Nate Silver's Blog, page 51

We Fixed An Issue With How Our Primary Forecast Was Calculating Candidates’ Demographic Strengths

Welp, this is never fun. We discovered an issue with how our primary model was making state-by-state and district-by-district forecasts. Specifically, the model was not properly calculating the demographic regressions that we use as a complement to the polls.

The top-line effect of fixing the error was not very large — for instance, Sen. Bernie Sanders’s chances of getting a delegate majority fell by around 3 percentage points, while former Vice President Joe Biden’s fell by around 1.5 points. In fact, it mainly helped the chances that no one will get a majority of pledged delegates (those chances rose about 3.5 percentage points), at the expense of any one candidate’s chances of getting a majority. The one candidate who did benefit was former New York Mayor Michael Bloomberg, whose chances increased from 7 percent to 9 percent). However, the changes are more notable in some individual states.

This problem was introduced into the model’s code on Feb. 5 (just after the Iowa caucuses); these functions had been working properly before then. The error was fixed as of the forecast we published just after noon on Feb. 19.

Here’s a look at what went wrong and what changed:

First, what is a demographic regression? To infer a candidate’s standing in all 57 states and territories at any given time, our model calculates a series of demographic regressions based on (i) the results of states that have voted so far and (ii) the polls in states where we have abundant polling. For instance, the regressions can figure out that Biden is strong in states with a large African American populations, and that Sanders is strong in liberal states. These demographic regressions are then combined with a geographic prior based on candidates’ home states and regions (for example, Sanders is assumed to be strong in New England). The result is then used as a substitute for polling in states where there are no polls and as a complement to the polls in states where there isn’t much polling. Nevada and South Carolina, for instance, have a fair amount of polling but not as much as the model would like, so the regression gets a small amount of weight in our forecasts.

What was wrong with the regressions? Basically, the regressions weren’t calculating at all, so the model was just defaulting to the geographic prior. (If you want to get very technical, when programming in Stata, please remember that local macros aren’t stored in the program’s memory when you execute another do-file from within the shell of a master do-file.)

What problems was this causing? By just defaulting to the geographic prior rather than also using demographics, the state-by-state forecast distributions were too compressed, or underdispersed. In other words, they weren’t recognizing fairly obvious demographic strengths and weaknesses of each candidate.

While relying solely on the geographic prior isn’t a terrible approach, especially for candidates such as former South Bend, Indiana, Mayor Pete Buttigieg who are strong in their home regions, it doesn’t capture some of the variations elsewhere in the Democratic electorate. Let’s take the presence of African Ameircans and moderate white voters in the South as an example. Because these two demographic groups are so important in that region, Biden is stronger there and weaker elsewhere than you’d guess from geography alone. The same holds true, though to a lesser extent, for Bloomberg. Conversely, Sanders has more strength in the West than you’d guess based on geography alone, because Western electorates tend to be liberal and to have a high number of Hispanic voters, and Sanders does well with both of those voting blocs.

As a result, the model was, for instance, underrating Biden’s chances in states like South Carolina and Alabama and underrating Sanders’s chances in Nevada and California. Since the demographic regressions are phased out in states with a lot of polling, the effect was larger in states with less polling, like Colorado. On average, candidates’ projected vote shares changed by about 1 percentage point (e.g. shifted from 18 percentage points to 19 percentage points) as a result of the bug fix.

What was that about district-by-district forecasts? The model also uses the demographic regressions to forecast the results in individual districts. For instance, although Sen. Elizabeth Warren might not be strong in South Carolina overall, she could be poised to do fairly well in a South Carolina district with a lot of college-educated voters.

Without the demographic regression, though, the model defaulted toward using the statewide forecast in each district. It did still account for random variation between districts — so there might be simulations where, say, Warren got 12 percent of the vote in South Carolina but 17 percent in a particular district, thereby earning delegates there. (Democratic rules generally require that a candidate receives at least 15 percent of the vote to qualify for delegates in a state or district.)

Still, with the demographic regression not working properly, the model wasn’t accounting for enough district-by-district variation. As a result, it tended to underestimate the number of district delegates that a candidate would expect to earn when they finished with less than 15 percent of the vote statewide. This matters because about two-thirds of Democratic delegates are awarded by district rather than statewide. The ability of candidates to earn these district delegates makes it slightly harder for front-runners (e.g. Sanders) to accumulate runaway delegate margins, which in turn makes a “no majority” scenario more likely.

How can you prevent something like this from happening again? I don’t know. We’ve gotten our share of forecasts wrong and had models that, in retrospect, we wish we’d designed differently. But I usually have a good eye for when code changes or new data creates output that doesn’t look right because either the data or the code was incorrect. I didn’t catch it this time, perhaps because the code changes were introduced at the same time that we input initial Iowa results into the model, and Iowa had a much larger impact on the model than the bug we introduced.

But as a note to myself and to other people who program statistical models: It may not always be a good idea to introduce failsafes into your program rather than letting it break. For instance, our model had a failsafe to default to the geographic adjustment if it couldn’t calculate the demographic regressions. But because the geographic adjustment by itself produces reasonable enough (but far from ideal) answers, it became harder to notice the error than it would have been if the program had stopped executing or had produced self-evidently ridiculous forecasts.

So if you see something that looks weird in the forecasts, please let us know! Reader feedback alerts us to a lot of small issues (like polls that were entered incorrectly) and occasionally helps us to catch some bigger ones, too. My sincere apologies for not catching this bug sooner.

Will The Nevada Debate Be All About Bloomberg?

Welcome to FiveThirtyEight’s weekly politics chat. The transcript below has been lightly edited.

sarahf (Sarah Frostenson, politics editor): Let’s not bury the lede: Former New York Mayor Michael Bloomberg is joining his first debate of the cycle tomorrow night in Nevada, even though he’s largely skipped competing in the first four states. He’ll be joined by five other candidates: former Vice President Joe Biden; former South Bend, Indiana, Mayor Pete Buttigieg; Sen. Amy Klobuchar; Sen. Bernie Sanders; and Sen. Elizabeth Warren.

And things are sure to get heated fast. Sanders is currently the polling leader in Nevada and favored to win the state, according to our forecast, but Bloomberg is catching up and is now at 16.1 percent in our national polling average, essentially tied with Biden.

So what should we expect tonight? Attacks on Sanders’s newly minted front-runner status? Or will the country’s first look at Bloomberg overshadow everything else?

natesilver (Nate Silver, editor in chief): I’m glad that we finally have an opportunity to talk about Michael Bloomberg since no one else is paying attention to him.

sarahf: Haha, fair. There’s nothing the media loves more than making a non-story a story.

nrakich (Nathaniel Rakich, elections analyst): The safest bet is that Bloomberg will bear the brunt of most attacks, right? That’s how it has played out in the media this week, between allegations he created a hostile workplace for women and renewed focus on the racial dimension of the stop-and-frisk policing strategy, which Bloomberg supported as mayor.

natesilver: It seems like a safe-ish bet, especially since NOBODY SEEMS TO HAVE NOTICED THAT BERNIE SANDERS IS IN FIRST PLACE IN NATIONAL POLLS AND IS IN PRETTY GOOD POSITION TO WIN THE NOMINATION.

nrakich: Yeah, it’s a pretty sweet deal for Sanders that he has largely been spared attacks even though he’s the primary front-runner after winning New Hampshire (and, arguably, Iowa). He looks on track to win Nevada, too! He’s a more imminent threat than Bloomberg is to the other candidates.

ameliatd (Amelia Thomson-DeVeaux, senior writer): Honestly, I think Sanders is really lucky that Bloomberg made it into this debate — since it’s Bloomberg’s first debate, it seems like a safe bet that everyone is going to go after him.

sarahf: BUT at this point … shouldn’t we be paying attention to Bloomberg?

