So, uh, what, what, what's your take on this?

Yeah, I mean, people have been making up polling for some time.

Uh, I remember some cases in the 2018 primary cycle of, uh, users of the prediction market platform PredictIt talking on private chat groups, conspiring to create fake Twitter accounts that then would publish alleged surveys.

Uh, and the intent there was to manufacture a price movement into prediction markets.

So that has happened in every election cycle since then.

There's always a case, one or two cases of this, and then people get caught or told on, and they stop publishing it, and they fade away back, back into the ether.

So this- it's not totally, it's not totally new.

So is there anything we can glean from that?

So how did the polls get that race so wrong?

Well, the first thing I'll say is that primary polls, in general, are, like, pretty bad.

On average, going back to twenty-fourteen, my colleague Mary Radcliffe and I have done the math.

A primary poll is off on average by about thirteen percentage points, so you should, like, expect them to be wrong.

In this particular case, the wrongness of the poll seems to come down to two factors.

First, there was a lot of late movement in this race.

Uh, one survey I helped conduct, which measured preferences among the same people in July and in August, found, uh, about ten percentage points of movement in the past couple weeks of the race.

I'm curious, what do you think are the real-world consequences, either for politicians or voters, when the polls aren't as accurate as people expect them to be?

So let's just start with what last night's primary results tell you about how voters are doing and feeling right now.

Well, thanks for having me on.

Uh, I do think your introduction really makes this, uh, clear for us.

There's, um, a- You know, in, in some states you have Francesca Hong losing to Crowley, for example, a progressive losing to a moderate.

But then next door in Minnesota, you have a progressive, Peggy Flanagan, uh, beating her moderate so-called challenger by, like, uh, 30 points, 20 points.

Uh, and this I, I think helps us understand that this divide in the Democratic Party really isn't about progressive versus moderate.

Uh, the big thing that's driving change within the Democratic Party, and also for the, uh, midterm election, as that headline you showed at the beginning of this segment shows, uh, is whether or not someone says that the system needs major disruptive changes, or whether or not someone feels that the political system, you know, might be working well for them, or might be working good enough that, uh, these major, so-called major disruptive changes, uh, aren't needed.

Um, and, and on that, I, I guess when you kind of think about these anti-system voters, do you think they're a, a sort of special phenomenon on the left right now for the Democratic Party, or actually, are you seeing that across the board?

Basically, the, the Stevens versus El-Sayed makes no difference at all to your probability of, of Senate.

No.

And that's the other thing is, uh, if you are really splitting hairs about the probability of winning Michigan, then, uh, there is an on average effect of, uh, Abdul El-Sayed winning.

It's on the order of four or five percentage points.

So to spell that out, our current forecasts, including Abdul El-Sayed, is for Democrats having a 65% chance, 64% or 65% chance of winning the seat in Michigan.

Whereas if you replace the quantitative indicators for El-Sayed, so the polls and some other factors with those that you would have in a hypothetical with Stevens, Stevens has a 69% chance of winning.

So there's a five-point difference there.

Uh, is that basically post Iran, or is there more going on than that?

And don't believe the lying media.

Maybe he's not being constrained by public opinion the way we would want a president to be constrained by public opinion, but there are a lot of other actors that could be constrained, especially in their local districts or their state, and there's other groups, nonprofits, agency heads in Washington, D.C., or judges could also be affected by public opinion.

The political scientist V.O. Key wrote that public opinion offers levees to steer a river, and we can view the polls as being tools for identifying policy that Americans really, really don't want, and then as good stewards of the public opinion data fight for that.

V.O. Key, the political scientist, he coined the term echo chamber.

The public is an echo chamber.

And today, people wrongly argue that Key is offering an indictment of the public, that they're only parroting what they see on the news, but he's actually talking about an entirely different echo chamber, the echo chamber of the choices that are offered to them by the political party system.

That if you are only offered the choice between democracy and autocracy, for example, or really imperfect political and economic system and the wholesale sacking of the federal government, or if they're only given, maybe even sharper, a choice between two idiots, you can't really fault them for picking an idiot.

But will people vote on it?

[laughs]

... that's one seat.

That's ...

Sorry, that's three seats lost on the Democrats.

It would've only been one, now it's three.

Okay.

But then we have redistricting in Texas, in Florida, [laughs] in North Carolina, Ohio, Missouri.

Yeah

What does your latest polling find on that, and what's your general sense of it?

Uh, the House I think is probably a foregone conclusion for the Democrats, if I'm being honest with you, Greg.

I mean, the generic ballot in our survey is D plus seven.

There's a pretty predictable historical pattern where the party out of power gains ground as the midterms go along.

So the party out of power is the Democrats.

We should expect them to gain at least a couple points in the generic ballot over, over the next couple months.

Um, and the Democrats are emerging from this referendum in Virginia with maybe a net add in their number of seats from the redistricting wars that Donald Trump started.

