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 state polls, Clinton running 5.8 percentage points ahead of Obama 2012

August 16th, 2016, 5:41pm by Sam Wang

There is a lot of media drooling over polls showing Donald Trump cratering in state after state. I find this gloating to be unseemly. Here at PEC, you can do all your gloating in one go, saving time for other reactions, like schadenfreude.

Plotted below are median Clinton-minus-Trump margins in all states for which August polling is available, plotted against Obama-over-Romney margins at the same point in the 2012 campaign. This horizontal axis quantity is better to plot than the final Obama-Romney election margins, which include undecided voters’ final commitments – not a fair comparison.

The data come from RealClearPolitics.

Clinton is overperforming Obama in 15 out of 17 states, the significant exceptions being deeply Democratic Maine and New York. Overall, the difference is a median of 5.8 +/- 1.4% (estimated one-sigma SEM).

This is very similar to the picture in July, before the conventions. I wrote then that Trump needed to recover disaffected Republican voters in red states. At that time, his red-state leads were smaller than Romney’s 2012 win margins by a median of 9.3%. Hillary Clinton’s blue-state leads were smaller than Obama’s wins by 1.9%. The difference was about 7%, comparable to the value of 5.8% given above.

Recall that the popular vote margin in 2012 was Obama 51.1%, Romney 47.2%. If Clinton’s strong performance were to persist until the election, 12 weeks from now, the popular margin would end up at approximately Clinton 54%, Trump 44%. If we assume a disaffected 6% go to other candidates, that leaves Clinton 51%, Trump 41%, 8% Johnson/Stein/McMullin.

I find this quite amazing. After attacking Gold Star parents, advocating Second Amendment remedies, and other stuff too tedious to recount, Donald Trump will probably still end up with at least 40% of the popular vote. That is an impressive testament to the partisan polarization that has developed since 1992.

The other feature of the graph above is its correlation coefficient, +0.87. Voters are entrenched in the same positions as 2012 – and earlier, for as long as we’ve had the now-familiar red-state/blue-state arrangement. To most Republican voters, any candidate they field – whether it be Mitt Romney or Donald Trump – is preferrable to a Democrat.History of electoral votes for Clinton

Tags: 2012 Election · 2016 Election · President

105 Comments so far ↓

  • Greg Gross

    So, Sam, is the cake actually fully baked?

    • Sam Wang

      Probably, though the random-draft parameter and Bayesian prior I am using both follow the Eisenhower-era (“slow rising”) option. The polarized option gives more baking power.

    • Ash

      Hillary leads looks pretty convincing. But we have to agree that this is not a conventional election. Negative for both candidates outweigh positives. I wonder if next 85 days may bring any surprise in shape of leaked documents/emails.

    • A New Jersey Farmer

      So perhaps it’s just matzoh at this point. The cake will come later.

    • Matt McIrvin

      There seems to be a media narrative developing that Trump’s latest campaign shakeup and his Milwaukee speech constitute the long-awaited “pivot”, he’s going to be on message from now on, and his big comeback is just around the corner! (They had to do this eventually just to keep it interesting.)

    • Kevin King

      Matt, from what I’ve seen in the press, the shakeup is being seen as Trump’s repudiation of the concept of a pivot altogether. In other words, he is going to continue what he’s been doing. Hard to see how he’ll win in that case, given how his campaign had performed in the general do far, absent some unforeseen contingency.

  • Paul Benjamin

    There’s hardly anything “seemly” about this election. I’m not into gloating. My Midwestern origins demand modesty and avoidance of braggadocio.

    I am not feigning objectivity or, for that matter, modesty; I just hope Trump loses. I still think that governance, whether limited or more expansive, needs to be serious, competent, and, rational. Your competence and rationality are refreshing and I am thankful that you’re doing this.

  • seeker

    Nice comment about gloating and schadenfreude.

  • Michael K

    So wouldn’t we expect the current meta margin to be higher than it was at this point in 2012, by a similar amount? Wasn’t Obama ahead around 2% or 3% in the meta-margin at this stage in 2012?

