Princeton Election Consortium

A first draft of electoral history. Since 2004

Slow news month ahead?

March 26th, 2016, 1:03pm by Sam Wang


I’m on a low-posting regimen for a little while. Basically, I think the Democratic and GOP primaries are settled.

That said, why don’t you comment on which of the following would be most interesting:

  1. An explicit calculation of what the probability is of Hillary Clinton getting a majority of pledged delegates by June.
  2. What kind of swing in opinion, in remaining states, it would take for the probability above to flip to favor Sanders.
  3. Same question as #1, except for Trump.
  4. What swing it would take to keep Trump below 50% of delegates.
  5. Analysis of long-term indicators for November election.
  6. Senate and House outlook.
  7. Dissecting the work of other websites and journalists.
  8. ???

Tags: 2016 Election · House · President · Senate

63 Comments so far ↓

  • counsellorben

    I have changed my mind.

    After the Wisconsin primary, I would be interested in seeing a statistical analysis of the probable disparate impact the new WI voter ID law had in the 2016 primary upon different groups of voters.

    Obviously, a confounding factor for the D primary is the lower turnout rate versus 2008 in prior primaries/caucuses.

  • Ed Wittens Cat

    Dr. Wang:
    Trump is now apparently turning his attention to delegate math–
    http://www.bloomberg.com/politics/articles/2016-03-31/trump-priebus-said-to-meet-in-washington-after-pledge-backtrack
    will this make a difference?

  • Tim

    Sam is clearly confident that Trump will get to 1267, otherwise the outcome couldn’t be a settled question. But I don’t see the analysis – can somebody direct me or give me the 5 cent version?

    Thanks!!

    • bks

      You mean 1237. Speaking for myself, not Sam, I would say that if the rules were administered fairly, Trump would have a lock on the nomination. The problem is that in the real world there is agency. The GOP can change the rules (e.g. provide a religious exemption to delegates that allows them to vote their conscience on the first ballot) or make post facto changes (e.g. SC delegates go to Kasich because Cruz and Trump reneged on their GOP oath). We are not used to the party hierarchy declaring open war on a leading candidate and this latter phenomenon is causing a great deal of uncertainty in the prediction markets if not in the actual process.

  • Ed Wittens Cat

    can i add an 8?
    im late to the party but maybe u guys have already discussed this– didnt the 2011 Citizens United decision pave the way for Trump style candidacies?
    and does the GOP realize that?
    I mean, i cant see a President Koch (extreme lack of telegenicity)– but a President Gates or (my personal favorite) a President Jolie could be in the offing.

  • bks

    How are the live-phone pollsters doing versus the Internet pollsters?

  • Vicente Piedrahita

    5 & 6, especially 6. Thanks regardless!

  • whatever next

    5 & 6 – given that you said ‘long-term indicators’, there is value in doing 5 now even though the race is far from settled on who the nominees are.

    It’s not that 1-3 are less interesting – probably more interesting right now in a way – but I can’t see what information there is that will satisfy a data wonk like yourself Sam, and also those skeptical to the point that they see signs of momentum with Sanders greater than the somewhat out-of-date polls suggest (especially in the light of his Washington victory, which I think is hugely impressive).

  • Karel

    Dr. Wang,
    8-
    After reading a Wash Post article today about the Clinton email mess, I think it could really hurt her chances either for primary or more likely in November. What would you estimate the likelihood is she gets defeated in November if this blows up more? Thanks !

  • Alphonso

    5&6

  • counsellorben

    I would opt for 5, especially an analysis based on some of the well-known models for predicting presidential election outcomes.

  • Richard Subber

    Nos. 5 and 6, please

  • Ed Wittens Cat

    An extra large helping of 6 please.
    Thank you for allowing me to comment, Dr. Wang.
    I just remembered you dislike proxies…but evr one i kno uses Tor, Tails and VPNs in this Brave New Surveillance State.
    It’s good to see so many back from 2012.

  • Barry Rigal

    Re the number of seats affected by gerrymandering.

    Maybe I’m not looking in the right places but it would be helpful to have an article on what if anything will change following the next census. And when will the supreme court consider the issue next?

    I’d like to see more on this but I can prove Sam Wang wrong on at least one point. It is FEWER than 11 seats not LESS than 11 seats at stake.

  • Art

    6, plz.

    • Richard

      Ditto. The primary zoo is largely over; what happens with Congress remains important.

  • Joseph

    Okay, I’ve got one, and it’s right in your wheelhouse, Doc: How many elections for House and Senate are likely to literally turn from one party to the other on the basis of obvious gerrymandering this election?