Is that the sort of thing that can help, though? It, it sort of seems like that's not really going to do much.

What does your latest polling find on that, and what's your general sense of it?

Uh, the House I think is probably a foregone conclusion for the Democrats, if I'm being honest with you, Greg.

I mean, the generic ballot in our survey is D plus seven.

There's a pretty predictable historical pattern where the party out of power gains ground as the midterms go along.

So the party out of power is the Democrats.

We should expect them to gain at least a couple points in the generic ballot over, over the next couple months.

Um, and the Democrats are emerging from this referendum in Virginia with maybe a net add in their number of seats from the redistricting wars that Donald Trump started.

Is that the sort of thing that can help, though? It, it sort of seems like that's not really going to do much.

Ah, I like that.

I figured it was Saturday, so let's have some fun.

Look, the Cook Political Report is great, good, good ratings analysts, but one common, uh, factor with these ratings is that, uh, they're, they're a little lagging.

The, the race raters take some time to update the race ratings, uh, based on, on-the-ground sentiment.

So what I have found cycle after cycle is that as things move in one direction, the race raters take a couple of months, uh, to, to really pick up on those changes.

And over the last couple of months, there's been, of course, a, a lot of changes in the national political environment, um, that make things look more Democratic.

So I think these adjustments they're making are a signal, uh, that the trend we're seeing in American politics is just towards the Democrats, and that makes the Senate opportunity pretty, pretty bullish for them.

What, what, what in your opinion is beyond the pale?

Elliott, can you talk about what you found?

Yeah.

I keep seeing evidence in these surveys that just gets worse for President Trump every time we take a new poll.

Um, so the trend here, in other words, is deeply negative, and it has been for a while.

It's one thing to be at minus two on border security, your previous best issue, for example, but when that number has declined 12 points over the last year, and that's...

I, I mean, to be honest, right, that's probably the one policy area where Trump is delivering on some of his promises, there being no, um, no significant traffic crossing the, the border, at least that, uh, shows up in the statistics.

That's a real indictment of the presidency overall, and really, it only gets worse from there.

Why is that happening? Is it because Trump is such a profoundly damaged figure, and is it because polarization's worse? What, what explains the compression and timetable here?

Elliot, can you talk about what you found?

Yeah.

I keep seeing evidence in these surveys that just gets worse for President Trump every time we take a new poll.

Um, so the trend here, in other words, is deeply negative, and it has been for a while.

It's one thing to be at minus two on border security, your previous best issue, for example, but when that number has declined 12 points over the last year, and that's...

I, I mean, to be honest, right, that's probably the one policy area where Trump is delivering on some of his promises, there being no, um, no significant traffic crossing the, the border at least-That, it shows up in the statistics.

That's a real indictment of the presidency overall, and really it only gets worse from there.

Why is that happening? Is it because Trump is such a profoundly damaged figure, and is it because polarization's worse? What, what explains the compression and timetable here?

You said what we saw recently was a directional shift.

Yeah, Brooke, there's this long term historical trend, right, where after a new president takes office and the subsequent election that the other party gains a lot.

In this case, the shift was actually larger than most pollsters expected and larger than the historical pattern.

To put some numbers on this, Abigail Spanberger, the Democrat in Virginia who won, she won by about 15 points.

That is about a 10 percentage point swing to the left since the last election, since 2024.

Usually, the swing is about half that large, so this indicates a pretty big victory for the Democrats, and they are moderate.

To our bigger question here, do moderates do better in elections? Well, this from the surface would seem to suggest they do, right? That you run Abigail Spanberger, you run Mikey Sherrell in New Jersey, two moderate candidates, they do really well.

So that suggests, big surprise, that politics is actually local.

When you say, you know, underwater by 30 points or 20 points, you're not saying that 80% like it or 70% like it, right?

In this case, if it's 20 points underwater, 60% of people would oppose the policy and 40% would support it, yes.

YouGov did this, a great recent study, they're a pollster, earlier this year where they asked about pretty much every Trump policy under the sun and, and things you can characterize as the economy, 36% of the public supported Trump on those policies.

The numbers are similarly dire if you look at the other policies of the Trump administration early on.

His healthcare policy, supported by 27% of the public.

I have found that deploying of troops to help with immigration enforcement, what Trump's doing in Chicago, what he did in LA and Oregon, is opposed by at least 60% of the public.

And if you ask about Marines in particular, it's a policy opposed by 65%.

If you had, like, $40 million, how would you use it to win voters in places like where you grew up?

Yeah.

So 11 percentage points better.

Um, they ended up doing, again, you know, an eight point margin in the H- in the House, and that was a shift of, um, of double digits as well.

So again, these numbers are pretty close.

At this point in the 2026 election cycle, Democrats are doing 16 percentage points better than they di- uh, than, uh, Kamala Harris did in 2024.