    Am I missing something? Is the electoral math much worse for Clinton in a closer race? Or is this just a reflection of states which have sparse polling or no polling at all?

  • Ryan W.

    Of your last paragraph, our deepening condition of partisanship has me very worried for the health and future of our nation. Guilty as anyone, I’ve been a straight-party guy since reaching voting age, but I’m considering “crossing over” as almost an act of civil disobedience in an effort, a small one-person effort, to help keep us from destroying ourselves.

    • Sam Wang

      Whichever way you cross, if you simply refrain from voting for your own party, it has half as much effect as actively voting for the other major party. Similarly, your vote is weakened by voting for a third party.

    • Ryan W.

      True, though I was thinking more of active engagement with the other major party than simply voting. Things like changing registration to opine and help form the party platform. It seems we’re spending too much time judging one another, too little time trying to understand each other. Apologies for not articulating better.

    • Lance S.

      Actually it may not be as hard to vote other than straight party as you think. For example, here in NC Steve Troxler is an excellent Commissioner of Agriculture who replaced a corruption-tainted Democrat. He’s done a great job and is well regarded by the states’ farmers. I have no trouble voting for him at all. In 2012, much as I hate to admit it, I voted for McCrory for Governor, since he had been a centrist mayor who worked to build bipartisan consensus in Charlotte, while we had just had several Democratic governors who were corruption tainted, and their candidate that year was not all that impressive. I had no idea McCrory was going to change into a rubber stamp for every right wing extremist idea from the legistlature. The point, though, is if you really look you can often find people of the other party in down ticket races who actually are the better candidates.

  • Olav Grinde

    “Donald Trump will probably still end up with at least 40% of the popular vote. That is an impressive testament to the partisan polarization that has developed since 1992. … To most Republican voters, any candidate they field…is preferrable to a Democrat.”

    The fact that Donald J. Trump has the support of 40 % of American voters is indeed depressing and terrifying. But I think it goes deeper than you suggest; Trump’s support signals that the USA is almost ready for an authoritarian leader.

    Trump may not win in 2016, but I think it highly likely that a similar candidate (Cruz?) will be victorious in 2020 or 2024.

    In fact, given a Black Swan event – such as a full financial collapse in October, or an ISIS-inspired attack with a dirty bomb in NYC or Washington DC – I would not entirely discount the possibility of a Trump victory already this year.

    • 538 Refugee

      Yes that level of support is depressing but remember what is propping it up. Us white guys are losing our majority. Some of us are OK with that since we realize there is only one race of humans. ;) I think this is more a last gasp than a sign of things to come.

      I think we are seeing the political landscape shift a little to the left and I don’t think it’s going to go back anytime soon.

    • Matt McIrvin

      I thought the Orlando nightclub shooting was going to be Trump’s ace in the hole, but his reaction to it just hurt him (his first response was to congratulate himself on Twitter).

      The Clinton campaign has stressed Hillary Clinton’s experience as a foreign-policy hand, her presence in the Situation Room during the raid on bin Laden, etc., and Trump’s volatility. It bothers leftist voters who already see her as a Bush-like warmonger, but I think the message is intended to neutralize the modern Republican advantage on national security and portray Clinton as the leader who would be steady in an international crisis. One of the effects is to inoculate her against something scary happening between now and November.

    • Josh

      To echo what others are saying: every four years the electorate gets 2% less white. Or, put another way: if Gore and Bush had been campaigning with 2012’s electorate, Gore would likely have won FL by 3-4%.

      So things are clearly moving in one direction, and not the other.

    • Ed Wittens Cat

      Clinton is defo not insulated against fallout (sry, pun) from a dirty bomb.
      2/3 of #US respondents believe in continuing to bomb ISIS even tho 1/2 dont believe it will eradicate the group.
      Trump is betting on terror– its a pretty safe bet

    • Brian Tucker-Hill

      Right-wing ethno-nationalism has some support in the United States, as it does in every other developed country. But the idea it is actually gaining support is not well-supported by the available evidence.

