Skip to main content
ABC News
A State-By-State Roadmap For The Rest Of The Republican Primary

Three weeks ago, when we last took a detailed look at Donald Trump’s quest to win 1,237 delegates, his path looked rocky but endurable. The panel of eight experts FiveThirtyEight assembled projected Trump to wind up with 1,208 by the time California and four other states finished counting their votes on June 7, a number that would leave him tantalizingly close to clinching the Republican presidential nomination — probably close enough that he’d be able to get over the hump by persuading some uncommitted delegates to come his way before the convention.

Since then, Trump has gotten mostly bad news. Last week, he lost Wisconsin, which our panel originally considered a toss-up state leaning in Trump’s direction. Then this weekend, he was shut out of delegates at the Colorado state convention. He’s also had a couple of minor setbacks; Trump got no delegates in Utah when we thought he might get a few. All told, Trump would finish with 1,175 delegates if he hits our panel’s original estimate in the remaining states. That’s far enough away from 1,237 that winning over enough uncommitted voters will be challenging, especially given Trump’s lack of success at finding pro-Trump delegates.

But we also have a lot of new information at our disposal. Trump’s polling has held up well in the Northeast, and he has a good chance to beat the panel’s original projections in New York and Connecticut. On the flip side, his loss in Wisconsin bodes poorly for his performance in Indiana, another state we originally had as leaning toward Trump. So it’s time to revisit our projections, going through the remaining states one at a time. (Given how much delegate rules vary from state to state, there’s really no avoiding this level of detail, much to the bane of my editor.)

This time, it’ll just be me (Nate) instead of the panel, but I’ll aim to compensate by issuing three types of projections for each remaining state:

  • First, a deterministic projection, which lists the single most likely outcome in every state, in my view. Say there’s a state with 50 winner-take-all delegates, and we give Trump a 60 percent chance of winning it. Since the win is more likely than not, we’d score all 50 delegates in his column.
  • Second, a probabilistic projection, which hedges its bets. In the aforementioned example, Trump would have a 60 percent chance of winning 50 delegates and a 40 percent chance of winning no delegates, which works out to an expectation of 30 delegates.
  • Finally, a path-to-1,237 projection, which is not the same thing as Trump’s best-case scenario. Instead, it’s roughly what I consider his path of least resistance to finish with 1,237 delegates exactly, not counting uncommitted delegates.1 It’s up to you, dear reader, to think about how realistic it might be, or not — I’ll chime in with some thoughts at the end.

Where they’re available, I’ll use FiveThirtyEight’s poll-based forecasts to inform these projections. (Specifically, I’ll use an average of the “polls-only” and “polls-plus” versions of our model.) However, there isn’t a lot of recent polling in some states, so instead I’ll make inferences from Trump’s performance in nearby states, the state’s demographics, and so forth, tending to give a fair amount of deference to the panel’s original forecasts.

The first stop on our tour is Cheyenne, Wyoming, where Republicans will hold their state convention this weekend.

  1. Wyoming
  2. New York
  3. Connecticut
  4. Maryland
  5. Delaware
  6. Pennsylvania
  7. Rhode Island
  8. Indiana
  9. Nebraska
  10. West Virginia
  11. Oregon
  12. Washington
  13. New Jersey
  14. Montana
  15. South Dakota
  16. New Mexico
  17. California


Saturday, April 16
Bound delegates available: 14
Original Trump delegate projection: 1
Allocation method: State convention

Wyoming is one of three states, along with Colorado and North Dakota, where Republicans don’t hold a presidential preference vote. Instead, Republican voters elect delegates through a series of local and state conventions. So far, such conventions have gone terribly for Trump in other states. He got shut out of delegates in Colorado, and while delegates aren’t officially bound to anyone in North Dakota,2 the overwhelming majority chosen there are likely to prefer Ted Cruz.

