President Trump’s tendency toward misstatement, fudging and outright deception is well-established, and in its most nefarious forms, it tends to focus on untrue accusations against his opponents or those who hold him accountable for his behavior. Over the weekend, for example, he claimed that the New York Times had made up a source for an article — a source who, it was quickly revealed, had spoken to the White House press pool on background.
One of Trump’s most consistent and less important disinformation habits, though, is misrepresenting poll data. During the Republican presidential primary in 2016, he lauded every poll that showed him in the lead. When those polls began to show him trailing Democrat Hillary Clinton and, then, broadly disapproved of by the American public once he took office, his embrace of poll numbers became more selective. By which we mean he selected only those outlier polls that presented him in a favorable light to lift up in his speeches or on Twitter.
On Tuesday morning, though, that habit took on a new challenger: how the Republicans are expected to fare in the 2018 midterm elections.
There’s a lot to parse in this tweet — the investigation by special counsel Robert S. Mueller III is run at all levels by lifelong Republicans, for example — but let’s focus on that bold claim about the polling. Trump is suggesting (although vaguely enough that he can deny it if need be, as is his practice) that Republican performance in midterm polling has prompted Mueller’s team to double down in its efforts to undermine his presidency. Trump’s goal is to continue to cast that investigation as riddled with partisan bias, and he uses that new Republican lead as an exclamation point.
But there is no Republican lead in the polls.
Last week, a survey from Reuters and its polling partner Ipsos did show a Republican lead in what’s called the generic congressional ballot, that “Will you vote for the Democrat or the Republican in your local House race in November?” question. Democrats have led on this question in most polls pretty consistently, so the sudden Republican lead was a surprise.
Those who pay close attention to the polls — and who want to convey what they say accurately — tend to be skeptical of sudden changes in one survey. And that’s what happened with the Reuters-Ipsos poll. On May 13, the Democrats had a 4.2-point lead. On May 18 — five days later — the Republicans had a 5.6-point lead. That’s a shift of 9.8 points over the span of five days in a poll that reports five-day averages of its polls.
Charles Franklin, director of the Marquette Law School Poll, noted on Twitter that there was a possible reason for that sudden change: The Reuters-Ipsos poll suddenly shifted its estimate of who would turn out in November. Pollsters have to use models of who is expected to turn out to properly weight the poll results; if you poll half Democrats and half Republicans for an election in which 80 percent of voters are Democrats, your poll will be way off the mark. The Reuters-Ipsos survey suddenly shifted to include a lot more Republicans in the mix, Franklin found, and the generic ballot shifted to the Republicans.
Then the mix shifted again — and the Democrats regained the lead in the poll.
The Republicans had a lead for four days in one poll, and that lead is now gone.
That there can be significant variations in individual polls is why it’s recommended to look instead at polling averages, aggregations of different polls. RealClearPolitics has a running average that includes Reuters-Ipsos numbers. The generic ballot has been tightening over the past few months in RCP’s average, with a surge to the Republicans in recent days that’s probably driven in part by the Reuters-Ipsos poll. Averaging a variety of polls, though, it’s clearly not true that the Republicans have taken — or are taking — the lead.
Even this average should be taken with a grain of salt. At this point in the 2010 cycle, the Democrats had a lead in the generic ballot after months of a Republican lead. This was not an indicator that the Republicans were going to fare badly in the midterms, though; they ended up winning the vote nationally by a nearly seven-point margin. They didn’t take the lead for good until June 1.
Something similar happened in 2014. At this point, the Democrats had a slight lead. By Election Day, the Republican lead was six points. They took the lead for good on Sept. 7!
Even if the Republicans were taking a lead now, in other words, the past two midterm elections have suggested that it might be worth employing skepticism about the significance of that success.
But the Republicans haven’t taken a lead, as of writing. The president’s point could have been made without claiming that the Republicans were taking the lead; he could as well have said that Mueller’s investigation was biased and “meddling,” out of worry about the narrowing generic-ballot question.
It still would have been an unfounded accusation, but at least it would have been an unfounded accusation that wasn’t layered with an inaccurate representation of the poll numbers.