He’s tied with Biden in our national polling average!!

[Our Latest Forecast: Who Will Win The 2020 Democratic Primary?]

natesilver: Of course people should be paying attention to Bloomberg, Sarah! But these things are also self-fulfilling prophecies to some extent. The fact that he’s gotten so much more attention than, say, Buttigieg or Klobuchar or Warren is something that media critics should think about. If you take off that media critic hat, he’s obviously a real contender for the nomination now.

ameliatd: I wonder if the attention on Bloomberg will make it harder or easier for the other candidates to stand out. Warren, for instance, could really use a strong performance a la Klobuchar in New Hampshire. I could see it going a couple of ways — maybe being able to attack Bloomberg on stage gives her a little viral moment. Or maybe the fact that everyone will likely be going after Bloomberg makes it harder for her to steal the spotlight.

kaleigh (Kaleigh Rogers, FiveThirtyEight contributor): That’s true, especially when people are eager for the field to start winnowing. Adding a new face might feel like going backwards, and could certainly be distracting.

natesilver: It wouldn’t necessarily surprise me if Bloomberg goes after Sanders in a big way, by the way. His team has certainly been escalating conflict with Sanders. And it plays into his message that he’s the first train leaving the station if you want an alternative to Bernie.

What Bloomberg DOESN’T want is Buttigieg or Klobuchar to surge. He wants to keep things a bit muddled.

Nor does he want a Biden comeback narrative to start brewing. So if he can make news by attacking Sanders or otherwise giving a memorable performance, that probably works for him. It’s not unlike Trump in 2016.

sarahf: And maybe this is less likely, but Sanders could go after Bloomberg. It’s probably too good of an opportunity for him not to at least try and land something. There will literally be a billionaire on stage who has spent a lot of money to buy access for his bid for the presidency. (In fact, Bloomberg has probably spent the most on his campaign out of any other presidential candidate … ever?!?!)

But OK, wait wait. What’s the basis for the idea that we’re NOT paying enough attention to Sanders?

natesilver: That he’s first in national polls and that he’s actually, y’know, had people vote for him and won states, which is what these election things are supposed to be about.

Nobody in the mainstream media has the right to complain about paying too much attention to the polls now, though. Because so far, the Bloomberg thing is entirely based on the polls when there are other candidates who have demonstrated actual support.

kaleigh: But it’s not as if Sanders’s success hasn’t been covered. Certainly there’s room to talk about more than just the front-runners?

There’s also the fact that the vast majority of Democrats have not yet voted. There’s still a race here.

natesilver: I mean, in some narrow but valid sense, Buttigieg is the front-runner (he has the most pledged delegates) and yet, he isn’t getting a ton of coverage.

sarahf: Kaleigh makes a good point, though. Sanders’s successes have been covered. What’s difficult, though, is they haven’t been portrayed as decisive victories. But I kind of get that. Sanders and Buttigieg both “won” Iowa — we’ll see if the recanvassing efforts change that. And he didn’t win the New Hampshire primary by that large of a margin (1.3 points).

So there’s this other narrative emerging around the field being divided and no one candidate having a firm grasp on things — our own forecast puts Sanders’s chances of winning a majority of pledged delegates at roughly the same odds as no one winning a majority — so maybe this is the year every political journalist’s dream comes true … a cOnteSTEd convention … uh, the troll cap is too much work.

nrakich: Sarah, I think some Sanders skepticism is warranted. He has done well in the first two contests but hasn’t dominated, and as you point out, Sarah, our model doesn’t think he has a mortal lock on the nomination.

On the other hand, I think the tone of the coverage of Sanders’s win (covered as if it were a loss) and Klobuchar’s third-place finish (covered as if it were a win) was topsy-turvy.

sarahf: That’s fair, but Klobuchar did manage to double her support in New Hampshire, going from about 10 percent in our polling average to winning 20 percent of the vote. That’s pretty impressive.

natesilver: With Klobuchar, I have mixed feelings. On the one hand, I’m just not sure that a fifth-place finish followed by a third-place finish should get you all that much attention. On the other hand, she fits the traditional credentials of a party nominee (not too old or too young, not “too” far to the left or too centrist, experienced) more so than most of the remaining candidates do. And people seem more inclined to chase shiny objects.

sarahf: But to step back for a moment, the New Hampshire debate was one of the more consequential debates of the cycle, yeah?

Probably means Nevada will be a nothingburger, but I am intrigued to see what happens there as Nevada is so weird to poll.

ameliatd: One big question for me is how many Nevadans will actually tune into the debate, too. In general, they tend to be less politically engaged than New Hampshire or Iowa voters. So that could dampen the impact.

natesilver: I guess I think that might make the debate more impactful, in some ways? If people are less tuned in, they may not be as far along in their decision-making process, which could make the debate more likely to sway their opinion.

ameliatd: It’s also worth noting that thousands of Nevadans will already have “caucused” by the time the debate starts. Early voting started on Saturday and ended on Tuesday.

kaleigh: More than 26,000 Nevadans voted in the first two days alone.

nrakich: I do think the debate will have more of an impact nationally than it will in Nevada, to Amelia’s points. Think about how the New Hampshire debate turned into a discussion of South Carolina issues!

And Bloomberg isn’t contesting Nevada, so obviously his debate performance will only affect his numbers in other states.

sarahf: So how confident are we about the situation in Nevada? Our Nevada forecast gives Sanders about a 7 in 10 chance of winning there. Biden is the second most likely winner with a 1 in 7 chance, but Warren is close on Biden’s coattails. Buttigieg and Klobuchar aren’t too far off either.

nrakich: Personally, I’m not confident. The Nevada polls are all over the place.

sarahf: Could a strong second-place finish by someone in Nevada overshadow a Sanders win? Similar to what we saw in Iowa and New Hampshire?

nrakich: Yes, I think if someone surprising finishes second (Warren? Steyer?), it will be New Hampshire all over again, where that person gets more favorable coverage than the outright winner (assuming it’s Sanders).

natesilver: I mean … it’s another state where media expectations (BERNIE CERTAIN TO WIN) seem a little out of line with the reality (probably Bernie, but high uncertainty). If I were a Sanders voter, I’d be annoyed at how Sanders always seems to be the victim of the expectations game.

ameliatd: We also haven’t talked about the fact that this is the first state with a significant population of voters of color. So it’s the first real chance to see if candidates like Buttigieg and Klobuchar can do well among Latino/black/Asian voters — if they do, that would obviously be a big deal.