Wyoming isn’t really going any better for Trump. It’s already halfway through its process, having elected 12 delegates — including just one for Trump — at county conventions on March 12. (We don’t count these March delegates in the totals listed here.) If anything, Trump was lucky to get even one delegate given how lopsided the vote was in most counties, meaning that a shutout is possible when the state convention picks 14 more delegates on Saturday. Revised Trump delegate projections: deterministic 0; probabilistic 1; path-to-1,237 1.

New York

Tuesday, April 19
Bound delegates available: 95
Original Trump delegate projection: 71
Allocation method:

  • 14 delegates awarded proportionally based on statewide vote, with 50 percent winner-take-all trigger.
  • Three delegates in each congressional district (81 total), split 2-1 between top two finishers, with 50 percent winner-take-all trigger.

Ahh, New York. So often the center of attention in everything else, now the center of attention in presidential politics, too. Politicians — they’re just like us! We get to see them awkwardly pandering to local customs and stuffing their face with fatty food.

But I digress. The headline is that Trump is very likely to beat our original projection of 71 delegates in New York. The mechanics of this are a little complicated to work out, however. I’m going to go through New York in some detail because many of the same principles will apply in California and other states.

The delegate rules themselves are tricky in New York. Fourteen delegates are awarded based on statewide results; they’re allocated proportionally,3 but they become winner-take-all if a candidate achieves at least 50 percent of the vote, a threshold that Trump exceeds (although barely in most cases) in almost every recent poll. The other 81 delegates are awarded, three at a time, in each of New York’s 27 congressional districts. In most cases a candidate gets two of three delegates for winning a district, but he gets all three if he wins the district with a majority of the vote.4

I’m not going to overthink Trump’s 50-percent-plus standing in statewide polling: He has yet to get a majority anywhere else, but New York is his home state. The more difficult question is how this translates to congressional districts; if Trump winds up with (for instance) 54 percent of the vote statewide, in how many districts will that translate to a majority?

You should be wary of quick-and-dirty answers from polls that break out the results by region. For one thing, the sample sizes on those regional breakouts are usually quite small. But also, the precise way the regions are defined is important; they may not tell us very much about the deep-blue districts we’re interested in.

One recent poll, for instance, broke out results from “New York City and Long Island” and showed Trump beating his statewide numbers there. Here’s the problem with that: New York City is overwhelmingly Democratic, while Long Island5 has a fair number of Republicans. (To put things in perspective: There are 11 times more Republicans in the Long Island-based 4th Congressional District than there are in the 15th Congressional District, in the South Bronx.) Among the registered Republicans who live in either Long Island or New York City, about 60 percent are in Long Island. Thus, Long Island will make up a majority of the sample in a poll of Republicans in New York and Long Island. However, Long Island has only four congressional districts — as compared with New York City’s 11 or 12 — and therefore a relatively small minority of the delegates at stake.

The ways around this are to take an extremely large sample so that you can actually pinpoint those blue districts, or to do some sort of statistical modeling. One pollster, Optimus, took the former approach, surveying more than 14,000 (!) New Yorkers in a “robocall” survey and providing a breakdown of results by congressional district. That poll had Trump finishing first in 26 of 27 districts (losing one Manhattan district to John Kasich) and getting a majority in 15 of 27, along with a majority of the vote (just barely) statewide. Such a result would earn Trump 82 of 95 delegates.

Note, however — especially since this has implications for other states — that the Optimus poll did not have Trump doing especially well in extremely Democratic districts in Manhattan, the Bronx, Brooklyn and Queens. Instead, he performed extremely well in Long Island, where the poll had him with 57 percent of the vote — along with the Republican-leaning NYC borough of Staten Island, where he had 68 percent.


The poll also showed a lot of variance in Trump’s vote from district to district in New York. Therefore, even if Trump gets 55 percent or 60 percent of the vote statewide, he’ll probably be under 50 percent in a handful of districts, preventing a clean sweep.

But unless Trump has trouble getting his voters to the polls because of New York’s strict registration laws, he doesn’t have a lot to worry about. He should beat our original estimate of 71 delegates unless he finishes below 50 percent statewide. Given the recent polls, our models calculate there’s less than a 10 percent chance of his failing to hit that majority threshold. Revised Trump delegate projections: deterministic 85; probabilistic 83; path-to-1,237 91.