    I suppose technically it’s a #6 choice, but a bit more specific in focus.

    • Sam Wang

      Senate: zero, the boundaries are fixed.

      House: from gerrymandering, less than 11 seats (see page 36 of my gerrmandering article). How much less, it’s hard to say – it depends on how much voters have rearranged themselves and their opinions since 2012. However, there’s an additional 9-10 seat disadvantage that comes from population patterns. So that’s 20 seats total that go in favor of Republicans due to existing asymmetries.

      For Democrats to retake the House, in 2012 they would have needed 53.3% of the two-party national popular vote. By now, that number has probably declined – let’s say to 52.0%. (Gotta estimate that better, but how?) House vote approximately follows Presidential vote. Obama won 53.7% of the two-party vote share in 2008, and 52.0% in 2012. It seems possible for Clinton to get to 52.0% this year, especially with Trump in the mix.

  • Mark S.

    Senate and House (#6). Specifically: which elections are the closest, and where would a donation have the most impact for someone who wanted to help a certain party (you can guess who) re-take the Senate (and House)? Four years ago, I think, you did this. It was very helpful!

  • Bil Carson

    I’m ready to move on to #6 already!

  • Nick

    I think #1 and # 2 are most interesting, it seems like Sanders is putting up a much stronger fight that pundits have been predicting, even granted that he was expected to win Washington, Alaska, Utah, Idaho etc… I don’t think pundits were expecting these kinds of margins, 65% in diverse Seattle as of last I checked, 70+% in Wash and around 80% of the vote in the less populous states out west is comparable to what he was getting in his home turf small New England, to me this suggests he has real energy and I wonder I we will see a Sanders surprise. Wisconsin seems like another highly likely Sanders win… granted NY and CA and PA and NJ and MD may be more challenging, but it might not be to much of a stretch to see a tide change…

    • Sam Wang

      That is crazy talk. OK, evidently I have to do this.

    • Nick

      Great, thanks Sam! There just may be enough people out there that need to see just why it is crazy talk.. (and seriously, what pundit was predicting a 72-28% blowout in WA? Maybe a 60-40 or something like CO would have been expected….And before Utah the poll that I saw was predicting something like a high 50s for Sanders there, not something approaching 80%, was MI really just an abnormal situation…? )

    • truedson

      With 93% of the vote in Washington, the grand total is 23,500 voters for the Democrats. And that is the biggest state tonight, with 17,000 going to Sanders. In a blue state with a population of 7 million. Sorry if I am not impressed.

    • Nick

      That count that Truedson refers to was caucus delegates not voters, and voter turnout was actually quite high, many more voters than counted caucus delegate votes. I think the 72% in Washington really is pretty impressive and its hard to not take it that way, WA is not a small state like some of the others that have been that lopsided. Like I said I wouldn’t have been surprised with a 60-40 CO style split, but 72-28 (even after Mar 15) seems pretty impressive. I see the case for it being hard for Bernie to pick up 58% of the remaining votes, if the polls are accurate, but I do wonder about the accuracy of polling (obviously a huge underestimation in Utah where they were putting Bernie in the high 5os). I does bother me just how dismissive pundits are of Bernie’s campaign out of hand when he is clearly running a more effective race than any republican challenger to Trump ( I couldn’t find anything at all about WA on 538 today for example).

      To keep it balanced dismissiveness coming from some Bernie supporters about Hillary is also bothersome. I hope- as someone who almost certainly will vote for the Dem winner that the two sides can come together well, its lucky from the Dem perspective that they will most certainly face a weak republican candidate.

    • Matt McIrvin

      Those small vote numbers reported in Washington are state-level delegates, not total caucus voters everywhere, right? I’ve been seeing some confusing and contradictory discussion on this subject.

      Caucus states are confusing things.

    • Matt McIrvin

      There’s lots of more or less content-free Bernie-mentum chatter going on in the political media right now, basically because three states that were really good for him happened to be on the same night.

      He’s running out of states like that, though. Oregon is one; he’ll do extremely well there (though I’ve seen no polling). He can win Wisconsin but it’d probably be a near thing where he gets a little over half the delegates. He’ll probably do well in Indiana, Montana and the Dakotas.

      But he’s probably going to lose big in most of the big Mid-Atlantic primaries, some of which are coming up very soon. If I were his campaign manager I’d go all out for maximizing his share in California; it’s the one Hail Mary left before you get into trying to swing superdelegates.

    • 538 Refugee

      The important part of this primary is the states Clinton has beat Sanders so far. Ohio, Florida, Virginia, North Carolina. What does it matter if Sanders takes California with 100% of the vote? Hillary still beats any Republican candidate there. In fairness, Hillary’s ‘southern firewall’ was in states the Democrats aren’t likely to carry in 2016.

