Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

In South Carolina, Trump could win all 50 Republican delegates

February 16th, 2016, 9:45am by Sam Wang

(Updated February 17th to include post-debate data. The conclusions are unchanged.)

The Republican primary in South Carolina is this Saturday, February 20th. Donald Trump continues to lead the pack, with 36 percent support in four post-N.H. surveys 34 percent support in two surveys done after last Saturday’s debate. His closest competitor is Ted Cruz, Marco Rubio, or John Kasich, depending on the survey. Trump’s median lead over the second-place finisher is 18.5 17 percentage points.

South Carolina’s Republican convention delegates are assigned on a winner-take-all basis, either at the state level (29 delegates) or by Congressional district (21 delegates in 7 districts, 3 delegates each). As a general rule, Congressional districts within a state behave similarly in a primary. Consistent with this, in 2012 Congressional Districts 2 through 7 showed an average margin of 14.7% between the first-place candidate (Newt Gingrich) and the second-place candidate (Mitt Romney), with a standard deviation of only 5.1%. The outlier was District 1, which includes Charleston, and which Romney won by a 2.9% margin.

District 1 is probably biased toward the less-evangelical candidate (such as Romney in 2012). In this year’s field, Trump, Kasich, and Bush should split the non-evangelical vote, and so District 1 might not differ that much from the six other districts. It is distinctly possible that Trump will win all seven districts – and all 50 of South Carolina’s convention delegates. That would bring Trump to a total of 67 out of 103 delegates awarded so far, or 65%.

After South Carolina, Nevada votes on February 23rd. Nevada awards its 30 delegates with genuine proportionality, so Trump should get about 10 delegates there. That would put him at 58% of delegates heading into Super Tuesday.

On the same day as the GOP primary in South Carolina, Democrats vote in Nevada. However, it’s a hard state to poll and so there is no good data. Looking ahead, the Democratic primary in South Carolina is on February 27th. In five recent polls taken since the New Hampshire primary, Hillary Clinton leads Bernie Sanders by a median of 19 percentage points. If this holds up, she would win with close to 60 percent of the vote. Like most Democratic primaries, South Carolina’s delegates are awarded proportionally, not counting superdelegates. So she should make up all the ground lost in New Hampshire, plus a few extra delegates.

Tags: 2016 Election · President

65 Comments so far ↓

  • GEinNY

    Who should Dems be “rooting” for? I wouldn’t mind if Trump or Cruz won the nomination. I would be less happy if Rubio or Kasich won.

    • Anthony Shifflett

      In that case, root for Cruz. He’ll be a disaster in the fall…

    • Matt McIrvin

      Depends on whether you’re thinking in terms of guaranteeing a loss, or in terms of minimizing harm in case of a win.

      I personally think “guarantee a loss” is a risky situation, since anyone who gets a major-party nomination can likely end up elected President in the event of some unexpected calamity.

    • Sam Wang

      GEinNY could root for Kasich, whose presence in the race helps keep the field divided. A divided field favors the nomirnation of Trump, an event that GEinNY favors.

    • bks

      Put me down for wanting a Trump sweep in SC. I’m hoping it will force the others to start saying things like, 9/11 happened on Bush’s watch and Planned Parenthood did a lot of good for women’s health. You know, true things.

    • Olav Grinde

      BKS, isn’t that a bit extreme? Calling for truth-in-politics??

      I thought fact-checking had long since gone out of fashion – and on the rare occasion when it is done, it seems relegated to a peripheral article, rather than front-page claims that “Candidate XX lied!”

    • Owen

      Matt McIrvin, exactly. Even Rubio could win if Wall Street collapses in October or the Iran deal falls apart in an ugly way.

      So it’s best to root for the least bad Republican, which is obviously Trump. Problem with that is if Trump mows Hillary down in the debates in Autumn.

      But hoping for Rubio just because he’s such an obvious loser in November is no good. A bizarre extremist wingnut like him would be terrible for the country.

    • Matt McIrvin

      Trump is actually great at stomping on other candidates’ lies. Of course he tells the most ridiculous whoppers of all, but the other Republicans seem reluctant to call him on them.

    • MPP

      Cruz seems probably the least electable. Trump is more electable but also does not have an ideology other than chasing popularity and attention, which means he could sometimes be an ally to Democrats.

      Rubio and Bush are about equal I think – Rubio is deceptive because his conservatism is much more extreme than he has been portrayed so far. Bush has name baggage.

      Kasich is most electable, but will have a much harder time winning the primary than Rubio.

      Ideal outcome, IMO?

      Trump, Bush, Kasich, Rubio, Cruz

      Unlikely to occur though, so I guess we can just hope for Trump to do well and the rest to be tightly packed, but with Bush doing better than Rubio is ideal.

