Click the map to create your own at 
Protected: Two ways to estimate primary outcomes without polls (transcript)
This content is password protected. To view it please enter your password below: Password:...
Trump expands the battleground…to Utah and the Deep South
Historically from 1952 to 2012, the likely range of movement in two-candidate margin from this time until Election Day has been 10 percentage points, which is the standard deviation from the 16 past elections. Therefore, even though Clinton currently leads by a median margin of 7 percent (12 national surveys) and would certainly win an election held today, she could still lose the lead, and from a purely poll-based standpoint, is only narrowly favored to be elected President in November (probability: 70%).
It is also the case that Clinton is the only candidate who is poised for a blowout. Her “plus-one-sigma” outcome (current polls plus one standard deviation) is a popular vote win of 58.5%-41.5%. Trump’s plus-one-sigma outcome is a narrower win, 51.5%-48.5%.
I should point out that the last four elections, from 2000 to 2012, have been far less variable than I have calculated above. They show a standard deviation of 4 percentage points. These have been polarized years. But considering the upheaval in the Republican Party, a little voice tells me to open my mind to a wider range of possibilities…including a Trump win.
Of course, the Presidential race is played out through the Electoral College, which is composed of winner-take-all races. The basic effect of the Electoral College is to amplify the difference between the two candidates.
The map at the top of this post lists states as being uncertain by either of two criteria:
- The median of state polls since February is within 10% for either candidate; or
- If there is no polling, then the election margin in 2012 was between Obama +7% and Romney +13%.
Criterion #2 is based on the fact that Clinton-v.-Trump is currently polling about 3 percentage points more Democratic than the Obama-v.-Romney vote in 2012. So a Romney +3% state would be right on the edge at this moment in time.
Obviously, this highly provisional map is simply a starting point. The list of uncertain states will change as more polls become available. For now, it looks like Democrats will have at least 262 electoral votes, and Republicans will have at least 122 electoral votes, with 154 electoral votes up for grabs. It takes 270 electoral votes to win, so Republicans face an uphill climb.
Another way to look at the data is to force a win for whichever candidate is leading in each state. This does not take into account all the possibilities. But it does give the mode – the single most likely combination of wins and losses. That mode gives a total of Clinton 364 EV, Trump 174 EV, very close to Obama’s win in 2008. This is totally consistent with the fact that Obama won the popular vote by 7% – the same as Clinton’s national poll margin today. The map would look like this.
If you do not like how one particular state is colored: that does not matter. You are getting lost in details. What matters is the overall picture: if the election were held today, the probability of a Clinton victory would be over 99 percent, and few people would care which way Utah or Missouri went. However, the election is still nearly six months off. Republicans’ main hope lies in the possibility that opinion will move across the board nationwide.
So far, I have taken a fairly conventional approach to this year’s race, and assumed that any swing would be equal across the board. But Trump has upended the national Republican party. So it is argued that he has the potential to scramble the national race as well – and win states that were previously out of reach for Republicans.
We can start looking for evidence of a Trump Scramble by plotting current state polls against 2012 election results:This plot is based on polling data from RealClearPolitics and HuffPollster. In these 15 states*, which include all the usual suspects, the Clinton-Trump margin is 3.0+/-4.4% (median +/-estimated SD) more Democratic than Obama-Romney. Overall, it is consistent with an across-the-board shift. Trump has not scrambled the map yet.
There is one massive exception: Utah. In 2012, Utah went for Mitt Romney by a 73%-25% margin, the largest margin in the nation. A recent poll shows Clinton up by 2%. This is a single poll and therefore subject to much uncertainty. But even if it is off by 10% (i.e. Trump actually leads by 8%), it is still over eight sigma away from the other 14 states. I am really not feeling the null hypothesis there.
From this I would say that Utah Republicans probably dislike Donald Trump, enough to express a preference for Hillary Clinton. Considering how radioactive Clinton is among Republicans, that is really saying something. It’s hard to imagine that Utah voters like either option much. No matter who wins Utah in November, they may well have lower-than-usual turnout.
Are there other exceptionally anti-Trump states? It depends on why Utahns dislike Trump. If it’s connected with the fact that 60% of them are Mormon, the only other state that comes close is Idaho, which is 24% Mormon.
GOP primaries could provide another clue. Ted Cruz won Utah with 69% of the vote compared with Trump’s 14%, a massive blowout. However, in Idaho the Cruz-Trump vote was only 45% to 28%, so that state seems safely Republican. One could conceivably imagine trouble for Trump in other states where Cruz was probably ahead in the primary: Montana (Google Correlate-imputed lead, Cruz +45%), and South Dakota (Google Correlate, Cruz +46%). Even one general-election poll in either state would test this idea.
On the Democratic side, no hugely anti-Clinton states have emerged. The closest is New York, in which Clinton is at +20%, 8 points worse than Obama’s showing in 2012. However, New York is Trump’s home state more than it is Clinton’s. So that could just be a favorite-son effect.
*NY, MI, NH, PA, VA, OH, FL, NC, AZ, GA, MO, IN, MS. WV, and UT.
Topics:
Interesting way to set the starting map. When will we start seeing the full projections of the EC? I assume the state-by-state polling isn’t plentiful enough yet. But are you also waiting until Bernie drops out?
“Party, a little voice tells me to open my mind to a wider range of possibilities…including a Trump win.”
That liberal ‘hair on fire’ thing? 😉 Seriously, my gut feeling is that Trump really, really, really doesn’t understand women and just how much his attitudes towards them is going to come back to haunt him. I could see Hillary possibly outperforming her polls among woman who are afraid to publicly admit support but will feel empowered in the privacy of the voting booth. That said, I’ll be nervous for a while on this one until the campaigns reach steady state.
People thought Utah would go strongly for Romney but he just won the same share as any historic Republican. So will Trump.
We Utahns are good at sticking together and expressing disappointment but when the time comes to vote, Utah Republicans don’t vote for Clintons. Bill came in third here in 1992.
So priors should be strong and attention to transient emotions should be weak. Don’t be fooled by one poll taken while all the authorities were backing Cruz. Utah will turn out to vote the same as it always does in presidential races. Trump will win by about 40%.
And it’s not just Utah that will return to priors and reject transient springtime poll numbers. Florida will be razor sharp again. Mississippi will be solidly red by a narrow but rock solid stable margin. GA, SC, and NC will be similarly back to their usual red form.
The only states I see being more interesting than usual are NJ and AZ. Trump could scramble those two but good.
