The 2016 presidential election saw the reliable predicting power of polls come under increased scrutiny. The previously infallible institutions seemingly did not foresee the election of Donald Trump, causing many pollsters to soul-search in preparation for the 2018 midterms. This was not the case in Arkansas. The Arkansas Poll has become a paragon of stability since its inception in 1999. In light of the upcoming 20th anniversary of the Arkansas Poll, I sat down with Dr. Janine Parry, its founder and director, to discuss the Poll, and politics in Arkansas as a whole.
What does it mean to be the director of the Arkansas Poll?
“It means that for 20 years I have collected questions, developed questions, and come up with the ideas for what to include in the Arkansas Poll. We’re an annual poll, so some questions are asked every year, some are just one or two years. As for polling, I arrange for the calls to take place. We used to have a call center on campus, but that died about 8 or 10 years ago when the University closed the center. Now we use an outside center. Then I take all the data which comes in a big spreadsheet of 50-60 questions. We usually poll 800 residents in the state of Arkansas. I tabulate all those results and then relay it to the media, to journalists, students, lawmakers and other policy-wonk types. That mostly happens in October or November, but if a policy comes up during a regular legislative session, we can provide reference to previous polls.”
You’ve described the Arkansas Poll as a “Public Public Opinion Poll.” What is the benefit of having a “public public opinion poll?”
“What it means in short is that we’re totally transparent, about the questions we ask, the answer options to those questions, the responses we get, and the sample we collect. We make it easy for people to see. We show how that sample is different from the average Arkansas adult, who we try to represent, and also whether or not that sample is different from the average Arkansas voter. We don’t pre-screen for voters, but we do determine whether or not someone intends or is likely to vote. Everything is public, and our data is posted on our website. The respondents are still anonymous. We don’t have our thumb on the scales at all, which makes it totally public.”
What kind of trends are you noticing in the state of Arkansas in the 20 years since the Poll began?
“The one change that I think is the most interesting, and the one that generates the most conversation statewide, is this slow, steady move, ideologically and with respect to partisanship, in the Arkansas electorate. It’s always been somewhere between 14 to 16 percent of Arkansans identify as ideologically liberal. For many years, it was about 40 percent of Arkansans who identified themselves as moderates, and then 20ish to 30ish percent who identified themselves as conservative. Ever so slowly, the moderate and conservative proportions have traded positions, so that now Arkansans are more likely to identify themselves as ideologically conservative than they are as moderates, or certainly liberals. In 2010, that started to translate to Arkansas’ partisan identification as well, as an increasing percentage of our respondents started to identify not so much as state-level Democrats who voted for Republicans occasionally or even always at the national level. Increasingly leading up to 2010, our independents, started to say when harder pressed that they “leaned to the right.” Starting in 2010, our election outcomes took a parallel turn with a steep increase in the number of votes being cast for Republicans for almost every position all the way down the line. That’s how Arkansas’ partisan earthquake happened. That has been the biggest thing we watched over time.
What do you think is the cause of that?
The data only suggests that the party shift was due to the ideological shift. What we know from other scholarship is that Americans in general are experiencing a time of heightened partisan polarization. For a long time, Arkansas elections were isolated from national elections because our media markets were different. For that reason, we were still engaged in shaking hands and kissing babies, in what you would call traditional “retail politics.” After money exploded in elections in 2010 and then really exploded in 2012, Arkansas was no longer insulated in that way. Arkansans finally got the memo that their conservative orientation is probably most suited to a Republican vote, even in those state and local level offices.
Have you noticed anything significant in the Arkansas Poll in the wake of the 2016 election?
Our transformation actually came at a different time. We were competitive until 2000, even for a presidential election. We were a battleground state as recently as 2000. We only had a 9-point gap in 2004. In 2008, we went to the double-digit land of the South in terms of percent voting for Democrat, and percent voting Republican. I would say that Arkansas’ real transformation actually came in the elections of 2010, 2012, and 2014. There’s a mix of things happening there. It’s not just the media markets and the explosion of money, but you have to add in the other pieces: Obama, the Tea Party reaction’s to Obama, Obamacare, and the financial meltdown of 2008. The proverbial “perfect storm” of things that were happening nationally had consequences in Arkansas that were only possible on the scale that they happened because we hadn’t been transformed at the same time as our Southern peers. There was no one more ripe, other than perhaps West Virginia, for total, utter transformation.”
2016 saw many polls reevaluating their methods due to some incorrect predictions. You spoke about the stability in Arkansas. What happened to cause those incorrect predictions, and what kept Arkansas’ predicting power consistent?
