The Bulwark Podcast - Ann Selzer: How Could This Be?
Episode Date: November 3, 2024Few states are as red as Iowa, and yet the legendary Iowa pollster Ann Selzer found that Kamala has leapfrogged over Trump to take the lead there. The turnaround is due to women—particularly women 6...5 and older, who previously tilted toward Trump, but now favor Kamala 63% to 28%. Iowa's new strict abortion law could be a factor. J. Ann Selzer joins Tim Miller for a special Sunday pod. show notes Des Moines Register story on the Iowa Poll Des Moines Register story on Iowa's congressional races
Transcript
Discussion (0)
Hello and welcome to a bonus Sunday edition of the Bullard Podcast.
I'm your host, Tim Miller.
We got the shock seltzer Iowa poll last night, 47 to 44 among likely voters in favor of Vice
President Kamala Harris in Iowa.
So obviously, we had to have an emergency pod with the Queen herself, Jayanne Seltzer,
president of the polling firm Seltzer & Company.
She's best known for conducting the Iowa poll for the Moyn Register.
Thanks for coming on the Bullwork Podcast.
Thank you, Tim.
Good to be with you.
No tiara for the Queen today.
I leave my tiaras and my crystal balls at home.
Okay.
Well, let's just start first with what happened to the poll.
I got to tell you, I looked at it three times before I realized that Kamala was winning.
I just saw 47-44 and I was like, that's pretty good for the vice president down three.
And then it took me about two minutes to like register that she was up three.
So anyway, I'm wondering just at the top level what you saw as happening in the poll and
then we'll kind of go through the particulars.
I don't think anybody would look at the numbers and not be in a somewhat state of shock.
We started interviewing Monday night, so Tuesday morning I walked in, my assistant was already
crunching numbers and she said, did you see the unweighted data?
I go, yeah, I'd like to see the weighted data.
So we're always very cautious about embracing things too much.
I'm always telling my clients, don't get married to these numbers because they can change.
But all they did was confirm what we were seeing in the unweighted data after Monday
night that Harris was in the, the unweighted data, uh, after Monday night,
that Harris was in the lead in Iowa.
And that was nobody in the right mind would predict it.
Our methodology is to make no assumptions and we made no assumptions. And here we are.
Yeah.
I mean, quite a remarkable change from your poll over the early summer that had the president Biden down 18
To a poll in September that had it had it significantly closer to this one
Like what what are you seeing as having been the impetus for such a dramatic turnaround?
You know, I can't give you a data-driven answer which would be my preference
I can't give you a data-driven answer, which would be my preference. What it seems like, given that just this morning, we released the findings from the congressional
races, and we are now showing two Iowa districts tilting toward the Democratic candidate, and
that includes a big lead in the first district, which is 53 to 37.
That district in particular has the driving issue from what I read has been the abortion
issue.
It was over the summer that Iowa's six-week ban on abortion went into effect after all
the court challenges were taken care of. So that's the decision being made, the law going into force, and then people living with
it for a while, and it being a strong, sort of compelling thing from Iowa's, in the congressional
districts, Iowa right now, all four congressional districts are held by Republicans.
So to me, it seems like to a certain extent, there's something organic going on because
the presidential candidates certainly aren't spending any money or time here.
And it may be in fact that these congressional races are driving the presidential race toward
the Democrat.
I want to get into the congressional races a little deeper, but just a couple of the
things at the presidential level first.
The gender gap is particularly noteworthy and you look at independent women, older women
in particular.
Talk about what you found on that, which obviously could also speak to the abortion question.
The pollsters who work with contests that are going to be decided by a gender gap typically
say you need to win with women more than you lose with men.
And we're seeing that in these data right now.
So men are 52-38 for Trump, but the margin is wider, 56-36,
a 20-point gap with women for Harris.
The gap for women aged 65 and over, and this, you know, the age 65 and over group, I've
been watching this whole cycle because they, in past cycles, have tilted more Republican,
but now they are strongly tilted toward Harris.
And the margin among women 65 and over is more than two to one.
It's 63, 28.
Wow.
Again, that's sort of a jaw-dropping number.
