Latest / Houston's Morning News on KTRH / Mark Jones - Rice University Joins Houston's Morning News
Transcript
- 0:00Speaker 2: Ah, yes.
- 0:01Speaker 1: Let's talk about some polling data. 622 is our time.
- 0:05Speaker 1: Here on Houston's Morning News. Polls, you know, I'm not
- 0:09Speaker 1: real wild about polls. Polls have been just horrible. It
- 0:14Speaker 1: seems like the last few election cycles, they've just been way, way, way,
- 0:18Speaker 1: way off. So when I see these polls, I just
- 0:21Speaker 1: kind of, I go, well, we'll see. The only poll
- 0:25Speaker 1: that really matters is the poll on election day. Mark
- 0:28Speaker 1: Jones joins us at Rice University. You've done a few
- 0:30Speaker 1: polls in your day, Mark Jones. What makes for a
- 0:32Speaker 1: reliable poll and what makes for a not so reliable poll?
- 0:37Speaker 2: Well, I think one thing that we've seen is a
- 0:40Speaker 2: proliferation of polls by groups that have an agenda. That is,
- 0:44Speaker 2: the polls have actually become part of the actual campaign.
- 0:47Speaker 2: This could be the campaigns themselves releasing the poll or
- 0:50Speaker 2: groups that support the polls or groups that may appear
- 0:54Speaker 2: independent but actually are not independent. They're the ones supporting
- 0:57Speaker 2: the polls. Those are the ones that you have to
- 0:59Speaker 2: be really skeptical about. Then you have another group of polls.
- 1:03Speaker 2: Those are being done by a lot of groups from
- 1:05Speaker 2: outside of Texas that are coming in to poll in
- 1:08Speaker 2: this race, in part just because the Paxton-Talarica race is
- 1:12Speaker 2: generating such nationwide attention. Those polls, every pollster has their
- 1:16Speaker 2: own methodology. You mentioned the New York Times-Siena poll. They've
- 1:20Speaker 2: adopted a new methodology for this cycle. I think it's
- 1:24Speaker 2: a situation where they're going to be actually really right
- 1:27Speaker 2: and everybody is going to say, wow, they got it,
- 1:29Speaker 2: or they're going to be really wrong. And people are
- 1:31Speaker 2: going to say they need to revisit their methodology and
- 1:34Speaker 2: how they approach these polls.
- 1:35Speaker 1: Well, what is their methodology? What has changed about the
- 1:38Speaker 1: New York Times-Santa Paula methodology?
- 1:42Speaker 2: It's a much more complex system of waiting. That is
- 1:45Speaker 2: they tried when they get the respondents that respond, they
- 1:49Speaker 2: try to wait them on what they think the electorate
- 1:52Speaker 2: is going to look like across a whole host of dimensions.
- 1:55Speaker 2: Once you start waiting in that complex way, your methodology
- 1:59Speaker 2: is really having a big effect on the result you get.
- 2:03Speaker 2: They also are using pretty small sample sizes in like
- 2:06Speaker 2: the 600 to 700 range, which generally is pretty small
- 2:10Speaker 2: for a statewide race. It's not to say wrong or bad,
- 2:14Speaker 2: but it's one where you're likely to have a little
- 2:16Speaker 2: more air than if you say we're doing a more
- 2:19Speaker 2: traditional 1,000 or 1,200 person surveyed. But New York Times-CNN
- 2:23Speaker 2: does a lot of polls across all of the countries.
- 2:26Speaker 2: So when they did their most recent release, it was,
- 2:28Speaker 2: I think, half a dozen races across the country, five
- 2:32Speaker 2: or six, meaning that they're spreading their resources a little
- 2:36Speaker 2: more thin because they're trying to do races across the
- 2:38Speaker 2: entire country.
- 2:39Speaker 1: Okay. When you do a poll, it seems like you
- 2:45Speaker 1: talk to either likely voters or– I guess there's two
- 2:49Speaker 1: different categories, right? People who are likely to vote and
- 2:52Speaker 1: then just the general opinion people have about certain candidates.
- 2:55Speaker 1: How do you go about trying to determine if a
- 2:58Speaker 1: likely voter indeed is someone who's likely to turn out
- 3:01Speaker 1: and vote?
- 3:03Speaker 2: Right, so at this point, registered voter models, you really
- 3:06Speaker 2: shouldn't be doing them. Some people use registered voter essentially
- 3:11Speaker 2: polls about people's opinion earlier in the cycle. But right now,
- 3:14Speaker 2: you should be focusing on likely voters. Likely voters are
- 3:17Speaker 2: based usually on a couple of different factors. The polling
- 3:20Speaker 2: firm determines, looks at their past voting history. Did they
- 3:23Speaker 2: vote in 2024? Did they vote in 2022? And use
- 3:27Speaker 2: that type of information to say, okay, are they a
- 3:29Speaker 2: likely voter? And then within the survey itself, they're asked
- 3:32Speaker 2: usually one or more questions that tries to gauge how
- 3:36Speaker 2: likely they are to vote in their mind and then
- 3:38Speaker 2: what their overall enthusiasm is about voting. You put all
- 3:42Speaker 2: those factors together and you can get effectively a model
- 3:46Speaker 2: of this person is very likely to vote, this person
- 3:48Speaker 2: may vote, and this person is very unlikely to vote.
- 3:51Speaker 2: And then you essentially design your– you write up your
- 3:55Speaker 2: report and do your analysis accordingly based on those different
- 3:58Speaker 2: probabilities of voting.
- 4:00Speaker 1: I'm guessing, Mark Jones, that you probably look at a
- 4:01Speaker 1: variety of different polls and maybe average them out in
- 4:03Speaker 1: your mind. And maybe you look at ones that you
- 4:06Speaker 1: see that have been historically accurate in the past. Based
- 4:08Speaker 1: on what you're seeing in the Senate race here in Texas,
- 4:12Speaker 1: the polls you look at in the governor's race, what
- 4:15Speaker 1: is it telling you about the likely outcome?
- 4:19Speaker 2: Well, so when I'm looking at the polls and we're
- 4:23Speaker 2: doing polls ourselves, I do polls at Texas Southern University
- 4:26Speaker 2: and University of Houston that I collaborate with colleagues there.
- 4:29Speaker 2: The polls we've been doing tend to have James Salrico
- 4:34Speaker 2: up by a very narrow margin, one or two points,
- 4:37Speaker 2: something that may or may not exist on Election Day.
- 4:41Speaker 2: Where I differ, I think, a lot more from, say,
- 4:43Speaker 2: this New York Times-Santa poll is I see Greg Abbott.
- 4:46Speaker 2: in a difficult situation that is only about maybe four points,
- 4:50Speaker 2: five points ahead of Gina Hinojosa. But as long as
- 4:54Speaker 2: Republicans turn out to vote, I think Abbott is still
- 4:58Speaker 2: in a relatively safe place in terms of winning. Definitely
- 5:01Speaker 2: not down, say, four points like in your term, Santa Haddon.
- 5:05Speaker 1: Okay. Good to hear from you, sir. Thank you. Appreciate
- 5:08Speaker 1: your time. That's Mark Jones from Rice University.