sarahf: That’s right. And at this point, the crosstabs suggest that only Sanders and Biden have a lead there, right?

ameliatd: Well, like Nathaniel said, the polls are kind of all over the place — but yes, among Latinos at least, Sanders and Biden tend to do best.

natesilver: Keep in mind that the Latino population is pretty young, and that Sanders obviously does well with young voters.

nrakich: Yeah, I wonder if Buttigieg will bust out some Spanish during the debate. He recently released a Spanish-language TV ad that he himself narrated.

kaleigh: He’s been running a fair amount of Spanish-language Facebook ads, too.

nrakich: That said, when Beto O’Rourke, Cory Booker and Julián Castro spoke Spanish in the first debate, some people saw it as a stunt. A YouGov poll in June found that 37 percent of Hispanic Americans found it respectful when a presidential candidate spoke Spanish in a debate, and 27 percent thought it was pandering. White Americans (who as of 2016 were still a majority of Nevada Democratic caucusgoers) thought it was pandering by a nearly two-to-one ratio. And Equis Labs, a polling firm that specializes in polling Hispanic Americans, told me that only 29 percent of Hispanic registered Democrats in Nevada say Spanish is their preferred language.

ameliatd: I know we’ve heard quite a bit about health care in these debates, but I bet it’ll come up again tonight, since Sanders is being attacked by the state’s biggest labor union over Medicare for All — and other candidates, like Klobuchar and Steyer, have been jumping on that bandwagon recently.

sarahf: OK, let’s talk candidate strategy for a moment. Bloomberg is a bit of a curve ball.

He isn’t contesting Nevada, so his strategy is to … stop a debate surge from happening and somehow make a plea for voters to consider him, not in South Carolina (which votes next), but hold out for Super Tuesday instead?

But what does this mean for the other moderate candidates, especially Biden? Biden at this point still leads Klobuchar and Buttigieg in national polls — does he need to nip Bloomberg’s momentum in the bud?

And then what does this mean for Sanders? Does he just hope that no one attacks him directly and let the moderates duke it out?

I guess what I’m really asking is … is Biden toast?

Or does this debate have the biggest stakes for him?

natesilver: Why would Biden be toast? He’s tied with Bloomberg in national polls. And he has an opportunity to get a boost with a win in South Carolina or, less likely, Nevada.

If he loses in South Carolina, he might be toast.

sarahf: What if he finishes in fourth or fifth in Nevada, though?

natesilver: Meh, who cares? South Carolina is coming in a week.

Biden’s media narrative is awful as can be now and it probably doesn’t get any worse if he finishes fourth or fifth in Nevada.

nrakich: It might be good for Biden that Steyer didn’t make today’s debate? Not that Steyer has turned in super compelling debate performances, but Steyer might be Biden’s most direct competition in South Carolina.

sarahf: There has to be some consolidation around a moderate alternative to Sanders, right? Buttigieg, Klobuchar, Bloomberg and Biden can’t all stay in and actually mount a credible campaign against Sanders, can they?

nrakich: They can if they all have visions of emerging as the nominee after a contested convention!

natesilver: Or candidates might think “Sanders v. Bloomberg could go very nuclear, so I need to stick around as the least radioactive option.”

That seems … quite plausible? We probably have to assume that Bloomberg will drop a lot of negative ads on Sanders at some point?

nrakich: He should probably do so sooner rather than later. There are lots of delegates up for grabs on Super Tuesday, and Sanders is currently riding high.

ameliatd: Biden has to be hoping that Bloomberg takes a beating in the debate, though, right?

natesilver: Sure, I think that’s right, Amelia.

But Biden also doesn’t want one of Buttigieg or Klobuchar to surge.

kaleigh: Sarah, if we’re assuming Sanders lands the role of the progressive candidate, does that mean you think Warren is DOA?

sarahf: That’s an interesting question, Kaleigh. She’s currently in third in our polling average in the state.

sarahf: And the labor union there didn’t criticize her health care plan as directly as it did Sanders’s.

nrakich: I think Warren needs to show a pulse in Nevada. She’s already basically conceded South Carolina, canceling a bunch of her TV ad reservations there.

sarahf: So I think there’s still room for a Warren comeback maybe? She definitely was pitching that after her disappointing finish in New Hampshire, but as Amelia mentioned earlier, at this point only Sanders and Biden have demonstrated that they can build more diverse bases.

ameliatd: Warren’s argument has basically been that she’s going to make a comeback on Super Tuesday. But if she has another disappointing finish in Nevada, that could definitely slow her down even further.

natesilver: I dunno. Warren is not THAT far behind since nobody is THAT far ahead. Obviously a win or a strong second-place finish in Nevada could reverse the narrative about her, though.

kaleigh: So who has the most to gain (or most to lose) in tonight’s debate? Biden? Warren?

natesilver: The debate is about Bloomberg, like it or not. It just is. The media is fucking obsessed with Bloomberg. And it’s his first debate. He’s 90 percent likely to be the headline, positive or negative.

sarahf: Yeah, he’s likely to be this chat’s headline.

natesilver: wE aRe PaRt Of ThE PrOBLeM.

nrakich: I mean, I think that is somewhat justified for this debate specifically.

We’ve heard Sanders, Warren, Biden, etc., say the same thing a zillion times. We have no idea how Bloomberg will hold up under pressure. So he is the one we will likely learn the most about tonight.

kaleigh: But Bloomberg kind of has nothing to lose at this point. He’s skipping Nevada, and negative or positive, the focus on him only further legitimizes him as a candidate.

sarahf: I actually think he has a lot to lose.

nrakich: Oh, he definitely has a lot to lose. He’s been pumping his message out to people totally unanswered thanks to his millions of dollars in TV spending!

sarahf: Yeah, that’s what I was going to say. At this point, he’s largely gotten to control his brand.

nrakich: Until this week’s bad headlines, people haven’t heard a bad word about him.

And a big known unknown for Bloomberg is how good of a debater he is. His team is already lowering expectations.

ameliatd: Yeah, I agree with Nathaniel. Whatever you think of the fact that Bloomberg hasn’t appeared in a debate so far, this is his introduction to a debate audience — and maybe more importantly, his rivals’ first chance to attack him in a debate setting. Which is crazy, since we’ve been covering these things for EIGHT MONTHS.

natesilver: I think he has a lot to lose but also expectations (there’s that term again!) seem to be fairly low, i.e., the media assumes he’ll be a mediocre debater.

nrakich: I think there’s room for him to be a decent debater but also for this debate to represent the first mass airing of anti-Bloomberg grievances to the American public.

Like, even if Bloomberg has answers for the criticisms levied against him, it’s the first time many Americans will hear those criticisms.

natesilver: Bloomberg is the one candidate that sort of unites all the others against him. He’s trying to elbow out the other candidates in the moderate lane. And he obviously has beef with Sanders, and the race could easily come down to those two.

kaleigh: That’s true. I just wonder if a debate-long attack from, say, Sanders will make it seem as if Bloomberg is the de facto moderate candidate.

ameliatd: That’s an interesting point, Kaleigh — maybe Biden does run the risk of seeming even weaker if the debate is largely between Bloomberg and Sanders.

natesilver: Also, Bloomberg has not been a longstanding and loyal member of the Democratic Party. He isn’t owed any particular favors or deference. So I’d expect a higher aggression level than we’ve seen in past debates.

nrakich: Oh, I disagree there, Nate.