That is not my point. Now matter how inaccurate that Utah poll is, there is almost certainly an unusual deviation in your state from the last election. Sure, the winner could be the Republican nominee. But I rather doubt that Utah will once again be the most Republican state in the country in November. Something is going on there.
As for priors…you and the Argental Satan!
Brian, your comment convinces me Republicans no longer view Utah as safe. Otherwise, why make it the focus of a typical Republican fantasy-election agit-fiction comment? I’d say that’s an interesting piece of evidence we have right here…
Sam: In 2010 polls showed a close, competitive race for Senate if the Republicans rejected long time senator Bennett and didn’t even allow him to run in the primary. We had a serious experienced conservative Democratic state senator running. Once the primary was over, the Republican base voters reconciled and the Republican Mike Lee won by 30%.
When ultra-right Republican Jon Huntsman Jr. ran against incumbent moderate (for Utah, still very conservative) Republican Olene Walker and the Republican party denied her the chance even to run in the primary voters were deeply disappointed. According to polls and our former-US Attorney Democratic candidate was polling even. Republican voters reconciled and soon Huntsman won by 16% in November.
(I know Huntsman has national audiences fooled. When he had a chance to actually govern, he made Mitt Romney look like Bernie Sanders.)
The same thing happened in UT1 and UT2 congressional in 2002 and UT3 in 2008. And in many other elections, we see the same effect.
Republican Utahns have a deeply loyal streak to favorite primary candidates and take a while to come around, but they eventually vote for the Republican. My experience — and you can’t have a model without priors — trained on Utah elections indicates that Trump will win the same 40%ish margin Republicans usually do for president.
I don’t know if that will be Trump’s highest in the country but as usual it will be close to highest.
Argental Satan’s problem isn’t relying on history to train his model or priors. He plainly has a partisan axe to grind against Trump, though he doesn’t explain just which issues it is based on. And he surrounds himself with similarly hysterical writers producing free flowing vitriolic evidence-free insults. If one has an agenda that prevents him from seeing things objectively or even considering any outside viewpoint, he’s going to have trouble building a useful model.
Some: I’m a lifelong Democrat and therefore can’t tell you what Republicans think in their dirty secret hearts but they don’t look worried.
This isn’t the first time I’ve heard somebody claim that Trump could win New Jersey. Where does that come from? Chris Christie? Trump as a quasi-local? Some notion of the state’s general temperament? There are few general-election polls there, but the ones I’ve seen show Clinton leading about 50-35%.
Probably from Trump. He has consistently claimed that he will mobilize the Reagan Democrats, and will win blue states such as NY and NJ.
I just looked it up: New Jersey is 58.9% non-Hispanic white. That’s difficult territory for Trump. Not impossible: he still seems competitive or dominant in several states that are lower. But in an already historically liberal state?
I do not think that there is a chance in a billion that Trump could take NY or NJ. The discussion on traditional blue states where Trump may have an opportunity that differs from other Republicans are MI, PA, OH which while being blue, have not been the deep blue of NY or NJ.
” Trump could win New Jersey. Where does that come from?”
Trump and his campaign manager Manafort.
“already historically liberal state?”
I think it is more properly an historically liberal state since 1996 in the sense that when presented a Republican from Texas, Plains, or Mountain West it rejected them by significant margins.
“I do not think that there is a chance in a billion that Trump could take NY or NJ.”
If Trump gets numbers close to like Guiliani and Bloomberg did in New York City for 5 elections, then he will win New York State. I don’t think that is one chance in a billion. It might look unlikely right now, but I wouldn’t dismiss it out of hand.
New Jersey was already polling as Trump’s strongest state of all the primary states before Cruz and Kasich dropped out – with Trump getting 70% of the vote. Many New Jerseyans are ex-New Yorkers from places like Queens and the Bronx and identify with Trump where they don’t identify with a George W. Bush. I’d also note that for a “Blue” state, New Jersey is actually overwhelmingly non-partisan, with around ~47% of voters not affiliated with a party (partisans are ~33%D, 20%R).
I personally think Trump has as good a chance in New Jersey as he does in PA, MI, and NH. I would also agree with his campaign manager Manafort that he also has just as good a chance in Delaware.
I’d just say that is the view of the candidate and his campaign, and so far they have surprised nearly everyone with the actual results they achieved. Maybe not Prof. Wang, but everyone else.
“traditional blue states where Trump may have an opportunity that differs from other Republicans are MI, PA, OH”
Except for Presidential elections, these states are not particularly Democratic. They have frequently had Republican governors and senators and constitutional row officers (Attorney General, Treasurer, etc.), and Republicans have controlled the state legislature and house delegations much more often than not since 1994. I think people are too focused on presidential elections when classifying states. This leads to a tunnel vision where we think a map is fixed and simply shifts as the national partisan mood shifts. This goes wrong when things change because of a change in circumstances.
Its not much of a stretch to think that if a Republican were running for President who is similar to the many local Republicans who hold office that they could easily carry these states.
NJ is Lucy holding the football for the GOP.
“Lucy with the football for Republicans” is how I’d describe Pennsylvania: it always seems like a close swing state, but a Republican presidential candidate hasn’t won it after the landslide of 1988. And since the heavily African-American, underserved urban precincts of Philadelphia tend to report late on Election Day, it always seems like it gets pulled out at the last minute after trending red all night long. Maybe this time…
“And since the heavily African-American, underserved urban precincts of Philadelphia tend to report late on Election Day”
Actually, since the Philadelphia voting system was modernized about 12 years ago under a great group of reform County Commissioners, this is no longer the case. Philadelphia now is typically among those to report first. In the recent primary, Philadelphia reported first, and the last reports came from rural Republican counties in central PA.
As late as 2002, election results were still being reported and tabulated by hand on manual spreadsheets in Philadelphia from each precinct and ward for each race and then sent to the County Commissioners and the Elections Department. They were never even entered into a computer, because if you asked for results (and I did every two years for all races), you got a copy of these hand tabulations.
Also, the reason Republican presidential candidates tend to lose in Pennsylvania was suburban vote splitters who rationalized in their minds voting say Gore for President and Santorum for Senate along with their local Republican congressman or Kerry for President and Specter for Senate along with their local Republican congressman. The percentage of these vote splitters in the electorate is as high as 20% in Bucks, Montgomery, and Delaware Counties, as well as wards in Northeast and South Philadelphia.
These vote splitters are also why talk of gerrymandering of Districts 6, 7, and 8 around Philadelphia ring hollow to my ears. Attractive Democratic candidates for other offices are perfectly capable of carrying these districts with 55 or 60 or 65% of the vote even as the usually Republican congressman gets the same percentage in reverse.
