Princeton Election Consortium

A first draft of electoral history. Since 2004

The cake bakes – is Rubio cooked?

February 27th, 2016, 5:05pm by Sam Wang

(delegate calculation corrected 2/28)

(See the comment by MAT on how both Democrats and Republicans benefit from a brokered Republican convention.)

Until recently, the most likely way to give someone other than Donald Trump a majority of delegates was to narrow the field of candidates drastically. But the Republicans didn’t do that in time for Super Tuesday (also see this look at the political twists and turns in the NYT). Now what?

As Marco Rubio and friends go on the attack against Trump,  is Rubio about to gain new life? Or is he cooked?

Super Tuesday is just three days away. Although the 12 states that vote are not particularly strong for Trump, they could conceivably give him a majority of the delegates awarded so far.

First…let’s note that some people seem to think Rubio is still in the mix to get a substantial number of delegates.

This breathless commentary suggests the feeling that candidates have risen and fallen a lot. That feeling is strongly contradicted by the data.

Candidate numbers have been anything but volatile. National surveys don’t tell us about individual states, but they do give us a barometer of public opinion. By this barometer, Donald Trump has been ahead of all other candidates since July 1 of last year. He has only accumulated support since then. And since late December, his median national support has stayed in the 35-40 percent range.

Trump’s trajectory is notably more stable than Mitt Romney’s in 2012. During the corresponding period, Romney was repeatedly overtaken by other candidates: Newt Gingrich, Rick Santorum, Herman Cain, and Rick Perry. Only in March did Romney assume a durable leading position.

Based on the graph above, it wouldn’t be surprising if Rubio’s attacks had little or no effect. Trump’s supporters seem to be quite dedicated. We will see what the coming wave of super PAC-funded ads can accomplish.

Even if there is movement, it probably won’t have much effect on Tuesday’s results. On that day, 12 states will vote to commit 624 delegates, or 25.2% of all convention delegates. Add that to the 133 (5.4%) delegates from the four early states of Iowa, New Hampshire, South Carolina, and Nevada, and that gets up over 30% of delegates. These layers of the cake are largely baked.

In early states, poll medians mostly predicted the early-state outcomes:

Year Trump polls actual Cruz polls actual Rubio polls actual
IA 26.5% 24.3% 23.5% 27.6% 18% 23.1%
NH 31% 35.3% 11% 11.7% 14% 10.6%
SC 33% 32.5% 18.5% 22.3% 20.5% 22.5%
NV 42% 45.9% 20% 21.4% 19% 23.9%

For these candidates, the average difference between polls and outcomes was 3.0 percentage points. Statistical variation would have predicted differences of just 1-2 percentage points, so there was some additional inaccuracy. Generally, all three candidates outperformed their polls: Trump by a median of 1.7%, Cruz by 2.6%, and Rubio by 3.4%. This might be because undecided voters committed at the last moment, or because of strategic voting by supporters of lower-tier candidates. Evangelical support is thought to be hard to capture in surveys, and those voters often don’t like Trump.

Whatever the reason, Cruz and Rubio have overperformed by an average of 1-2 percentage points relative to Trump. The only place where such a small difference could alter the #1 finisher is Minnesota, where one lonely poll shows a three-way tie (Rubio 23%, Cruz 21%, Trump 19%). Even that doesn’t matter since Minnesota will give them each a proportional share of delegates, which might not be altered.

In the rest of the states, Cruz is ahead in two (Texas and Arkansas) and Trump is in first place in eight: Alabama, Alaska, Georgia, Massachusetts, Oklahoma, Tennessee, Vermont, and Virginia. That would suggest some big headlines for Donald Trump on Wednesday.

Trump has attained this level of dominance with average support of only 31% in these states. That is 10 percentage points weaker than his national support. In contrast, Cruz and Rubio’s Super-Tuesday averages are within 1 percentage point of their national numbers. So Trump could do even better in the weeks ahead, especially as contests shift to winner-take-all races.

This statement from Costa glosses over the importance of the actual rules, which are pseudo-proportional. This, combined with continued division in the field of candidates, has serious implications.