Pete I didn't actually look at this. Do you do a religion cross-staff? Do we know like
our church going? I do wonder how that cuts, particularly among women.
Dr. Cate I can't give you among women, but I can give you evangelicals, which is the
group that has been consistently and staunchly supportive
of President Trump.
And that's more than three to one.
So they haven't shaken off of their affection for the former president.
But like, as people are looking at this, you know, everyone's trying to explain it.
Like, it's just such a dramatic result.
I guess I should have noted at the top, you've been gone contrary to the conventional wisdom
several times.
The Obama caucuses in 2008 were right about that.
I remember the feeling in the pet of my stomach still seeing your poll in 2016, showing Donald
Trump winning by a wider margin than expected significantly in Iowa, which turned out correct.
I just insert here, I have a nickname as I have a
couple of nicknames.
One is the outlier queen because people keep
seeing my numbers not match others.
So I've been in this seat before.
Uh, the second is starting with the
senatorial race between Bruce Braley and Joni Ernst.
Our poll was the only one showing Joni Ernst
with a comfortable lead.
And I became known among those campaign workers
as the harbinger of doom.
The one time that was not on my side.
That was my final cycle as a Republican in good standing.
So I was actually consulting on the harbinger of good news
side of that Senate race.
Yeah, so you've been here before in the app.
But anytime there's an outlier, look,
sometimes they're just outliers, right?
Sometimes they're just like, that's just statistical modeling.
Sometimes they're just outliers that don't turn out to be harbingers of doom or glory.
But your outliers have tended to direct towards what the end result was going to be.
But when there is an outlier, people are like, okay, what could it be?
And one thing that there was a series of New York Times, CNN polls today that Nate Cohen observed was that among older
voters, like there's a Democrats are more likely to like answer,
right? Like there potentially is a, like, do you worry at all
about kind of response bias where like older Democrats, for
whatever reason, are more interested in taking the poll?
Well, our method controls for that.
Our method is we're looking to end up with 800 likely voters, but for people we talk
to who don't qualify as a likely voter, that is they've already voted or they say they
will definitely vote.
That's the two ways you get into my pool.
Then we go ahead and collect your age, your sex, and what county you live
in, I think race as well. So we have demographics for the full Iowa adult 18 and over population,
and we weight that to the known population parameters. So now we have a cross section
of all Iowans, and we can extract from that the 800 and I think it was 808 people who qualified.
So if older people are more likely to vote, and they are, they show up in my voter sample
in more plentiful numbers. So for example, we talked to, had contacts with 1,038, all Iowans, in order to get enough likely voters at 808.
So we've controlled for that idea of response bias by making sure they're not overrepresented
at the state level, and then they are what they are at the likely voter level.
That would control for the fact that the likelihood of older voters, the correct amount, the likelihood
is in the sample, but does it control for the fact that older voters who are more likely
to vote for Donald Trump might not be as likely to answer?
I mean, I guess the non-response bias is the question.
Yeah.
Well, you can't control for that.
If we have too many 65 and overs, we weight them down so that they're not overrepresented
at the state level.
They might look like a larger population than they are among the likely voter sample, but
that's because they're more likely to vote.
One of the other things, as you mentioned, you're the outlier queen, so you're comfortable
with this, but Nate Silver wrote in his Silver Bulletin last night, talked about the cojones
that you had to drop this poll. And he said,
basically, that what we've seen out there in the industry is
that there's a lot of what he calls poll hurting, right?
Whereas people are waiting things to 2020 recall vote or
waiting things to this or that in order to not have a result
that ends up is this much of an outlier.
Clearly, that's not something that ever crosses your mind, but I'm curious what you think
about the idea that other people are hurting and that might explain why you're so unique
right now.
I am grateful to have clients who we work intimately together.
They know my method.
They know my history.
And so, the Des Moines Register had every right to say,
this is too far off.
Nobody will believe it.
We're going to can it.
But they don't.
And I think they think like I think,
is we've taken our very best shot at what we think
is what's true out there.
We've tweaked a little bit here and there.
There are things that we decided that's a corner cut that we don't want to do.