Our track record is enviable, which I’m very thankful for, but we’ve blown it a couple of times in our history. This year we were a little off on the governor’s race. We projected that Asa Hutchinson would get 59 to 60 percent of the vote, and he got 65 percent. That’s as wide of a margin as we’ve been off in several election cycles. We think that’s due to a higher than average report from our sample of having a college degree which would make our projection show more votes for a liberal candidate. That may have something to do with the number of cell phones in our sample, but that’s just speculation. FiveThirtyEight has given us some shout outs for accuracy though! The thing to keep in mind for any election cycle is that poll predictions are like the weather: we only remember the times that they fail. The national polls were exactly correct in predicting that Hillary Clinton would get a majority of the vote, because she did, by almost the exact margin that was predicted. The modeling for the electoral college was off though. It was really close, and within the margin of error, but those particular states that were wrong were so close to call that it truly could have been one way or the other. No one wants to be wrong, which leads to a phenomenon called “poll convergence,” in which outliers are removed. That’s not good data. It’s scary to be out there by yourself as a poll because you are going to be held accountable, but it’s also scary to think of how outside groups invest in races based on polling. If a candidate is not doing well in a poll, they may lose funding because nobody wants to donate to a losing race. We are deeply uncomfortable with the ethical dilemma of our polls influencing a race, but that’s part of it.”
Does the 2018 Poll suggest anything about the future with regards to Arkansas?
Arkansas is going to be solidly red for the foreseeable future. When some Southern states start to shift back because of their growing Latino populations or pushback from highly-educated white women in the suburbs in states like Virginia, North Carolina, and Texas, Arkansas will be 20 to 30 years behind that, because we don’t have a big urban area. We’re also typically second to last in terms of proportion of college educated residents. This is our foreseeable future – overwhelming Republicanism with a capital R. With that said, we partnered this year with the Northwest Arkansas Council for their Engage Arkansas project to do an oversample in Northwest Arkansas. It showed that while Northwest Arkansas is every bit as Republican as the rest of the state, it tends to be more progressive. This is because its education levels are higher, and/or its income levels, and because of the rapidity of its growth. As the population picks up here, that’s probably where you’ll see the change. It’ll be supplemented by the Democrat-leaning areas in central Arkansas, and maybe some majority African-American counties in the Delta, but certainly on some issues, Northwest Arkansas is more liberal than the rest of the state despite being every bit as Republican. I think that’s really interesting. You see that mainly with issues related to the gay community. Arkansas has transformed dramatically in a short period of time, as has the country, on that issue, but Northwest Arkansas was double digits more progressive on those issues than the statewide average. Even if you’re Republican-leaning, you tend to be more progressive on social and moral issues in Northwest Arkansas.
Speaking toward the future, how are polling methods going to change in the 21stcentury? Obviously cold-calling landlines isn’t going to be sustainable, so what does the modernization of polling look like?
“Increasingly it means computer-aided interviewing and computer-aided polling, and not just with dialing a selection of numbers, but actually conducting the poll. That’s how people do nationwide polls overnight. That can be really useful. However, it also tends to mean that the samples are biased to people who are home, which isn’t a great snapshot of the public. 60 percent of our calls are made to cellphones now, as opposed to landlines. We do five or six callbacks to the same number, because we want to sample a random collection. We don’t want to just sample people who are home. We call at different times of the day, because otherwise you’re not sampling double income families, or people coming back from childcare, or people taking care of a sick relative, or old people, young people, and all people in between who are just busy. The Arkansas Poll is going to continue to use live callers, which means our interviews last anywhere from about 14 minutes to about 45 minutes long. Our response rates are comparable to the rest of the nation, but we don’t have much of a drop-off rate. Once Arkansans are on the phone with us, they’ll keep talking to us! In for a penny, in for a pound! Our call center does tell me that Arkansans are really nice!”
As of this interview, we’re on pace to have the largest class of female lawmakers ever on a national level. You wrote a book called Women’s Rights in the USA. What is the state of women in politics in Arkansas? Has that been reflected in Arkansas?
“The state average percentage of females in state legislatures was expected to go from 25 percent, which is an all-time high, to 38 percent which we’ve never seen. Arkansas isn’t the first, but we’re not the last. Arkansas is just under the national average. Despite the wave of women nationally, we’re going to stay about the same, but that’s not an insignificant trend.”
Lastly, what can Arkansans do to aid in the process of the Arkansas Poll?
“If you hear Arkansas Poll, please answer! We’re non-profit, non-partisan, and the research helps us understand each other, and helps us understand this group project called the republic. To me, that’s worthwhile.”
RELATED: Arkansas Poll Released, Shows Increased Concern for Climate Change