In 1996 I started comparing the swing in state polling to national polling. And it was odd: the state polling never shows as big a lead as the national polling did, and in ’96 polling pretty badly missed on the high side.
The basic theory behind the analysis is two fold:
- There are many more state polls than national polls, which makes them less susceptible to being distorted by outliers.
- There are more pollsters – so you are not dependent on about 5 pollsters for the state of the race.
Here is a summary since 2000.
It was significantly better in 2000 and 2012. In 2000 the state polling always showed a close race. In 2004 national polling was better. In 2008 they were both pretty close, but there was a late decline in state polling that the national didn’t see. In 2012 the state polling was SIGNIFICANTLY better.
This shows the moving average as of the morning of November 5th. By the end of the day it would predict an Obama lead of 3.1 – much better than the national polling average did. The diary I wrote this in – of which I am proud – is
here:
So this is the data as of this morning:
For the first time Trump actually leads in the 5 day moving average. There are reasons to be suspicious of this number – it is heavily weighted to Emerson numbers, and those have leaned right. The 10 day contains better news – suggesting a 4 point lead. Almost exactly where Obama was at this point.
One more thing to say about 2012. On the eve of the first debate Obama’s state numbers looked very close to his 2008 numbers. He was, I think, headed for a significantly larger margin than he wound up with.
And then the first debate happened…
Random thoughts.
Seems to me that comparing mid-September polls 2016 to Oct-Nov polls 2012 isn’t a strong illustration. Not denying that the first 2012 debate had an impact on the polling (and it’s also possible that Benghazi kicked in as an issue in the minds of voters at that point). But that impact faded by election day. A trend that only ABC and Pew managed to detect sufficiently.
Looking back at those October 2012 polling numbers I can see why Mitt was so optimistic and am perplexed as to why I was continuously unconcerned in real time. While inconsistent on my part, it’s possible that I didn’t bother to look at the October 2012 polling. Or I was throwing in some other factor which is never a good idea if such a factor isn’t articulated and subsequently verified.
Since at least 1984, both teams always overestimate how well their nominee will do in the first debate and underestimate how well the opposing nominee will do. What generally gets overlooked is the performance bar required for the respective nominees. It’s higher for the lagging candidate and lower for the leader. However, complacency on the part of the leader (GHWB in ’92 and BHO in 2012) gives the opponent a huge opening. One that’s difficult to close in the subsequent two debates. Similarly, merely holding one’s own for a lagging candidate isn’t nearly good enough.
At this point, HRC is still in the lead. If just barely. A problem I see for both HRC and Trump is that their standard debate strategy from the primary debates isn’t what either of them need in facing her/his current opponent. If that’s what both of them go into this one with, it will be dueling smirks and filibustering. At the end of which HRC pulls out her “he’s being mean to the girl” schtick and then she wins.
I suspect the permutations are what gave you comfort. The state that pushed Obama over 270 – Colorado – was won by over 5.
In a simplistic way, Romney would have had to win by 1.3 to flip Colorado and actually win the election. Put another way Obama had a 1.3 point margin from the Electoral college, and that showed up in the probabilities.
But 2012 was way closer than most Democrats are willing to accept. It does appear – and I wrote about this – that there was systemic bias in the polling as a result of cell phones. But this was gone by 2014 – and in fact polling missed in the opposite direction as it had in 2012.
Debates. There is a tendency for them to benefit the candidate that is behind because they tend to bring home partisan holdouts.
In 2012 the first debate remade the race.
This debate is the most unpredictable of my lifetime. And I suspect it will matter. Often the debate after the debate is more important – and I agree with you HRC will play the women card after the debate.
Don’t know what you mean by “permutations” and I don’t read polls for comfort. In fact, it was reading polls that led me to begin contributing to blogs in the first place because I thought I might have something to contribute. Back in 2001-2 I only casually read a few blogs, mostly for some relief from the unrelenting pro-war, pro-GOP position that the MSM had adopted. In the lead up to the 2002 mid-terms, the so-called leftie bloggers were claiming that Democrats were poised to make gains. I kept shaking my head and wondering what numbers they were looking at and doubted my ability to read polls because I’d never actually used anything I’d learned in all those stat classes. I was seeing losses in 2002. So, I jumped in when they had been wrong and I had been right.
