Q&A with UNH Survey Center Director Andy Smith
Few people have their pulse on the people of the Granite State quite like Andy Smith. As director of the UNH Survey Center for over a decade, Smith's job is to determine what people all over the state think about a huge range of issues, from which candidate they're supporting in the presidential primary to what they think of a proposed retail development down the street.
The Ohio native began his polling career while a graduate student at the University of Cincinnati, from which he holds three degrees, including his Ph.D. in political science.
He left his post as director of the Ohio Poll in 1999 to head up the UNH Survey Center, drawn by the magnetic pull of New Hampshire's biennial elections and its significance to the presidential primary.
Employing professional student and non-student interviewers, the Survey Center is best known for its local and national political polling — from its quarterly WMUR Granite State Poll to its presidential primary polling for news outlets such as CNN, Fox News and USA Today — but the brunt of its work actually comes in conducting surveys for students and faculty of the university, the state and local governments, and nonprofits and local businesses in New Hampshire.
Q. What is the Survey Center?
A. We're an academic and public opinion survey research institution. We serve multiple clients. In order of importance, first are the faculty and students of the university. We do quite a bit of research for the university.
The second big audience or client would be state and local governments in New Hampshire. We do a lot of projects with local governments, everything from master plan surveys to gauging the public's willingness to build a new high school or if there's any controversy over a major retail development, we poll to see what kind of public support there is for that. We also do projects for the state of New Hampshire, both survey research projects, and also some unusual ones. We do a survey every year on seatbelt usage. We actually send people out to randomly selected spots and observe how people are or are not using their seatbelts. We also work for a lot of the agencies within state government — the Attorney General's office, office of the Secretary of State, the Supreme Court, Health and Human Services, Education, Department of Transportation — most of the agencies within state government.
And then the third area is primarily media work — the election polling, public polling, and that's the thing we have the most visibility in. It's not the major percentage of our revenues, but it is certainly the most visible.
We run the Granite State Poll, a quarterly survey that we've been running for more than 10 years now. That's a nice survey because it feels the pulse of public opinion in New Hampshire. It's what we call an omnibus survey — if there are organizations interested in statewide surveys, if they don't have the money to do a whole survey themselves, they can buy questions on that Granite State Poll, and that's really helpful for the state government, but for not-for-profits in particular, where they might not have the budget for a whole survey.
Q. Where do you get your funding?
A. We don't get any direct funding or support from the university. We have to generate all of our employees' salaries and interviewers' expenses, so it's like running a small business within the university. We do have a surplus in most years that allows us to mainly help out students to do work — if we have students or young faculty members who don't have funding to do research, we use our surpluses to help them do their work.
Q. Why are polls important?
A. It depends what kind of polls you're doing. If it's work you're doing as an academic researcher, you're looking for input from people on their lives; political research, you're trying to figure out what makes people vote the way they do; sociologists are trying to understand stresses on people's lives. There's a whole host of reasons why people would want to study other people, and surveys are one of the most effective ways of doing it. There are drawbacks of surveys like there are drawbacks for any methodology, but the bang for the buck is they are one of the best ways we can generalize information about people to bring back to the general populations.
Q. What are some of the drawbacks to surveys?
A. One of the drawbacks is that it's somewhat of an artificial situation. It's artificial in that we assume as researchers that the people we're talking to are aware of the issues we're asking about and they have some real interest and opinions on them, which isn't always the case, so we have to be careful not to create what we call pseudo-opinions.
Another issue is what we call social desirability. People want to look good to the people they're talking to on the phone, so we always get people telling us they're going to vote when they actually aren't going to vote. We have to be careful about those sorts of things.
Then there are some practical issues. Because people have call waiting and caller ID, and not having a landline, the response rate has declined. The big national media polls are only getting response rates in the 5 to 10 percent range. We have to work harder to get a hold of people.
That's one of the methodological changes in the past few years — we make more phone calls than we used to. We have trained refusal converters — if you hang up, we'll call back and have somebody who's trained at trying to get you to repeat the survey. There's a number of things like that we've been doing to change our best practices, that the whole industry of public opinion research has been working on.
One of the things we've seen across the industry is something called IVR, or interactive voice response, the robo calls. The response rates on those are atrociously low, in part because it's easy to hang up on a computer but it's harder to hang up on a person.
Q. What media outlets have you done polling for?
A. We've been conducting polls for WMUR here in New Hampshire for going back in the 1970s or the '80s. We've been conducting polls for The Boston Globe for the past seven or eight years. During the presidential primary, we partner with CNN and Fox News to do polls in New Hampshire for the primary. We've also conducted polls in other places, in Philadelphia for the Philadelphia Inquirer, and a TV station in Philadelphia, KYW. We've partnered with USA Today as well, for a number of organizations, but mostly we try to stay in New England. This is the area we know best.
Q. In designing a poll, do you work with the organizations to write questions that optimize responses?
A. Absolutely. The way I try to do it is work with organizations on a decision-making model. What I want them to think about are not the questions themselves, but the research behind it. Sometimes we can find out that the information already exists and actually talk people out of doing surveys and save them the money.
What I then will do is, if you need to do a survey, just tell me the things you're interested in. Don't write them as a survey question — what is the information you need to know? Then I or my staff will turn them into survey questions. Then we pretest those questions with the same population we're going to be interviewing to make sure they mean the same thing to the respondents as they mean to us.
One of the major things we're good at is developing questionnaires that are both nonbiased that also get you results you can use, rather than nice-to-know information. I really try to encourage people when they're developing a survey that they're only asking questions that they specifically have a use for, not just asking questions because it's nice to know the answer.