Basically, its not possible to say that African Americans in Philadelphia are the secret sauce for Democrats winning Pennsylvania presidential elections, because those very same African Americans do not present an impediment to Republicans regularly winning Senate seats and state row offices. The real secret sauce is Democrats have run candidates for President far more attractive to moderate or fiscal conservative/social liberal suburban swing voters than Republicans have.
What percentage of Republican voters will vote for the Libertarian candidate?
That’s a good question, but it is incorrect to assume he will draw mainly from the GOP. In the 2014 VA governor’s race, the Democrat was the second choice of many Libertarian voters. Remember that Libertarians have many positions that will appeal to some liberals who don’t like Clinton.
Arizona went from Romney by 8 pts but is now polling 3.5 to 7 pts in favor of Clinton. Guessing that Trump’s toxicity with the Hispanic vote has a lot to do with that swing. Which suggests that John McCain has just committed political suicide by aligning himself however tepidly with Trump. Will be interesting to see if Trump does to the GOP nationally what Pete Wilson and Prop 187 did to the Republican Party in California. Are there down ticket races in Arizona and Texas worth following as tests of this?
I guess McCain knows more about Arizona politics and has internal polls, that he is forced to endorse Trump says a lot about his struggles with primary challenger Kelli Ward, as well as Trump’s strength in the state. AZ has little to no chance of going Democrat. Putting a lot of faith in one poll, so far out is not wise.
Prop 187 won with some 59 percent of the popular vote in California. It also resuscitated the political career of Gov. Pete Wilson, who campaigned for it, and carried to victory at least five Republican congressional candidates, which helped the party win a congressional majority in 1994.
LondonBob: The success of Prop 187 in revitalizing the CA Republican Party was extremely short lived, and many people see it as the nativist overreach that finally made the GOP close to irrelevant in CA.
Texas currently has only one competitive House race (per Cook), and that is TX-23 where former rep. Pete Gallegos is trying to knock off Will Hurd. I think at this point Gallegos is probably a narrow favorite, given that this southwest Texas district is liable to go massively against Donald Trump; it is a majority-Hispanic district and Trump only won two extremely-rural counties, neither of which cast more than 80 votes for the Donald; there is actually one county in this district, Zavala County, where not one single ballot was cast in the GOP primary on March 1st).
Other than that, there are few statewide races this year. There is a race for Railroad Commissioner and some judicial races I think. Locally, the entire state House is up for re-election. There is that crazy lady in East Texas who thinks Obama was a gay prostitute, running for the State Board of Education. Also, Austin and some other cities have city council elections.
Almost all of Texas elections are held in off-years (one of the reasons why the GOP dominates the state is that they are only on the ballot when nobody votes).
California began leaning Democratic in the modern Presidential vote era in 1984. The last GOP Senator elected was Wilson in 1988. The last GOP majority in the Legislature was 1970. The last GOP majority in the US House delegation was in 1958. The GOP is irrelevant now in California because Prop 187 was overruled by courts and a significant section of the Republican base moved out of the state with the decline of the defense industry in southern California and were replaced as residents by foreign immigrants with significantly less affinity for the GOP.
https://en.wikipedia.org/wiki/Political_party_strength_in_California
Hey Sam — I love the blog and followed it religiously for the last election. I noticed the meta-margin graph goes back to late May or so for 2012. When do you anticipate starting the meta-margin and EV distributions for 2016? Maybe around the conventions, though the candidates are more or less chosen already? Or when polls become more plentiful?
Agree with Ryan. Izzit time soon, Sam?
Favorite son effect? While he certainly has name recognition in NY, is there any evidence that it translates to likability (or votes)?
60% Trump 25% Kasich 14% Cruz in GOP primary.
bks, this discussion is about the general election. When counting all voters, New York is solidly in the Democratic side of the tally. DT winning a majority of a minority party is not the same as being a favorite. NY State has not voted for a R presidential candidate since 1984. The lowest margin for D’s since then was 1988 at +4%, the next lowest was +16%, and the median is over +20%.
Read the question again JerryA. I didn’t say he would win NY, just that his name recognition has already generated a supermajority of GOP votes in NY. I certainly can’t produce evidence from the future.
The reason for Wlezien and Erikson data looking so different post 2000 could be the earlier start of saturation election coverage.
Mid-20th-century election coverage was probably a few columns-inches in the middle pages of local newspapers and a few minutes on Cronkite. Now we have 24 hour TV and web coverage which may be leading to ever earlier decision making by the electorate.
In spite of the upheaval in the Republican party, if anything the coverage this election has been even more intense, which would presage an earlier hardening of opinions and a smaller shift in polling between now and November.
I thought I’d check on Utah in non-Romney years, and for the three elections before 2012, the vote averaged 67% Republican, 28.9% Democratic. So, it’s probably normally more of a 40% margin than a 50% one, really, but the significance of that swing still stands.
Quinnipiac polls out. Trump up in Ohio, tied in FL and PA.
Clinton might not be leading in the key states.
https://www.qu.edu/news-and-events/quinnipiac-university-poll/2016-presidential-swing-state-polls/release-detail?ReleaseID=2345
The national average would look very different without the CNN/ORC poll since every other poll has both candidates mired in the 40’s where they would be expected right now. Every year, someone is always way off. Maybe it’s their turn. So far it has been NBC/WSJ who has been understating Trump support levels in the primaries.
Also, the ISideWith map predicts Trump trouble in Utah based on specific weakness in Salt Lake City and its southern suburbs. I’d begin with theorizing that has to do with emerging liberal culture in Salt Lake City making it much more like Denver and Albuquerque, and Trump turning off a certain percentage of well-to-do Mormons in the suburbs.
I view it as bad procedure to focus on single polls (yes, I made an exception for the weird Utah result, which seemed notable). My usual policy is to delete such comments, which are susceptible to motivated reasoning. For example, in Ohio there were 4 post-March polls with a median of Clinton +5.5%; your new information makes that Clinton +5% (n=5).
The idea that Utah has generically swung toward Democrats by fifty points since 2012 is not plausible. Or do we suppose Senator Mike Lee is suddenly vulnerable?
I would also note that although the Democratic primary has been decided, it is not over. There are many Bernie supporters not yet ready to say they’ll vote for Clinton over Trump. I expect this will change, but feelings are intense on this subject right now.
Even that weird Utah poll showed either Cruz or Kasich beating Hillary Clinton by tremendous margins. It was about Donald Trump specifically.