  • Amitabh Lath

    I vote for Senate and House outlooks (#6) knowing full well that the uncertainties will be large (especially House). As you say, Clinton and Trump are settled and deciding if they are at 3 sigma or 3.5 sigma is tedious and useless waste of resources.

    As for critiquing other analysts (#7), why bother? Most people who comment here are numerically savvy enough to see through the “it’s still anyone’s game folks” and “Kasich has a shot at the Acela corridor” malarky.

    Another benefit of #6 will be to help direct money are other resources to races on the edge.

    • Matt McIrvin

      Most people who comment here are numerically savvy enough to see through the “it’s still anyone’s game folks” and “Kasich has a shot at the Acela corridor” malarky.

      Maybe, but it’s nice to have an authoritative source to link to when people talk nonsense elsewhere.

    • Amitabh Lath

      PEC is good, but will never fix the “someone is wrong on the internet” problem.

      I have been chuckling at Douthat’s “Trump could never ever ever” articles since September. Also when Nate Silver predicts number of delegates to 4 significant figures. Or any recent article about how Ryan didn’t exactly say no forcefully enough…Ugh.

      I suspect a serious look at House and Senate races in April could cause the topic to be talked about and be prominent in the national conversation over the summer and people to donate when it could actually help.

  • JayBoy2k

    I do believe that the cake is baked and if it turns out to become, unbaked, you can open discussions at that point.
    It does not make a lot of sense to go after the House and Senate races because the Presidential race does heavily influence the down ticket races and frankly we will have months to discuss these after much more data is available.
    I vote #7…

  • ArcticStones

    Sam, I would like your analysis of the following scenario. If, shortly before November, there is a far worse financial meltdown than the last one, how do you think that would influence:

    1) A Presidential election between Hillary Clinton and Donald Trump

    2) Control of the Senate

    3) Control of the House of Representatives

    4) State-level races.

  • Mark F.

    If Trump is trailing Clinton by double digits before Cleveland, expect a huge effort to change the rules to unbind delegates on the first ballot and ditch Trump, regardless of whether he has 1237 delegates or not. Some delegates may be Trump delegates in name only.

  • Scott Pauls

    Of course you forgot:
    9: Profit!

    However, I’d love to see 4 and 6. For 6, it would be great to look at the House/Senate projections conditioned on a Trump or other nomination. Do you have a way to estimate the impact of a contested convention? I don’t.

  • Adam

    6, 5, 4 in that order

  • Mary

    #1, #5, and #6 in no particular order.

  • Josh

    We can never have too much of #7.

  • Eric Walker

    Number 5, analysis of long-term indicators for November election, is interesting, but I suspect mainly for the size of the EV win. But number 6, Senate and House outlook, is, I think, where the deep interest has to be. I was just reading that the Democrats are concerned about money for the Senate campaigns (since most big Republican money is going to bypass the hopeless presidential campaign); will that likely shiuft current sentiments and polling?

  • Mark Buckley

    Number 5 for both the Presidential outcome and current Red Senate seats in Obama States

  • Matt McIrvin

    Everyone’s going to be talking about the unlikely long-shot possibilities for the primaries. Speculation about some sort of plot to unseat Trump on the Republican side is rife, and it all just seems really unlikely to work even if Trump doesn’t get to a majority of delegates, nor is a data-driven approach likely to be very useful in dealing with it.

    I say, pivot to the general, and not only the presidential race but the Senate and the House.

    Speaking of unlikely scenarios, the one in which a third-party run somehow leads to a deadlocked Electoral College and a President anointed by the House needs more of a critical eye than it’s gotten so far. I think a lot of people really don’t see why this isn’t a probable outcome.

  • Dan Kohn

    Please post about how catastrophically Trump would have to tank before the House might legitimately swing to Democrats.

    • Sam Wang

      That’s a super-interesting problem. However, it’s hard because one either needs to do district-by-district handicapping, or rely on an accurate estimate of how much gerrymandering/structural effects are left over from the Great Gerrymander of 2012.

    • Phoenix Woman

      One also needs to know how many races will have Democratic candidates, and how many Republicans will be running unopposed. Hillary’s busily raising money for the downballot races for that ver reason.

  • truedson

    Oh dear, they all sound interesting. I don’t think I can choose!

  • Jason Bennett

    I’m a little surprised to see you say that the Republican race is settled – it seems just as unsettled as ever to me!