    • Olav Grinde

      American authoritarianism need not be rooted in ethno-nationalism.

      While Trump’s appeal among “angry white men” has been pointed out repeatedly, I think it would be a huge mistake to assume that this is the only possible base of support for a future American authoritarian leader.

      The fruitful ground of wide-spread fear and anxiety may well be rooted in economic uncertainty – for instance after a full-scale financial collapse.

  • Rey Howard

    Sam, I have a question about the “Median EV estimator” graph at the end of this post. Why are the red and yellow triangles asymmetrical across a horizontal axis? This is not the case with the similar “History of the Metamargin” graph, the triangles on which *are* symmetrical across a horizontal axis.

    • billsct

      I believe the assumetry occurs because the predicted strike zone takes into account the history of the Metamargin (i.e. The trace of the meta margin from the beginning of the cycle. )

    • S L Hugh

      I think it is because the error estimates come from permutation analysis, rather than being a computed standard error. The metamargin lives in a fairly symmetric zone, that is, it is as likely to diverge in one direction as another. The electoral vote median (340), on the other hand, has more room to go down than to go up, that is, it is more likely to move towards the 50-50 point (270 EV) than to become still more asymmetric.

    • Matt McIrvin

      Once Hillary’s win gets much above the electoral count of Obama in 2008 (365 EV), we start getting into states that are so deep-red that it would take huge popular-vote margins for them to flip. (Even if Clinton were leading with Reagan ’84 margins, around 60-40, she wouldn’t be getting 49 states.) That puts a harder ceiling on the EV count than on the Meta-Margin.

    • BillSct

      I menat to say “EV Estimator” not “Metamargin”. Think about it this way, the assymetry of the strike zone reflects the dip in the EV estimate that corresponded with Hillary’s email issue before the RNC convention.

    • Rey Howard

      Interestingly, since I made this query the asymmetry has completely gone away — despite the fact that it’s been present for quite some time. Sam, can you clue us in as to what adjustment you made? That change isn’t likely to be a coincidence.

    • Jeremiah

      @Rey Howard

      Sam posted something about this last night but it went away. Apparently the asymmetry was caused because the red states were still using 2012’s election data and hadn’t been updated to reflect the new polling. This has been fixed now. Of course, 2012 numbers would be more favorable to Trump.

    • Rey Howard

      Got it. Thank you, Jeremiah!

  • John A.

    I’m disappointed that Trump’s not lower.

    Hillary’s not higher because her own negatives (real and manufactured) plus sexist voters are keeping Trump @ 40%

    Don’t overlook people discounting Hillary because she’s a woman.

    A man with her credentials would easily be getting 70% with the gaffes Trump has been making.

    • Mark F.

      You Democrats are something. Your gal is winning and you are upset that she is not winning by more. Please.

    • Ketan


      I think you misunderstand the 40% discussion. (Or I am misunderstanding you.) If Obama were leading Romney with Romney having only 40% of the popular vote, the “You Democrats” would not be upset that Obama wasn’t getting more.

      Trump has deeply offended many people, across parties, across backgrounds, across time and space. The idea that 40% of Americans are crypto-Trumps and agree with his statements goes beyond this election.

    • Gregory Scott

      The reason for running up the score is that it takes a presidential landslide for the Democrats to take the rather gerrymandered House. That’s the game. With the House, Democrats can legislate. Without it, no one can. (Assuming Trump is indeed toast.)

  • Amitabh Lath

    Looking at this scatter plot just I’m just itching to dig deeper. What is it about the internals of SC and WI that causes them to have the largest deviations from the diagonal?

    Maybe google correlate can help? Do LV in WI, SC (and maybe GA, VA) have similar search patterns?

    • PR

      I have a separate hypothesis for each state. In SC, ~66% of the electorate is made up of non-whites and college whites, which is the same percentage as in Pennsylvania and in the US as a whole. We know Trump is underperforming with these groups, so their (relatively) high percentage in SC is dragging down his numbers there.