Tuesday, April 26
Bound delegates available: 28
Original Trump delegate projection: 19
Allocation method:

  • 13 delegates awarded proportionally based on statewide vote, with 50 percent winner-take-all trigger.
  • Three delegates in each congressional district (15 total) awarded winner-take-all based on district vote.

Five Northeastern states vote on April 26. The most important is probably Pennsylvania, for reasons I’ll explain in a moment. But Pennsylvania’s delegate rules have a complicated twist, so let’s start with Connecticut, another state where Trump has a chance to beat our original projections.

Connecticut’s statewide delegates are awarded proportionally, but (as in New York) they become winner-take-all for a candidate getting a majority of the statewide vote. So, does Trump have a shot of getting to 50 percent? According to the only recent poll of the state, from Emerson College, he certainly does; that survey has him with 50 percent of the vote exactly. Our polling average in Connecticut rounds Trump down to 48 percent because Emerson’s polls have had a Trump-leaning house effect, but it’s obviously a close call as we await more polling.6

And there’s some further good news for Trump. Connecticut awards three winner-take-all delegates by congressional district. Because its congressional districts are fairly homogenous — especially relative to New York’s diverse ones — they’ll probably be swept by anyone who dominates the statewide vote, as Trump looks like he will.

In some sense, those results would speak to the primacy of geography over demography in the GOP primary. Demographically, Connecticut is a lot more white collar than the areas where Trump has excelled. But Trump got almost 50 percent of the vote in neighboring Massachusetts, and it looks like he’ll get there in New York also. Revised Trump delegate projections: deterministic 23; probabilistic 24; path-to-1,237 28.

Listen to the latest episode of the FiveThirtyEight politics podcast.



Tuesday, April 26
Bound delegates available: 38
Original Trump delegate projection: 31
Allocation method:

  • 14 delegates awarded winner-take-all based on statewide vote.
  • Three delegates in each congressional district (24 total) awarded winner-take-all based on district vote.

Trump is an 86 percent favorite to win Maryland according to our polls-only forecast and has a 71 percent chance according to polls-plus. That range of odds sounds about right to me. Trump did not do especially well in neighboring Virginia, where he won with only 35 percent of the vote. But whereas Marco Rubio emerged as the principal alternative to Trump in Virginia, it’s not clear whether Kasich or Cruz is the more viable non-Trump in Maryland; they’re roughly tied for second place in our polling average. That could make it hard for the anti-Trump vote to consolidate.

Kasich and Cruz could have more success at denying Trump delegates at the congressional district level. (Maryland awards three delegates to the winner of each district.) Trump did poorly in the Washington, D.C., suburbs on Super Tuesday — winning, respectively, just 22 percent, 26 percent and 29 percent of the vote in Virginia’s 8th, 11th and 10th congressional districts. Kasich is perhaps an outright favorite over Trump in Maryland’s suburban 4th Congressional District, with the 5th and 3rd potentially competitive. Cruz is probably Trump’s main competition in the 6th Congressional District in Maryland’s Appalachian panhandle and the 1st Congressional District on its Eastern Shore, although Trump will probably win both — perhaps by huge margins, given his performance in similar areas in Virginia.

With such diverse districts, a clean delegate sweep for Trump would be hard to pull off. Winning six of eight districts (along with the statewide vote) would constitute a good day for him and would win him 32 delegates, just slightly ahead of our original estimate of 31. In the absence of detailed polling, I’d probably bet on Trump’s getting five of the eight districts instead, which would be worth 29 delegates. However, there’s downside for Trump if he loses the statewide vote, which can’t totally be ruled out, according to the polls. Revised Trump delegate projections: deterministic 29; probabilistic 27; path-to-1,237 32.


Tuesday, April 26
Bound delegates available: 16
Original Trump delegate projection: 15
Allocation method: Winner-take-all based on statewide vote.

