On this week’s episode of my podcast, I Have to Ask, I spoke to Harry Enten, a senior writer and analyst at CNN, and the network’s political data guru. Prior to joining CNN, he covered politics for FiveThirtyEight. Below is an excerpt from the show, which has been edited and condensed for clarity. In it, we discuss why next week’s California primaries may determine control of the House, whether pollsters have corrected some of the mistakes they made in 2016, and the most—and least—politically effective ways Democrats can attack Donald Trump.
You can find links to every episode here; the entire audio interview is below. Please subscribe to I Have to Ask wherever you get your podcasts.
Isaac Chotiner: Why is the primary in California on Tuesday so important?
Harry Enten: I think it’s important for a number of reasons. For those who are not familiar with how California elects its elected officials, essentially you have what is called a Top Two primary. That is, all the candidates, regardless of their party affiliation, will run next Tuesday, and the two top vote-getters in each congressional district will advance to the general election in November. That means that in a number of districts, there could be two Democrats advancing or two Republicans advancing. And because a lot of Democrats believe that their path to a majority runs through California, there are a number of cases where there are Republican incumbents or seats that are held by Republicans where Democrats, at this point, look like they could get locked out. That is, there’ll be two Republicans advancing to the general election in November.
How many seats are there where that’s a real possibility?
I would count three. There’s Ed Royce’s district, there’s Dana Rohrabacher’s district, there’s Darrell Issa’s district. Those are three seats where I think there are a lot of Democrats very, very worried about what’s going on.
These are districts that Hillary Clinton did quite well in, especially compared to a normal Democrat, or let’s say in a year in which Donald Trump wasn’t the Republican nominee. So, they would at least have a good shot at picking up those seats. But of course, if they don’t advance to the general election then their chance is no shot. There are no write-ins, there are no do-overs. It’s whoever advances, the two candidates who advance, those are the two that face off in the general election in November.
How did it get to this point with Democrats, where they haven’t been able to find a candidate to be the standard bearer for the party in the district? Do you think that this would be happening elsewhere if we had this situation where you have the top two vote getters in more states? Or was there a specific screw up by the party?
No. 1: It’s the fact that you’re dealing with a lot of Democratic enthusiasm, right? That is, a lot of people want to run, a lot of people want to get involved. No. 2: I think on both sides, although in this particular case on the Democratic side it’s especially coming back to bite, is you don’t have as strong of a party as you may have had, say, 20 or 30 years ago that could have directed [that energy] and say, “This candidate has to get out, this candidate has to get out so we focus in on this candidate.”
We have seen some instances of that, where some candidates were forced out when they were showed the polling and [told], “hey, you’re just going to be a spoiler.” But the fact that we haven’t seen more of that is an indication that the major party structure that we’re used to being strong in this country isn’t as strong as you might expect right now.
Democrats need to have a net gain of 23 seats to win the House. People like yourself have estimated that the national margin for the House of Representatives—that’s the combined vote from all 435 districts—will have to be around six or seven points for Democrats to probably take the House. How do you come up with that number of six or seven percent nationally, and how accurate is it?
There is a wide margin of error around that estimate, so it shouldn’t shock anyone if Democrats, say, win by five and then take control of the House, or it shouldn’t shock anyone if Democrats win by eight or nine and then they don’t take control of the House. Essentially what we’re doing is we’re building a model that takes into account the past congressional vote in that district, the past presidential vote in that district, whether or not an incumbent is running, and then perhaps some additional variables and basically saying, OK, if we play out the election any number of times based upon those different district factors, and then how much of a national swing we’re expecting, and simulate that out, this is the number that most frequently results as the tipping point whereby control goes from Republicans to the Democrats.
And that that’s an important thing to note: that there are many more districts nationally speaking that lean Republican, at least on the presidential level, so that even though Hillary Clinton won the popular vote, say, in 2016, Donald Trump won the majority of congressional districts and that’s part of the reason why we expect that Democrats are going to need to win the popular vote by a significant margin in order to take control of the House.
A lot of Democrats have been drawing a distinction between what we see in these special elections and what we’re seeing in the polls, even though the polls have been not great for Trump. The argument is essentially that what’s happening in these special elections is a real sign of a wave election. Where do you land in this debate?
I should point out that so far in federal special elections, those for Congress and the Senate, Democrats have been out-performing the partisan baseline based upon the last two presidential elections by about 17 percentage points. That’s significantly better than, say, a 6- or 7-point lead on a generic congressional ballot, as is currently the case. Now the question is, do those special elections mean more than the generic congressional ballot? Maybe, although the sample size is quite small on that.
In 2006, Democrats were outperforming the baseline by about 15 percentage points and they only ended up winning the national House vote by eight. So just because Democrats are running away in the special elections doesn’t necessarily mean that they’re going to run away with the House in the fall, though it’s certainly not a bad thing for Democrats. How Democrats do in the fall and how they do in special congressional elections are certainly correlated.
President Trump’s approval rating has ticked up a few points to about 42 percent.
I have a few theories, not the least of which is that they’re not talking too much about taxes anymore and they’re not talking too much about health care. What’s in the news today, for instance, is Roseanne, Roseanne, Roseanne. And you know, the fact is that most people expect Donald Trump to be talking about Roseanne. That’s not a big surprise to his supporters. When he went after health care, I think that went against that populist screen. So, the fact that we’re on these more ridiculous issues, I think that has certainly helped Donald Trump.
The idea being that that’s kind of baked into the cake with his popularity?
Yeah, exactly, that it’s baked into the cake. People expect Trump to go on Twitter and do things that most presidents wouldn’t do. What they didn’t perhaps necessarily expect was that he was going to go in and be an arch-conservative when it came to dealing with the health care issue, and I think that surprised a lot of people and why his approval ratings dropped during that debate.