      In Wisconsin, Charlie Sykes, the influential king of Republican talk radio there, has been anti-Trump from day one. He led the charge to rally behind Cruz for the Wisconsin primary and I’m sure he’s been bashing Trump on his show every day. This has probably killed Trump’s numbers among Wisconsin Republicans.

    • Bela Lubkin

      Paul Ryan (R-WI) was the VP candidate in 2012. While this was not sufficient for Romney/Ryan to win the state, it probably did boost the R ticket by a few points, enough to show an anomaly in this graph.

      PR’s comments about this year are also apropos. Trump lost badly to Cruz in this year’s WI primary. I can’t guess whether there was an organic component to this — some natural reason for Wisconsinites to particularly dislike Trump — or whether it was all due to talk radio, but some small lingering effect might still be showing up.

    • Amitabh Lath

      Thanks PR, Bela, demographics definitely look like the right answer but then why no effect in IA, NV, MO (although it might explain ME).

      As for the Ryan effect in WI, yes. But then what is it about WI that goes for Ryan/Cruz and not Trump? Look at other OH, IA (IL is probably not a good state to compare with being Obama’s home state).

    • Will Hutchinson

      To add to the comments below, DJT did not help himself in the Wisconsin primary by trashing Gov. Scott Walker, who is very popular among the Republican base. He is probably more popular than Paul Ryan. Walker has shown he has an acute sense of revenge. I would not be surprised that it is manifesting itself in this election.

  • June bug

    “The fact that Donald J. Trump has the support of 40 % of American voters is indeed depressing and terrifying. But I think it goes deeper than you suggest; Trump’s support signals that the USA is almost ready for an authoritarian leader…I think it highly likely that a similar candidate (Cruz?) will be victorious in 2020 or 2024..”

    I’m trying to be a bit more optimistic than Olav. I believe (or want to) that the 40% are aging, and thus will represent an ever smaller percentage of the voting population in future elections. (fingers crossed)

    • George

      Both aging out and being demographically overwhelmed. Once they are gone, there will still be a rump group of xenophobes, but we might see a situation where the millenials and post-millenials start diverging within groups over issues other than guns, gays, abortions and skin color

    • Amitabh Lath

      Yglesias and recently Krugman have hypothesized that Trump’s support is largely based on racism. In other words, it is the “hidden variable” driving the 40% of the elctorate that Sam identifies (with tax/immigration reform and other Republican issues as cover).

      Of course one way to test this would be for the Republicans to nominate a candidate of color while the Democrats fielded a WASP. Failing that, google correlate?

    • Amitabh Lath

      Yes, I see the flaw in my experimental methodology.

    • Ed Wittens Cat

      if Project Ivy works as intended the GOP will never hold a branch of government again.
      The demographic timer will go off early.

    • ajay

      “Of course one way to test this would be for the Republicans to nominate a candidate of color while the Democrats fielded a WASP. ”

      Clinton is a WASP (a Methodist, to be precise) and Trump is a candidate of color, that color being orange. Which, I am reliably informed, is the new black.

    • Jerry S.

      We have to wait through the demise of my cohort, older Gen X’ers born prior to 1975.

      Younger Gen X who came of age during bill Clinton through Millenials are stronger Dem voters than those of us who came of age during Reagan/Bush.

    • Phoenix Woman

      Look at the Kentucky gubernatorial election of last year.

      White voters went for Matt Bevin even though they knew he’d kill Kynect and make it much harder to get Medicaid. Why? Because he embraced the anti-gay county clerk Kim Davis, that’s why.

      People of color never screw themselves in this way. They don’t prioritize social/racial issues over their own economic interests. Yet whites do this all the time.

    • Craigo

      Phoenix Woman: Attempting to boil down complex situations to simple “economic interests” explains why liberals are so confused by this behavior. Many downscale whites derive psychic benefits from racism and supporting socially conservative candidates, and value high-status group membership over improving individual circumstances. It’s only irrational if you assume that narrowly materialist views are universal.

  • smartone

    , Donald Trump will probably still end up with at least 40% of the popular vote. That is an impressive testament to the partisan polarization that has developed since 1992.