He has been arguably Democrats’ most important donor in the last few election cycles.

He helped many members of Congress get elected, for instance.

I would argue that it could actually be tricky for someone like Klobuchar or Biden, who are establishment-aligned and came up through the big-donor system, to go after him directly.

natesilver: OK then they deserve to lose, to be honest.

nrakich: I mean, sure, maybe.

natesilver: They should also be going after Sanders directly.

nrakich: But I think only Sanders and Warren are capable of really taking the gloves off.

All the other candidates probably are hoping Bloomberg runs super PAC ads on their behalf in the general.

kaleigh:

Politics Podcast: Can Michael Bloomberg Win?!

By Galen Druke, Clare Malone, Nate Silver and Micah Cohen, Galen Druke, Clare Malone, Nate Silver and Micah Cohen, Galen Druke, Clare Malone, Nate Silver and Micah Cohen and Galen Druke, Clare Malone, Nate Silver and Micah Cohen












 












More: Apple Podcasts |
ESPN App |
RSS
| Embed


Embed Code

Former New York Mayor Michael Bloomberg has surged to 16 percent in our national polling average. In this installment of the FiveThirtyEight Politics podcast, the crew discusses Bloomberg’s chances of winning the Democratic nomination. They also debate whether Sen. Bernie Sanders is a lock to win the Nevada caucuses.

You can listen to the episode by clicking the “play” button in the audio player above or by downloading it in iTunes , the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen .

The FiveThirtyEight Politics podcast publishes Mondays and Thursdays. Help new listeners discover the show by leaving us a rating and review on iTunes . Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.

Do You Buy That … A Contested Convention Is A Real Possibility?

Does Sanders Have A Ceiling? Maybe. Can He Win Anyway? Yes.

The Sen. Bernie Sanders rally that I attended on the evening before the New Hampshire primary drew a reported 7,500 people — about twice as many as his actual 3,867-vote margin of victory in the primary the next day. I say that not to endorse crowd sizes as an alternative to the polls. (Despite the large crowds, Sanders slightly underperformed his polls in New Hampshire, in fact.1) Nor do I mean to imply that Sanders won in New Hampshire because of the rally. (It was held before a largely student audience at the University of New Hampshire — people who were already likely to vote for Sanders.) But it does go to show how razor-thin the margins have been so far in the primaries. The voters who pushed Sanders past former South Bend, Indiana, Mayor Pete Buttigieg in New Hampshire might only halfway fill a college hockey rink.

[Our Latest Forecast: Who Will Win The 2020 Democratic Primary?]

But the rally was also impressive. It was full of star power, including speakers like Rep. Alexandria Ocasio-Cortez, former Ohio state Sen. Nina Turner, actress Cynthia Nixon and political activist Cornel West — all of whom were introduced to the crowd like heavyweight boxers — along with a concert by The Strokes. In between the big names, Sanders organizers gave students detailed voting instructions.2 Everything was tightly scripted.3 It was a show of force.

What the rally largely lacked, though, were attempts to persuade voters who weren’t already aboard the Sanders train. On the contrary, the emphasis was on turning out the faithful, and the faithful were all presumed to be on board with Sanders’s lefty platform. Nixon, for instance — who earlier had drawn a round of boos for a brief reference to Hillary Clinton, which she quickly shushed — said that nominating a moderate candidate would ensure that nobody showed up on Election Day.

But moderates were once a source of strength for Sanders. Four years ago, Sanders won voters who identified as moderate by 20 points in New Hampshire, about the same as his overall margin of victory in the state over Clinton. This time around, Sanders finished third among moderate voters, getting 16 percent of the moderate vote compared with 27 percent for both Buttigieg and Sen. Amy Klobuchar.

So, let’s talk about the “c” word: ceiling.

Does Sanders’s support have a ceiling?

I shudder to ask the question in part because of bad memories from four years ago, when theories about Donald Trump’s ceiling were a big reason that people like me initially dismissed his chances in the primaries.

In Trump’s case, though, there was at least some polling-driven evidence of a ceiling. He tended to lose ground in polls that asked about hypothetical head-to-head matchups against other Republicans, for instance. And his favorability ratings among Republican voters were quite low for someone who was leading the field.

There isn’t much evidence of this for Sanders. His favorability ratings are roughly as good as any other Democrat’s — and often the best in the field, depending on which poll you look at.

It’s also worth mentioning that Sanders gets a lot of support from younger African Americans and Hispanics, making his coalition among the most diverse in the race. Granted, he does have very little support from voters over the age of 65, but of all demographic deficiencies, that may be one of the easier ones to overcome. There are plenty of young voters in every state, provided you can turn them out.

Additionally, a set of YouGov polls last week showed Sanders winning in hypothetical head-to-head matchups against every other Democrat — narrowly beating former Vice President Joe Biden and Sen. Elizabeth Warren while more clearly defeating Buttigieg, Klobuchar and former New York Mayor Michael Bloomberg. (There are some qualifications to that YouGov poll: Buttigieg and Klobuchar still have fairly low name recognition, and earlier polling that tested head-to-head matchups hadn’t shown Sanders doing so well, especially against Biden.)

But if you look at the actual behavior of voters in Iowa and New Hampshire — and the most recent polling on how voters in the next states to vote are reacting to the Iowa and New Hampshire results — there are a few troubling signs for Sanders, including some evidence of what you might call a ceiling. In no particular order:

In both Iowa and New Hampshire, Sanders did relatively poorly among late-deciding voters. In New Hampshire, he got 17 percent of the vote among voters who decided in the last few days before the primary, as compared to 35 percent among voters who decided before then. And in Iowa, he got just 13 percent of late deciders, versus 28 percent of early deciders. These numbers are a potential hallmark of a campaign that emphasizes turnout over persuasion. This kind of campaign identifies its voters early and turns them out, but it doesn’t have a lot of voters drift into its orbit late in the race.
In Iowa, where voters could realign to second choices if their candidate didn’t clear the viability threshold (usually 15 percent of the voters at their caucus site), Sanders gained relatively little from this process, going from 24.7 percent on the first-alignment vote to 26.5 percent on the final alignment — a gain of 1.8 percentage points. (By contrast, Buttigieg added 3.8 percentage points to his vote total via this realignment process despite starting out with a lower vote share than Sanders.) Sanders’s gains were considerably less than our model expected, too. It thought that if Sanders had around 25 percent of the initial vote, he would have wound up with about 30 percent of the vote after realignment. This realignment process doesn’t take place in most states — although it does occur in Nevada and a few states later on that use ranked-choice voting — but it’s nonetheless a negative sign for Sanders as it would seem to indicate that he is relatively few voters’ second choice. That means he might not stand to gain as much as other candidates when opponents drop out or fall in the polls.
Based on the evidence we have so far, Sanders has gotten relatively little bounce in the polls from his outright win in New Hampshire and popular-vote win4 in Iowa. In our national polling average, he is at 22.9 percent, only slightly higher than he was in our final average before Iowa, when he was at 21.7 percent.
Even in fairly liberal states like New Hampshire, voters seem to prefer a more moderate candidate in the abstract. Nearly two-thirds of voters in the New Hampshire exit poll said they preferred a candidate who could beat Trump to one who agreed with them on the issues. (Note that a lot of voters see Sanders as electable, and he polls pretty well in head-to-head polls against Trump.) And 50 percent of voters said Sanders’s positions were too liberal. Meanwhile, the combined vote shares for Buttigeig, Klobuchar and Biden (52.6 percent) considerably exceeded that for Sanders and Warren (34.9 percent). No, it’s not quite as simple as there being two distinct lanes (left and moderate) with no overlap between them. But as the primary has evolved, the electorate has behaved more and more as the lanes theory might predict — Buttigieg and Bloomberg have gained ground as Biden has declined in national polls, for instance.
Finally, turnout in the first two states has been a mixed bag. Turnout in Iowa was 176,000 people, about what it was in 2016 but well below 2008 levels and less than what most observers expected. Turnout in New Hampshire — about 300,000 voters in the Democratic primary — did end up being record-breaking. The Iowa numbers probably weren’t as bad as they looked, though. The caucuses did not receive their typical share of media attention, thanks to a busy news calendar involving impeachment, the Super Bowl and other stories. Also, Iowa has drifted red, so there are simply fewer Democratic voters in that state than there once were. But if the Iowa numbers weren’t as bad as they looked for Democrats, the New Hampshire ones weren’t as good as they seemed. Democrats were not going up against a competitive Republican primary as they were in 2008 or 2016, which helped boost turnout in a state where independents can vote in either primary. Nor was turnout particularly high as a share of registered voters as the number of registered voters has grown since 2008.

So overall, while Democratic turnout has been just fine, it has not exactly been revolutionary. It may even be that the Sanders campaign — if it has a highly loyal but relatively fixed number of voters — prefers lower turnout overall, since that means its base will make up a higher share of the electorate. In Iowa, with its relatively low turnout, 24 percent of voters were under 30. In New Hampshire, meanwhile, voters under 30 accounted for just 13 percent of the electorate, down from 19 percent in 2016.

Think less about ceilings, more about volatility

Over the course of building our primary model late last year, I grew less skeptical of Sanders’s chances. In fact, our model is now quite bullish on Sanders, having him as by far the most likely Democrat to win a majority of pledged delegates — although the most likely scenario is that no one wins a majority (meaning a contested convention is possible).

Ideally, working on a model helps you to see a race with a fresh set of eyes and reexamine premises that might be outdated. Relevant to Sanders, we found that endorsements had less predictive power than they had once had, no doubt in part because adding data from 2016 (when Trump won the GOP primary and Sanders was competitive against Clinton despite receiving little support from party insiders) undermined the “Party Decides” theory of the race. Conversely, we found that fundraising was more predictive than we had previously assumed.

I don’t want to go overboard here: The media probably still overstates the importance of money overall, and for the most part, the best indicator of a candidate’s position in the race is simply his or her polls. Nonetheless, Sanders’s lack of endorsements is only a marginal reason to be worried about his chances, while his excellent small-donor fundraising is a reason to be optimistic.

We also found in building the model that the concept of “ceilings” and “floors” tends to be rather fuzzy; like “electability,” it’s something that would benefit from more precise specification. For instance, we found no evidence that candidates who were frequently cited in polls as voters’ second-choice picks were more likely to increase their support than ones who weren’t. (Although that data is hard to come by for previous election cycles, so I wouldn’t take that conclusion as definitive.)

But we did find that for candidates who carved out their own space and had few close substitutes, the polling behaved differently than it did for those who had a lot of competition.

To categorize candidates, we rate them along four dimensions:

How liberal or moderate they are on social policy.
How leftist or centrist they are on economic policy.
Whether they portray themselves as “insiders” or “outsiders.”
Whether they tend to be technocrats (who mostly appeal to elites and college-educated voters) or populists (who mostly draw support from the working class).

And in this conception, Sanders is something of an island unto himself. Warren is close to him on the issues, but she is less anti-establishment and more reliant on support from college-educated voters than he is. None of the other major candidates are very close to Sanders at all.

This means someone like Sanders has few direct competitors, so his polling swings tend to be smaller. If things are going poorly, there are fewer places for his voters to go because there isn’t a clear alternative to him. But someone like Sanders also tends to pick up less support from other candidates when things are going well because he’d represent a big leap for, say, a college-educated, moderate Klobuchar supporter.

Conversely, a candidate like Buttigieg — or Sen. Kamala Harris, before she quit the race — can experience more polling booms and busts. In trying to be an acceptable choice to everyone, these candidates have high upside potential, but because they may be relatively few voters’ first choice, they can also have a low floor.

To put it another way, instead of thinking of hard ceilings, it’s probably best to think of candidates as being either low or high volatility based on the amount of competition they face. It’s also best to think of this as being a dynamic status that can evolve over the course of the race. For example, Biden once seemed to have little direct competition, but the growth in support for Bloomberg, Buttigieg and Klobuchar has suddenly made the moderate/establishment lane far more crowded.

As to whether you’d prefer to be a low-volatility or high-volatility candidate, that depends on a lot of factors. But being a low-volatility candidate definitely helped Sanders during a rough stretch of the campaign in early October when Warren was surging in the polls and he took a break from the trail following a heart attack.

At the time, I thought some of his supporters would jump ship to Warren, perhaps putting her in the overall driver’s seat for the nomination. Instead, Sanders’s base largely stuck with him, and he received endorsements from Ocasio-Cortez and others. And by late November, Sanders had overtaken Warren and reclaimed second place in national polls. It was an impressive show of loyalty from his supporters.

So even if you do want to think in terms of ceilings instead of volatility, know that ceilings imply the existence of floors — and low ceilings and high floors generally go together. Having a high floor can be helpful, too, especially in a multicandidate race where the support of 20 or 25 percent of voters can be enough to lead polls and win states.

Sanders has several ways to win

To state the obvious, no one knows with much certainty how the Democratic race is going to turn out. But Sanders will likely have some paths to victory available almost no matter what:

If everything stays the same, Sanders could win the nomination with a plurality of pledged delegates. It remains quite possible that some candidate — whether it’s Sanders or someone else — will have some kind of polling surge in the race soon, making the top of the field less crowded. But suppose that doesn’t happen and the race muddles along roughly as it is now, with Sanders at around 25 percent of the vote and several moderate candidates with around 15 percent each.

That would actually be a pretty nice scenario for Sanders. He’d get delegates almost everywhere, whereas the moderate candidates sometimes wouldn’t, depending on whether they hit the threshold required to win delegates in each state and district. So you could go into the convention with a scenario like: Sanders has 40 percent of pledged delegates, one of the moderates (Biden, Bloomberg, Buttigieg or Klobuchar) has 30 percent, while another moderate candidate has 25 percent, and Warren has 5 percent.

