He’s springboarding off of something WaPo reporter Robert Costa said a few weeks ago:

The most important voter in this movement, uh, when I travel around the country, is the previously disengaged voter. They’re almost a nonpartisan voter, but they’ve given up not just on the political process, but they’ve disengaged from civic society. They don’t really follow politics. If that’s a real coherent voting block, then Trump — regardless of the polls — will have a shot in November — and regardless of all the mistakes — because that’s a huge block. There’s so much of this country that rarely, if ever, votes, and if — for some reason — they come to the polls in droves, that changes everything.

Here’s Rush today, citing Salena Zito’s recent column about driving across Pennsylvania and seeing one Trump yard sign after another:

Call ’em the silent majority, the silent number or whatever. We’re gonna find in November just how many of them there are. We’re gonna find out in November how many of them show up and vote. We’re gonna find out a lot of things in November, ’cause I guarantee you these people are not being polled. They’re not being reached. And in an even greater sense the people responsible for polling and the editors and producers of major media networks. They’re not interested in these people…

So they are out there lurking, and every presidential year comes along and they stay home because it’s more of the same. They don’t have to a political party. The Tea Party, maybe, was a vessel for them. But Trump has come along and has ignited them. Trump has come along and reenergized them, and that’s who they are.

He’s arguing that the polls are skewed, not because the pollster’s putting a thumb on the scale to favor Democrats and depress Republicans but because the best guesses of pollsters as to what the electorate will look like this fall are all wrong. There’s supposedly a massive pro-Trump movement out there that isn’t showing up in the data but might show up on election day, enough so to erase a 10-point Clinton margin in Pennsylvania. If you simply must unskew, I prefer that argument to the more traditional dopey nitpicking about sample sizes and partisan splits, etc. For one thing, for Trump to erase the sort of deficits he’s staring at in some swing states, you need a Big Explanation for why so many different surveys are systemically wrong. Rush’s idea at least attempts that, even if it’s little more than a deus ex machina at the end of a very depressing play. And hey — there’s some truth to the idea that there are millions of voters out there who stayed home in 2012 who might be willing to turn out for a more populist, Perot-like candidate. Not enough of them in the states Trump badly needs to win the election, as it turns out, but they’re out there.

Let’s think this through, though. Why would the “silent majority” not show up in polls, as Rush speculates? Disaffected non-voters all have phones. It’s possible that they might be called by a pollster and not show up in the final data because they’ve told the pollster they’re an unlikely voter, but … why would they tell him that? The point of Zito’s column isn’t that Trump fans are hard to find but rather the opposite, that they’re easy to find because they’re not shy about their support for him. As such, you’d expect they’d be telling pollsters that they will vote, that they’ve become likely voters thanks to the magical charisma of Trump, which means they’ll probably show up in the final data after all. One reason to exclude them anyway would be if they haven’t registered to vote yet but plan to; if so, that’s a problem in itself as we’re getting closer to Election Day and some of them might miss the registration deadline. Another reason their Trump support might not show up in the data is if there’s some sort of “Bradley effect” happening in which people who secretly plan to vote for Trump are embarrassed to confess that to pollsters. (There’s evidence that the Bradley effect, if it ever existed, ceased to exist years ago. Obama’s actual vote share in 2008 almost perfectly matched the final RCP poll average.) To believe that that’s what’s responsible for Trump trailing, you need to believe that in Pennsylvania something like one in 10 people polled are lying outright to pollsters for fear that someone they’ve never talked to on the phone before and never will again might think less of them for stating their honest preference of candidates. Does that seem plausible? As I recall, there was no evidence of a systemic “Bradley effect” in the Republican primaries, with Trump consistently overperforming his polls. If anything, he underperformed in some states like Iowa.

Another problem: Why assume that unlikely voters are uniformly or overwhelmingly pro-Trump? How many disaffected non-voters, whether Bernie fans, disgusted centrists, minorities, or young adults, will come to the polls this year to vote against Trump? His favorable rating is toxic nationally and has been that way for more than a year. Democrats have used his positions on immigration for months as motivation for Latinos to register and make their opposition to Trump known in November. Hillary’s going to have some unlikely voters at the polls too this fall, and every ballot they cast reduces the edge the “silent majority” is supposedly going to give Trump. Plus, never forget that if the election comes down to a battle of which side can do a better job of registering and turning out voters, Democrats would be heavily favored thanks to Hillary’s superior ground game. There may well be many hundreds of thousands of pro-Trump voters out there who don’t pay much attention to elections and would like to vote Republican, but as I said above, some won’t get the chance because the GOP never got around to steering them through the registration process. Meanwhile, Hillary may end up turning out more regular Democrats than Obama did, which would further cut into Trump’s supposed advantage among irregular voters. In fact, despite all the hype in the primaries about Trump bringing in new voters and expanding the party, the data shows that most of his “new voters” were already registered Republicans who voted in general elections but typically skipped the primaries until Trump ran this year. He brought new voters into the primaries, in other words, but there’s no reason to think he’s bringing them into the general. Frankly, he may be a better turnout machine for Democrats, insofar as he’s giving millennials who are iffy on Hillary a reason to vote for her as the perceived lesser evil, than he is for Republicans.

But if all of that’s too complicated, here’s a simple reality: Trump trails by far enough in some national and swing-state polls that realistically no “silent majority” effect can make up the difference. It’s one thing to fret about “skewed polls” or flawed turnout models when some polls have the race dead even, as was the case with Romney in 2012. (Rush’s take on that election just before the big vote, by the way, was “Everything — Except the Polls — Points to a Romney Landslide.”) When he’s getting thumped in poll after poll of Pennsylvania, though (and possibly Florida), and sporting a 35/63 favorable rating? C’mon. The silent majority isn’t going to save him. He needs something to change among the vocal majority, and soon.