    With the current delegate projections (http://www.centerforpolitics.org/crystalball/articles/assessing-trumps-path-to-1237/ http://www.vox.com/a/republican-primary-2016-projections) showing Trump on the edge of getting a majority, and with Cruz doing his best to recruit individual delegates, it looks possible, or even probable, to me that Trump will miss his threshold, and then have the convention turn on him on the second ballot.

    I guess my answer is #3 and #4, because it seems to me that a small swing could change everything.

    • Sam Wang

      To be honest, I agreee that the GOP delegate calculation is worth doing well. To my surprise, I don’t see it out there. I suspect the reason is that it’s bad for business at a place like ESPN to say things are probably settled.

    • AySz88

      I have to admit, these name checks of “ESPN” (I’m guessing you mean Silver and Enten) make me really curious about the details behind the criticism.

      I did also notice that there are several counts (Jason linked two more) which tend to say “close” rather than “settled”. So I’m presuming good faith for now…

  • Rachel Findley

    I’d be most interested in the Senate outlook, and also the long-shot possibility of turnover in the House. And also in long-range estimates for the presidential election in the fall.

    I agree, the current Presidential front runners are likely to get the nominations one way or another. I’m still campaigning for Sanders in California, as his policies are the ones I like best. After the nomination I’ll work for the Democratic nominee, and for whatever Senate races best combine my true values with a strategic assessment of impact.

    Oh yes, I’ll take a look at control of state legislatures for the gerrymandering party that starts after the 2020 census.

  • mediaglyphic

    i think #3 and 4 would be interesting, at the very least it would give us some numbers to hang on as we get near noisy season. is it possible to do this using an automated script, or is it too hard because of the CD detail needed?

    • Sam Wang

      No, there is an easy approximation for CD estimation. The whole thing is at this point a variation of the general election script. I haven’t automated the GOP delegate estimator simply because it seemed like overkill, especially since I don’t do this commercially. Probably I’ll do it manually, then update it every few weeks.

  • Bill Herschel

    6 A thousand times 6!

  • Robert Egenolf

    I would appreciate a detailed explanation of #1 and #2 to help Bernie supporters get a reality check and some thoughts on #6

  • Jared Harris

    House and Senate commentary is by far the most interesting to me at this point. I guess polling is probably very thin so far, but there’s historical and demographic data, plus whatever useful information can be extracted from the primaries.

    I agree with the New Jersey Farmer and Michael Hahn that this is the actual focus of Republican establishment anxiety. Conversely, composition of the Senate and House has a bigger effect on the governance we get than who wins the Democratic or Republican nomination (though I agree we know that already).

  • Sean

    Can we choose more than one?

    I like 5 and/or 6 myself.

    But I’d like to add that in 2012, your site helped recalibrate my thinking about political reporting and punditry. You often pointed out instances where a writer was conflating their “sense” of a campaign with actual trends supported by data. For better or worse, this was a new concept for me.

    As I thought about it, and as you continued to call attention to it, it became clear that data analysis can helpfully shift the focus from settled horse race to issues, or to races where outcomes are actually in doubt. It can change the way one thinks about news coverage and also the way or ways in which one channels their own efforts and resources as an activist/engaged voter.

    So I am always grateful when you look closely at current news reporting and its relationship to data – feels like a public service in some sense. It certainly helped me.

    So 7, too :)

    Sean

  • Abraham Rash

    I think an analysis of how Trump’s turnout has been compared to the polls would be interesting. I know he was off in Iowa at a minimum, but has he settled down since then?

    Is he a true ‘wildcard’ candidate, whose electoral performance will be a test for poll-based models? Or is he a sheep in wolf’s clothing, and actually as predictable as the candidates in 08 and 12?

  • Dave Wink

    7 – dissect the work of others. Without doubt. As you say, these races are done. I would find it enlightening, though, for you to poke holes in analysis that says otherwise.

    • Sam Wang

      This is a time-tested genre. In previous years there’s been no shortage of such targets. This year is no different!

  • A New Jersey Farmer

    I think the Senate and House projections would be interesting, especially because much of the GOP panic seems to be how to hold the Senate and minimize the carnage in the House if Trump is the nominee.

    • Michael Hahn

      I agree!! It is really too early yet for explicit analysis of the presidential race, as much will depend on exactly how the Republican race shakes out, and how deep the divisions within the party really turn out to be.

      But I would really like to see an analysis of the House and Senate races might shake out. To me, this is at least as important as the presidential race.

    • Dave Wink

      Aren’t the Congressional and Presidential races linked to some degree? If it is too early to project the President, it should mean that you also can’t predict Congress to the degree that Presidential candidate impacts Congressional elections.