Super-Tuesday support for Trump, Cruz, and Rubio add up to an average of only 69% of votes. John Kasich, Ben Carson, and undecideds make up the remaining 31%. As I have written, this division is a real problem for Republicans who oppose Trump. The reason has to do with minimum thresholds, written into state-level rules, to get any delegates at all.

Eight Super Tuesday states require a candidate to get at least 13 to 20 percent of the vote in order to get any statewide delegates. Cruz appears to be below threshold in Alabama, Georgia, and Vermont. Rubio appears to be below threshold in Alaska, Tennessee, Texas, and Vermont. And then there are Congressional district-level delegates, which account for about one-third of the delegates to be awarded. Here, the rules are even worse for runners-up: in addition to the threshold requirement, the tendency is to give two delegates to the top finisher and one delegate to the second-place finisher. At the district level, Cruz and Rubio are about equally likely to fall into the third-place abyss.

If I use current polls and feed them into the delegate selection rules using the same simulation I have previously coded, I get the following distribution of delegate totals for the 11 states that vote on Tuesday.

(Note: the following calculations were corrected on Feb. 28th to account for pledged party officials and to leave out Wyoming.)

As you can see, the number of delegates drops off fairly steeply for candidates with lower vote share. This dropoff is a direct consequence of the rules, which strongly favor the candidate with the largest fraction of the vote. Combined with early-state delegates, Trump would have a cumulative total of 369 out of 695 delegates, or 53%. So there is a very good chance that Donald Trump will end up with over 50% of cumulative delegates the end of Tuesday night.

(I also note that we now have a firm number: an average vote share of 30% on Tuesday is approximately enough to get Donald Trump 50% of cumulative delegates.)

The other estimated cumulative delegate totals are: Cruz 218, Rubio 104, Kasich 18, and Carson 11. Ted Cruz was already planning to stay in the race past Tuesday; these outcomes should ensure it.

After Super Tuesday, conditions get even tougher for Rubio. The next few winner-take-all states favor Trump. In Ohio (66 delegates), Trump holds a lead of only 3.5%, so attacks on Trump might flip Ohio – to Ohio Governor John Kasich. That would get Kasich almost caught up with Rubio in delegates. In Florida (99 delegates), Trump leads Rubio by 18 percentage points, making that race a probable humiliation to Rubio in his own home state. If indeed Kasich wins and Rubio loses, look for Kasich to stay in the race to the convention.

In light of this, the best hope for Rubio is to drive Trump’s support down by a few percentage points. It’s a desperate move, but it might delay Marco from being baked into the cake just yet.

Tags: 2016 Election · President

51 Comments so far ↓

  • JayBoy2k

    I saw this quote today from a Hugh Hewitt Article:
    ” Those who accept get tough questions in a respectful, even amiable, environment. We will know soon enough who is the standard bearer, so I think cries of foul or doom are premature. If Trump wins Texas, he knocks out Sen. Ted Cruz. If Trump wins Florida or Ohio, he knocks out Sen. Marco Rubio or Gov. John Kasich respectively. If Trump doesn’t win all three, we go to an open convention as no one will have the necessary 1,237 delegates to win the nomination. That isn’t punditry. It is math…..”

    Is Hugh correct? Trump could beat Rubio and Kasich in their own states on Mar 15 and would not be able to accumulate more than 1237 before the convention? I am not sure that Hugh is the expert on polls and delegate math projections. I believe Sam has that title.

    • 538 Refugee

      If Trump loses all three of those states that would show an erosion of his support going forward. I don’t think it means he mathematically would precluded from clinching though.

      Texas is still proportional so it probably wouldn’t be a total loss. Ohio is narrowing and winner take all so that could hurt in delegate count but keeping Kasich in the race might be good for Trump to siphon Rubio votes. Trump still up big in Florida though. Losing that would really hurt his chances in delegate count and show him as vulnerable to Rubio.

  • whatever next


    This graphic shows things well in Trump’s favor – but also quite finely balanced between a majority and a plurality of delegates for Trump – with Cruz actually closer to Trump than you might think.
    However, the yuge caveat is that nearly all polls are 100% before the 10th GOP debate, and subsequent lines of attack, outrage over KKK endorsement etc.