And I got it right at the Presidential and Congressional level for the next five elections. If I’d been looking for comfort, I would have blown 2004 and 2010. What I suspect I got wrong in my read of the 2014 polls was using 2012 lenses when given a choice between barely okay Obama Democrats vs. less than okay Republicans, Democrats could still prevail. But doing more of the same after 2012 instead pivoting left, made barely okay not seem good enough to bother voting for.
Agree that Democrats took far more comfort in the 2012 EC vote than was warranted. Disagree that it was really close. 2000 and 2004 were really close (the latter only at the presidential level because GWB was so dreadful).
I probably disagree about the impact of primary and presidential debates. It’s more what is seen than heard and is the only opportunity that voters have to make a visual, side-by-side comparison of the candidates. That’s what hurt Gore in his first debate even though there was no contest between what the two said. By your standard (brings home the undecided partisans for the lagging candidate), the 2004 debate should have won it for Kerry. It didn’t because Bush added 12 million votes to his 2000 tally and Kerry only added 10 million votes to Gore’s tally. (My interpretation doesn’t assume that the GOP has a 2.5% advantage with the general electorate.)
I meant that HRC will play the woman card in the debate. A variation on that in her last debate with Sanders was “I’m not a natural at his like Obama and Bill” delivered as if one could see her teary eyes. If Trump tries to wing-it, he’ll step right into HRC big muddy.
Seems to me that the second and third debates only gain in importance if the leading candidate blows the first debate. But difficult to say when the margin between leading and lagging is quite small. This isn’t 2000 for Trump or 2012 for HRC.
Permutations – the number of ways to put 270 EV’s together.
But that in an of itself is kind of mistake. Xenocrypt on twitter has made this case effectively. States don’t move independently of each other. It is unlikely – very unlikely – that someone wins by even 1 point and loses the EC.
In historical terms a win of less that 4 is close. I don’t agree with you about that. Bush won by 2.4, Obama by 3.8. I don’t see that as a huge difference.
Debates had enormous impact on the primary fight I was most involved in. They certainly mattered in the ’92 NH primary fight. Debates mattered in the general election in ’60 (though the extent is overrated) in ’76, in ’80 and in ’04.
The debates offer a chance to Trump to appear not crazy – there is a parallel to Reagan I think. The underlying dynamic is that it is hard to win 3 elections in a row, and most of the economic models suggest the GOP should win. The right track numbers are terrible.
Okay — and agree that states don’t move independently of each other, but that’s less true in counting up EC votes today than it once was. Because so few red and blue states have the potential to flip.
If I were within the GOP inner circle, I’d be mighty ticked off that the party was so ill-prepared for this election. OTOH, those wrong-track numbers aren’t exclusive to the policies of the Democratic Party, and therefore, the CW about a third term are weaker than usual. Plus, at a personal level, Obama has continued to enjoy a high rating. One he may be trashing with his push for the TPP — in appearances with Kasich no less.
I’m incapable of wrapping my mind around how those so infected/infested with hatred of the black man in the WH can view someone like Trump as acceptable. Other than racism, he doesn’t represent anything that they have been claiming for decades is important to them. I suppose sticking it to Democrats is a powerful motivator, but even Wallace offered more than that in that he only rejected the civil rights position of the Democratic Party but was fine with everything else it stood for at the time.
It’s always difficult to reject the notion that a tradition like the debates aren’t important. We think they are, but there’s little agreement as to why and how. GWB got through six of them with little command of public policies against two opponents that knew their stuff but lacked that generic personable/likeable quality. In a near tie among the electorate, likeability may be the tiny secret sauce an idiot GOP candidate needs to win. Seems to me that Trump’s likeability factor is too specialized and not generic enough.