Would convention delegates try to deny Sanders the nomination when he had a fairly clear plurality? I don’t know. (Our forecast model doesn’t try to predict the outcome of a contested convention.) It would be a heck of a story to cover. Sanders certainly would have a decent shot, though. And if Sanders had a very clear plurality — say, 47 percent of the delegates, while the next-closest competitor had 28 percent — his chances would be stronger still.

As the rest of the field slowly consolidates, Sanders could gradually increase his vote share just enough to win a narrow majority. Even if the moderate lane narrowed to just one or two alternatives later on in the race — say, at some point in March or April — Sanders would still be in a pretty decent position. He would probably have a head start on the competition by having won a lot of delegates on Super Tuesday and in the first four states while the rest of the field sorted itself out. Contests up to and including Super Tuesday account for 38 percent of all pledged delegates, so this matters a lot.

Also, even if there is some upward resistance to Sanders’s numbers — more than there might be for the average candidate — it isn’t likely to be absolute resistance. Case in point: Sanders improved his support from 15 percent in national polls for much of last year to the low-to-mid 20s now. Without those gains, Sanders might be in the fairly difficult position that Warren now finds herself in, following third- or fourth-place finishes in Iowa and New Hampshire.

Making slow-but-steady polling gains was roughly the path that Trump followed in 2016 to win the Republican nomination, too. True, Trump had one major advantage that Sanders didn’t: the presence of winner-take-all states, especially later on in the race. (All Democratic states use proportional delegate allocation for candidates who clear the 15 percent threshold for getting any delegates.) Still, Trump gained ground later in the race once Republican voters realized that they faced a choice between Trump and a contested convention (which might nonetheless have resulted in his nomination). Democratic voters might act similarly. Maybe a voter would prefer Buttigieg to Sanders in the abstract, but if a Buttigieg win would require a contested convention while a Sanders win would not, she might feel differently.

Sanders could easily win a one-on-one race. Finally, suppose that we do wind up with a two-candidate race fairly soon; Biden loses South Carolina, for instance, and quits the race, and the large majority of delegates on Super Tuesday go to either Sanders or Bloomberg.

This seems to be the outcome that a lot of moderates I talk to are rooting for, but it could also fairly easily lead to a Sanders nomination. If Sanders’s main opponent was Bloomberg, for instance, he’d play perfectly into Sanders’s messaging about the corrosive influence of money on the democratic process. Bloomberg also has a lot of baggage that has been somewhat unscrutinized because of his late entry into the race. Against Buttigieg or Klobuchar, meanwhile, Sanders would probably have the more diverse coalition, and he’d also have an organizational advantage against someone like Klobuchar, who is only now starting to raise serious money.

I don’t know who would be favored in a head-to-head matchup between Sanders and another Democratic candidate, especially a resilient Biden, or if it somehow came down to Sanders and Warren. (As the YouGov polling shows, these are potentially tougher matchups for Sanders.) But the bottom line is this: Even if Sanders is far from the textbook nominee — and even if he’s likely to have some trouble winning new voters to his side — all of the other candidates have a lot of problems too. Sanders is in the strongest position for now, and he has a high floor of support that should win him delegates almost everywhere, while the rest of the field is a mess behind him. Ceiling or not, that’s why you’d rather be in his position than anyone else’s.

Politics Podcast: Post-New Hampshire, ‘No One’ Is On The Rise

By Galen Druke and Nate Silver and Galen Druke and Nate Silver












 












More: Apple Podcasts |
ESPN App |
RSS
| Embed


Embed Code

Since the New Hampshire primary, the chance that no Democrat wins a majority of the delegates in the presidential primary has increased, according to our forecast. In this edition of “Model Talk” on the FiveThirtyEight Politics podcast, Nate Silver and Galen Druke discuss the reasons for that and answer listener questions.

You can listen to the episode by clicking the “play” button in the audio player above or by downloading it in iTunes , the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen .

The FiveThirtyEight Politics podcast publishes Mondays and Thursdays. Help new listeners discover the show by leaving us a rating and review on iTunes . Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.

Politics Podcast: Sanders Won The New Hampshire Vote. So How Did ‘Klobucharge’ Win The Narrative?

By Galen Druke, Clare Malone, Nate Silver and Geoffrey Skelley, Galen Druke, Clare Malone, Nate Silver and Geoffrey Skelley, Galen Druke, Clare Malone, Nate Silver and Geoffrey Skelley and Galen Druke, Clare Malone, Nate Silver and Geoffrey Skelley












 












More: Apple Podcasts |
ESPN App |
RSS
| Embed


Embed Code

In a late-night installment of the FiveThirtyEight Politics podcast, the crew reacted to Sen. Bernie Sanders’s win in the New Hampshire primary. They also looked at the parts of the electorate that powered former South Bend, Indiana, Mayor Pete Buttigieg and Sen. Amy Klobuchar to strong second and third-place finishes, respectively, and asked what comes next for the various campaigns.

You can listen to the episode by clicking the “play” button in the audio player above or by downloading it in iTunes , the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen .

The FiveThirtyEight Politics podcast publishes Mondays and Thursdays. Help new listeners discover the show by leaving us a rating and review on iTunes . Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.

Sanders Is The Front-Runner After New Hampshire, And A Contested Convention Has Become More Likely

The Democratic primary is in a confusing state at the moment. And our forecast model is a little confused, also. It’s making a couple of assumptions about how the polls may react to New Hampshire that may not be entirely right. The model is also limited by the lack of polling in states that vote after New Hampshire, most notably Nevada and South Carolina. So we’d encourage you to take the model with a large grain of salt until some of that post-New Hampshire polling comes in.

But the two takeaways that the model feels most confident about are two things that I’m happy to vouch for:

Model takeaway No. 1: Sen. Bernie Sanders is the most likely person to win the Democratic nomination.
Model takeaway No. 2: The chance of there being no pledged delegate majority — which could potentially lead to a contested convention — is high and increasing.

I’m going to be relatively brief here as I’m writing this at 2 a.m. But let’s take the Sanders conclusion first. The model’s contention that he’s the closest thing to a front-runner we have in this race seems inescapable to me. Sanders won the popular vote in each of the first two states (and he may eventually win the state delegate equivalent vote in Iowa). He leads in national polls (having recently overtaken former Vice President Joe Biden). He has raised a ton of money. He polls fairly well in Nevada (or at least he did back when people bothered to poll it). And he has a reasonably diverse coalition that should net him at least some delegates in almost every state and congressional district.

There are also some negatives for Sanders. While he won New Hampshire — although pledged delegates were split evenly between him and and former South Bend, Indiana, Mayor Pete Buttigieg, which the model gives Buttigieg a tiny bit of credit for1 — his 25.7 percent of the vote there underperformed our projections by 2 to 3 percentage points. By all rights, New Hampshire ought to have been a fairly strong state for him (as Iowa should also have been). And although Sanders leads in national polls, he averages only 22 percent of the vote in them, unusually low for the national leader at this stage of the race. Between the slight underperformance in New Hampshire and a couple of mediocre polls coming in for Sanders while our model was frozen awaiting New Hampshire results, he actually fell slightly in the forecast from where he had been 24 hours earlier.