  • Petey

    “In South Carolina, Trump could win all 50 Republican delegates”

    Yup. That was my thought too on seeing the PPP poll.

    And while 3/1 – 3/8 allocation isn’t quite as crazy as that, as we all know, it’s not really proportional either. This particular crosstab from the PPP poll is a bit of indirect evidence that confirms my (long held) feeling that Trump is going to sweep through the South like Sherman in early March.

    • Mark F.

      If he wins the “winner take all” primaries in Ohio and Florida on March 15, the anti-Trump Republicans will all be very close to being screwed.

  • Some Body

    Erratum: The 20th is less than a week from now.

  • Billy

    Why is there so little talk of the Nevada Dem caucus on the 20th? There’s a curious lack of polling there.

    • Sam Wang

      Nevada is super-hard to poll. High transient population, so many voters’ phones don’t have local area codes. Hispanic voters are also hard to reach, especially men.

    • Matt McIrvin

      It’s too bad, since the two recent polls I’ve seen both show Clinton and Sanders in a dead heat.

  • Amitabh Lath

    Harking back a couple of posts — about the GOP deadlines for coalescing about one (or two) Trump alternatives — I suppose one silver lining of an SC blowout by Trump would be that this culling happens before Feb 29.

    But, from current polling it is quite possible that all the non-Trump candidates including Carson manage double digits (but not quite 20%) and no one gets a pink slip.

    • JayBoy2k

      March 15th is Florida. There is no way that Rubio or Bush will pull out before then. Trump and Cruz are not going anywhere. With a recent bump to Kasich, and Carson with money to stay in and carrying a message for his supporters, there is no reason for anyone to drop before March 15th primaries.
      Even if we get down to Cruz, Rubio and Trump, the split will give Trump a bunch of delegates going into the convention.
      I think that Democrats should hope that Rubio does poorly.

  • bks

    Over drinks, aides for other campaigns speak wistfully about a brokered convention — once the pipe dream of journalists and Ron Paul supporters, but now a scenario many are counting on as their last-ditch safety net. State officials wonder aloud what life for them would actually be like if Trump became president.

  • Anthony Shifflett

    Sam Wang – love your stuff.

    Yep – I totally am hoping for a muddled outcome here. Trump 1st with Cruz 2nd and the rest outside of Carson huddled within 2 or 3 percent. That will do it.

    Preferred order: Trump, Cruz, Kasich, Rubio, and then Bush. All really close. That will keep this going awhile…

    • C.S.Strowbridge

      “All really close. That will keep this going awhile…”

      That’s the nightmare scenario for the GOP. Trump wins and the rest are within the 15% to 10% range. None will leave and Trump will dominate Super Tuesday. By that point, it might be too late to stop him.

  • Matt McIrvin

    There’s some kind of buzz going on about Marco Rubio fever sweeping South Carolina, but I don’t see it much reflected in actual numbers.

    • 538 Refugee

      Not indicated in three polls released with yesterday’s date. You have a source at all?

    • bks

      There was a large last-minute shift in 2012 in SC from Romney and Paul to Gingrich and Santorum that was not picked up in the polls until 3 days before the primary:

    • Sam Wang

      A Republican debate on January 17, 2012 probably triggered that sudden shift toward Gingrich, which in the data is visible in the very first polls that included January 18th.

      As 538 Refugee points out, this year after Saturday’s debate, there has been no obvious bounce for any of the top three candidates. Maybe Kasich picked up five points or so, which is significant. So far it doesn’t look like Anthony will get what he wants: the rank order (average of ARG and PPP) looks like Trump, Cruz/Rubio, Kasich, Bush. Trump leads his nearest competitor by 17 percentage points, still enough for him to probably get all 50 delegates.

    • Matt McIrvin

      Just this kind of questionable Politico article:

      I think the guy went to a rally and got excited.

    • Petey

      Sam, not about Rubio, but:

      It’s an outlier, but there is a shred of evidence that the debate hurt Trump. That CNN/ORC poll from yesterday claimed a 9 point drop for Trump between their pre and post debate samples…

    • Froggy

      If Marco Rubio fever strikes, be sure to drink plenty of water.

    • 538 Refugee

      On the Republican side the race has settled into a steady state. Carson rose when he was unknown but fell back once people knew who he was. After all the debates and exposure I don’t think there are any surprises like this left. At most I see share moving if/when people start dropping out.

    • 538 Refugee

      So, of course NBC claims that, at least according to their poll, Ted Cruz has overtaken Trump at the national level. They say it is because while Trump has a high floor, he has a , wait for it, low ceiling.