OTOH, Trump isn’t competing with Gore or Kerry. HRC’s natural likeability factor may be as narrow as Trump’s. A difference is that he’s going with what he had and HRC is going with a hundred makeovers to improve hers. So, he ends up with a veneer of authenticity and she ends up looking fake. That authenticity factor also feeds into the fact that Trump won the GOP nomination with no help from the party and with far less money than the others spent and it’s highly unlikely that HRC could have won the nomination on her own. It was too transparent in this election cycle for voters not to notice that and they do tend to hold that against a nominee that couldn’t make it on his/her own.
Of course the state polling shows less volatility over the national figures. You are comparing data (national) with a mean (states) or a mean (national) with means of means (states). That smoothing is built into aggregating the states.
In the end, it is the states (and their electors) that matter and in this election year the differential proportion of demographics that Trump has offended as compared with ones he hasn’t interrupts comparisons with previous years. It’s going to be very noisy in comparisons.
Errors in national or state or comparing between them likely are going to be randomly distributed from election to election unless there is something systematic (requiring an explanation) going on.
It does not make sense to me (YMMV) to do other with states than assign the electoral votes and the resulting popular vote based on state popular margins and margin of error.
The saw about the lead narrowing the closer to the election seems from your data to be not the case. What does happen is uncertainty starts being apportioned between the candidates.
The moving average smooth the trends over different time periods. Trump coming on top of the moving average is what one would have expected from media coverage in the past 2 weeks or so. What has been fascinating to me is the rearrangement in Silver’s report of the “Path to Victory” states as the average floats up and down. The moving average represents the electoral votes in the current states in the Path to Victory being swapped in and out as quanta of population. Those swaps have ramifications in the sense of the downticket races geographically.
And states with large minority populations will have more friction to movement than states that are more homogeneously white. But the homeogeneously white states are pretty much Trump’s base for now anyway (I would like to see that proved with data arranged in Silver’s Path to Victory order and partitioned by ethnic and female demographics and conservative/non-conservative splits.
Given the polarization in this race, I’m not sure how much the debates will move sentiment compared to previous elections. Who is it whose opinions will be expected to swing as a result of debates? What ideas will drive the debates, or will it just be a matter of who falls to the gotcha? With Trump and his voters, how would one know?
Xenocrypt has done some serious analysis trying to explain the shifts from ’12 based on demographics.
It doesn’t work. The shifts in these states can’t be explained by demographics. He ran a linear regression based on share of vote of college education, Hispanic population, income and a few others.
The r squared is .18.
In Texas Clinton is down 6, AZ and GA are both close. Clinton runs well ahead of Obama in a number of states. But she also runs behind Obama in a number as well. The oddest is Florida, where she runs between 3 and 5 points behind Obama. Given the demographics Florida should be moving the other way. Similarly, you don’t see as big of a shift in North Carolina as you do in SC or Georgia.
The reason state polling average are less volatile is there are 5 times as many – which is why I started doing it.
The Hispanic vote remains about the most interesting in a way. Trump isn’t really running that far behind Romney. And there is only 1 state (Florida) where the shift from ’12 to ’16 matters more than 1 point. Over half the Hispanic Vote is in 2 states: California and Texas.
This is based on the Univision polling and comparing to the 2012 exit poll. In most states Clinton really isn’t getting much more additional advantage. In Arizona for example she runs only 5 points better than Obama did among Hispanics.
AZ .36
CO .42
FL 3.06
NC .88
NV 1.03
OH .64
I’m not sure what the situation with Hispanics is from your description. Are Hispanincs mostly disconnected from the Republican Party (except for anti-communist Cubans) or have they not been mobilized by the Clinton camp? Are there additional votes to be gotten by Election Day? Where is Trump’s ceiling with Hispanics? With African-Americans for that matter (he seems to be working the cultural/religious anti-LBGTQ angle there.
I think the point re: Hispanics is Obama pretty much maximized the advantage in ’12. So there isn’t much incremental advantage to be had from them in the states that matters.
There are about 20% of Hispanics who are Republicans.
Which surprised me.
The African American pitch I think is actually aimed at college educated white.
Nate is now perplexed. Election Update: Democrats Should Panic … If The Polls Still Look Like This In A Week .
A week is before the first debate. So, I guess I’d say if the polls look like this going into the debate on 9/26, the debate will indicate whether panic is a proper response.