The season of people freaking out about individual SHOCK POLLS has officially begun.
Mitt is Mormon, and I think it was pretty well accepted that gave him a big boost in Utah.
A fairer year to year comparison might be to Utah in 2008, when McCain squeaked by with a 28% margin over Obama.
Andrew: Salt Lake City has been extremely lefty for decades. We had a mayor leading national protests against Dubya, a flotilla of gay and lesbian state legislators when gay marriage was an issue (one now mayor), and a strong reliable blue lean at every level. I can go canvassing for the Sierra Club and every house where anyone is home wants to sign a postcard petition and a large share of them want to hand me cash money.
We can’t outvote Utah County, though. Not remotely.
I’ll repeat a point I made earlier: I’m not sure we can assume the undecided group will be split symmetrically between the candidates. It is a very large group in most recent polls (I’m not sure if this is typical of this stage in the campaign, but with two almost universally known candidates, an undecided group approaching 20% in many polls—even 34% in one recent NH poll—seems a bit high). There is an argument to be made that this group has many Trump voters, who would not tell so to a polster, as part of it (see: “Bradley Effect”). If that is the case, I’d also try to see where Clinton is polling above (or at least very close to) 50%. It’s probably less of a concern now, before polls switched to LV, but something to bear in mind.
I could be wrong but I believe one of the takeaways from poll aggregation is that the closer we are to the election, the less likely there are to be truly “undecided” voters and so the chance they’re going to unevenly split in favor of one candidate over the other is low. Back in 2004 I believe Dr. Wang (incorrectly) predicted that undecideds would break late for Kerry. http://election.princeton.edu/2004/11/03/wrong-assumptions/
True, but there are circumstances that make things different. Trump is now painted in many media accounts (correctly, in my view) as an unacceptable choice in November (though right-leaning media sources make the same claims about Clinton as well). Also, misogyny is an important ingredient in Trump’s appeal, running against a woman. Now, some major past polling failures (and by that I mean failures of the average/median of polls) have been retroactively explained by surveyed voters falsely claiming to be undecided, when in fact they planned to vote for the candidate considered unacceptable, but did not want to reveal their choice to a stranger on the phone. This sort of factor could well be in play in this election round, but wasn’t meaningfully present (or cancelled out on both sides) in 2004, and any other run-of-the-mill campaign.
There are ways to test for whether this effect is in place in advance. Especially telling would be a visible difference between the results of polls using different methodologies (live polling vs. IVR; web polling would depend on the specific methods used). My point is that we should all keep our eyes open and not be lulled too much by mere difference in topline numbers between Clinton and Trump.
Again Pollster’s customizable charts are pretty fun for taking a quick look at questions like this.
Right now Clinton is +6.2 in the all-polls aggregate:
http://e.huffpost.com/screenshooter/elections.huffingtonpost.com/pollster/embed/ss2/2016-general-election-trump-vs-clinton/20160510190151445.png
If you eliminate all the live interviewer polls, that goes down to +3.9%:
http://e.huffpost.com/screenshooter/elections.huffingtonpost.com/pollster/embed/ss2/2016-general-election-trump-vs-clinton/20160510190323907.png
So–maybe? Probably too soon to tell if this is a persistent effect, or to put an estimate on the magnitude if it does persist.
Our host’s great mistake in 2004 was assuming a greatly asymmetrical break in undecideds. Since then he’s avoided that kind of thing.
Matt—see my reply to Brian above.
Brian Tucker-Hill—yep. To be sure, the same gap could have another plausible explanation (cell-phone only housholds), and above all, we have some months to wait and see, but I’d keep an eye on that, anyway.
“Now, some major past polling failures (and by that I mean failures of the average/median of polls) have been retroactively explained by surveyed voters falsely claiming to be undecided, when in fact they planned to vote for the candidate considered unacceptable, but did not want to reveal their choice to a stranger on the phone.”
Do you have a link for that? I don’t follow things terribly closely, but the only related thing I remember is the (actual) Bradley effect manifestly failing to materialize in the 2008 election after being the subject of intense speculation.
I think Morning Consult claimed early on that live-interviewer polls are undercounting Trump’s vote because of a Shy Trumpster effect. But Morning Consult’s numbers are actually close to the average of all polls now. It looks as if Clinton gets a ten-point gap from live-phone polls, a six-point gap from Internet polls and almost no gap at all from automated-phone polls.
But the automated phone polls are just Rasmussen, Gravis Marketing and Emerson College Polling Society. Rasmussen and Gravis are both famous for having fairly bad track records in past cycles.
I actually voted in the election that is the second-most-famous example of the Bradley Effect: the 1989 Virginia governor’s race, in which Doug Wilder (the first black governor since Reconstruction) was heavily favored to beat Marshall Coleman, whose ad campaign had thinly coded racial dogwhistling (“is that the kind of man we want for governor?”) Wilder won, but considerably underperformed his polling, and the election was unexpectedly close.
The discrepancy showed up even in exit polling, but I’m less impressed by that fact now than I was at the time, knowing what I know about exit polls.
There’s some controversy over whether the Bradley Effect actually consistently existed, but most analysts seem to think that at the very least it doesn’t exist today: people voting against black candidates find a socially acceptable rationale, they don’t just lie. But shy Trumpsters may be another story. The thing is, there are so many Trump fans who are the opposite of shy.
It’s also interesting to look at the Republican primaries to see if there’s any hint of Trump overperforming his polls there. If I recall correctly, Ted Cruz was the one with significant overperformance until the later primaries, in which Trump started pulling ahead of his poll numbers… but he was also overwhelmingly winning in those contests, by a rapidly increasing margin, and Sam suggested that the polls were just failing to capture last-minute time variation.
And see this from NYT (Th. Edsall) today: http://www.nytimes.com/2016/05/11/opinion/campaign-stops/how-many-people-support-trump-but-dont-want-to-admit-it.html
The Bradley Effect and Shy Tory Effect (UK Parliamentary election, 1992) are the two classical examples I had in mind, of course. Edsall cites more evidence from Pew. Social desirability is (potentially) a factor this time around, in ways it wasn’t in recent years, even with Obama on the ballot.
The way I understand the phenomenon (and that’s just my subjective interpretation, to be sure), it’s not just the identity of one candidate on the ballot; it’s also the level of explicit or near-explicit bigotry the other candidate is willing to use in his campaign).
That’s a lot of ink spilled for an effect that is probably no more than two percentage points.
Sam—That analysis of yours looks at race (not gender), and, more importantly, doesn’t try to quantify the level of racism employed by the other candidate (not that I know how to quantify that). Considering the utter shamelessness of Trump’s all-round bigotry, I’m not all that sure it will extrapolate to our scenario. But again, when we get to October or so, we’ll have relatively clear indicators: difference between polls by methodology and comfortable Clinton leads where she nevertheless remains somewhat below 50% in key states. If we see both of these (and that’s an if all right), I’d consider that a possible sign of a systematic error in polling, possibly going beyond those 2%.
Love your analysis. I don’t see how any turmoil in the Republican Party translates into more votes for Trump. It would seem that it would translate into fewer votes. I also wonder what effect the hardened opinions on both these candidates has on the polling. To get a big enough swing in the polling for a Trump win, there would have to be a big opinion change on Clinton. That seems rather unlikely given her 25+ years in the public eye.
Hi Sam,
Keep up the good work.
I have a question. The media will spend a lot of time talking up the notion of an October Surprise. Is there any statistical evidence that such a thing actually exists, or is it a largely made up explanation ex post for how a race turned out?
Before we get buried by the onslaught of media stories I thought it would be worth knowing what the data actually says, if anything.
Of all the polls done in FL Trump is up in more than he is down, especially if you ignore the obvious outlier done by some mysterious lobbyist group.
The PPP and Quinnipac today are suggesting Trump is consolidating Republicans now he is the presumptive nominee, I assume this process began before IN voted. The NBC in IN (48-41) also supports this.
Can you tell me where you’re getting your data that says Trump is up in FL polling averages? On both RCP and HuffPost, over the last three months, Clinton leads by between 4-5% on average. If you discard older polls and only look at the last month, it becomes Clinton +6-7%, which would put FL just to the right of the country as a whole.
Please remember that single polls have almost no predictive value, especially this early on. Polling averages give a much clearer picture.
RCP. Twelve reputable polls done, AIF isn’t reputable, shows Trump up in 7 of them and HRC up in 5. Can’t be bothered to do the average. At this juncture in time I would guess it is too close to call, but I would expect Trump to take it relatively comfortably come the election.
I think the point of using polling averages is that it allows one to use all of the polling data. If you start making qualitative judgments about which poll to use and which not to use, it defeats the purpose.
Well you are free to construct your own polling average from the twelve polls done by recognised pollsters. I just can’t be bothered when a cursory glance tells me it is margin of error stuff.
Pollster lets you customize and then link charts. Here is one for Florida with AIF removed:
http://elections.huffingtonpost.com/pollster/2016-florida-presidential-general-election-trump-vs-clinton/edit#!hiddenpollsters=associated-industries-of-florida-r&maxdate=2016-05-10
It still shows Clinton +2%. Note the concept of a MOE is a bit difficult to apply to polling aggregates–if this was a lot of polls and right before the election, a +2% in the aggregate would imply a very high chance of winning.
My policy is not to pick and choose which polls I think are valid. You get into trouble that way.
http://www.realclearpolitics.com/epolls/2016/president/fl/florida_trump_vs_clinton-5635.html
Hillary has a clear lead among polls done since the start of the primaries. I guess you could go back even further into 2015–which you apparently did, in order to get your results–but I’m not sure adding a Trump-Clinton general poll from August of last year really helps your case…
IIRC, Sam’s quadrennial general-election analysis has a way of aggregating polls that de-weights weirdo outliers without having to designate certain polls as Bad; I think it involves using medians rather than means.
Nate Silver’s method is instead to do all these calculations of weights and house effects to discount for every single poll, and it doesn’t do any better, so simple is probably best.
Well, I don’t know much about statistics being a political science/history/MA/MBA guy, but I’ll tell you, here in Northern Virginia you DO NOT see any Trump stickers as of yet. None. It’s uncanny, nothing.
I have lived here since 1989 and by this time in the cycle you traditionally see large numbers of bumper stickers, etc., but this year no Trump. Not a one.
Over on Washington Monthly/Booman Tribune Longman sees the making of a blowout, and I think that might occur too. Not reflected in the polls yet, but for those of use that go by instinct, gut feel and smell tests, this sure has the makings of a blowout.
I agree with the above commenter Brian that Utah will remain Republican. But the mere fact that Hillary is within 10 is just astonishing. Really.
If Democrats had a really popular candidate, then we’d have the makings of a 1932 style blowout. But “what ifs” are pointless.
One thing I do know is that smarter Republicans are understanding that the stench may stick to them beyond this election.
Demographics are destiny, and I believe demographically the composition of the nation currently changes at .5 percent per year, but that is soon to go up to almost 1 percent. This kind of thing puts them in a real straightjacket, and given the composition of THEIR support, and even tighter straightjacket.
It’s not a given that we have to march forward in time with the two same political parties. The Federalists and Whigs disappeared. We have had the same two major parties since the 1850’s, and so we may be seeing the back side of the Sixth Party System here in the states.
I would love to see Sam model that, one of the main variables would be – I believe – political positioning. The Federalists are the more apt comparison to the modern day, as they were ground into dust by a number of variables, with one of the main ones being “inability to adjust to changing culture.” The Whigs were ground into dust by slavery, so I don’t think that’s an apt comparison.
So, from a Sam perspective, n=1 in terms of examples, but maybe after this election is all over maybe Sam, N and Amitabh Lath can give modeling political extinction a good try. Who knows?
I read this site every day, and am rather bummed when there’s not a new post. Knowledge is power – keep up the excellent work Sam.
I don’t know about that. Groups that vote at much lower rates historically take a very long time to and sometimes never overwhelm groups that vote at higher historical rates. And then there’s the effect of voting at a higher rate to preserve political power. The current demographic changes are that the younger folks vote at much lower rates than young folks of the past did.
Tony, Sam and N. are august company indeed. They have both invented powerful prediction engines, while I have but kibitzed.
An extinction-level event would have to raise the level of intra-party toxicity so high that members overcome institutional lethargy and jump. Slavery did that to Whigs. Do you see that level of man-the-ramparts, oil-the-muskets acrimony down there in VA?
Amitabh, I think immigration might be that issue. There is a real split within Republicans on the issue, and at least one side feels very “man-the-ramparts” about the issue (quite literally). The other side may not feel as strongly about the issue per se, but they may feel pretty strongly about continually losing national elections over the issue.
And in fact we have a recent example along those lines in the California Republican party.
Trump yard signs are everywhere in northern Massachusetts/southern NH. If you looked just at those you’d think Trump was winning New England, even though he’s way behind in both states.