    • Olav Grinde

      “Trump has a high floor…and…a low ceiling.”

      I am trying do dwell on that image. Thank you!

    • Petey

      “meh, outlier.”

      I hear you, Sam.

      But I respect Hart/McInturff a hell of a lot more than Morning Caller.

      And I did have ‘pundit’ intuition that Trump may have have been in a bit of a ‘danger zone’ on Monday. Which is dead-center of that poll’s sample.

      (Though I did respect every other pollster on Tuesday and go overboard in declaring Trump the nominee.)

      But I don’t dismiss Hart/McInturff lightly. Yes, we need more data to show they’re not simply an outlier. But my spidey-sense ‘pundit’ thing that was going on Monday is back again. This thing really feels like it could be fluid. Important stuff really has happened in the last 5 days.

      Need more data.

    • Sam Wang

      Like the national Morning Consult poll, South Carolina polls have not shown any swing. Four surveys.

    • Petey

      Again, I really do hear you. Genuine outlier. But the only ‘new’ polls of those I don’t consider ‘garbage polls’ is the fresh Selzer, and the older PPP.

      And, maybe SC is different than national!!! (Of course, if we even bother to stipulate that to be true, SC results may change national numbers with momentum effect.)

      Just sayin’ that big political / tribal things are going on with Trump in last 5 days. On the sharp lookout for any data to show if any voter fallout. Hart/McInturff shows a certain possible result I’m watching out for. But still definitely an outlier…

    • Matt McIrvin

      I think Morning Consult has a systematic house bias toward Trump; they’ve posted some of the most wildly pro-Trump poll results, and articles explicitly saying that they think everyone else is shortchanging Trump.

      That said, I don’t believe that poll with Cruz in the lead either. There are a bunch of polls with date ranges overlapping it that are not consistent.

  • A New Jersey Farmer

    Cruz would likely turn off moderate Republicans and any Democrats, while Trump could actually attract some moderate Republicans since he really is one.

    • bks

      Yes, I find it hard to believe that Bush/Kasich/Rubio supporters would automatically move en masse to Cruz, or that Bush/Kasich/Cruz supporters would move to Rubio. A large proportion of voters just want to back the winner.

  • Amitabh Lath

    CNN just polled Nevada and it does not look good for the anti-Trump brigade.

    Maybe this is an outlier, and he’ll do mid-30’s in NV. But at that point even the consolidation of anti-Trump efforts into one candidate might be considered a long shot.

    • Petey

      Based on the polling evidence, (with the PPP poll showing Trump beating or running even with all the other candidates in a 1 on 1 playing a big part), and knowledge of the GOP delegate allocation process and front-loading:

      The Petey Election Desk™ officially called the nomination race for Trump yesterday.

    • Olav Grinde

      However, that CNN/ORC poll does look good for Bernie – the poll has Clinton 48%, Sanders 47%.

      No idea whether these numbers are even remotely reliable. As Sam Wang and others have pointed out: Nevada is a notoriously difficult challenge for pollsters.

    • Amitabh Lath

      True, NV polling is tricky. On the other hand Trump has his name in huge glowing neon plastered on the side of a tall tower right there on the Vegas strip. Can’t miss it.

      I don’t see Kasich advertising a free buffet.

  • Amitabh Lath

    I saw that NBC/WSJ poll showing Cruz > Trump and my first thought was: I bet it has a small sample size. And sure enough, it has N=400.

    That’s a 5% statistical uncertainty and given Trump’s average is 37% this is barely a 2 sigma fluctuation from that.

    In fact a few scatters like this give confidence that maybe the polls really are normally distributed and one can put some confidence in the mean (or median).

    If out of a hundreds of polls not even one showed this kind of fluctuation one would start to suspect that the pollsters are starting to cook the books to agree with each other.

    • JayBoy2k

      Is there a point where a polling organization just states the obvious, which is I changed so many factors in midstream that , frankly, my poll results are worthless? I just went to the site and saw this:

      ” Another possible explanation for Trump’s decline in the new NBC/WSJ poll is an increase in “very conservative” Republican voters from January’s sample.

      If the current poll is re-weighted to reflect the ideological composition from last month, the GOP horserace numbers are: Trump 26 percent, Cruz 25 percent, Rubio 18 percent and Kasich 13 percent — so Trump is ahead by one point, but still down from January. “

  • Olav Grinde

    There are ten fascinating Quinnipiac polls released today. As Dr Wang and others previously have pointed, general election polls this far out are of limited value. But still, these are very surprising results!

    • Amitabh Lath

      Useless, but amusing. I expect the main use of these polls will be to try and reinflate the sagging campaigns of the three remaining establishment Republicans.

