For most of this election cycle, analysts operated under an assumption that the Democratic nomination wasn’t really Hillary Clinton’s to lose, because she couldn’t really lose it. 

But two polls now show her support among Democrats falling below 50 percent, and another poll shows her behind Sen. Bernie Sanders in New Hampshire by an almost statistically significant amount.  I don’t think many analysts would have predicted either of these things.  I certainly wouldn’t have. Given this, I would be remiss if I didn’t re-evaluate my other preconceived notions about the race.

After reviewing the evidence, I’d still make Clinton the overwhelming favorite to win the Democratic nomination.  I’d put her chances somewhere between 80 and 90 percent. This is in line with where most other analysts seem to place the race.

Even though Clinton’s chances of being the Democratic nominee have declined, I still think that she’s the overwhelming favorite, though my reasons differ from other analysts’ views.  Before I give my own thoughts, it’s worth running through some of the other arguments floating around out there.  This is an important exercise, because if you get the reasons that she’s likely to be the nominee wrong, you increase your chances of missing it when the odds turn against her.

1) “Her polling is fundamentally different than in 2007.”  Back in 2014 and early 2015, some observers liked to point out that she held large leads in the polls in 2007, yet still lost the nomination. The standard counter – which I liked to employ – was that her leads in 2014 and 2015 were simply in a different league.

It’s true: Clinton’s lead over Barack Obama peaked at some 28 points in October 2007.  During the summertime, it was around 18 points.  In August that year, her lead in New Hampshire was seven points, and her lead in Iowa was just two points.  By October, they were 19 and seven points, respectively. 

By contrast, Clinton’s lead was in the neighborhood of 50 points for most of 2014, and has been somewhere in the 40s for much of this year.  In Iowa, her leads have been around 50 points for most of this campaign, and she sported 35-point leads in New Hampshire as late as June.  So it was perfectly fair to say that her leads were, in fact, different in magnitude from 2007.

No longer. Her average lead in national polling is 24 points, with a distinctly downward trajectory. New Hampshire has become a virtual tie. Her lead in Iowa is relatively robust (24 points), but her trajectory is downward.

Obviously the discussion is about where things will be next winter, not about where things stand today.  But the point is, her polling right now looks an awful lot like it did in 2007.

2) “Late entrants are disastrous.” This isn’t a major argument, but it sets up some of my responses to other arguments in Clinton’s favor.  It’s true that every cycle sees a last-minute entry: Rick Perry, Fred Thompson, and Wesley Clark, for example.  For all, their first days were their best days.

But Perry, Thompson, and Clark were not, say, sitting vice presidents of the United States. Clark had no political experience, and it showed.  Thompson was a one-term senator who had been out of office for five years, and it showed . Rick Perry’s major gaffes might have been avoided if he’d run earlier, but they’d also have been avoided if he’d spent a decade on the national stage.

That’s what makes a potential Joe Biden run, or Bernie Sanders’ rise, so intriguing. These are candidates with lengthy political experience, who aren’t likely to make the sort of mistakes that sank other summer candidacies.  That’s not to say Biden would win, it’s just to say that I don’t take previous late entries to be all that informative about how a Biden candidacy might go.

3) “The party loves her, and the party decides.” “The Party Decides” is a popular book among academics and analysts, and for good reason.  It gives a reasonably compelling account of the way party insiders influence election outcomes. 

At the same time, we simply don’t have very good data on party selection processes.  A lot of what goes on in party decision-making occurs behind a curtain; as one of my friends put it, endorsements and so forth are really only ripples in the curtain. We also don’t have much data on these processes, as binding primaries are a fairly recent innovation for choosing nominees. Unsurprisingly, there are other political science narratives that emphasize candidate and voter agency more strongly.

Perhaps the best argument here, though, is that 2016 really might be different.  I say this as someone who is fairly allergic to “this time it’s different” arguments.  This model, however, didn’t perform all that well in 2008, and came a few thousand Michigan and Ohio primary voters away from being severely tested in 2012.  Super PACs and the Internet both weaken party control to a certain degree. Whether they represent a paradigm shift remains to be seen.  But at the very least, they empower outsider candidates, and provide a mechanism for late entrants to catch up quickly.

4) “Her fundraising/organization make it difficult for another candidate to catch up.” This is really just a specific version of No. 3 above. There are three related crucial factors at work here.  First, candidates receive diminishing returns on their money.  Think about it: The first time you see ads, you’re probably entertained.  By the 50th, you’re reaching for your remote.  It probably doesn’t matter whether Clinton has $100 million or $500 million by the time the Iowa caucuses roll around.

Second, this is a place where super PACs probably are game changers.  In 2000, John McCain’s candidacy folded in part because he didn’t have the money to compete in late states.  In 2016, that just wouldn’t be a problem.

Third, remember, a candidate doesn’t have to have the ability to win South Carolina in February.  She just needs to win Iowa or New Hampshire.  A win there brings with it national prominence, favorable coverage, and money. 

So why, after all of this, do I think Clinton is the favorite to win?

Back in 2014, I was having drinks with a friend, and he asked me what it would take for Clinton to lose.  I gave it a lot of thought, before coming up with essentially this response:  First, she would have to see her favorability ratings tank. Second, and relatedly, she would have to fall behind the Republican candidates, or at least into a tie with them.  As long as she was a sure-fire winner, the Democrats would not switch horses.  But if they started to get nervous, they might start to look around.

It doesn’t surprise me that this has happened, though the extent of the collapse in her favorable ratings did surprise me, as well as how quickly the national polls have closed. But she still has avoided the third prong of my scenario.

The third prong goes like this: Hillary Clinton is broadly acceptable to the Democratic Party, but she doesn’t have a natural demographic or ideological base of support – perhaps among white urban and suburban women, but they don’t comprise anywhere near a majority of the vote in today’s party. This was her problem in the early contests in 2008: She finished third in Iowa and Barack Obama and John Edwards peeled away working-class and liberal voters.  Obama was later able to break off African-American voters as well, putting together a winning coalition.

There isn’t an Obama-like candidate out there this cycle.  Instead, to defeat Clinton, a variety of candidates would have to break off different constituencies.  So Bernie Sanders breaks off liberal voters, Joe Biden breaks off white working-class voters, then someone like Deval Patrick breaks off African-American voters.  That could prevent her from winning the nomination.

The problem with this scenario is that neither Biden nor Patrick (or someone like him) is running.  Sanders is doing surprisingly well with liberal whites, but he probably runs into a wall on Super Tuesday.  If Biden gets in he could steal working-class whites from Clinton’s 2008 coalition, but her remaining coalition of non-white voters and urban/suburban women would probably be enough for her.  Until a candidate demonstrates that he can break non-white voters out of Clinton’s coalition, she will probably win a war of attrition.

I do say “probably,” however.  Remember that in 2008, African-American voters remained in Clinton’s camp until it was clear Obama was viable – she led by around 20 points in South Carolina in October 2007. If Clinton loses Iowa or New Hampshire, or both, it scrambles perceptions of her and upends the Democratic contests in ways that are really difficult for us to predict.  But that’s just a caveat.  For now, I think that absent an indictment, she remains the solid favorite for the nomination.