    So what difference will these things make?

    Well, for most Trump loyalists, probably not a lot – they stay with DT through thick and thin. However:

    1. A significant minority of Trump supporters are basically decent, if disgruntled, patriots who are far from being white supremacists – especially true of many genuine evangelicals. Although Trump himself has attained almost cult status amongst supporters, I surmise it is likely he will leak a couple of % over a combination of the KKK thing and wider doubts creeping in as to his suitability for President. Most existing supporters will stay, however, so keep him at 29% average across the Super Tues states.

    2. The sustained, apparently more convincing, attacks made on Donald, along with the KKK issue, will prevent late deciders breaking for Donald – suggest this could mean his overall support moves only a little, from 29% to 31% (2% added being 10% support of late-deciding 20%), compared to, perhaps, 10% for Rubio and 5% for Cruz – taking them to average 28% and 25% respectively. This might seem excessive – but remember late deciders in the first four contests – even Nevada – were shown to have broken AGAINST Trump, even without the current furore over his bankruptcies, hitting the little guy, KKK etc. Rubio has been the greatest beneficiary of late deciders up to now, even before he found his boxing gloves. Not everyone will like his new style – but he does manage to keep smiling with it all.

    3. DIFFERENTIAL TURNOUT – the anti-Trump tide hoped for, and perhaps partially in evidence in Iowa where a turnout smashing all previous records ultimately delivered AGAINST Trump, may be more in evidence now as the reality of a Trump nomination – and all it means for conservatism – kicks in. The attacks in recent days will only have reinforced this – whilst they may not keep many Trump supporters at home or get them switching their votes, it could re-energise the ‘non-Trump’ voters to come out in record numbers. I couldn’t begin to put feasible numbers to this – with no doubt widely varying effects statewide – but suggest it’s not completely unfeasible to see Rubio ultimately match Trump’s vote shares tomorrow.

    Remember the fallout from the pre-New Hampshire debate wasn’t fully seen until the NH polling booth itself.

    In conclusion, I’d wager Trump will definitely carry Vermont, Massachusetts and Arkansas (thanks to Mike H), but elsewhere is more to play for than the last polls have recorded. He’s a strong favourite in Georgia, and fairly evenly split opposition here will probably deliver him a win here – even if on a vote share lower than other states he fails to win. A 3rd-place finish in Texas won’t look good on him, although anti-Trump late deciders in Texas may coalesce round Cruz and keep Rubio in 3rd.

    I rest my case, and look forward to Tuesday night.

    • whatever next

      I missed one important point above, in Trump’s favor. This should have been a postscript to point 2.
      That is that the early voting will no doubt be shown to heavily lean Trump’s way when counted in Super Tues states, and will go some way to cancelling out points 2 and 3. Therefore Trump topping the overall vote must still be regarded as inevitable.

      Also, given that Trump was at 31% average – but a good few points below this in the most populous state by far – Texas – he obviously averages a little above 31% everywhere else and the others a little lower. This might seem like splitting hairs, but will make the difference between winning and losing a couple of states.

  • MAT

    Going back to Sam’s article, this seems to me one of those very rare times that the GOP & Democratic establishments are in alignment. Both benefit by a brokered GOP convention.

    The GOP establishment benefits in that it’s about their only hope at this point to avoid the Trump Train and all the ensuing damage that will do to the rank & file GOP establishment members in the future. So they need to do whatever they possibly can to keep Trump below the magic 50% mark going in and stacking the delegate decks as I described in a previous comment.

    On the other side, the Dem’s can only win with a brokered GOP convention – first, Trump is very, very unlikely to win the nomination if he doesn’t win on the first ballot. The convention will be filled with a majority of non-Trump loyal delegates (even some of those bound to him on a first ballot), many of them old hands that know the rules inside and out. Unless Trump can peel enough outsider delegates from whatever Cruz & Carson may be holding, he’s toast. I think in this case the GOP would prefer dealing with a sitting GOP Senator they may detest to Trump, and will make Cruz some sort of deal, like a VP slot, where he can be hid away in a corner and they don’t have to deal with him all that much.