So an even larger shift in the special election index.

Now, uh, I, I wouldn't say this is particularly bullish for them, um, a- at least, like, in comparison to 2018.

And so, i- isn't this kind of becoming a recurring theme here?

Certainly not if they disapprove of the job that, that he and his party are doing.

Yeah, so a lot of the Trump voters, I might say, uh, that were young people, uh, in 2024 but were Biden voters in 2020, they're not Republicans.

They might vote for Republicans but they don't identify as strong Republicans.

They might not be firmly in that camp in the next election.

Uh, w- w- we think that for a couple of reasons.

First, on policy preferences, they're not conservative.

Um, they favor a lot of wealth distribution, in particular they favor high taxes on wealthy people, um, and they a- are also oppo- opposed to a lot of the action, um, in the Middle East, especially, in Israel and Gaza.

How are you thinking about the broader economic numbers in terms of inter- in terms of our sprint toward 2026 here?

Everyone should read that too.

He, Elliot Morris is the author of the Strength in Numbers Substack.

Welcome to Fast Politics, Elliot Morris.

Hey, thanks for having me back, Molly.

So, I always like to, just for everyone who's not completely read up on this, I've been a long time fan of Elliot's work doing polling, statistical, sort of the more math side of politics.

I think that's fair.

Um, and I'd long wanted- That's a very generous way to put it, Molly.

Oh, yeah, 100%.

No. No.

So here, Johnson again conflates releasing the grand jury testimony with releasing the files.

Note again the effort to portray Trump as supremely transparent.

Like I said, the message does appear coordinated.

They've got to be looking at similar data to what you're talking about here.

And, and let me ask this, Elliott, if I could, there's gonna come a point when House Republicans get even more nervous about this, aren't they? Yeah, and I think that point's gotta come pretty soon.

Uh, so we know that the, that Trump's really unpopular with all voters, but he's also unpopular with Republicans, the types of people that vote in the primary elections.

This cuts against his m- mystique, his political mystique, at a very deep level.

Right.

Um, the thing that I think he's indexing on is the previous data.

I think people in the White House see that Trump has traditionally been, uh, approved of on immigration in general.

And when it comes to quelling violence in the street, they have some experience dealing with that as well in ways that they think, um, go, goes well for them.

And, and I'll also just add, like the previous data we had on LA was about the protesters themselves.

YouGov also asked, "Do you approve of the protesters doing the, you know, the things they're doing?" And, and Americans said no.

By a nine point margin, they said no.

Is there any historical precedent here, for example, as it relates to Kenosha? Um, the last time there was something similar, uh, that Donald Trump is kind of hanging his hat on or that we can, um, or that we can, uh, uh, compare these results to?

Can you tell us what you found?

Sure.

Well, look, the motivation here is that there's an assumption that the politics of sending in the National Guard and the Marines, uh, against these protestors, assumed to be, like, violent rioters by the administration, at least that's what they want to sell, there's an assumption that that's good politics for Trump and for the administration.

Um, and that at least doesn't seem to be the case immediately.

Uh, so YouGov did some polling overnight, uh, and they asked the question, "Do you approve or disapprove of Donald Trump sending the National Guard, uh, to quell protests in, uh, California?" That's, that's all they asked.

And this question is seven points underwater, so 38% approve and 45% disapprove of this.

Uh, YouGov asked the same question, they asked, "Do you support sending the Marines in?" And even more are, are disapproving of this.

What did you find there?

What we've sort of heard is that 70% of the country is sort of the mo-... is more centrist.

Hmm.

I don't know.

I, I'll just add some nuance here.

So, like, usually people don't have moderate opinions on a bunch of different issues and you call them moderates.

That's really rare.

Most people don't pay attention to politics, and they feel very intensely about a couple of issues, and then they pick countervailing opinions on a bunch of stuff.

When you look at these protests, what do you see in the data on the protests?

Can you talk about Elon Musk's popularity and whether it was a mistake or how much of a mistake it was to have him be the face of the Republican effort to win in Wisconsin?

Yeah, that's a great question because I think it is probably a, a growing liability for, for Republicans, and it seems like they're noticing this.

There was some reporting in The Daily Beast out yesterday with, uh, Republicans in, in close competitive races saying they would help, they were hoping for a blowout in Wisconsin so that they could make the case that he sort of toxic to Republican efforts.

Uh, and the polling data suggests that, that he is.

Uh, so one of the last pieces I wrote at 5:38, uh, at ABC News, um, w- was a comparison, uh, of the popularity of, of all of the different policy actions by the Trump administration in the first month.

Um, so from January 21st to February 21st.

Um, when we looked at that, uh, we could see, y- y- you know, these polls ask Americans, "Do you support, uh, the president's actions on Israel, on the Gulf of Mexico," or you know, whatever.

Yeah.