Does this suggest anything to you about what Democrats’ midterm messaging should be?
Health care is something that certainly seems to work. That is a big driving issue for Democrats to get them to the polls. That’s doesn’t necessarily mean they should be arguing, say, for a Medicare-for-all package, but [they could] say hey, these guys tried to mess with and take your healthcare away. If you look at the Pennsylvania 18th special election when the Republicans were talking about the tax cuts there, it simply did not work. It didn’t work. So, going after Republicans on tax cuts, in my opinion, is also a pretty good idea.
How much data do we have on a president’s approval rating versus his party’s performance in the first midterm after the president’s first election?
Well, we’ve got a lot of data on that. And we do know that how many people approve of a president and how his party does in the midterm election is certainly correlated. Now again, I hate to say it, but there’s a wide margin of error around that. But generally speaking, a president who has an approval rating in the low 40s, his party would be dead meat in a midterm election. Dead meat. Now, there are a number of reasons to perhaps suggest why President Trump’s Republican Party won’t be dead meat, but historically speaking they’d be dead meat.
And so, what may save them are the factors you were talking about earlier?
Well, there are a number of factors. The ones I was talking about earlier, [and] also the fact that [with] President Trump and Republicans, there’s been an increasing divergence between what percentage of all Americans approve of the job the president is doing, and what percentage of registered voters or likely voters do. And we’ve seen that in a number of polls this year where Trump’s approval rating among, say, adults is 41–42, but in that same poll his approval rating among registered voters is closer to 44–45.
Now 44–45 still isn’t great for a president, but if you could imagine the president’s approval rating, say, ticking up an extra point or two, then all of a sudden Trump’s approval rating among registered voters is in the mid-to-high 40s, and that in itself is not emblematic of a wave election. So that’s one thing I would certainly be keeping my eye on.
As the Republican Party keeps winning over white Americans with less education and losing more educated suburban voters, wouldn’t that play to Democrats’ favor in terms of winning registered voters versus all adults?
Well, there’s certainly that factor that’s working in the Democrats’ favor, but there’s also the other factor working in the Republicans’ favor, which is Republican support is increasingly amongst the oldest of Americans, and they tend to turn out in much higher number than the younger Americans, who are the most likely to dislike Donald Trump. And there’s also the fact that Republicans are increasingly relying on white voters versus the overall population, and white voters tend to turn out in larger numbers than non-white voters do. So, there are a couple of moving parts there, but most of the moving parts are going in the Republicans’ direction, at least in this day and age.
What are the three most interesting Senate races in your mind?
I mean, to me the most fascinating race is probably Tennessee. This is a state that Donald Trump won by over 25 percentage points. It’s a state that hasn’t voted for a Democratic presidential nominee since 1996; hasn’t voted for a Democratic Senate candidate since I think the early ’90s. And Phil Bredesen, who’s the former governor of that state, has generally been polling ahead of the likely Republican nominee, Marsha Blackburn. And if Democrats can win there—they need a net gain of two in order to pick up control of the Senate—and then hold on and, say, win in Nevada and in Arizona, well then, all of a sudden you’ve got a net gain of three, and that means you could afford to lose one of the many states which have Democratic senators right now but where Donald Trump won in 2016.
Two, I’d say is Indiana. Joe Donnelly running in that state. Not a lot of polling there. Donnelly won in 2012 because, among other reasons, Richard Murdoch, the Republican nominee, was flawed. He was one of the candidates who made an allusion to rape on the 2012 campaign trail. And then the third contest I’d probably keep an eye on, just because I’m always fascinated by the state of Florida, is the state of Florida given that the incumbent, Bill Nelson, who won his first term in 2000, is facing off against Rick Scott, who is the current Republican governor. And Rick Scott is going to spend so much freakin’ money, and the question ultimately is: Can he spend enough money to win that race because he’s going to vastly outspend Nelson? And if he does, if Republicans are able to win in Florida, that probably closes off the path for a Democratic majority. But if Nelson holds on there, perhaps it means that money isn’t everything.
Polling and pollsters were criticized a lot in 2016 for not getting the result right. When you look back at the polls in November of 2016, are there one or two things that you think were done incorrectly by pollsters, and are those mistakes fixed now?
I think that there are two things to keep an eye on. Number one is to make sure that pollsters, when they can, weight to education. There were too many white college graduates in the polls as opposed to white non-college graduates. And although perhaps once upon a time education levels weren’t the most predictive of a vote, now it certainly is predictive, especially among white voters. So hopefully, we’ll be seeing more pollsters adjust for that factor going into 2018 because obviously white, non-college voters will be much more supportive of the President than white college voters.
And then I would say the second thing which, perhaps, seems simple enough is that on the state level, the pollsters just need to poll to the end. There were a number of states where pollsters tended to pull out because perhaps they thought that Donald Trump was cooked in that state, and perhaps he wasn’t as cooked as some pollsters thought.
So, tell us: How does a political analyst and data junkie like yourself spend an average work day?
Let’s see. Yesterday I was basically on the census website scraping data about citizen voting age, population. I was also looking at FEC returns, reports from 2006. From 2008 onward, you’re able to get them a lot more easily. They changed the system over so I was basically going race by race, candidate by candidate to make sure I had the correct filings for each one. And then in the evening, I think I was talking with a fellow numbers analyst that we mutually know by email and via text, just saying this is what we’re seeing, let’s see if we can get this spreadsheet together, I’ll give you this spreadsheet if you give me your spreadsheet. And it’s just a love affair.