Brian: Immigration has been a much bigger issue in the past. Living in New Jersey we occasionally take the kids to the Statue of Liberty and Ellis Island, and the articles and cartoons of that era (turn of the 20th century) make it clear the fight was truly vicious and beyond racist.
As a fraction of the population, the immigrant group was larger than the 11M under discussion now, and the country was much more homogenous. Both parties survived that. Comparatively, the immigration fight today is tame. I do not think it’s enough to deal a death blow to an institution as huge and pervasive as a national political party.
Tony, I’ve seen a few Trump bumper stickers, and I think one yard sign. There is a guy one street down that usually has giant signs up for Republican candidates, but his yard has, so far, remained devoid of signage.
I do not believe VA to be a swing state any more. The DC suburbs of VA are densely populated, wealthy, highly educated, and culturally diverse, a population that votes D pretty reliably, and drags the rest of the state along with it.
Massachusetts is a super-blue state, but our Republicans are Trump-crazy. I think being always surrounded and outvoted by Democrats is one of the things that makes Trump appeal to Northeastern Republicans; his outrageousness and scary authoritarian edge are like a thumb in the eye to everyone around them.
Amitabh Lath, what you say is true, but the demographics of the country is changing – prior immigration was from Europe. This is different.
This *MAY* be the beginning of an extinction level event. These things take time.
I think the Federalist analogy is the correct one. They went out because of things like opposing the vote for non property holder white men, and put themselves into the position of actually asking people for votes in the very early 1800’s whom ten, 15 years before that they didn’t even think should be voting. Remember, while George Washington actually said he wasn’t affiliated with any political party, his policies somehow always worked out to be Federalist.
The current party system is long in the tooth. If the stench sticks, then we could see a slow motion collapse over the next ten to twenty years.
It seems inconceivable now, but in 1607 when the English landed at Jamestown you can be certain that Powhatan and his boys and girls didn’t foresee their own doom. Disease surely helped, but things happen.
@Tony: Anglo-Americans feared immigrants from Southern and Eastern Europe as much as they now fear Hispanics and Muslims. Remember that America turned away European Jews fleeing the Holocaust just as some of us now seek to turn away Syrians fleeing their civil war. Italians and Greeks and Poles and Jews only gradually because “white”.
@emmy
Saying that youth turnout is much lower today than in the past is simply incorrect.
https://en.wikipedia.org/wiki/Youth_vote_in_the_United_States#/media/File:Voting_Trends_by_Race_and_Age.jpg
With the exception of the 1998 and 2002 midterms, youth turnout has consistently been between 20% and 25% going back to the Nixon Administration.
I think what’s more interesting in the recent polls is that Hillary’s numbers are somewhat more volatile than Trump’s. If she’s down and Trump’s up, it’s usually because her numbers plunged in a state, not because Trump’s went up. He seems mired at 45% as his, er, ceiling, while Hillary bounces from 38-50, which seems more volatile than head to heads of the recent few Presidential cycles.
I’m guessing this has to do with how the polls treat undecided and “other”, and the fact that Bernie Sanders has a lot of supporters who aren’t convinced he’s done yet.
…Also, compare Clinton vs. Trump (still dropping) with Sanders vs. Trump (still rising/steady). Trump gets almost the same percentage in both comparisons, but Clinton’s lead is much smaller because there’s a huge amount of “undecided/other”, especially in some polls that don’t push those people as hard. There seem to be a lot of people who will pick Sanders but won’t state a choice when it’s Clinton vs. Trump. They may be hoping for a third-party candidate of some wort or thinking about writing in Bernie Sanders.
Trump is consolidating some support, but I think a lot of this is that the Sanders campaign’s attacks on Clinton are genuinely hurting her not just with the left but with centrists as well, because they’re painting her as dishonest and corrupt.
The WSJ just endorsed Hills.
Does that change anything?
It probably helps Trump. Seriously. How does “Wall Street Loves Clinton” play in Jacksonville?
The Wall Street Journal is owned by Rupert Murdoch, as is Fox News. In the past the opinion sections of these outlets have generally been supportive of the Republican nominee. If they switch sides for the 2016 election that could affect tens of millions of readers/viewers.
Probably a wash. It makes the Trumpenproletariat love him more, but might swing a few Very Serious Centrists away from thinking Trump isn’t so bad after all.
I’d think the causation arrow probably runs more the other way (i.e. endorsements are more a symptom of weak support, and less a cause of it). So you could take it as some evidence of problems, albeit weak.
I doubt it “changes” anything.
To be honest, even to the extent you could extract information from it, I don’t think it’s anything that the constant “split in the GOP” stories haven’t already told you. It’s essentially just rephrasings of the poli-sci ideas regarding acrimoneous primaries + extreme candidates creating problems in the general election.
Not exactly; Bret Stephens, writing for WSJ, endorsed Clinton. That’s not an endorsement from the paper.
Mark F: There are an awful lot of retirees in FL whose savings depend on the stability of Wall St., and Trump has been saying some pretty scary things.
@Matt McIrvin If “Trumpenproletariat” is your coinage, then very well played, sir.
@Leading Edge Boomer: Not my coinage. Forgot where I saw it.
Anyway, this is a heartening map, but my hunch is that Florida still belongs in the uncertain category. If Trump can get Florida, that brings the Democratic total down to 233 and he has many more paths to win, though the odds are still against him.
Perhaps it’s the cynic in me but does anyone believe that some pollsters (e.g. Rasmussen and Q’piac) place their thumb on the scales this far out in an election to make it seem like a horserace and to generate private polling contracts? As a race gets closer, they reflect potential outcomes more accurately. Just wondering.
Sam, I love the site and spend a great deal time on it. Thank you!
I would be willing to bet my house that Hillary beats Trump in Florida.
considering how close Florida has been in recent election cycles, jharp, I’d bet something a little smaller on Clinton taking Florida in the general.
Let’s all not get too excited about the current polling. The lead for whoever’s ahead shrinks anyway over time, the question before us is whether Trump can find 8+ points between now and roughly September or whether Clinton can expand her lead. Either would be historically unusual, but this race is kind of that way all around.
“The lead for whoever’s ahead shrinks…” is not true. This has happened in 10 out of the last 16 elections, which is in the suggested direction…but not distinguishable from chance.
