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

South Carolina (D) primary – open thread

February 27th, 2016, 5:04pm by Sam Wang

[HuffPollster results page]

Be nice to the Sanders supporters. Tonight starts what might be a challenging week for them.

10:00pm: Wow, Clinton really overperformed. Polls showed her ahead 62%-25%, but she is winning 74%-26%. This might be an example of what I wrote about back in 2008 (“The Exuberance Of Likelier Voters“). When a win is predicted to be lopsided, it is common for the margin to be considerably larger than predicted by polls.

Back then I estimated the increase as being a factor of 1.2. Applied to tonight’s numbers, the poll-based margin of 37% would become a margin of 44%. The actual margin is 48%, not that far off.

Tags: 2016 Election · President

21 Comments so far ↓

  • sjw

    Exit polls have Hillary beating Sanders 84-16 with black voters.

    “But beyond the primary math, it raises a kind of conceptual issue about Sanders’s political theory: Given the centrality of African-American voters to the Democratic Party, it’s hard to envision a left-wing political revolution taking place in the United States that they have so little enthusiasm for.” (Matthew Yglesias, Vox)

  • Amitabh Lath

    The SC polls said Clinton/Sanders were 65/35 while the results are closer to 75/25. I remember long ago there was a PEC discussion on how polls go awry when the margin is large. Polls are better when the margin is small. Is there a rationale?

    • 538 Refugee

      Less interest in putting time, money and effort into a race that isn’t close?

      Not sure why folks are interested in calling the race when it’s about delegate count.

    • Amitabh Lath

      I suspect the larger errors in SC are due to something inherent in polling itself. The instrument goes into a nonlinear regime.

    • Matt McIrvin

      People like winners. Knowing your candidate is going to get crushed is probably a huge demotivator to turn out.

      I recall hearing that if you ask people who they voted for after the election is over and results are known, the winner always gets a bonus of several points over the actual vote.

  • 538 Refugee

    Us Sander’s fans never really expected him to do as well as he has done so far. But 75/25? Excuse me while I go treat my wounds with alcohol.

    • Matt McIrvin

      I heard African-Americans in SC turned out at a level ABOVE 2008, which is astounding. They’re mobilized this year. But white Democrats didn’t turn out in large numbers anywhere, not for Bernie or Hillary. The Sanders fans could maybe work on that…

  • P G Vaidya

    I imagine the complete political change Bernie is talking about will take about 16 years to accomplish, (Just like Goldwater 1964 to Reagan 1980). I can see Bernie playing the role of Yoda, inspiring new leaders and new followers.

    • bks

      At least we can mention socialism without people fainting.

    • Olav Grinde

      I sometimes joke that, yes, I am in favour of publicly owned highways, airports and shipping ports. Heck, I even support those socialist national parks!

      However, on one key point I do depart from current American practice: I do not believe that the losses of too-big-too-fail banks should be socialized.

  • bks

    I don’t have any data to back it up, but my impression is that the candidates’ final percentages (both parties) are remarkably well determined after just 1 or 2% of precincts have reported. Not well enough to distinguish 2nd from 3rd in the GOP SC race, but still pretty good.

    • Olav Grinde

      I would think that this very much depends on which precincts are counted first. I seem to recall primaries and general elections where the eventual loser at first looked like they were trouncing the winner. Atypical precincts are not going to be predictive.

    • Owen

      Try following a statewide general election in Missouri or Illinois or Pennsylvania and you will be relieved of the illusion that you can predict the final result from the first 1% or even the first 25%.

    • Sam Wang

      It actually is possible to do in a general election by correcting for local partisan tendencies.

  • Mark F.

    Yes, it’s about delegates, but after Tuesday Clinton will probably have such a huge delegate lead, it will be virtually impossible for Sanders to catch up. Especially not when you are losing black voters by such huge margins. Senator Sanders gave it a good try, he is to be congratulated for that. But it’s now really too late to turn things around.

  • The Raven

    No. Richard Wright must be rolling over in his grave.

  • Fritz

    Is primary turnout predictive of general election success? IOW, should Dem’s be worried about disparity in turnout?

  • Froggy

    Looking forward to Tuesday, reports are that the Sanders campaign has decided to run no ads in Alabama (53 non-superdelegates), Arkansas (32), Georgia (102), Tennessee (67), Texas (222), and Virginia (95). Instead they are focusing resources on Colorado (66), Massachusetts (91), Minnesota (77), and Oklahoma (38), and Vermont (16), where the terrain is more favorable for them.

    Essentially then they are conceding to Clinton the lion’s share of the 571 delegates in the states they are not actively contesting, and are pushing to win states with 288 delegates available. This is more a strategy of trying to appear viable as opposed to one of actually remaining viable (although I’m hardly in a position to argue with how the Sanders campaign is allocating its limited resources).

    • Roke

      Sanders seems to be going for a few “feel-good-victories” so that he can keep his campaign alive and wait for more favorable states after Super Tuesday. (Or at least people on social media keep making this claim)

      But the question is: are there enough Sanders favourable states that can swing the delegate count in his favour? I doubt that.

    • Josh

      It’s even worse than it looks because delegates are assigned proportionally, and even if Sanders wins states like MA and MN, he’s only going to win something like 55-45%, meaning Hillary will get almost as many delegates as he does. By contrast, states like GA and AR and TN are going to go 70-30 for Hillary, so she’s going to be running up the score. For all intents and purposes, the race will be over after Tuesday.

Leave a Comment