      It may help two or maybe all three keep the lights on through March. Of course this is good news for Trump (who is the only loser in this set of polls).

    • Olav Grinde

      I was actually more struck by the poll numbers showing Bernie doing better than Hillary against the various Republican candidates. Yes, I realize that’s in advance of the general election’s hundred-million dollar GOP propaganda campaign. But still…

      I also thought the different between today’s Quinnipiac number and yesterday’s numbers from the comparable USA Today/Suffolk polls rather interesting.

      I do wish there was more public debate between pollsters delineating why their polls are better than their competitors’…

    • 538 Refugee

      The Republicans have been running against Hillary for 8 years so no surprise. Bernie is not as well known yet and the Republicans have been too busy eating their own to pay him much mind yet.

    • Matt McIrvin

      People have been doing these head-to-head polls for a while, and there’s a lot of noise mixed in with the signal. In general, there’s very little pattern to whether Sanders would do better or worse than Clinton against the Republicans, but there’s always somebody cherry-picking the latest data point to claim something either way. Sanders supporters are particularly eager to jump on the ones that go their way.

    • Matt McIrvin

      …Here are Huffington Post’s aggregations of these national head-to-head questions:

      The main thing that comes through is that there just aren’t enough polls asking about Sanders.

    • Froggy

      Matt, I’ve seen a pretty consistent pattern over the last couple months of Sanders doing better than Clinton in polls of possible general election matchups, though I still see such polls as essentially worthless at this point. (Back in late summer paying attention to such polls would have led you to the conclusion that Carson was the strongest general election candidate in the Republican field.)

    • Sam Wang

      As we have seen in the case of Trump, favorable/unfavorable numbers can change substantially. General election matchups can change too.

      Once we are in a general election campaign, there is plenty of time to call attention to details regarding Sanders, who is atheist, socialist, and would be the oldest person to ever be a major nominee. Finally, there is the anti-Semitism factor. I do not doubt that Republicans would prefer Sanders as an opponent in November.

    • Olav Grinde

      Froggy, that’s true. The problem is that, since last summer, Ben Carson has opened his mouth on at least a couple of occasions. His current poll numbers reflect that.

    • Amitabh Lath

      Josh Marshall at TPM has an interesting take on Sanders in the general. The GOP has not been attacking Sanders at all, but if he is the nominee they would go full McCarthy red baiting, painting him as an old school hippy commie pinko. And there is a lot of old Bernie video, audio to help them do this.

      Marshall sums it up: ” But it’s important to remember that it’s not “Bernie Sanders”, it’s “Bernie Sanders” after a fusillade of attacks based on a life history of a sort of politics that is very alien to a lot of Americans.

  • A New Jersey Farmer

    Predictwise now has Trump below 50% and Rubio rising. Tomorrow’s polls should be interesting.

    • Froggy

      Trump’s absolute high on Predictwise has been 52% (just before the Iowa caucus), and he’s been rising pretty steadily since a certain debate dispelled with the notion that Rubio was rocketing up to being the front runner. However Rubio does appear to be consolidating the establishment support, and on Predictwise it looks like my man Jeb! is going down for the last time. Adios, dulce amigo.

    • Amitabh Lath

      Predictwise concerns me. From a quick look it appears the algorithm looks at several betting sites, as well as polls. I don’t know what the actual algorithm is that kneads all these inputs into an answer, but I would be **EXTREMELY** worried about correlations between the various inputs. Some of the same people that bet on the Iowa Electronic Market probably also go on Betfair, etc. And they probably bet after looking at polls. So large amount of correlation. Which is ok if you know about it and there are ways to correct for it.

      But if you assume all inputs are bringing in independent information when in reality they are highly correlated, then your output will have large, sharp, unphysical (unpolitical?) swings.

      And in fact you do see the GOP nomination plot on predictwise having almost vertical lines.

    • Froggy

      Here’s more of an explanation of what PredictWise is doing: . Clear as mud. I just use it as a shorthand way to see how the betting markets are moving.

  • 538 Refugee

    I read an article saying Bush is spending a lot of his money in SC and mainly attacking Kasich? He’d need to pick off a full half of those votes to even move him up to third. All of them might get him to a distant second. But given Kasich has been courting the civilized, rational voters I don’t see how attack ads will work much. Or is he proving he can be mean spirited also hoping to pick off Trump and Cruz supporters yet not directly attacking them? ;) I think the article attributed this to a Kasich advisor but I can’t find it now. I believe he said something like the Bush campaign is spending $28 million. Enough to seriously hurt Bush going forward if he doesn’t finish well in SC.

    Or I could be even more cynical and think Bush has been promised he will win a brokered convention and he is now trying to mop up THAT competition?

Leave a Comment