    So when Trump loses on the 19th ballot or so, as is likely, his supporters pitch a royal fit as their revolution got stolen and stay home in anger in November.

    The Dems have a smooth running convention, where Bernie gets on stage and says how wonderful Hillary is. The contrast between competency and chaos couldn’t be clearer. In the best possible scenario Trump announces an independent bid, and both him and the GOP nominee go down in flames while spliting the vote.

    So I’m pulling for Trump @48%.

  • E L

    I have had this statement posted above my kitchen sink since Bush v. Gore in 2000: “History is the autobiography of a madman.” —Alexander Herzen.

  • bks

    Speaking of models:
    A Stony Brook University professor of political science named Helmut Norpoth, who developed an almost perfect statistical formula model that predicts who will become president, has declared that if Donald Trump wins the Republican nomination, he will almost certainly become President of the United States.

    • Sam Wang

      Maybe he knows this guy. Seriously, though, models like this are experiments to test his ideas, starting with past outcomes as evidence. In the true sense of prediction, I am sure his model has not predicted even one election yet. November would be the first test.

    • pechmerle

      Or these guys (from WSJ 11/7/2012):
      “Sheldon Jacobson, a computer scientist and elections forecaster at the University of Illinois, at Urbana-Champaign, told The Wall Street Journal he expected Obama to get 281 electoral votes despite tying Romney with 50% of the two-party popular vote, and losing in Colorado, Florida, North Carolina and Virginia. Steve Rigdon, Prof. Jacobson’s University of Illinois forecasting colleague, was the only forecaster surveyed to predict a Romney victory, basing that in large part on Rasmussen Reports polling data, which he said he considered “the most reliable.” Rigdon predicted that Romney would win 273 electoral votes to Obama’s 265.

      They aren’t out with anything on their website( yet this year.

    • Amitabh Lath

      Polisci guys plus a little curvefitting; not a good idea. Add Neural Nets and things go haywire. Boosted Decision Trees are the new goat entrails.

  • bks

    The Kasich strategy is not so far fetched:
    The scenario requires Ted Cruz to underperform on Tuesday, when seven Southern states, including his home state of Texas, go to the polls. It also requires Marco Rubio to lose his home state of Florida on March 15 and Kasich to win Ohio simultaneously. In that scenario, Weaver said Kasich would likely be the only candidate with a path to take on Trump.

  • mediaglyphic

    Trump seems to be cozying up to David Duke and the KKK. There seem to be some information that Fred Trump was arrested in a KKK rally when he was 21. Does this hurt Trump? Or are the people supporting trump comfortable with the KKK?

  • A New Jersey Farmer

    I can’t imagine the GOP wanting to have a contested convention and believe that they would swallow hard and coalesce around Trump. What would they gain if he has a plurality, but the party takes away the prize and gives it to Rubio because they think he’s more electable? Doing that would render him unelectable because then Trump’s supporters would demand an independent run or they would stay home.

    The Super Tuesday thresholds will matter, but the polling and Sam’s analysis point to Trump gaining a majority of delegates.

    • Tim

      By the time of the GOP convention, isn’t it too late for a potential third party candidate to get on ballots?

    • RJ

      There would be plenty to gain by “giving the prize” to Rubio as he is definitely more electable and he is a Republican unlike Trump. By Trump’s supporters staying home, Rubio stands to gain even further because many of them would have voted Democrat since that is where Trump really leans.

  • JayBoy2k

    Thanks for the reality check. Josh is delusional. and the Times article is excellent. I may have to reassess Trump. Anyone who is disliked by the Koch brothers, Romney, Rove, and McConnell can not be all bad.
    This schism in the GOP would seem to overcome whatever other advantages Republicans might have in this election. Brokered convention ?? What happens to the very large enthusiastic chunk of Trump supporters/voters? Do they just fall in line behind what ever nominee the party establishment dictated? and if McConnell follows up with Anti-Trump ads in the general…
    This would all seem to improve Hillary’s chances.

    No matter what.. a once in a lifetime political event. It will get very interesting in Mid to Late June.