One more simple state before we get to Pennsylvania. Delaware awards 16 delegates to the statewide primary winner. The words “we could really use a poll of Delaware” have apparently never been written before on the internet, according to Google. But, we could really use a poll of Delaware! Until then, I’ll assume Trump is favored based on his polling leads in Pennsylvania and Maryland. Delaware’s votes are surprisingly high-leverage — based on the ratio of voters to delegates, a Republican vote in Delaware is worth about four times more than one in Florida, which voted on March 15. So it will be interesting to see if one of the candidates — perhaps Kasich, who has some endorsements from local Republican leaders — will make a point of campaigning in the state. Revised Trump delegate projections: deterministic 16; probabilistic 13; path-to-1,237 16.


Tuesday, April 26
Bound delegates available: 17
Original Trump delegate projection: 16
Allocation method:

  • 17 bound delegates awarded winner-take-all based on statewide vote.
  • Three unbound delegates from each congressional district (54 total) are directly elected on the ballot.

Why does Pennsylvania only have 17 delegates? It doesn’t; it has 71. But only 17 of them are bound by the results of the primary; they’re awarded to the statewide winner. That winner is likely to be Trump, who is an 83 percent favorite according to our polls-plus forecast and 94 percent according to polls-only.

Those probabilities conceal a few differences in the polls, however, with recent surveys showing Trump getting anywhere from 33 percent to 48 percent of the vote. Toward the lower end of that range, Trump is very beatable; toward the upper end, he’s almost certainly not. One thing that helps him, however, is that there’s not a clear second-place candidate, with Cruz and Kasich (who was born in Pennsylvania) tied at 24 percent in our polling average. The results could wind up looking something like Illinois, where Trump won fairly easily with 39 percent of the vote without a clear challenger. Revised Trump delegate projections (bound delegates only): deterministic 17; probabilistic 15; path-to-1,237 17.

The stakes in Pennsylvania are higher than the 17 bound delegates imply, however. Voters will also elect 54 unbound delegates — three from each congressional district — directly on the ballot. See here for a sample of what this looks like: Unlike in other states where delegates are directly elected, such as Illinois and West Virginia, there’s no guidance on the primary ballot as to which candidates those delegates might support.

However, slightly more than half of Pennsylvania’s unbound delegates said they’d support the candidate who wins the popular vote statewide or in their districts, according to a survey conducted by the Pittsburgh Tribune-Review. True, those delegates could change their mind later, with an April primary being a fairly distant memory in the event of a July contested convention. But particularly if Trump (or some other candidate) wins the state emphatically, he could get a couple of dozen unbound Pennsylvania delegates to go along for the ride. At the very least, it’s worth thinking about the bloc of Pennsylvania delegates separately from other categories of unbound delegates; they’ll potentially be more amenable to Trump and are an underrated means by which he might get to 1,237 delegates if he pulls up a bit short after California.

Another question is whether any of the campaigns are capable of directing voters toward their preferred unbound delegates through their ground games, such as by circulating lists in advance of the primary via door fliers or social media. Cruz has pulled off analogous tactics at state conventions, but those involved hundreds of highly informed voters in a captive environment, instead of what should be in excess of a million voters in a primary.

Rhode Island

Tuesday, April 26
Bound delegates available: 19
Original Trump delegate projection: 10
Allocation method:

  • 13 delegates awarded proportionally based on statewide vote.
  • Three delegates in each congressional district (six total) awarded proportionally; they’re split 1-1-1 between the top three finishers unless candidates receive more than two-thirds or less than 10 percent of the vote.

Rhode Island, the final April 26 state, is one of the most proportional on the calendar — unluckily for Trump since its working-class demographics, which resemble some of the areas in Massachusetts where he got a majority of the vote, are strong for him. Three delegates from each of its congressional districts will be split 1-1-1 unless one of the candidates finishes with more than two-thirds of the vote or one of the candidates finishes with less than 10 percent of the vote. Because of this quirk, it could be hard for Trump to meet our original target of 10 delegates even if he has about half the vote statewide, although he should come close. Revised Trump delegate projections: deterministic 9; probabilistic 9; path-to-1,237 10.


