Is Donald Trump inevitable?

Republican presidential candidate Donald Trump spoke during a campaign rally at the Macon Centreplex in Macon, Ga., on Monday.Branden Camp/AP

Four months have now passed since Donald Trump first rocketed to the top of the Republican primary polls, and he still hasn’t come down — despite all the dismissive media coverage, a challenge from fellow outsider-candidate Ben Carson, and a stunning array of gaffes and false claims.

There’s still time for his high-flying wings to melt, especially with Republican Party insiders talking about an anti-Trump “guerilla campaign.” But Trump’s surprisingly durable success is starting to flip expectations. While only months ago many wondered if Trump could really win, now the question many are asking is “How could he not?”

What is the case for Trump’s inevitability?

It starts with his unassailable position atop the polls.

Since July, Trump has led nearly every national poll by an average of 11 points. That lead is about to get even stronger now that Ben Carson’s short-lived challenge is souring. Several recent polls show Trump with a 20-point advantage over the scrum of second-tier rivals, currently headed by Ted Cruz and Marco Rubio.

There are times when top-line poll numbers like these can actually hide signs of weakness, like if Trump’s supporters were all located in one part of the country or one demographic group. But that doesn’t seem to be the case with Trump, who’s popular with a broad array of Republican voters, including men and women, the young and old, the lower income and middle class. What’s more, his national appeal translates seamlessly to the state level, giving him a slim advantage over Cruz in Iowa and wider advantages over opponents in the other key early states like New Hampshire and South Carolina.

To top it off, Republican voters see Trump as the most electable candidate, and not as some kind of long-shot outsider. In a recent Washington Post-ABC poll, 38 percent of Republican-leaners said Trump had the best shot at being elected president; no one else came close.

Have any candidates in Trump’s position ever lost?

Presidential primaries don’t happen all that often, so there aren’t a lot of examples to choose from. But looking back over recent cases, it’s hard to find candidates who have lost after amassing a large, longstanding lead the way Trump has in the months just before primary season.

Howard Dean may be the closest example, given that he spent the last quarter of 2003 firmly atop the polls, before collapsing in the early primaries. Even then, however, Dean never looked as strong as Trump. He wasn’t on top for nearly as long, and his average advantage during that period was closer to 6 points, not 10 or 20.

The great comeback stories in modern presidential history usually don’t involve the toppling of a formidable and long-dominant front-runner.

Jimmy Carter, Bill Clinton, Michael Dukakis, and Mitt Romney all surged late to claim victory, but without a giant to slay. Carter and Clinton benefited mightily from the fact that their most popular adversaries decided not to run (Ted Kennedy in 1976 and Mario Cuomo in 1992). In Dukakis’s case, a path was cleared by the scandal engulfing Gary Hart. And Romney was always a very close second.

So how can Trump lose?

It’s possible we’re paying too much attention to the early polls. Stacks of political science research have shown that early-stage primary polls do a really poor job of predicting the eventual winner.

One reason is that people tend to make up their minds at the last minute. Despite the regular debates and constant media coverage, a huge proportion of voters simply won’t decide until it’s time to vote. So when pollsters call them months beforehand and ask them who they support, they may give an answer — but it can’t tell you much.

Trump, moreover, is perfectly positioned to benefit from this quirk in early polling. His fame gives him wide name-recognition, making him the easy choice of people who want to get off the phone but who haven’t yet researched the other candidates.

And even if the polls are right today, there’s still time for people’s opinions to change. Trump’s enemies are counting on it, including party insiders who fear that Trump’s nomination would hand the general election to Hillary Clinton — not least because he has alienated many Hispanic voters with his virulent anti-immigrant rhetoric.

To avert a Trump candidacy, some conservative groups have already begun running anti-Trump ads, and a former RNC staffer has setup an organization called “Trump Card LLC” to take Trump down.

These kinds of moves should matter, at least according to the political science research showing that nominations are generally determined by party insiders, not voters. The way it works is that once the party has coalesced behind a favored candidate, it signals its decision to voters using resources and endorsements. That’s why endorsements are generally considered a good predictor of primary outcomes.

By these lights, Trump is in trouble. He is not the establishment choice, and he hasn’t gotten any mainstream endorsements.

So which is it: inevitable or unlikely?

Take your pick. That’s the amazing thing about this race.

All the research says that Trump should lose. Dark-horse outsiders just can’t muster the force to beat out party-anointed candidates, who benefit from pooled resources and public endorsements. If Republican insiders could unite behind a single favorite — likely Marco Rubio or Jeb Bush — it’s possible that person would unite all the non-Trump voters in a winning cause.

Then again, Trump may prove the whole theory wrong. No outsider candidate has brought his combination of independent wealth, media celebrity, and signature (anti-immigrant) message.

So powerful is Trump’s appeal that he might well beat out a consensus insider, even if everyone else dropped out of the race. An Economist/YouGov poll from last week showed that in a head-to-head race, Trump would beat Rubio handily, 57-43.

For now, it’s hard to say which of these positions is right, but clarity is coming. When the primaries and caucuses begin, in about nine weeks, we’ll quickly find out whether Trump is as inevitable as he seems today, or as doomed as history would suggest.