  • hermit

    MAT, I can totally see Trump’s people missing many nuances, but from what I have read in the rules, the RNC delegates don’t really need to show up for the first ballot–their votes can be counted starting now, as there is no wiggle room.
    What’s more, don’t think for a second that Trump will bother with fighting for his own delegates at the convention if they slip away. He will go to court so fast Reince’s head will spin at record speed, and the whole convention will stop in its tracks. It’s gonna be huge.

    • MAT

      Interesting fact. The delegates aren’t really bound. The current rules expire at the ‘beginning’ of the convention, and the rules for governing the convention are adopted by majority vote of the delegates. If the GOP wanted to stage a coup, they could free the delegates (Rule 40 is also likely to be an issue, which says a nominee has to win at least 8 states). It would be a total sh**st**m, but if a majority of the delegates on the floor weren’t personally loyal to Trump, it’s possible.

      I first heard this idea floated on a radio interview by Michael Steele, who as a former chairman of the RNC presumably knows what he’s talking about.

    • Sam Wang

      Now, that is an interesting point, though the version I heard – from Ben Ginsberg, Bush/Cheney lawyer – is a little different. He said that only the first-ballot vote is bound. However, other votes are not: rules votes and credentials votes. He thought there would be lots of struggles at this stage.

      The likelihood of a convention coup turns on how many delegates are true loyalists to their pledged candidate. I bet TheGreenPapers and FHQ have something to say on this point. Also, as you say, it would be a total fiasco for the party to stage such a coup, especially if they don’t have an agreed-upon endpoint.

    • MAT

      But assuming they stay bound, it all comes down to if Trump is over 50% at the convention. If not, then that’s when all the shenanigans come into play.

      My dad was in the Jaycee’s for years, and they had brokered conventions every year to elect the national president. It was great fun. But the frontrunner on the first ballot who fell short almost never, ever was the final winner.

    • Olav Grinde

      But but but isn’t the GOP the party of law and order? You know, government shutdowns and that sort of thing. And gerrymandering cartographers who compete to see who can make the most concavity-pocked boundaries for congressional districts?

  • MAT

    One of my pet theories for awhile now has been that when the GOP establishment finally woke up to the fact that they were staring a Trump nomination in the face, they’d consider a 3rd party run of their own. Not to have any real hope to win the Presidency, but to motivate moderate R’s to turn out & vote for Senate candidates, rather than stay home because they can’t bring themselves to vote for either Trump or Clinton. Also, the smarter ones will realize that should Trump win the Presidency, the Republican Party won’t continue to exist in a form where establishment players are welcomed. So it’s a fight for political survival, they just need someone willing to take one for the team as a trade off for a few months of getting headlines. Someone like Newt Gingrich.

    Right on schedule this appeared in Politico today:

    • 538 Refugee

      The Republican party was declared dead in 2008. Then came 2010.

    • whatever next

      One problem with this theory is the idea that Trump would be bad for the establishment if elected. Yes, he would be more arrogant / independent than average – but Trump would be the King of horse-trading / backroom deals that his supporters allegedly loathe and blame for them getting left behind in society.

      He’s more political than ‘the politicians’ he claims to castigate – the difference is he believes that he can do it more effectively. Although, as long as he can fulfil his latest and greatest ambition ,he probably doesn’t really care.

      Odd paradox that his supporters don’t want to open their ears to, even with plenty of Trump past history to highlight his true unprincipled character.

    • Brian


      The problem for the GOP establishment with a Trump presidency is that it ruins their business.

      The Republican Party is a system where pols are elected running on issues that appeal to white working and professional classes and then go to Washington and implement the opposite policies. The plutocrat and big monopoly or government protected business elite pays for the campaigns as long as the policies are favorable to them.

      Most Republican establishment workers take jobs with Congress, think tanks, and Republican administrations for a while and eventually retire into protected industry PR or lobbyist jobs where they make big money for their connections. That money is much more than the total investment in campaigns and it represents the payment for the life work of Republican establishment insiders.

      Most Republican voters actually prefer the policies of the Democratic Party. But the Republican establishment is lying to them constantly about what they intend to do when they reach Washington. And that lying is successful. The establishment believes the lying succeeds because white identity politics is holding it together and nobody in sophisticated Washington could be foolish enough to implement white identity policies so it’s all safe.