    To put this in an even more terrifying perspective- Bill Clinton was elected with 43% of the vote in 1992.

    • Matt McIrvin

      That was a significantly three-way vote, with Ross Perot taking votes probably from both parties. The plausible scenarios for a Trump win this year always involved third-party votes eroding the Democratic nominee’s support, and I don’t think that’s changed–it’s unlikely, though, since, again, Johnson and Stein together seem to take votes from both parties in rough proportion; including them in a poll doesn’t change the Clinton-Trump margin much.

  • Lil Sister

    Looking enviously at the power of one vote, I wonder what state’s voter has the lowest power. Would it be my state — California?

    • Judd Conway

      But a change in California’s winner, as unlikely as it is that one voter can effect it, could change the entire election, while such a swing in, say, Delaware is much less likely to.

    • Michael Ralston

      Well, California has the fewest EVs per capita, the most people, and is one of the safer states.

      Maybe DC or one of the really deep red states manages to beat it by virtue of being even harder to flip + irrelevant if it does flip?

  • Joel

    Are there fewer congressional polls this year? The earlier conventions have thrown off my timeline but it feels like there would be more by this point.

    • Jeff

      Some primaries still need to happen so hard to poll ones without defined candidates.

    • Sam Wang

      No. You just started paying attention sooner this year! It’s been quite a spectacle.

      To broaden the question a bit…

      In the RCP database there were 19 national Trump-v-Clinton polls with an end date in July 2016, compared with 26 Romney-v-Obama polls with an end date in July 2012.

      Generic House preference: 15 polls in 2012, 13 polls in 2016.

  • fred flint

    I think everyone is missing something. This is not 40% of the population. It’s 40% of voters. It’s no secret that a disproportionate number of old white conservatives vote.

    Cheer up people. It’s not nearly as bad as some of you seem to think. All the numbers show that the country is moving to the left. It is not going to happen overnight and a lot of poeple need to die first because their attitudes will never change.

    • Mark F.

      I don’t know. I bet Clinton would be neck and neck with John Kasich or Marco Rubio. And remember, a lot of people thought the country was moving sharply left after Goldwater. The Democrats should have won 1968 easily, right?

    • Ketan

      I laud your attempt to cheer me up! Thanks!

      However, here’s a recent poll from Gallup from “all adults;” not just registered voters or likely voters.

      32% favorable for DJT. It’s only a few percent lower than similar polls on RV/LV. (And hasn’t changed much over the last year i.e. not just bounce or single issue thing.)

      I don’t see any numbers telling me that the country is moving to the left. (But please share some!) I see a GOP House/Senate as well as 4,125 of 7,383 state legislators.

      And I don’t know how to be quantitative, but I see the GOP moving right, not towards the center.

    • Craigo

      To win the Republican nomination, Kasich and Rubio would have to do and say things that would make average voters like them less, and do and say them under a very bright spotlight. Candidates who don’t seem like a serious threat to win don’t attract serious attention from the general electorate.

      Cf. Clinton’s sky high approval ratings before she was the prohibitive favorite for the Democratic nomination, and Republicans remembered that they were supposed to hate her. Or the tendency of outgoing president’s approval ratings to rise in their last few months of office. It becomes “safe” for opposing partisans to approve of them at that point.

    • Jeremiah

      @KETAN I found this from Gallup which shows some drift towards liberal ideology in the self-identification of the population:

    • Ketan


      Thanks! Very interesting info and it led me to this link which shocked me.

      People really don’t change their lib/con identification (much) as they get older; even as the parties and issues have been changing..

  • Ben

    Sam: Barry Goldwater earned 38.5% of the vote share in 1964, and he lost by the widest popular margin in modern US history, didn’t he? So is Trump’s 40%, disappointing as it is, that much more a signal of modern partisanship than whatever drove Goldwater’s share?

    • Matt McIrvin

      The most lopsided vote in an essentially two-way election since the founding of the modern parties was Warren Harding over James M. Cox in 1920, and even there, Cox got 34.2% (and still won the South). To see bigger blowouts than that you have to go back to James Monroe.

