For the better part of a week, President Trump’s campaign and its allies have been trying to combat the notion that his poll numbers are in the tank.

They’re still coming up empty.

Faced with several ugly polls, they’ve argued that the polls are skewed, dispatched their own dodgy spin on the new numbers and pointed to a pair of polls that don’t prove their point at all.

First came Trump’s response to a CNN poll that showed him trailing presumptive Democratic presidential nominee Joe Biden by a 14-point margin. That margin is bigger than that of most any other poll, but a memo Trump furnished from his pollster John McLaughlin, who has his own recent record of huge polling misses, wasn’t exactly compelling.

In light of that poll, the head of Trump’s favorite cable network, One America News Network, previewed a coming poll that he said “looks as though it will be in favor of” Trump. Except it wasn’t — not really. That poll from another highly suspect pollster showed a 50-50 dead heat in Florida, which Trump won in 2016. Despite OANN’s best efforts to spin the results in a pro-Trump direction, it soon pulled the poll entirely. (The poll somehow now appears on its website with different numbers than originally reported — a six-point Trump lead.)

On Sunday came the latest iteration. Trump campaign lawyer Jenna Ellis appeared on CNN with media analyst Brian Stelter and again attacked the CNN poll. Again, though, the argument was fallacious.

Ellis claimed the CNN poll was “junk” because it included adults who aren’t registered voters. “No pollster in the world would stand behind this as a legitimate poll, according to industry standards,” she said, Politico reported. Except as Stelter rightly noted, this is not at all uncommon with polls — including Washington Post polls — and the 14-point lead the poll showed was based only upon registered voters, not all adults.

Ellis wasn’t done with Stelter, though. Later in the day, she tweeted a new poll from Arkansas that she suggested was a counterpoint to all the other polls. Except the poll showed a near dead heat in the deep-red state — a state Trump carried by a cool 27 points in 2016. (Indeed, people were questioning this poll Sunday, but that was because it was suspiciously bad for Trump.)

“Hey @brianstelter … still wanna stand by CNN’s junk science?” Ellis tweeted. “None of the polls except yours show such false numbers.”

(It’s worth noting that while CNN’s 14-point margin was larger than those of most other recent polls, others including a Washington Post-ABC News poll have shown Trump down double digits, so it’s hardly without compare.)

As I noted at the time, it seemed Ellis had perhaps mistaken the Arkansas poll for a national one — in which case a near dead heat would indeed have been good for Trump. But she soon came back with a different defense: That she had tweeted the wrong poll.

This time she tweeted the same message on a poll written up by the Washington Examiner, with a correction: “NOTE: Earlier I tweeted a different poll than what I cited today on CNN. Updated. Because I actually correct things, unlike CNN. ;).”

Except the poll she tweeted wasn’t even a presidential preference poll. Instead, it was a poll of who people thought would win the presidential race. The poll from Zogby Analytics showed people thought Trump would win by a 51-43 margin.

If you want a great example of how little that means, just look at what happened in 2016. As The Post’s Dave Weigel notes, the final CNN poll that year, in fact, showed two-thirds of people thought Hillary Clinton would win.

Even those numbers, though, appear to reflect Trump’s worsening position. For months, Democrats have been rather skeptical that their party would win the presidency, probably owing to what happened in 2016 when they thought they would win and lost. CNN’s Harry Enten calculated last month that, on average, 55 percent of people in polls thought Trump would win, despite Biden’s consistent leads in national and swing-state polls. In other words, the poll Ellis tweeted not only didn’t tell us anything about whom people were actually voting for; it also suggested Trump might have actually regressed slightly on this question, too.

The Examiner piece did briefly mention a previous Zogby poll that had showed a dead heat between Trump and Biden on candidate preference, but that poll is out of step with the vast majority of reputable polls. Rather than link to that poll — or a piece on it which had also been written up by the same Examiner author — Ellis linked to this article.

Alas, it was apparently the best argument she could find that Trump isn’t actually down in the polls. The fact that these are the kinds of arguments that Trump and his top allies are falling back upon — rather than citing reputable pollsters and valid, logical arguments — doesn’t exactly speak to his political strength.