      Take the last R debate. Cruz and Rubio promised to pass Gang of Eight amnesty and flood the country with triple the immigration there is now. But they disguised the promises with tough talk and the “path to citizenship” distinction. Cruz and Rubio are, in fact, to the left of Hillary on immigration and said so explicitly at the debate, though they covered it with dog whistles. They figure that by announcing white identity undertones they can scam the boorish working class voters they need.

      Trump poses a problem for them because the donor classes aren’t going to provide golden parachutes for policies that actually help the working class. Furthermore once the policy changes are in the open it’s going to be hard to keep up the lie that Republicans care about middle America. Trump will end the white identity dog whistle. Racism won’t be enough to win elections anymore. And even winning elections, Republicans won’t be able to turn around and sell policy to the craven cartel of highest big business bidders.

      That’s why the Republican establishment would much, much rather have Hillary than Trump.

  • bks

    If the delegates break as in Sam’s simulation the pundits and GOP mainstreamers are in for a rough Wednesday morning. Will anyone outside the Cruz camp plead with Rubio to drop out? Seems unthinkable.

    • Olav Grinde

      Breaking News: Reince Priebus has just held a press conference and apologized to the American for the Republican Party not having a qualified candidate in this election.
      John Kasich, Ben Carlson, Ted Cruz, Marco Rubio and Donald Trump have all announced their withdrawal from the race.

  • AJ

    What would happen if you applied a sort of stress test, assuming the Rubio/Cruz assault knocked off 10% of Trump’s support putting him at ~28/29% of the vote. Then redistributing that proportionately. How large an impact does that have on the delegate distribution?

  • Olav Grinde

    Rubio is toast – if the recipe calls for that, then he is in the cake.

  • Olav Grinde

    If the recipe calls for toast, then Rubio is in the cake

  • E L

    Since the most probable way to defeat Trump is at a contested convention, Trump’s opponents need to keep him from a majority of the delegates. Super Tuesday’s Cruz-Rubio delegate count in your calculations is slightly higher than Trump’s so Super Tuesday might be a good day for the anti-Trump forces who want a contested convention.

    • Olav Grinde

      If Cruz holds enough delegates to be king-maker, then I have a hard time seeing him throwing his weight behind Rubio.

    • E L

      Rubio to Cruz: Start fitting yourself for a Supreme Court robe, Ted.

    • Josh

      Yes, I think this is right. Maybe not Rubio, but certainly Cruz and Kasich have calculated that by staying in the race they’re far more likely to force a brokered convention where either A) they can play kingmaker or B) they could even horse trade their way to the nomination. I actually think, as Sam points out, that Rubio is in the *worst* position, as he’s the only candidate who doesn’t have a real chance of a state in the near term.

  • LondonYoung

    The year is 2016. I would think any decent analysis of the GOP nomination process starts with feeding poll numbers into codified rules for delegate allocation.

    But it seems that only PEC has done this.
    Is this true?
    If so, how can this be true?

    • Sam Wang

      I think The Upshot might have done this, not entirely sure.

    • MarkS

      Yes, the Upshot has done the math, complete with a very cool interactive graphic that plots delegate totals vs time, with sliders that control average winning percentage of each candidate before and after March 6. I presume the people behind that get paid for it, unlike Sam. But the businesspeople running most of the MSM believe that paying clueless pundits large sums brings a better return on investment than paying numerists, even though the numerists will work for much less.

    • bks

      Has Nate Silver’s secret sauce gone rancid?

    • LondonYoung

      that Silver didn’t even bother just seems so … weird

    • fledem

      I have done this . See below

      Sam is really missing why 3/1 will be volatile.

      Let’s take Georgia as an example:
      Right now the polling average there is:
      Trump 37, Rubio 21, Cruz 19.67
      Georgia has a 20% threshold both statewide and by CD.
      On those numbers Cruz is below the threshold.
      So you get the following:
      Trump 50
      Rubio 26
      In point of fact Cruz would likely be above 20 in some of the CD’s which would result in 1-1-1 split, but the key point here is Trump gets a significant majority of the delegates.

