the case that Quinnipiac’s evaluation of Biden’s approval has consistently trailed the FiveThirtyEight average. That’s true, but a difference of four points last September with a poll with a 2.8-point margin of sampling error is quite different than the 10-point spread of this week’s offering.
The memo also quotes an article written by CNN’s Ariel Edwards-Levy last year, noting how methodologies in how pollsters ask about approval affect responses.
One explanation for a wide divergence in approval, Edwards-Levy suggested, “is that these polls aren’t telling contradictory stories so much as one, somewhat more nuanced one, along these lines: About one-eighth of the public doesn’t hold particularly strong views about Biden, but if prompted, they’re probably lukewarmly positive on balance.”
In response to the Quinnipiac poll, Edwards-Levy pointed to that same idea.
Among Democrats, 11 percent of respondents to Quinnipiac’s poll had no opinion. Among independents, 19 percent did.
One aspect of the poll not highlighted by Dillon was instead seized upon by conservative media: It found that only 28 percent of Hispanics viewed Biden’s job performance with approval. Within that group, though, more than 1 in 5 had no opinion. Quinnipiac also didn’t include Spanish-language contacts in its outreach, excluding a large group of Hispanic respondents.
There’s another hint that this poll result is particularly noteworthy: the White House decided it worthy to take note of it. Again, this is almost certainly primarily a function of keeping its narrow margins in the House and Senate from sprinting away from its policy priorities. But it is also probably to some extent wishcasting, convincing themselves that the situation isn’t that bad.
Even without the Quinnipiac result, though, things look bleak for Democrats. Last November, I plotted the changes in House seats in midterm elections relative to presidential approval ratings. In Gallup polling, my baseline, Biden then was about where FiveThirtyEight has him now.
What does that suggest about the House results? Something somewhere between a wave and obliteration.
The thing about Trump’s memo in June 2020 was that it took a valid point — polls had underestimated Trump’s support as in fact they did — and blanketed it with animus and dubious rationalization.
The Dillon memo leaves out the animus.