Tuesday, May 3
Bound delegates available: 57
Original Trump delegate projection: 37
Allocation method:

  • 30 delegates awarded winner-take-all based on statewide vote.
  • Three delegates in each congressional district (27 total) awarded winner-take-all based on district vote.

Indiana, usually an afterthought in electoral politics, played an important role in the 2008 Democratic primary. The state could be a pivot point again this year. It has a date to itself on the primary calendar on May 3 and may end Trump’s winning streak as the Republican contest moves out of the Northeast. Just as important, it has an aggressive delegate allocation method, with 30 delegates awarded to the statewide winner and an additional 27 to the winners of each of nine congressional districts. After California, in fact, Indiana is probably the most important state remaining.

Last month, our panel gave Trump an average of 37 delegates in Indiana, which implied that he’s the favorite to win there. I don’t think I can agree with that after Wisconsin, however. The states are relatively similar demographically. In Indiana, as in Wisconsin, Trump doesn’t have much support from statewide elected officials. Moreover, the Midwest as a whole has been a middling region for Trump. Earlier in the calendar, he got away with some wins in the Midwest with a vote share in the mid-to-high 30s. Now that the field has winnowed and Republican voters have learned to vote tactically, he’ll often need 40 percent of the vote to win a state instead.

Still, some caution is in order. There’s been no polling at all in Indiana. Perhaps Trump can hope that his momentum from April 26 will carry him to victory, or that Kasich will drain a few votes from Cruz in counties that border Ohio. My deterministic projection has Trump losing — although salvaging a few congressional districts — while the probabilistic one is more equivocal. The path-to-1,237 projection has Trump winning, almost out of necessity, because it will be hard for him to carve out a path to 1,237 delegates without the Hoosier State. Revised Trump delegate projections: deterministic 9; probabilistic 22; path-to-1,237 48.


Tuesday, May 10
Bound delegates available: 36
Original Trump delegate projection: 1
Allocation method: Winner-take-all based on statewide vote.

Nebraska has every appearance of being a strong Cruz state. He’s performed well everywhere else in the vicinity, most notably in Kansas, where he beat Trump by 24 percentage points. Yes, most of those earlier results were caucuses — Nebraska holds a primary — but we expect Nebraska to take its place as part of an anti-Trump wall of states that cuts through the prairie. Revised Trump delegate projections: deterministic 0; probabilistic 4; path-to-1,237 0.

West Virginia

Tuesday, May 10
Bound delegates available: 34
Original Trump delegate projection: 33
Allocation method:

  • Three delegates awarded winner-take-all based on statewide vote.
  • 22 statewide delegates, along with three delegates in each congressional district (nine total), are directly elected on the ballot. These delegates may be bound or unbound depending on how they list themselves on the ballot.

Trump has performed well in counties bordering West Virginia. And the economy of the region, heavily affected by outsourced jobs, bodes well for Trump. This is another state where Trump has a good chance to get at least 50 percent of the vote.

The catch is that only three delegates are awarded in West Virginia based on the popular vote. The other 31 are directly elected, in a process that somewhat resembles Pennsylvania’s but with some important differences that increase the potential for chaos.

One difference is that delegates may choose to affiliate themselves with a candidate; if so, the affiliation is listed on the ballot, and the delegates are bound to that candidate at the convention. Other delegates may run and be elected as uncommitted, however.

Another difference is that the ballot is much more disorganized. Voters have to pick not only three delegates from their congressional district but also 22 statewide delegates from a list of 220 candidates!