      By contrast, if you allocate the undecided evenly for Cruz, Rubio and Trump this becomes your polling average
      Trump 39
      Rubio 23
      Cruz 21.8
      As a result, this is your delegate split:
      Trump 30
      Rubio 23
      Cruz 23

      In this case Trump is far short of a majority of the GA delegates.

      This will be repeated in state after state on 3/1.

      This is a LOT of volatility in delegate result for not much difference in the statewide vote %.

      On 3/1 the story of the night IF Rubio and Cruz are able to get above the delegate thresholds will be the rising chance of a contested convention. This will be wrong in a way – as the winner take all states on 3/15 and later make this less likely than people thing. But the media will run with this story.

    • Sam Wang

      This is not the way I think about this problem at all. It’s easy to get lost in details like this, and lose sight of the big picture.

      Thresholds are important, and their overall effect contributes to the uncertainty in the total. It’s like my original 2004-2012 meta-analysis of the Presidential race – one has to aggregate everything to get a clear picture. The median estimate is the important thing. That’s why I don’t report individual states.

      To my thinking, the way to address this concern is to run the calculation repeatedly, as my script does, allowing natural variations from the polling median. That gives a distribution of outcomes, which can then be used to calculate statistical distributions of likely outcomes. These outcomes would contain a multitude of results that pass above and below the threshold. To be honest, I decided this was not worth the trouble to do. The overall uncertainty is about 10-20 delegates above or below the median that I present.

  • JWY

    Would it better for everybody in the GOP race to stay in it if the goal was to not beat Trump but to get to a brokered convention?

    • 538 Refugee

      I don’t think so if you read the way the it is broken down with the way the Republicans stacked the primary deck. It favors the front runner to encourage others to drop out. You would need a more even vote split for your scenario to work out.

    • Roke

      I think Kasich and Carson have to drop out for a brokered convention to happen. As of now I think that Trump is on the path to a majority at the convention. When winner-takes-all kicks in Trump should be near unstoppable.
      The convention should be fun to watch. Interesting times indeed!

    • Sam Wang

      Mostly agree, though depending on the next 1-2 weeks, Cruz or Rubio might also have to drop out to get a contested nomination at the convention.

      I try to avoid using the word “brokered” because there don’t really seem to be brokers any more – just the candidates and their proxies. Republican Party officials and actors don’t have that much influence at this point.

    • JWY

      But if you are in the non Trump camp and strategy says brokered convention is likelier with more candidates running, then you stay in regardless of the relative lack of delegates in your favor.

    • MAT

      The key to a ‘brokered’ convention are two things – first, not all ‘bound’ delegates are loyal to the candidates. Using NC as a sample, there are 6 bound delegates that are state GOP officials. Presumably as they are establishment by definition, they probably wouldn’t support Trump longer than they have to.

      2nd, other delegates ‘pledge’ themselves to a particular candidate prior to being selected as a convention delegate – but – this is where vetting is needed by the candidates, which is where an insider, who knows the rules can play games. If the party members were smart, they’d start declaring as Trump delegates. If a campaign doesn’t pay close attention, all sorts of mischief can occur down the road, and this seems like exactly the sort of detail the Trump camp might not pay attention to.

      Lastly, national delegates are usually elected by district & state level delegates, and these are almost always longtime party stalwarts who would be only too happy to elect a secret Rubio supporter as a Trump delegate. If I was a GOP operative, this is how I’d attack the problem.

      Disclosure- I ran the national convention delegate selection process for the NC Democratic Party in 2008.

    • Some Body

      JWY: That would be useful until March 14th (and assuming at least one, and preferably two, non-Trump gets over all delegate thresholds). But when winner-takes-all contests become frequent, you’d still need to narrow the opposition to a single candidate for optimal results.

    • JWY

      Thank you for the clarification. Clearly I haven’t been paying enough attention to prof Wang’s repeated explanations about the nonproportional nature of this process.

    • Josh

      Mostly just echoing every else–until state go winner-take-all I think that’s correct, but once we get deeper into March/April a couple of candidates would need to drop out so that somebody could win enough of the winner-take-all states to prevent Trump from hitting the 50% of delegates mark.