Taking the Pulse of the State

Camilla Shumaker: Welcome to Short Talks from the Hill, a podcast from the University of Arkansas. My name is Camilla Shumaker, and today I’m talking to Janine Parry, professor of political science. Welcome Janine.

Janine Parry: Thanks so much. I’m glad to be here.

Camilla Shumaker: We’re going to talk about polls today. So you do the Arkansas Poll. Tell me about that.

Janine Parry: The Arkansas Poll is an annual public opinion poll of adult Arkansans. We take the pulse of the state with respect to politics and policy. I’d add that we’re still the only wholly public public opinion project, meaning everything we do is totally transparent. I think that’s important. We also are turning 20 years old this year and roughly speaking that appears to make us one of the oldest in the country.

Camilla Shumaker: So why is it important to regularly take the pulse of the public?

Janine Parry: The reason we started the poll, those of us who were involved in it wanted to offer the public and policymakers and certainly also journalists a more frequent portal into public preferences than elections allow. And so it was, that was a real opportunity, so there was definitely a public service component. In addition to that, in political science as well as in a good portion of psychology and sociology and a few other fields, public opinion research is really valuable of course in terms of publication and scholarship.

Camilla Shumaker: Over the past 20 years of polling, what large trends have you seen?

Janine Parry: In most respects what I see is stability. So when you look at some of these eternally hot-button policy issues, say abortion regulation or gun regulation, there are little bounces up and down over the years, but we ask the same question almost every year on both of those issues, and we see very little movement in terms of the public’s preferences. That tells me that in any individual year I’m not getting noise. There’s great comfort in the stability, like okay we’re really measuring something real here.

The most significant change however is the one that we’ve seen in partisan identification. So when we ask “Do you consider yourself a Republican, a Democrat, an independent, or what?” From 1999 all the way through until 2010 the division was fairly even: about a quarter Republican, about 40 percent Democrat, and 35 to 40 percent independents, and sometimes the Democrats and independents would switch.

But in 2010 it took a dramatic turn. We started to see an uptick in the Republican identifiers, but the main thing that happened was with that enormous group of independents who really drive statewide elections in Arkansas. They took a hard right turn in our second question, which is where we asked just the independents, “Well generally speaking—you can tell me—do you lean Republican, lean Democrat, or are you just independent?” And in 2010, completely in sync with the change in our election outcomes after more than a century of patterns, we saw them take a hard right turn, and whereas about a third had divided out into lean Republican, lean Democrat and independent for a while, for since the Arkansas polls inception, suddenly about half of the sample said “Well I’m leaning Republican.”

Camilla Shumaker: Are there any other questions besides that one that you’re interested in seeing the results from this year?

Janine Parry: This year I’m mostly just interested in being able to truly have 20 years, you know full stop, of data on every question.

Camilla Shumaker: You mentioned working with other researchers. Tell me a little bit about that relationship.

Janine Parry: What we do, and we started doing this early on, and it’s just really exciting because it exposes me to all the wonderful research going on, not just in other departments in Fulbright College, because this poll lives with the Blair Center in Fulbright College, but researchers all over campus. So we’ve had these what we call collaborating researchers and we offer them space on the poll for free which is quite valuable opinion data is, good opinion data is expensive.

So I’ve worked with accounting professors on questions regarding, regarding tax preparation, economists on water quality and perceptions of that. The education reform team did some very exciting stuff around the time that there was a real significant volatility in Arkansas’s K through 12 public school funding landscape. We’ve done things on the death penalty, things with folks over in psychology. So it’s just a great opportunity being exposed to other research questions and other research findings all of which have public policy applications.

Camilla Shumaker: Alright so talking about polls in general, what are the benefits and the challenges of doing research through polling?

Janine Parry: Well, I think the benefit is that we get a more regular snapshot of what the public wants, at least the public we can get to participate in a poll, which looks a lot like voters look, so not a perfect representation of the public at large, but as it turns out an incredibly accurate a snapshot of what voters look like. Plus elections are kind of blunt instruments, right, in a representative democracy sometimes we get to vote directly on issues but it’s an either/or. Particularly in our collaboration with researchers from other departments and colleges we can dive a little deeper into people’s preferences.

Camilla Shumaker: Tell me about the, the logistics of it. How do you actually get the information and what are the what are the kind of traditional techniques versus the new techniques?

Janine Parry: The main thing for us right now has been the transition to cell phones, and we have just slowly increased the percentage of the calls we make. We use live interviewers so right away we’re different from a lot of the less expensive projects that are happening out there, but our live interviewers are making a higher and higher percentage of calls to cell numbers instead of to landlines, because of course we all know that the people who have landlines are not representative of the broader American public at this point. It’s more expensive to make those calls but it’s really important for us to do in order to capture younger people, working people, parents.

We’re still maintaining traditional standards of household randomization and callbacks. So it costs a lot of money because it takes time to have a caller call one number because it comes up randomly and find that no one’s home and then just to keep calling that number instead of moving on to the next number. If you keep moving on to the next number what you have is a sample of people who are home, which is not a sample of voters and certainly not of the public so we have a standard now, I think, still, of five or six callbacks to the same number before we move on even though it’s quicker it’s more efficient to move on it’s not going to be as accurate so we don’t do it.

Same thing when it comes to the within household randomization, which is simply to say, and some of our listeners may have heard this before, somebody calls they want to take a poll and they might ask to speak with the adult male who has most recently celebrated a birthday and you think “What are you you’re gonna send me a card, what’s the deal? What are you selling?”

What we’re trying to do is offset the reality that women are both more likely to answer the phone and more likely to cooperate so we almost always at some point have to actively be pursuing male respondents. And then the birthday thing, right, is randomizing. If there are more than one adult at home we want the one just who comes up randomly.

Camilla Shumaker: How many responses do you usually get?

Janine Parry: We’re at 800 responses which gives us a margin of error of plus or minus 3.5 points, so that’s a pretty tight margin of error.

Camilla Shumaker: So how can you tell, if you come across a poll on the internet or something like that, how can you tell if it’s the information you’re getting is reliable?

Janine Parry: My best advice, and the thing, the rule I follow myself, is I should be able to see all the information about how the data were collected. Was it live interviews? Was it an online poll? Was it a random sample of the country or just of a state or a region? Was it done just among Amazon shoppers? The second screen I would use—and this should be available in the same place—is I should be able to see the questions and the response categories. Some people would consider that to be a little bit pickier, but I know enough from experience and scholarship to know that the way the question was asked and the response categories that were offered has everything to do with findings that were revealed, so I think that’s something that’s important to the methodology. The protocol, the interview questions should be available to people who want to consume that information.

Camilla Shumaker: So when is, roughly, when is the next Arkansas poll?

Janine Parry: We have traditionally released, I think every time, sometime the last week of October or the first week of November. So folks should look for information from us around that time this year.

Camilla Shumaker: And where can you find all the methodology?

Janine Parry: It’s on our summer report, which we release every year, or you can click all the way through to the complete protocol. All of those things are available in clickable links, in addition to our full datasets. All of those are available on the Arkansas poll website, and the quickest way to reach us is just to Google the Arkansas poll in quotes.

Camilla Shumaker: Thank you very much.

Janine Parry: It’s my pleasure. Thanks for having me.