Still, Sanders’s 38 percent chance of winning a majority of pledged delegates is far better than that of any other Democrat. He also has a 52 percent chance of winning a plurality of pledged delegates. Even if this isn’t the strongest possible version of Sanders, he’s come far closer to actualizing his potential than anyone else in the field. Furthermore, the tactical considerations of the race are setting up well for Sanders: The moderate “lane” is still very crowded and perhaps getting even more crowded (it’s no longer just Biden and Buttigieg, but also Sen. Amy Klobuchar and former New York Mayor Michael Bloomberg!), and Sanders has pulled well ahead of Sen. Elizabeth Warren in the progressive lane.

But New Hampshire is also good news if you’re hoping for chaos. Our forecast has the chances that no one wins a majority of pledged delegates up to 33 percent, its highest figure yet, and roughly double what it was before Iowa.

Almost everything went well if you’re rooting for a contested convention. Sanders won, but with a smaller share of the vote than the model expected. Moreover, the second- and third-place candidates, Buttigieg and Klobuchar, may or may not be poised to take advantage of any post-New Hampshire surge they get, having begun the evening at just 10 percent and 4 percent, respectively, in national polls, and not having any obvious strength in Nevada or South Carolina. Meanwhile, the two candidates apart from Sanders who had seemed to have built the broadest national coalitions, Warren and Biden, did terribly in New Hampshire. (Although the race is so wide-open that they can’t entirely be counted out either — especially not Biden — at least not until we see some Nevada and South Carolina polling.) Meanwhile, Bloomberg continues to rise in polls, including having his first polling lead of the campaign in any state, in an Arkansas poll that came in while the model was frozen.

Now then, what about potential shortcomings of the model? First, as I mentioned, the lack of polling data in post-New Hampshire states is a problem. That data should begin to trickle in over the next couple of weeks (although note that Nevada never gets much polling).

Also, the model may not be guessing right about the bounces that will emerge from New Hampshire, especially for Klobuchar. It does give Klobuchar some credit for substantially beating expectations (which it defines as national polls adjusted for regional factors) in New Hampshire. But media coverage on election night practically treated Klobuchar — not Sanders — as the winner in New Hampshire (to an extent that probably deserves a little mockery).

Klobuchar needs a huge bounce given that she’s at only 4 percent in national polls. But there’s the possibility of a virtuous circle for Klobuchar, where she rises (perhaps slightly at first) in the polls, that rise begets more media coverage, which begets more endorsements and money raised, which begets more media coverage, which begets a further polling rise, and so forth. I don’t know how high that possibility is. It probably isn’t that high, and even if it is, there are lots of scenarios where Klobuchar rises to, say, 15 or 20 percent in polls and never gets much further than that. Still, because her media coverage will probably wind up being more favorable than the model is assuming, the chances of New Hampshire serving as an initial spark for her to rise into contending status are higher than it probably assumes. More thoughts on all of this — and hopefully more polls — coming later this week.

The FiveThirtyEight Politics Podcast reacts to the New Hampshire results

28 Scenarios For What Things Could Look Like After New Hampshire

Manchester, N.H. — After an extremely confusing week in the Democratic presidential primary, I have some bad news: New Hampshire might not resolve as much as you think.

That’s for two reasons. First, our study of past primaries finds that the bounces that result from New Hampshire are only about half as large as the ones that come out of Iowa. The bounces from Nevada and South Carolina tend to be even smaller. Instead, it’s an event like Super Tuesday that has the potential to produce a big bounce. Our research finds that the symbolic importance of early-state victories diminishes as the race goes on, and the actual number of delegates at stake matters more, with the media giving more attention to contests in high-population, delegate-rich states like California and Texas, which both vote on Super Tuesday. New Hampshire is still getting quite a lot of press coverage, obviously, but it also has just 24 delegates.

The second reason is that the leader in New Hampshire polls, Sen. Bernie Sanders, may not see his overall position fundamentally change by whatever happens tonight — barring an extremely unlikely outcome like finishing in fourth place.

On the one hand, a New Hampshire win is largely priced in to Sanders’s numbers. Since our forecast has Sanders winning New Hampshire about 2 out of 3 times, his current chances of winning a majority of pledged delegates nationwide ( 46 percent) already account for his probable win in New Hampshire. Furthermore, our model — which attempts to simulate how voters and the media will frame the outcome in New Hampshire — assumes the magnitude of a candidate’s bounce depends on “expectations.” (This is defined as national polls adjusted for regional factors.) And since Sanders has become the national polling front-runner — and New Hampshire neighbors his home state of Vermont — a win here is largely expected for him.

On the other hand, the primary field is such a mess that a loss here would probably be survivable for Sanders, especially if it is the second-place candidate in New Hampshire polls — former South Bend, Indiana, Mayor Pete Buttigieg — who ultimately wins. Buttigieg, even after Iowa, is still only at around 10 percent in national polls and has a lot of work to do to appeal to the more diverse electorates in Nevada and South Carolina. Buttigieg would certainly become a major contender for the nomination if he won New Hampshire, although it would still be a wide-open process, with Sanders and other candidates having a pretty decent shot.

Let’s look at some numbers from our model … and then I’ll go ahead and add a major caveat to everything that I said above.

First, here’s a list of scenarios based on the winner and his or her margin of victory. I’m only listing scenarios if they occurred at least 100 times in 10,000 simulations (or 1 percent of the time), which means this chart only includes scenarios that involve Sanders or Buttigieg winning. (Collectively, Sen. Elizabeth Warren, former Vice President Joe Biden and Sen. Amy Klobuchar have about a 2 percent chance of winning New Hampshire — stranger things have happened! — but no one scenario involving them came up more than 100 times.)

How New Hampshire could affect nomination odds, Part I

Based on winner and margin of victory, according to FiveThirtyEight’s primary forecast

chance of winning the majority of delegates overall


Winner▲▼


NH Margin▲▼


Biden▲▼


Sanders▲▼


Warren▲▼


Buttigieg▲▼


Other▲▼


None▲▼






Sanders
large
12%
57%
4%
1%
2%
24%


Sanders
medium
13
53
4
2
2
25


Sanders
narrow
15
48
5
2
2
27


Buttigieg
large
13
31
5
18
4
29


Buttigieg
medium
16
34
5
11
3
32


Buttigieg
narrow
15
33
5
10
4
33

Specifically, we’re defining a “narrow” win as anything less than 4 percentage points over the second-place candidate, a “medium” win as 4 to 12 percentage points, and a “large” win as more than 12 percentage points. Scenarios are only listed if they had at least a 1 percent chance of occurring.

A narrow Sanders win (specifically, by 4 percentage points or less) would basically preserve the status quo; he’d go from a 46 percent chance of winning a majority of pledged delegates to a 48 percent chance. Bigger Sanders wins would improve his position, perhaps making him more likely than not to win a majority of pledged delegates. In a landslide win (more than 12 percentage points), for instance, he’d have a 57 percent chance of getting that majority.