So I don't lie in bed thinking, I wish we'd done something
different. It doesn't mean I'm not nervous about whether my poll will be
good or not good.
As for other people's methodology, to be honest,
that's the first thing I go looking for when a poll is published by anybody,
is to see what it is that they're doing.
And I kept my mouth quiet about this idea of waiting to the recalled vote.
So how did people vote in 2020?
I have a lot of questions about how that would work out actually, arithmetically.
But I have talked about my method as pulling forward.
And so what I want is my data to reveal to me without favor what this future electorate
looks like.
So when I hear people saying, well, you know, we're trying to figure out what this future
electorate looks like, and I kind of think to myself, why bother?
How could you?
And if you're looking at the past to say that's what the future is going to look like, to
me, that's pulling backward.
So you're doing something that will give you results that look like 2020, perhaps.
I want results that look like 2024.
I was listening to an interview you did back after 2020 when once again, you weren't quite
as much of an outlier in 2020, but it was a little bit of an outlier again.
It was less favorable to Biden than what the rest of the national polls were showing, particularly
in the region.
It turned out to look more like the electorate that you had prognosticated.
In a post, one of these post-op interviews, you you did worry about like how long this method will last, right?
Like that there are potential issues still with polling.
I've a pollster friend of mine who called me a couple of weeks ago and he's like, you
know, you're taking these polls too much at face value.
We really, it's so hard to poll 18 to 30 year olds right now and like, we don't know.
So I'm just wondering, like what are the anxieties that you have looking at the current polling
methodology?
Well, I'm guessing the anxiety level will tick up on Tuesday, but that's sort of par
for the course.
I've had a good track record and like to say, may it always be so.
But I do think, and I have said, if people looked at my methodology and it's published,
there's no secret sauce.
People sometimes think that.
It's published in every article in the Des Moines Register, right there, how we do it.
But you look at it on paper and you go, it's too simple.
This can't possibly work.
And so far it has.
But I'm prepared that one day it will not work and that I'll blow up into tiny little
pieces scattered across the city of Des Moines. that one day it will not work and that I'll blow up into tiny little pieces
be scattered across the city of Des Moines.
Well may that not be today. The other note you mentioned the congressional
numbers at the top there are four the Republicans hold all four districts in
Iowa right now. That Iowa third district which is around Des Moines you showed the
Democrat winning but by a margin that feels potentially possible it's a swing
district Des Moines has a lot of college-educated Republicans that are You showed the Democrat winning, but by a margin that feels potentially possible. It's a swing district.
Des Moines has a lot of college educated Republicans that are trending left.
Then you, as you mentioned, the first district, which is Iowa city, which has
a college, but then also some of rural kind of Eastern Iowa and some other
of the smaller cities out there.
And the quad cities.
Quad cities, Davenport.
That had a big lead for the Democrats.
And so I do wonder, do you look at that and think, hmm, I don't know, like this, I mean, obviously, you're not going to change your results.
But I just wonder if that first district congressional results gave you any kind of pause
about the overall makeup of the state? Well, I keep in mind that we had been seeing something
going on in the first district in terms of approval ratings, even for Joe Joe Biden for more than a year.
And so when we would do briefings with the reporters and editors, I kept saying, what's
going on in the first district?
There's something going on in the first district.
So this wasn't new, but it was something that caught our attention to have a lead that wide
for someone from the opposing party.
So we gave that extra attention and looked at it several different ways.
And in the end, we had no reason to doubt and keep in mind,
we wait by geography as well.
So there, there wasn't a way for us to be over representing that district, unless
they're more in that district intending to vote.
So you said Monday night, the unweighted was showing good stuff for Harris.
When you saw the final numbers, which I guess would have been what, Friday morning?
Correct.
I'm just curious, what was your mental state there?
Did you have any, were you starting to sweat at all looking at it?
Did you look at it two times like I didn't say, wait, she's really winning by three?
Well, we watch it day by day.
So it was an incremental difference on Friday morning.
I will say that my nephew is a senior
at the University of Nebraska
and currently interning at Gallup, interestingly enough.
And he wanted to come and I wanted to have him here for this.