      This election may be a test of how much of the general closeness of presidential elections is pure tribal affiliation, and how much comes from the fact that both parties are usually mounting a competent, well-funded national campaign of some sort.

    • Sam Wang

      Actually, the benchmark is George McGovern (D) in 1972: 37.5%.

      If I had to guess, I would think Trump could end up with 42% of the national vote. This would put him ahead of Goldwater 1964, McGovern 1972, Mondale 1984, and tied with Stevenson 1956. Yet it will be regarded as a huge loss – and it is, by current standards. It illustrates the extreme polarization that has arisen in the last 20 years.

    • Partha Neogy

      I see that partisanship and polarization are being offered as the explanation for the high floors that recent losers have achieved. Is there any evidence that partisanship/polarization is isotropic?

  • Phil B

    Campaign shakeup? On message? Yes, if the message is AM talk radio.

  • Scott

    Best opening paragraph ever.

  • Charles McCullough

    Total BS. This poll is a poll cat if you know what I mean. It stinks to high heaven. When a real and accurate poll was taken with a sampling of over 100,000 people Donald Trump was on top of Hillary by double percentage points. This has Trump 60 percent higher than Hillary.

  • Brian Tucker-Hill

    Random state-specific question: I have heard two theories about why Nevada is remaining relatively close in the polls. One is that Nevada is just tough to poll well. Another is that maybe Trump has some sort of special casino-guy appeal in Nevada.

    It probably won’t matter, but with Nevada it becomes easier to put together a winning map in which Clinton wins only one of PA or OH (and not FL, NC, or VA).

    So anyone have any thoughts about Nevada?

    • Fair Economist

      Nevada has a low rate of college-educated voters, who have been quite favorable to Clinton, and a bad economy, which is bad for the Presidential party.

    • Craigo

      Nevada polls frequently underestimate the Latino vote in general and the Spanish-speaking vote in general, and the Nevada Dems excel at field operations.

  • Joseph

    Re: the predicted 40% Trump figure, how much of this involves bringing people in to vote that would normally not vote, and seeing people sit out the vote who normally would vote? It sounds like a big percentage of the population, but is it really?

    Also, there are those who hod their noses and vote Republican, period.

    Finally, there must be some degree of “protest vote” going on. Some Sanders supporters, for example, or people who just respect the way Trump speaks his mind but don’t really agree with him. This last might be called an anti-politician vote.

    IOW, we can’t equate the percentage of people who vote a certain way with the percentage of people nationally that are either “Trump-like” or agree wholeheartedly with Trump.

    Just my 2 cents worth….

    • Josh T

      You can equate that 40% with people who find it acceptable for Donald Trump to be president. That’s all that matters, and that’s what I tell my Republican friends: you can try as hard as you like to convince me that I shouldn’t agree with the policies of the Democratic Party, but it’s irrelevant because there is no universe where Donald Trump is an acceptable candidate for president. I would never, ever vote for him. But unlike me, 40% of registered voters are willing to vote for him. And that correlates pretty well with the number of hardcore Republican voters. Disaffected Bernie voters are not going to vote Trump in any significant number. They will vote Green Party or Libertarian or they will stay home.

    • Bob Altemeyer

      For an analysis of that 40%, written ten years ago, see the website above.

    • Bob Altemeyer

      For an analysis of the core of that 40% see .

  • Andy

    Regarding retrenchment–in my view that is certainly worse since the advent of Fox and other right-wing outlets that use their own set of facts to refute reality. The first time I noticed was the right’s reaction to the Florida recount…so where Trump is concerned, you could list all of his missteps and 40% would respond by blaming the media for over-reporting his gaffes and underreporting hers. Just for starters. I guess I would say that Trump’s support amazes (and depresses) me, but what really makes me despair is the knowledge that with Fox entrenched, 40% is probably the floor.

  • Dan Albert

    ” That is an impressive testament to the partisan polarization that has developed since 1992.”

    You seem to view the polarization of our politics as a result of both parties chasing out moderates. If I am correct in this impression, please direct me to data to support that view.

    • Jeremiah

      Mann and Ornstein have done extensive work showing the Republicans are the party that has become more extreme. Summarized in this Washington Post article:

    • Ketan

      Polarization is not necessarily the “result” of or caused by the parties. The public’s polarization could cause the parties to move apart on ideological issues.

      The graph makes you want to blame the parties (and maybe R more than D) but, the voters did choose these folks.

    • Jeremiah

      I would have thought that the population over the decades has gotten more progressive overall. Therefore the polarization we see is a reaction to the perceived threat to the “traditional” way of life. There must be studies out there or ways to demonstrate this hypothesis.

    • Michael Coppola

      Actually, Ketan, many of those pols chose their voters rather than the other way around. We can debate the relative importance of various factors driving polarization, and I suppose that we’ll always be left with something of a chicken-egg question in the end, but we can’t deny that partisan redistricting creates a very powerful feedback amplifier.

    • Sam Wang

      Actually, partisan-gerrymandered districts are only a small fraction of lopsided districts – maybe about one-fifth of them. They do enhance the effects of polarization by creating environments where polarization can play out. David Wasserman had an excellent piece over at the Great Argental Satan’s site on the role of primaries in creating polarization.

      I think there are some big questions: why we have so many lopsided districts (geography? culture?) and where the enthusiasm for polarization is coming from (nationalized media and messaging?).

    • Michael Coppola

      Yeah, so sorting… But why are so many of us so bothered by people we never interact with? Guess maybe it’s because we never interact with them…

  • Carey Sublette

    Are there graphs of the movement of the Bayesian predictor over time for previous elections and this election?

    Inquiring minds want to see.

  • pechmerle

    Drew Linzer is finally out with his forecast for this year: As of today, 89% HRC wins the presidency. Over at
    Reassuringly close to PEC and Upshot, though he expects somewhat lower EV’s for HRC than PEC currently shows.

    • MarkS

      Details of his methodology are frustratingly opaque (as they were in 2012). PEC remains the gold standard in this regard.

  • ottovbvs

    This is not all amazing given polarization. Nor for that matter is it that much of a recent phenomenon. In 1936 Alf Landon pulled 36% of the vote. Who was going to vote for Alf Landon in 1936 but they did.

  • Amitabh Lath

    I am a little concerned about the LV filters used by pollsters. There are news articles about Democrats expanding their GOTV efforts much earlier than usual in states like GA. Usually these efforts do not start until the leaves have turned brown, but 2016 is different.

    If this is correct then asking (for example) did you vote in 2012? would filter out these newly registered voters who usually wouldn’t matter numerically at all but in the era of Trump might have a little extra motivation, and in fact looking at the August RV polls in GA you do see Clinton close or leading.

    • 538 Refugee

      I haven’t looked a a poll’s questions for awhile. Are there stats on how many newly registered voters generally show up? Seems like if you took the effort to register it might have been for a purpose. Especially this cycle. It seems the second question after ‘Are you registered to vote’ would be ‘For how long’.

    • Amitabh Lath

      538R, I don’t know exactly what the LV filters are, but I presume they are some sort of did you vote? or do you know where your polling place is?

      And in general they serve a good purpose but in case of a push factor of Trump’s magnitude (he is 1% in African American support) who knows.

      Adding to the general unease with poll numbers is that all the systematics that one hopes cancel out seem to be pushing one way. One side has no field operations, advertising, and large scale defections, negative press…

      The correct answer is that the polls should be picking all this up. But it feels messy this time around, less like a proper measurement and more like undergrad lab.

    • Ketan

      Here’s Gallup’s explanation of LV filters.

      Still looking for a more scientific (and more recent) explanation.

      I doubt that they end up being all that accurate; but sounds better than just using RV.

      In addition to GOTV, you have the problem that the demographic turnout might look different from 2012. So stuff based on the past (education, Hispanics) may be quite wrong.

      Sam, do you have opinion about LV vs RV?

Leave a Reply to Joel (Cancel)