Florida has the sixth highest Hispanic population of US states. 14% of the eligible voters are Hispanic. About 1/3 of those are of Cuban origin. Let’s say that 1/3 votes for Trump and, given Trump’s overwhelmingly poor support from Hispanics, the other 2/3 or 9ish% go to Clinton. Is that enough to make a strong difference which way Florida goes? (of course, this assumes high Hispanic voter turnout)
I think he pretty much has to win Florida
Hispanic turnout doesn’t crack 50%. It varies by ethnic background, with some Hispanic groups turning out pretty high, but many of the most numerous groups turn out at sub-30% rates, giving the whole category a low overall turnout pattern.
Until I see some September polling showing Clinton up 5 or something, I’m going to stand by “too close to call” on FL. Trump does need it to win, or even keep from getting killed in EVs.
On the other hand, I personally think Trump’s chances in New Hampshire and Virginia are close to nil, though people still think of them as swing states because they went Republican as late as the Bush era. I think Virginia is now a bluer state than Pennsylvania or Ohio.
Also, I am really feeling the complete lack of polling in Iowa and Colorado. In the most recent polls there, it was close enough to call them definite toss-ups… but those polls are from late last year, when the national gap between Trump and Clinton in head-to-head polls was quite small.
To state the obvious, Trump has unfavorable demographics, low enthusiasm among his own partisans, a hypersensitive temperament ripe for exploitation through provocation, and a dearth of knowledge all lined up against him. Clinton is a disciplined candidate likely to make few unforced errors. The criticism for Clinton from the left will also dissolve away soon.
The general election campaign will be vicious but the tactics employed by both sides will most likely offset each other. In these circumstances, I find it difficult to imagine how Trump overcomes the deep structural disadvantages faced by any Republican in a Presidential election to win the most votes in swing states. Perhaps the Senate contests will be more interesting than the Presidential contest this year.
Thomas Edsall in the NY Times today (5/11) has an article about Bradley effect (doesn’t call it that). He has charts and figures showing even Democrats and Independents harbor pro-European-descent ideas.
I guess he wants you to freak out over your nice Euro-American friends and neighbors secretly being Trump voters. Of course these same Democrats and Independent voters gave Obama over 300 EVs, but Edsall chooses not to mention that.
Amitabh—My concern here is that neither McCain nor Romney were employing anything resembling the levels of explicit racism (of all sorts… misogyny is particularly pertinent in this case) that Trump does, and it’s really the behavior of the candidate that benefits from the effect that triggers it (see the point on social desirability that Edsall cites from Pew). So I’d say we should keep a small reservoir of freaking-out ready for the months ahead, just in case, also in terms of how we read polls.
The jittery souls that represent the GOP establishment have presumably read the same public and internal polls, and they are worried. The key will be where the money goes–once it starts moving to House and Senate races, then we’ll know that the party has written off Trump. Plus, Trump said that he’s not going to back off his provocative comments. That will please his base supporters, but it won’t attract a large enough general audience.
Thomas Edsall in today’s (May 11) NYT is the latest peddler of the “Nobody expected Trump so watch out Democrats anything could happen!!!” column-filler.
So I went back to the PEC archives to see what was being said. We didn’t really start discussing the election until late Dec/early Jan. And the posts and all comments basically make it clear Trump was leading, expected to clinch the nomination. There are some discussions of “ceilings” and “floors” but Sam’s observation that most people are ignoring polls and looking at weirdness of Trump sums up the general consensus.
Trump’s rise was not a shock. It was the most probable outcome. Just as Clinton starting with at least 260 EV is now. It’s fair to argue with that assessment, but “Trump was a total surprise, anything is possible” is not a logical argument.
Edsall quotes a fairly shoddy analysis from Morning Consult. Do people really believe the claim that polls underestimated Trump’s support? Is this even worth writing about?
Edsall specifically is not credible, I agree. However the Times is the newspaper of record, and the sentiment expressed is widely held and spreading. Although the supremacy of polls is self evident (especially with your careful study of extrapolating uncertainties from May to November) as we know from encounters with anti-vaxxers and global warming deniers, the innumerate need to be countered.
It is interesting that the state primary polls show a much smaller split between Internet, automated-phone and live-phone polls than the national general-election polls do. I imagine the sample is just harder to weight demographically when it’s the whole country, which is probably one of the reasons aggregated state polls do so much better than national popular-vote polls do.
Nostradamus woke me from my nap to tell me to pay attention to this story:
http://www.nytimes.com/2016/05/12/us/politics/hillary-clinton-donald-trump-tax-returns.html
http://www.npr.org/2016/05/11/477700943/mitt-romney-trumps-decision-to-not-release-tax-returns-is-disqualifying
Just read the Politico article on the Battleground polls. One poll. Let the games begin.
http://www.politico.com/story/2016/05/polling-clinton-sanders-trump-223094
The HuffPo average just narrowed some more, to Clinton +5, with a couple of scary national polls.
I think this is at least mildly real, though it could be short-lived. It looks as if Republicans are starting to consolidate behind Trump, while Clinton still has to fight off “corruption” attacks from Bernie Sanders.
And there are near fist fights like what just went down in Nevada.
How about an analysis such as the following? I call it the apathy calculation.
Vary wR and wD and see what the error bounds are.
Assuming wR and wD are very close to 1 how far off would a calculation like this have been in previous elections? What’s the biggest source of error?
% CrusIIaVermis
% The Apathy Calculation. Assumes that most Americans don’t vote in the primaries, that most Americans will generally vote along party lines, and that a very small fraction of American will make the effort to vote in one party in the primary and then vote in the other party in the general.
% CrusIIaVermis
% The Apathy Calculation. Assumes that most Americans don’t vote in the
% primaries, that most Americans will generally vote along party lines,
% and that a very small fraction of Americans will make the effort to
% vote in one party in the primary and then vote in the other party
% in the general.
%
for ii = 1:50 % run on each state independently
stateIdentity = ii;
nR = 3; % number of Republican voters who have historically voted
% in the general election divided by the number of Republican voters
% who have historically voted in the primary election. Generate this
% number on a state-by-state basis.
nD = 3; % number of Democratic voters who have historically voted
% in the general election divided by the number of Democratic voters
% who have historically voted in the primary election. Generate this
% number on a state-by-state basis.
nTrump = 100; % number of people who voted for Trump in the primary in that state.
% Assume they’re unlikely to switch to Clinton in the general.
nR_other = 20; % number of people who voted for any other Republican in the primary in that state.
% These people might switch. Give them a probability (weight) of
% switching between 0 and 1. A probability of 1 means none
% switch and a probability of 0 means they all switch
wR = 0.3;
nClinton = 100; % number of people who voted for Clinton in the primary in that state.