And he walked in Thursday night and I had a folder,
I had to swear him to secrecy.
He has to take the oath.
And he opened the folder and he couldn't believe
what he was looking at.
And his mouth sort of fell down to the kitchen counter
and it took him a while to wrap his head around it.
So you're not alone in, first of all,
not recognizing what the poll numbers meant,
because your expectation was for a completely different difference.
We do a preliminary briefing with the reporters and editors on Thursday.
By the time we got there Friday, we're settled on a plan of action for how to get the stories out.
And I think I'd answered all of the questions people had about how could
this be to their satisfaction.
And now the voters will tell us.
Yeah.
We'll see.
You said it was changing incrementally through the week.
Was it moving one way or the other through the week?
You know, if you look at the rolling two day average or even a rolling
three day average, it's pretty flat.
There there's daily variation as there is with any poll.
The last one of these that you had a miss on was 04, John Kerry.
Does that one still haunt your dreams or do you have all the successes
since then washed that away?
I was just starting to get over it.
No, interesting.
I was told by someone who had access to exit polls that not only would carry
when Iowa, but by a wider margin than we thought I had an incident where I ran
into former governor Terry Branstad.
This is before his second reign.
And he saw me in a conference room where he was the president of the organization
and he walked in and said to me, long story but the short of it is that because we had
to stop polling then on Friday night, we couldn't have gotten the late Bush search.
There was a rally in Sioux City that he'd never seen anything like it in terms of the
work that they were doing with the rally
Participants to have them go out and get more votes
and he said you couldn't have caught that in your data and the fourth district which is where city is came in with a
big margin for Bush and
Governor Branstad claimed that rally was the reason Tom Daschle was evicted from
his Senate seat in next door South Dakota.
That was a gift to me for someone who understands where every vote is in this
state to say your poll couldn't have caught it.
It takes a few years for that pit in your stomach to resolve.
I have a lot more L's than that.
So you can, you work, you work your way through it.
All right, last thing, is there any,
I mean, obviously Wisconsin is not a mirror image of Iowa.
There are differences in demographics,
is there anything you think can be learned
about the other swing states from the data that you saw?
I said before, when people wanted to say,
well, if this is going on in Iowa,
this is what's going on in Wisconsin, that our state parties function very differently.
Iowa is so much more red than Wisconsin is.
Both houses of our state house are Republican by large majorities.
Every statewide executive that's elected by the entire state are Republican,
except for the auditor and the entire Washington delegation is Republican.
So it's hard to get more Republican than Iowa and Wisconsin is more purple than that.
I'm not going to say they're blue, but they've got, they've moved back
and forth more than Iowa has.
So I don't think I can make any claim about Wisconsin
unless we did our method in Wisconsin.
All right, well, the whites seem kind of similar though.
The types of people, the culvers,
the kind of type of food that they're taking in,
I don't know, I'm hoping to take some kind
of kind of cultural similarity
between the Badgers and the Hawkeyes,
but I guess we'll have to wait and see on Tuesday.
They do have a professional football team, which we do not have.
That is true. Thank you so much, Anne, for coming onto the Bullard podcast,
morning after the big poll, 47 for Kamala Harris, 44 for Donald Trump. We will see
how things shake out on Tuesday and hope to be in touch, I don't know, sometime
before the 2028 Iowa caucus.
All right.
Good to talk to you, Tim.
Thanks, Dan Seltzer.
We'll be back tomorrow for our normal Monday edition of the Bullwork Podcast.
Peace. to thinking that I thought
Maybe I'd get less stressed if I was tested less like all of these debutantes
Smiling for miles in pink dresses and high heels and white yards
But I'm not, baby I'm not No, I'm not, that I'm not
I've been tearing around in my fucking account 24-7 Sylvia plaid
Writing in blood on my walls cause the ink in my pen don't work in my notepad
Don't ask if I'm happy, you know that I'm not But at best I can say I'm not sad Cause hope is a dangerous thing for a woman like me to have
Hope is a dangerous thing for a woman like me to have
The Bullork Podcast is produced by Katie Cooper with audio engineering and editing by Jason
Brown.