A Buttigieg win, on the other hand, would cut Sanders’s chances of winning a majority of pledged delegates down to about 1 in 3. That’s not great for Sanders, although it’s hardly a disaster; the bigger dangers to Sanders have always been a loss to Biden (who could leap back ahead of him in national polls) or Warren (who could overtake him in the progressive “lane”).

And if we flip the race and look at things from Buttigieg’s perspective rather than Sanders’s, New Hampshire matters quite a bit more. A loss here would put him on the ropes, with other candidates (Biden, former New York Mayor Michael Bloomberg, Klobuchar or Warren) having a claim to being the most viable rival to Sanders, especially as they are polling better than Buttigieg in upcoming states. And even with a win here, Buttigieg would need a fairly big bounce to become the national front-runner — bigger than the one he got for sort of halfway winning Iowa. A big, emphatic win could do a lot more for him than a narrow one.

However, a Buttigieg win in New Hampshire would also increase the likelihood of no one winning a majority of pledged delegates and possibly producing a contested convention. So if you’re rooting for chaos, the scenario you should probably be rooting for is a narrow Buttigieg win, one that would weaken Sanders but that wouldn’t necessarily be enough to establish Buttigieg as Sanders’s sole rival to the nomination.

As for Biden, there just isn’t a great scenario for him either way. Sanders is stronger overall, but Buttigieg is more in Biden’s lane. Warren would probably slightly prefer Buttigieg winning, although the difference is not as large as I would have thought. Keep in mind that there’s also a fair amount of overlap between Buttigieg and Warren voters, in that both candidates primarily rely on college-educated Democrats.

We can examine more detailed scenarios, but they don’t necessarily reveal all that much. Here’s a table that accounts for the second-place finisher in the race in addition to the winner’s victory margin.

How New Hampshire could affect nomination odds, Part II

Based on winner, margin of victory and second-place candidate, according to FiveThirtyEight’s primary forecast

chance of winning the majority of delegates overall


winner▲▼


NH margin▲▼


2nd▲▼


BID.▲▼


SAN.▲▼


WAR.▲▼


BUT.▲▼


Other▲▼


None▲▼






Sanders
large
Biden
13%
57%
4%
0%
3%
23%


Sanders
large
Buttigieg
12
56
4
1
2
24


Sanders
large
Warren
11
59
4
3
2
22


Sanders
medium
Biden
16
47
4
3
2
28


Sanders
medium
Buttigieg
13
53
4
2
2
25


Sanders
medium
Klobuchar
13
55
2
3
3
24


Sanders
medium
Warren
11
53
6
1
3
26


Sanders
narrow
Buttigieg
14
49
5
2
2
27


Sanders
narrow
Warren
16
41
10
1
3
29


Buttigieg
large
Sanders
13
31
6
16
5
29


Buttigieg
medium
Sanders
16
34
4
11
3
32


Buttigieg
narrow
Sanders
15
34
5
10
4
33

Specifically, we’re defining a “narrow” win as anything less than 4 percentage points over the second-place candidate, a “medium” win as 4 to 12 percentage points, and a “large” win as more than 12 percentage points. Scenarios are only listed if they had at least a 1 percent chance of occurring.

Based on our research, the order of finish beyond first doesn’t actually matter very much — unless the race is so close that the outcome is ambiguous (as it is in Iowa). However, the share of the vote a candidate gets does matter, and candidates get a higher vote share on average when they finish in second place. So Warren could considerably help her chances by beating her polls and finishing second, for instance.

Finally, here’s a table showing the post-New Hampshire odds based on the order of the top three finishers. Again, you don’t see big differences here based on who finishes in second and third:

How New Hampshire could affect nomination odds, Part III

Based on winner, second and third place candidate, according to FiveThirtyEight’s primary forecast

chance of winning the majority of delegates overall


winner▲▼


2nd▲▼


3rd▲▼


BID.▲▼


SAN.▲▼


WAR.▲▼


BUT.▲▼


Other▲▼


None▲▼






Sanders
Buttigieg
Biden
14%
52%
4%
2%
3%
25%


Sanders
Buttigieg
Klobuchar
13
53
5
2
2
26


Sanders
Buttigieg
Warren
12
53
5
2
2
26


Sanders
Biden
Buttigieg
16
49
4
2
3
25


Sanders
Klobuchar
Buttigieg
12
55
4
4
3
21


Sanders
Warren
Buttigieg
11
53
6
2
2
26


Buttigieg
Sanders
Biden
19
33
4
11
3
31


Buttigieg
Sanders
Klobuchar
11
36
6
12
3
33


Buttigieg
Sanders
Warren
14
34
6
11
4
32


Buttigieg
Warren
Sanders
16
30
7
14
0
33

Specifically, we’re defining a “narrow” win as anything less than 4 percentage points over the second-place candidate, a “medium” win as 4 to 12 percentage points, and a “large” win as more than 12 percentage points. Scenarios are only listed if they had at least a 1 percent chance of occurring.

But here comes that big, gaping caveat that I mentioned before: The media often comes up with inconsistent and even arbitrary framings regardless of the results. It’s largely the media’s reaction that dictates the bounce in early states, where there are few actual delegates at stake. And the media seems to care quite a bit about who finishes third in New Hampshire. Does that mean the media doesn’t take Sanders and Buttigieg as seriously as it should? Well, probably, yeah!

Nonetheless, the buzz among media elites and campaign professionals regarding Biden, Warren and Klobuchar could vary a lot based on the exact order of their finish, and that could affect the outcome in a number of ways in subsequent states, more so than our model assumes. Keep in mind that our model is just making a fairly crude educated guess about these scenarios. The only way to know for sure about how big the bounce will be is to wait for the polls, and the model will rapidly adjust once post-New Hampshire polls are published.

Live From Manchester: The Politics Podcast Previews The New Hampshire Primary

By Galen Druke, Clare Malone, Micah Cohen, Nate Silver, Sarah Frostenson and Geoffrey Skelley, Galen Druke, Clare Malone, Micah Cohen, Nate Silver, Sarah Frostenson and Geoffrey Skelley, Galen Druke, Clare Malone, Micah Cohen, Nate Silver, Sarah Frostenson and Geoffrey Skelley, Galen Druke, Clare Malone, Micah Cohen, Nate Silver, Sarah Frostenson and Geoffrey Skelley, Galen Druke, Clare Malone, Micah Cohen, Nate Silver, Sarah Frostenson and Geoffrey Skelley and Galen Druke, Clare Malone, Micah Cohen, Nate Silver, Sarah Frostenson and Geoffrey Skelley












 












More: Apple Podcasts |
ESPN App |
RSS
| Embed


Embed Code

At a live show in Manchester, Hew Hampshire, the FiveThirtyEight Politics podcast crew previewed what to expect from the New Hampshire primary. They also debated the likelihood of a contested convention and played a round of “Guess What Americans Think.”

You can listen to the episode by clicking the “play” button in the audio player above or by downloading it in iTunes , the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen .

The FiveThirtyEight Politics podcast publishes Mondays and Thursdays. Help new listeners discover the show by leaving us a rating and review on iTunes . Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.