What Are The Polls Saying?

Start your day with TPM.
Sign up for the Morning Memo newsletter

We’re at the start of the first of two national party nominating conventions which will dominate the second half of July. We’re also just coming off Donald Trump’s choice of Mike Pence as his running mate and only days away from Hillary Clinton’s choice. All of which means we’re in a period both of intense media focus on the campaign, a series of key decisions from the nominees and what is often a period of substantial flux and instability in the poll numbers. Conventional wisdom suggests we won’t know how all of this shakes out until the first week or two of August. With that said, it’s a good time to look at the polls and see what they’re telling us.

Over the weekend and back into Friday we’ve had four premium national polls released by major national news organizations. NYT/CBS, NBC/WSJ, CNN/ORC and ABC/WAPO. The results are, respectively: tied, Clinton +5, Clinton +7, Clinton +4. The closest result is the oldest and the biggest Clinton lead is the newest. But we’re talking about four polls all of which concluded between the 12th and the 16th of July. So random variation and methodology likely account for more of the difference than the slight difference in time.

[ed.note: Since I wrote this poll two additional national polls have been released, slightly less premium than the other four but still respected pollsters. Those show Clinton +2 and Clinton +3. Those tighten the average slightly. But all the following still stands.]

What does this tell us: In all likelihood Clinton is ahead. The current PollTracker Average puts Clinton ahead by 2.9 points. But that trend estimate figures in three polls over the last two+ weeks from Rasmussen, all of which show Trump ahead. We know Rasmussen’s reputation from the past. It seems to be an even more Republican-outlier in its current incarnation. Other than Rasmussen, no poll has shown Trump ahead since an ABC/WaPo poll in mid-May at the height of Trump post-clinch bounce. If we remove Rasmussen from the trend estimate Clinton’s margin moves to 4.5 points.

There’s a been a constant debate over the Spring and early Summer about whether Clinton is ahead or whether it’s close to even. In substantive terms, it is undeniably true that the race is considerably closer than you might expect given how wild and transgressive Trump’s campaign has been after the primaries. Still Clinton is likely around 4 points ahead of Trump nationally. If you look at the state by state results, the key swing states show a fairly similar picture: a close race, but one in which Clinton seems to have a consistent if still narrow advantage.

Again, we’re about to enter what is usually a period of high volatility in the polls. We’ll probably have a much clearer sense of the structure of the race going into the fall a week or so into August.

Besides Clinton narrow margin, two things stand out to me about this race and give me a fair degree of confidence (but no more than that) that Clinton will be elected in November.

Look at the trend chart above. Through all the vicissitudes of the last 6 months, Trump has never moved into a lead or even a tie within Clinton. Not after he clinched the nomination in May or after James Comey’s press conference about the email scandal. Some of the ups and downs you see seem to be the result of news events. Others look more like artifacts of the schedule of polls. The rush of national news media sponsored polls over the last couple days have buoyed Clinton’s numbers. The high point she had about a month ago also came around the time when the same organizations all released polls. PollTracker treats the polls the same. But I have always placed more personal weight on these polls because of the track record of the methodology and organizations running them. So if you’re looking for reasons to think Clinton may be doing a bit better than the average suggests, there’s your reason. But it’s usually best just to look at the numbers in aggregate rather than scrutinizing them too closely because it’s easy to slide into a self-serving cherry-picking of data. But again: looking at all the numbers together, the trend estimate does not show Trump ever able to draw even or move into the lead. We’ve seen enough data at this point that this is not random and appears to be a highly persistent characteristic of the race. It tells me a simple fact: Clinton’s coalition is simply larger than Trump’s. That’s close to saying more people are supporting her than Trump (obvious), but not exactly. What I’m saying is that the communities and demographic subgroupings that Clinton is able to appeal to are larger than Trump’s. So even when she’s having a bad run and he’s having a good one, she always remains, though sometimes only slightly, on top.

Another way of looking at this is that for all the narrative about a close race Trump seems to have a very hard time getting much over 40% in the polls. Clinton gets down there with him sometimes too. But she also gets into the high 40s frequently. There seem to be technical markers keeping Trump from getting much over 40% and another preventing his trend line from ever moving ahead of Clinton’s.

The second thing that stands out to me is closely related to this point but comes at it from a different angle. The polls that have rich data on demographic sub-groups – whites, African-Americans, Hispanics, men and women, etc. – mostly show Trump underperforming against Mitt Romney. Indeed, this is basically true of all demographic groups. White men, white women, college educated whites – even with these groups where he is usually ahead he is almost always underperforming Romney.

Obviously, since Romney lost and the demographic groups that favor Democrats (specifically, racial and ethnic minorities and millennials) have gotten a bit larger in the last four years, Trump can’t really afford to underperform Romney anywhere unless he massively over-performs his numbers elsewhere. That doesn’t seem to be happening.

Now there is one technical issue to bear in mind here that probably isn’t an issue but might be: when we talk about ‘how Romney did’, we’re talking about exit polls. We don’t have the actually results of the election coded by race, ethnicity, age and gender. Exit polls aren’t pre-election opinion polls. The methodology is different. In theory, they should probably be more accurate since you’re asking people not what they believe they will do at some point in the future but what they actually just did. Still, it’s possible we’re comparing apples and oranges in comparing the two. I think it’s fair to say that most public opinion analysts and technical specialists do not believe that this is a significant issue – that we can compare the two more or less as we are. But consider it an edge case source of uncertainty.

Setting aside that issue (which we quite likely can) Trump simply isn’t doing what he needs to do with any key constituency to win the race. Indeed, he appears to be doing the same or worse and in some cases much worse than Romney in almost every key group. As you can see this amounts to looking at the building blocks rather than the full building as I did in point one. They line up pretty well, as they should. These are the numbers and the way of looking at polls that people who run campaigns tend to focus on.

What all of this tells me is that there appears to be a deep underlying stability in the race, one that persists in the face of a wide variety of news events, polling methodologies, pratfalls by the candidates and different modes of analysis that can be used to scrutinize the polls. That stability suggests a win that is larger than the scale Obama managed over Romney but not as big you would expect when the opponent is a demonstrably unfit pathological narcissist who would seriously threaten the safety of the country and maybe everyone in the country in a way that far transcends the bounded differences between Republicans and Democrats. It also isn’t nearly a big enough margin to make me feel like there’s nothing to worry about. One thing to bear in mind in all of this is that Trump has one very big advantage he has nothing to do with. It is always quite difficult for the incumbent party to carry a third term in the presidency. Absent everything else about this election that should make the GOP the big favorite. Obama’s rebounding popularity and a non-bad economy are countering that to some extent. But it’s important to remember those inherent advantages.

Now all this can obviously change. The notorious polling instability of the period around the conventions is the time it’s most likely to change if it is – either as a temporary blip or as a more fundamental shift in the race. (As I said, wait until the second week of August until you start to seriously freak or start high-fiving your friends.) But until I see a shift in those two technical markers I’ll be relatively confident that the race remains fundamentally stable, indeed, surprisingly so, and that that stability is in a place that points to a solid if not resounding Clinton win.

Latest Editors' Blog
Masthead Masthead
Founder & Editor-in-Chief:
Executive Editor:
Managing Editor:
Associate Editor:
Editor at Large:
General Counsel:
Head of Product:
Director of Technology:
Associate Publisher:
Front End Developer:
Senior Designer: