STEVE INSKEEP, HOST:
G. Elliott Morris has been listening along with us. He is co-founder of FiftyPlusOne, a poll-aggregation and election-forecasting website. He also writes the Strength In Numbers newsletter on Substack. Mr. Morris, good morning.
G ELLIOTT MORRIS: Steve, thanks for having me.
INSKEEP: All right. I know you're reading that survey at npr.org. What stands out to you?
MORRIS: I think the one number that your readers need to know about is the question that asks whether or not voters will be voting for Trump or against him. In your survey, 44% say that their vote in the midterms would be one against Trump as a check on him, as opposed to one for him - to enable him. That's 21%. So you've got about a 2-1 split here. In a midterm election, political scientists, where I sort of have one foot in their world - they say that these midterm elections are referendums on the party in power, with voters issuing these sorts of ballots against the direction of policy change or the general conditions of the country with the incumbent president. And combined with these numbers, yeah, that would be a disaster for Republicans if anything like this survey is accurate.
INSKEEP: Wow. So, like, a 2-to-1 advantage for Democrats if this is a referendum on the president. I want to add another complexity, though. When I've been out talking with voters, I do find some folks who are conservative who are disappointed in Trump, disappointed in these last two years, particularly disappointed in the Iran war. But I wasn't sure that they were going to go all the way over to voting for Democrats necessarily.
MORRIS: And they might just not vote, right? That's an important variable in this equation, too, as Domenico was just saying. Turnout in the midterm elections can drive a big proportion of the overall swing in the electorate, as well. In 2018, and when I was reviewing the data on this before this show, the turnout component of the change in the election result from 2016 to 2018 was about 30%. So that is to say, Republicans not casting their ballots decreased support for the Republicans about 30% of the overall swing that race, which was about 10 points - you're doing rough math with me, Steve...
INSKEEP: Yeah.
MORRIS: ...About 3 1/2 points. So, you know, it could be the case that Democrats are, like, really, really close in these key red Senate states and that turnout component tips them right over the edge. Or, hey, it might be the case that these voters do decide it's important for them to vote and issue their support for their president and Democrats don't flip those seats. That's the type of thing that happens...
INSKEEP: Yeah.
MORRIS: ...When these elections are so close.
INSKEEP: That's got to be the biggest undecided question - how many Senate seats flip? - particularly since, as you said and as Domenico said, they are mostly in red states, mostly in Trump-voting states. And then Domenico referred to the possibility of polling error here. All the polls show Democrats with a lead. It's a big lead in many cases, which is why people think that Democrats are likely to take the House. But it's much narrower in those red states. Am I correct that even a small polling error could have a huge consequence there?
MORRIS: So for your listeners, I'll just say the - one number. That's three. Three percentage points is the average error on Election Day for Senate polls in the midterms. So that's the average polling error across all states. Some states can have larger errors. Some states can have smaller errors. But it's 3 points. And importantly, we cannot predict ahead of time what direction those 3 points will go - which party it will favor. We've done the math on this.
INSKEEP: I just want to say that's - like, if we picked out James Talarico in Texas, for example, 3 or 4 points up in some polls, you're telling me that could turn out to be a Talarico loss or a huge Talarico win.
MORRIS: Yeah. Either direction. And it's not just Talarico. It's Ohio, Alaska, Iowa, Maine - right? - Michigan. All of these states are within 3 points. So this could be the case where Democrats win only 49 seats - right? - they only pick up two from what they have. Or if the polls are underestimating them by that historically average margin, 3 points, maybe they win 55. Maybe even they win 56. So, yeah, this is the type of volatility - I think Domenico called it a super (ph) volatility or some such. This is the type of uncertainty you get here when all the races are close and the polling error can act on all of them at the same time.
INSKEEP: Do you assume that the polls are finding all Trump voters, who sometimes have been elusive in the past?
MORRIS: So in our surveys at Strength In Numbers, the polls don't seem to be getting the types of Trump voters who live in rural areas. That could be a big factor. But they also seem to be overrepresenting more MAGA Americans with the voter - Republicans who voted for Trump in 2024 who now disapprove of him either dropping off of the survey or being downweighted for some complex statistical reasons. So, yeah, I'm not confident about the mechanics of the polling error in either direction. I just want us all to go into Election Day thinking, hey, these polls could be off by about as much as they normally are.
INSKEEP: Got it. G. Elliott Morris - Strength In Numbers. Thanks so much.
MORRIS: Thank you. Transcript provided by NPR, Copyright NPR.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.