% Assume they’re unlikely to switch to Trump in the general.
nD_other = 20; % number of people who voted for any other Democrat in the primary in that state.
% These people might switch. Give them a probability (weight) of switching between 0 and 1.
% A probability of 1 means none switch and a probability of 0 means they all switch
wD = 0.3;
nPredictedRepublican = nR*(nTrump + wR*nR_other);
nPredictedDemocrat = nD*(nClinton + wD*nD_other);
% This is a function that determines the electoral votes for that state
[nElectoralDem(ii),nElectoralRep(ii)] = DetermineElectoralVotes(stateIdentity, nPredictedRepublican, nPredictedDemocrat);
end;
nDem = sum(nElectoralDem);
nRep = sum(nElectoralRep);
% who wins?
There’s another state that bears watching. I learned some years ago that Texas has one of the nation’s most Republican Hispanic electorates. That’s one reason Republicans have done well there. Now, on RCP, David Byler has a map that lets you adjust turnout and vote share individually for black, white, Hispanic, and Asian/other. Raise the national Hispanic figure for turnout 2% to 55% and the Republican vote share down from 20% to 12%. Don’t touch any other group. What happens? Just two states switch to Democratic—N. Carolina, which is expectable—and Texas.
I know endorsements (or in the case of Romney, anti-endorsements) are hard to quantify. But here goes: Gallup survey of enthusiasm in swing states in May 2012 had 25% or Romney voters as “Extremely enthusiastic” and another 21% as “Very enthusiastic”. At the time this was considered soft support, lower than Obama enthusiasts.
However, since Trump has to turn out Romney voters at the very least, and Romney is actively campaigning against him, the question is what fraction of Romney enthusiasts (around 40%of all Romney voters) listen to Romney and shun Trump.
If it’s a quarter of them, then Trump loses 6M Romney voters. Even if it is just a tenth of them, he loses 2.5M.
mediaglyphic // May 16, 2016 at 8:06 am
“If he has in fact predicted all elections since 1984, that is an accomplishment. (am i calculating correctly that i think it would be .4% chance that this is random .5^8=.004). But of course we have not seen all the evidence that he in fact did forecast correctly well ahead of time.”
Two comments:
1) Several of those were not even close:
1984, 1988, 1996, 2008.
2) IIRC, a few more were predictable if one just bet on the polls:
1992, 2004, 2012.
That leaves 2000,
Now we know where + in polls+ comes from. I couldn’t even read it all.
https://www.washingtonpost.com/news/the-fix/wp/2016/05/12/this-professor-has-predicted-every-presidential-election-since-1984-hes-still-trying-to-figure-out-2016/?tid=pm_pop_b
If he has in fact predicted all elections since 1984, that is an accomplishment. (am i calculating correctly that i think it would be .4% chance that this is random .5^8=.004). But of course we have not seen all the evidence that he in fact did forecast correctly well ahead of time.
However on points 2 and 11 in his list, i think they point to a democratic win. There is no serious competition to Clinton and there have been foreign policy accomplishments.
It appears that the answers to these questions are depend on the observer. Since he is the forecaster the mechanism may be something other than these questions. Perhaps how he is answering them.
Whether each key supports or undermines the party in power could be a very subjective determination.
@mediaglyphic The probability of getting the prediction right will not be a completely random 50 percent because some years are much easier to predict than others. It is probably true that most years the prediction of the ultimate winner was above 90 percent whereas in other years (e.g. 2000) the result was definitely close to 50/50.
Can Hillary turn North Carolina blue? She’s leading in the polls and this won’t hurt:
You know a candidate is in trouble when sentences about him begin, “Even Richard Nixon …” And that’s what is happening with Trump and the Great Tax Return hoopla….
http://www.newsobserver.com/opinion/editorials/article77979962.html
North Carolina is almost certainly a must win for Trump. But not for Hillary (Obama lost it in 2012).
NC is insurance for Hillary in case she loses Ohio, Pennsylvania and Florida, which she might.
Utah is an outlier in so many ways. There is little reason to believe it will go the way it did in 2008 (McCain-Obama), because practicing Mormons didn’t loath McCain the way they do Trump. They can’t stand Hillary, either, but she’s a pretty run-of-the-mill, somewhat-left-of-center Democrat, and Utahns have lived through the past eight years of that and emerged relatively unscathed, thanks to an overwhelmingly GOP state legislature that passes laws which affect their lives much more materially than any POTUS. Trump, on the other hand, has qualities that are reminiscent of those practicing “priestcraft” in Mormon thought. His vainglorious narcissism, open boasting of sexual exploits, and outright perjury when confronted with his past, combined with a half-hearted, insincere attempt to embrace piety when asked about his religious beliefs, are all things devout Mormons really cannot tolerate. It’s not about being pro-Mitt. It’s truly about being anti-Trump. Only Trump could put Utah in play, something the Dems would do well to seize upon.
In the latest iteration of that Utah poll, Trump is significantly ahead. Looks like they might have reconciled himself to him. More polls would be better, though.
Anybody who actually thought Utah was in play was being very silly.
I didn’t think Utah was all that winnable, but I thought (and still think) that it might be possible to force Trump’s campaign to spend money there.
Sam and N,
Would love to see a version of the electoral college estimate based on Google Correlate in addition to the usual one based on state-level polls. Run them in parallel and/or use them in combination.
Marc
I don’t think the method would work. You need at least some prior results of contests between these two particular candidates in some states as “training” data. That works well in the primaries, where the voting is spread along some months, but in the general election it all happens on the same day.
Could use polling data to fill in missing states. That would be cool.
State poll aggregates always have a bit of a lag behind national polls because of the sporadic coverage of states, especially early in the season.
Clinton’s national lead has been narrowing, as Trump seals up Republican support while Clinton still has to fight off attacks on the left from the increasingly quixotic and angry Sanders campaign. Prediction: Over the next couple of weeks, as more state polls come in, we’re going to see the whole map shift significantly toward Trump, maybe even with a median Trump EC win, and everyone will really freak out.
These polls are a reflection of a united right vs a still divided left. Interesting that they started to shift as soon as Trump was declared presumptive nominee, and the right has coalesced really fast behind him. Does not bode well for Hillary should left fail to firmly unite. Maybe having a VP such as Warren will help unite the fractures? What do you think?
Things could get very interesting, especially with a possible Johnson/Weld Libertarian ticket, which might draw a significant number of votes from both Clinton and Trump. Jill Stein of the Greens might draw a number of Sanders voters as well.