It’s hard to know how much this might hurt Trump in a state that should otherwise be very good for him, but I suspect that our panel — which originally gave Trump 33 of 34 delegates — may have underestimated the risk. In West Virginia’s 2012 Republican primary, held at a point when the GOP race was no longer competitive, Mitt Romney won 70 percent of the vote. Nevertheless, five of his delegates lost out to uncommitted delegates and another two lost to supporters of Rick Santorum, who had suspended his campaign weeks earlier. There was also a massive undervote, with only about half as many votes cast for at-large delegates as permitted.

A well-organized candidate might be able to take advantage of this in various ways. For instance, he could recruit delegates whose names appeared higher in alphabetical order or were more ethnically appealing to West Virginia’s predominantly white electorate. Or the candidate might seek to circulate delegate lists to their voters7 or at least remind them not to undervote the ballot. Anything that requires organization poses a risk to Trump, and while I could see him nearly sweeping the delegates in West Virginia, I could also imagine a disaster where he bleeds a dozen or more delegates despite winning the popular vote. Revised Trump delegate projections: deterministic 29; probabilistic 26; path-to-1,237 33.


Tuesday, May 17
Bound delegates available: 28
Original Trump delegate projection: 12
Allocation method: Proportional based on statewide vote.

All right, a simple one for a change. I don’t expect Trump to do well in Oregon — the demographic models we originally drew up in January and February rated it as one of his worst states. But there aren’t a lot of delegates at stake, and they’re allocated highly proportionally. So Oregon just doesn’t matter much. It is interesting mostly insofar as it could give us a few hints about how higher-stakes Washington and California will vote later on. Revised Trump delegate projections: deterministic 11; probabilistic 11; path-to-1,237 12.


Tuesday, May 24
Bound delegates available: 44
Original Trump delegate projection: 17
Allocation method:

  • 14 delegates awarded proportionally based on statewide vote.
  • Three delegates in each congressional district (30 total) split 2-1 between top two finishers, with 50 percent winner-take-all trigger.8

Washington is not quite as proportional as it’s been billed. Its 14 statewide delegates are allocated basically proportionally.9 But it uses the New York State method for allocating its 30 congressional district delegates, generally splitting them 2-1 in favor of the district winner but giving all three to a candidate who wins 50 percent of the vote or more.

As I mentioned above, demographic models don’t expect Trump to perform well in the Pacific Northwest, but we don’t have many polls, or many results from neighboring states, to confirm that. The safest bet seems to be that given Cruz’s dominance in Idaho panhandle counties that border Washington, Eastern Washington will be problematic for Trump, with Cruz possibly achieving majorities in the 5th and 4th congressional districts. The rest of the state is harder to figure out, but I’m hedging slightly to the downside versus our initial projections. Revised Trump delegate projections: deterministic 15; probabilistic 15; path-to-1,237 18.

New Jersey

Tuesday, June 7
Bound delegates available: 51
Original Trump delegate projection: 51
Allocation method: Winner-take-all based on statewide vote.

Five states vote on June 7, when the big prize is California. I’m going to go at lightning-round speed through the other four; for various reasons, these states aren’t a major source of uncertainty.

New Jersey’s 51 winner-take-all-delegates, for instance, are almost sure to go for Trump, who was well ahead in polls there even before he was endorsed by Gov. Chris Christie.10 Although the New Jersey polling is a bit out of date, Trump’s strong numbers in New York and Connecticut polls bolster our confidence that he’ll win in the Garden State. Revised Trump delegate projections: deterministic 51; probabilistic 48; path-to-1,237 51.


Tuesday, June 7
Bound delegates available: 27
Original Trump delegate projection: 0
Allocation method: Winner-take-all based on statewide vote.

Montana, by contrast, looks like a safe Cruz state, both according to our demographic model and based on Cruz’s dominant performance in nearby states. Revised Trump delegate projections: deterministic 0; probabilistic 3; path-to-1,237 0.

South Dakota

Tuesday, June 7
Bound delegates available: 29
Original Trump delegate projection: 0
Allocation method: Winner-take-all based on statewide vote.

