President Trump’s fury was palpable last week after CNN released a poll that showed him trailing in his reelection bid by 14 points. He lashed out on Twitter and took the unusual step of threatening a lawsuit over the result, a move that might be called dubious in the same way that LeBron James might be called athletic. It is — but that’s underselling the point a bit.

Happily for the president, there exists an ecosystem of questionable actors willing to assuage whatever concerns might emerge. His repeated excoriations against “fake news” have made clear to his supporters that any reporting that may be in any way construed as negative should be considered untrue. That puts pressure on outlets such as Fox News that, although certainly more generous to Trump than other outlets, still at times include voices other than the most sycophantic. That’s a frustration to the president — and an opportunity by One America News, a tiny cable-network upstart that has earned praise from Trump for being unfailingly positive about his presidency.

Last week, we reported that OAN’s core mission of supporting Trump had been hampered by some bad news: a poll of Florida that the company’s chief executive pledged would offer good news for the president instead showed him in a tie with former vice president Joe Biden, the presumptive Democratic nominee, in the state. This state, remember, is one Trump won four years ago and is also his newly adopted home.

This was not positive news for Trump, despite the network’s initial effort to cast it as such. And because bad news is fake news, it got deep-sixed, the network’s report erased from its website and the video removed from YouTube.

That was not the end of the story, however. While reporting on another story, The Post’s Aaron Blake noticed that the network had created a new report based on the same poll conducted by Gravis Marketing — one that framed the same numbers in a significantly different way.

“In a hypothetical matchup,” host Patrick Hussion reported, “when [voters] in Florida were asked, ‘If you had to choose a candidate right now, for whom would you vote?’ the OAN-Gravis poll shows the president with a 53 to 47 percent lead over former V.P. Joe Biden there in the Sunshine State.”

That was accompanied by this graphic.

Trump won Florida by a little more than one point in 2016, so this would reflect a significant improvement.

But Hussion left out a very important bit of context for those numbers: The people being forced to choose a candidate were people who already had indicated that they were undecided in the race. A screen shot from the prior report shows the actual question, which applies only to “uncertain responses from Q2″ — meaning Question 2.

The report that was removed from OAN’s website last week made clear (unintentionally) that there were two questions in a row. The first, Question 2, asked whom respondents would support. The second, Question 3, forced those who were undecided to pick a candidate. When those two results were combined, Biden and Trump were tied at 50 percent, as the prior report showed.

Normally, pollsters report the results of questions like Question 2: For whom would you vote if the election were today? That’s why most polls have results in which support for the two candidates doesn’t add up to 100 percent; they include a “not sure/not decided” option.

That’s important because it’s necessarily the case that the results for Question 2 in this poll were more favorable to Biden than the results for Question 3 — that is, than the results that OAN reported. If Trump has a six-point lead among those undecided voters forced to pick a candidate, then Biden must have an advantage among the other group in order for the combined result to be a tie. How big Biden’s advantage is depends on how many people indicated that they were undecided. The bigger the pool of undecideds, the smaller Biden’s advantage over Trump in the poll — but the less representative the results OAN shared actually were.

It’s hard to overstate how dishonest this was. Imagine if you asked 20 people what kind of ice cream they wanted. Nine pick vanilla and seven pick chocolate. Four people have no opinion — maybe simply because they like both — but you make them pick. Three say they would take chocolate if they had to choose one or the other, while the fourth picks vanilla.

The original report claimed that vanilla and chocolate were equally popular. The new OAN headline is essentially that people forced to pick an ice cream flavor pick chocolate by a 3-to-1 margin.

If you’re curious whether this was perhaps a new poll, it wasn’t. The numbers reported for Trump’s approval were identical in both reports.

Here’s the new version of the report.

And here’s the original.

It’s the exact same graphic, with that “2%” indicator floating up over the red bar. Even without that detail, though, the fact that the numbers are precisely the same indicates that this is the same poll. The only time approval numbers remain unchanging time and time again is when they’re invented ones tweeted out by Trump.

OAN’s deceptive presentation of the poll doesn’t end with cherry-picking a result that casts the president in the most favorable light. Hussion also reported on how Floridians view the candidates’ ability to handle the economy.

“Now, when it comes to the economy, the poll also shows Florida voters are decisively more confident in President Trump’s economic policies than the direction Biden would go in if he were elected,” Hussion said. “The president pulls in nearly 50 percent of the respondents compared to Biden’s 44 percent.”

This is ... a five-point difference. This is not only not a demonstration of “decisively more confidence,” it’s almost certainly not even statistically significant. It’s also extremely bad news for Trump if the factor on which he’s predicating his reelection — his ability to keep the economy strong — has him under 50 percent against Biden in a state he needs to win.

This is fake news, folks. It’s invented, cherry-picked and deceptively framed information meant not to convey actual information but, instead, to placate the president and his supporters. It is to news what your mom’s evaluation of your performance in your Little League game was — reassuring but hardly objective. And not only is this report misleading and unethical, it makes very clear why OAN removed the initial report last week: It wasn’t positive enough for Trump. This one is, even if it’s dishonest, so it’s still online.

Yet somehow I am skeptical that Trump will reject OAN as an untrustworthy news source.