Q. What are your thoughts on the debate around "unskewed" polling?
A. This comes up every two years, not even every four years anymore — the political campaigns have a standard response to surveys. If it's a survey that shows their candidate doing well, they'll talk it up as much as possible to get positive media attention. If it's a survey that shows their candidate not doing well, they'll look for a reason to disparage that survey.
It's really a matter of shooting the messenger. I've been doing political polling since the '80s, and what I've learned is you tend to aggravate both political parties.
In 2006 in New Hampshire, the Republican Party was furious with us, they thought we were all Democrats over here, because that was a really big year for Democrats. That was just what the results were. In 2010, the Democrats thought we must all be Republicans over here because Republicans were polling so well. We were just reflecting what was going on in the electorate.
We've also seen a proliferation of polls. Now candidates have more polls they can pick and choose between. They are being used more as political weapons than they have in the past.
Q. Do you read Nate Silver's blog? (The New York Times' FiveThirtyEight, which averages various polls)
A. I look at it sometimes. There are some things that are interesting, but I have a real problem with his and others doing averages with taking high-quality polls done with best practices and combining those with those of poor quality. When you mix good stuff with bad stuff, it doesn't give you good stuff — I don't like the idea of mixing crystal-clear spring water with muddy water — you get out a puddle someplace.
Q. How accurate is the Survey Center?
A. It depends on how you calculate it. We've been historically the most accurate at predicting elections in New Hampshire, both in terms of picking winners and getting the margin of victory right.
In presidential primaries, it's a different sort of a ballgame where we are as accurate or more accurate. In 2008, we got the Obama/Clinton race wrong too, but nobody got that race right. There are things that make those kind of races tough to call, when a third of the electorate was making up their minds on Election Day long after we've quit polling. There are some challenges with primaries that make them difficult.
Q. What are some other challenges in primary polling?
A. The biggest reason is you don't have the party questions involved. In general elections, if you're a Republican, 90 percent of the time you'll vote for a Republican candidate, and if you're a Democrat, 90 percent of the time you'll vote for a Democratic candidate. We can usually tell how people are going to vote.
But in a presidential primary, all the candidates are from the same party. You're making the decision on ephemeral things: Do you like his personality? Are they attractive? Are they electable? And all those things can be impacted at the last minute by the campaign.
Presidential primaries are fought going into the polling booth. It's like taking kids to an ice cream shop. You don't leave the house saying I'm going to get a rocky road ice cream, you wait until you get there and you see what's on sale, you read the sign, you see what someone else is walking out with, you see that guy's got butter pecan. But no matter what happens, you'll be happy with whatever ice cream you get. No matter who wins your primary, it's going to be your party's nominee. The amount of thought and effort people put into a presidential primary — I've literally seen people at polling places flipping coins.
Q. What's happening with this year's New Hampshire gubernatorial race?
A. It's very close. A significant percentage of the voters are undecided because they don't know who the candidates are. Our most recent poll had Hassan up by two, my newer one on that race is it's really close.
I think that that race as well as others will be impacted by what happens at the presidential level. If the Obama or Romney campaigns nosedive in New Hampshire, that's going to hurt their party's tickets.
Q. Were you surprised at your poll that showed Obama up by 15 percent over Romney in New Hampshire?
A. I was surprised at the margin, but not terribly surprised. First off, you look at the internals there. It points out some serious problems for Romney. He hasn't locked down the Republican base like he needs to — he needs to get 90 percent of them saying they'll vote for him. Independents are breaking toward Obama by about a 2-1 margin, where they were split about two weeks ago. More problematic is the enthusiasm they seemed to have a lead in that is gone.
Q. How have you witnessed New Hampshire change politically since you moved here in 1999?
A. The state has been undergoing a slow-going political realignment where the entire electorate shifts over time. This has been happening not just in New Hampshire, but all across New England for 50 years or so.
The region was once the most Republican in the country, but starting in the '30s it began shifting to Democrat. There was Rhode Island in the '30s, Massachusetts in the '50 and '60s, Connecticut in the '70s, Vermont in the '80s, Maine in the '90s. Well, New Hampshire is the last of those Republican dominoes to fall.
The other thing about New Hampshire is it's the only state in the Northeast that's had significant population growth in the last few decades. Those people moving in are far less Republican, and those older Republicans are doing one of two things — either dying or moving south. Older, more solidly Republican voters are being replaced with younger, much-more-likely-to-be-Democratic voters.
When I moved here it was slightly Republican, to now I'd say it's about a 5 percent plurality for Democrats in the state. We're headed to be a solidly Democrat state, and in another 10 years we will be a solidly Democrat state.
Q. So our status as a swing state is in jeopardy?
A. It seems to be slipping some already. Obama won here by almost 10 points in 2008. It's looking like now he's going to win here again. We'll see what happens, but I hope we don't, because if we lose our swing-state status, it won't be much fun in the November elections.
Q. What are some of the biggest issues Granite Staters are worried about now?
A. Economy and jobs. It's by far the most important issue, and it has been for three to four years since the end of 2007 and into 2008. People's concern about the economy and jobs blows everything else away. Ten years ago, education funding was the most important issue — that's dropped off the table.
Q. Anything else you'd like to add?
A. One of the things to remember, when people think about polls, is that pollsters are far more cautious with the interpretations they make and conclusions they draw than the general public.
It's important when looking at polls is for the general public to understand the methodology and it's important that they investigate the methodology — the questions asked, how data were collected, those sorts of things. The Internet makes that information available. If you can't find it, that should make the consumer more dubious about whether or not to believe it.