The same goes for South Dakota, another winner-take-all state where Cruz is a clear favorite. It’s not that it’s impossible to imagine Trump winning South Dakota or Montana — those contests are two months away, and a lot could change. But if things have gotten so good for Trump that he wins South Dakota or so bad for him that he loses New Jersey, the outcome of the Republican contest should be fairly clear by that point. States like these aren’t tipping-point states, in other words. Revised Trump delegate projections: deterministic 0; probabilistic 3; path-to-1,237 0.

New Mexico

Tuesday, June 7
Bound delegates available: 24
Original Trump delegate projection: 10
Allocation method: Proportional based on statewide vote.

New Mexico has only 24 delegates, and they’re allocated proportionally, the only proviso being that a candidate needs at least 15 percent of the statewide vote to qualify. A February poll found Trump running behind his national numbers in New Mexico and trailing Cruz, and our demographic model suggests that New Mexico could be a poor state for Trump. But there isn’t enough at stake for that to matter much. Revised Trump delegate projections: deterministic 9; probabilistic 9; path-to-1,237 10.


Tuesday, June 7
Bound delegates available: 172
Original Trump delegate projection: 93
Allocation method:

  • 13 delegates awarded winner-take-all based on statewide vote.
  • Three delegates in each congressional district (159 total) awarded winner-take-all based on district vote.

I suspect the political commentariat hasn’t fully woken up to how monumental a finish California will be. Even if Trump is going gangbusters and meeting or exceeding his path-to-1,237 projections, a poor performance in California could leave him well short of 1,237 delegates. Or if he has a disappointing May, he could salvage a chance at clinching the nomination with a strong finish there.

The math in California is simple enough; 13 delegates go to the statewide winner, but the real prizes are congressional districts, which are worth three delegates each, for 159 total. Given California’s political and demographic diversity, it’s not wrong to think of the state as consisting of 53 micro-primaries. It’s daunting to get a handle on, but this chart might help. It breaks the congressional districts down into some broad categories based on their regions11 and how Democratic or Republican they are:12


As in New York, there are a lot of votes at stake in blue districts. There are nine districts — worth 27 delegates — available in the San Francisco Bay area, all of which are very Democratic. An additional 14 districts, worth 42 delegates, are up for grabs in Los Angeles County, where only 20 percent of eligible voters are registered as Republicans. There are also 10 strongly Republican or Republican-leaning districts in California, however, including several very Republican districts in Orange County.

Who’s going to do well where? I don’t think we have enough data to speak to that reliably. The highly rated Field Poll had Cruz running ahead of Trump in Los Angeles County despite trailing him statewide — big news if true, but as I said about New York, I’ll need larger and more precise samples before coming to very many conclusions. There are also differences of opinion about whether highly blue districts tend to favor Trump overall, disfavor him or something else. Personally, I’m of the view that we don’t really know very much about this and that the rules we learn in one state may not be applicable in others. There’s probably even an element of randomness given that district lines were mostly drawn with general elections in mind and may cut at odd angles across Republican groups of various kinds.

The top-level view in California isn’t all that helpful either, however. Trump leads in our polling average, with 36.5 percent of the vote, but Cruz isn’t far behind, at 30.4 percent. There are a lot of undecided voters, perhaps in part because Rubio voters, who were once fairly plentiful in California, are looking for a new candidate. While our polls-only forecast has Trump with a 65 percent chance of winning California, polls-plus has Cruz favored instead, figuring that Trump is running below his national polls in California (a bearish indicator) and that Cruz has almost two months to figure out how to pull ahead.

Speaking of which, California represents a potential data-mining opportunity for Cruz. Trying to track down Republican voters in East Los Angeles or Nancy Pelosi’s district in California is the equivalent of looking for needles in haystacks. Could his superior data and turnout operations be worth two extra districts? Five? Eight? It’s a factor that could swing the Republican race. For the time being, I’m splitting the difference between our two polling-based models and giving Trump about half the California delegates — close to where we had the state originally. Revised Trump delegate projections: deterministic 94; probabilistic 88; path-to-1,237 112.

So … drumroll please … here’s how our various estimates sum up for Trump:

April 16 Wyoming 1 0 1 1
April 19 New York 71 85 83 91
April 26 Maryland 31 29 27 32
Connecticut 19 23 24 28
Rhode Island 10 9 9 10
Pennsylvania 16 17 15 17
Delaware 15 16 13 16
May 3 Indiana 37 9 22 48
May 10 Nebraska 1 0 4 0
West Virginia 33 29 26 33
May 17 Oregon 12 11 11 12
May 24 Washington 17 15 15 18
June 7 California 93 94 88 112
New Jersey 51 51 48 51
South Dakota 0 0 3 0
Montana 0 0 3 0
New Mexico 10 9 9 10
Projected totals 417 397 401 479
Already won 758 758 758 758
Total delegates 1,175 1,155 1,159 1,237

Although our panel’s original estimates had Trump finishing with 1,175 pledged delegates, my revised deterministic projections have him at 1,155, and the probabilistic version has him at 1,159. I wouldn’t make a huge deal of the differences given the considerable uncertainty in the race, however. Basically, flipping Indiana from a probable win to a probable loss outweighs the gains I have Trump making relative to our original projections in New York and Connecticut. In other states, the differences from the original projections are minor.

At the same time, the path-to-1,237 scenario doesn’t look all that far-fetched — certainly not as compared with, say, Bernie Sanders’s quixotic path to catch Hillary Clinton in pledged delegates. Our path-to-1,237 path has Trump sweeping almost everything in the Northeast, winning Indiana and winning California by a solid-but-not-spectacular margin. I wouldn’t bet on that parlay at even odds, but it’s far from impossible. There’s also a reasonable variation; Trump could win slightly fewer delegates than I’m expecting in New York and Connecticut but make them up with a bigger win in California. So we’re not yet at the point where absolutely everything has to go right for Trump to clinch 1,237 delegates after California; although he can’t afford major setbacks such as losing Indiana or Maryland.

Keep in mind, however, that the question of whether Trump will get 1,237 bound delegates by California is not the same as the question of whether he’ll win the Republican nomination. If he’s close, Trump could still get some uncommitted delegates to come along with him — especially some of the 54 from Pennsylvania if he wins that state. He could also win on the second or later ballot in Cleveland, although I wouldn’t want to count on that if I were in Trump’s shoes.


  1. The way I’ve drawn this up, Trump’s path-to-1,237 projection is also deterministic; I don’t give him partial credit for winning states or congressional districts where he has an outside shot.

  2. Note that the delegates chosen in Colorado and Wyoming can be bound to candidates on the first ballot.

  3. Among candidates receiving at least 20 percent of the vote.

  4. Or if no other candidate reaches 20 percent, although that’s unlikely to happen.

  5. Meaning Suffolk and Nassau counties — not Brooklyn and Queens.

  6. Trump could also get a couple of extra delegates by holding Ted Cruz below 20 percent, the minimum to qualify for statewide delegates. My deterministic scenario in Connecticut assumes that Trump finishes just under 50 percent but that Cruz is held under 20 percent statewide.

  7. This can matter since there’s the risk of a candidate’s delegates cannibalizing votes from other delegates for the same candidate. Cruz has 36 at-large delegates listed on the West Virginia ballot, and Trump has 31, arguably giving voters too many to pick from. Kasich has too few statewide delegates (just 10) to wind up with a full slate, by contrast.

  8. Districts in Washington also become winner-take-all if only one candidate clears 20 percent — very unlikely in a three-person race.

  9. A quirk is that a candidate receiving less than 20 percent of the vote doesn’t receive statewide delegates; instead, the delegates they would have received become uncommitted.

  10. Christie had a 68 percent approval rating among New Jersey Republicans in a Quinnipiac University poll in November, even though he’s unpopular with voters in New Jersey overall.

  11. There are dozens of ways to classify the regions of California; this is my home-brewed version.

  12. Based on their Partisan Voting Index.

Nate Silver is the founder and editor in chief of FiveThirtyEight.