Behind the Numbers
An interview with pollster Brian Sanderoff
When New Mexicans want to know about the numbers behind local elections—or any numbers, really—they turn to Brian Sanderoff. You’ve probably read his name at one time or another in a newspaper or on a blog, or heard it on the radio. Sanderoff’s Research & Polling, Inc. conducts the vast majority of election polls throughout the state, most of which grace the pages of the Albuquerque Journal.
Sanderoff’s been in the political scene in New Mexico for nearly 30 years. He started out as a graduate student in political science who, in order to get credit for a class, joined former Gov. Bruce King’s campaign. When King won, Sanderoff worked his way up the ranks of local government and ran four state agencies before the age of 30 as a troubleshooter. His car was bombed by corrupt ex-employees he’d fired. He witnessed the prison riots of 1980 firsthand.
Eventually, after another few years in government service, Sanderoff left his job to start a polling company, Sanderoff and Associates, which later morphed into Research & Polling, where he’s been for 20 years. Known as the most reliable name in the business, when Sanderoff spouts a number, people listen. His polls almost always predict the outcomes of elections—in fact, he’s only been off twice … in more than 20 years.
Last week, Sanderoff took some time to chat with the Alibi over the phone about the upcoming elections, his polls and what it all really means.
How do you get the numbers you report?
Well, we're talking about political polls here, or election polls. We do a bunch of other kinds of polls, too, but speaking of the election polls, we start out with a sample. We have a list of all the registered voters in the state, and we also know all registered voters’ history. In other words, we know if a person is a fair-weather voter, or always voter, or just votes in school board elections, or just votes in presidential election years. We know whether they are registered and what their history is for voting in different elections.
The success that Research & Polling has had over the years in predicting the outcome of elections is partially due to how we generate our sample. Many national polling companies get a random sample of households and ask people, “Well, are you registered?” and then they ask them, “Well, how likely are you to vote?” and then they call that likelihood. Well, there is something in polling called a social stigma bias. We plan on going to church, we plan on starting a diet, we plan on training the dog--there are a lot of things we plan on doing and don't get around to. So what we do is generate a sample of known voters, and that allows for our accuracy to be greater than asking people to self-report whether they are registered and whether they are going to vote.
How many people do you usually poll? Does it depend on what you’re polling for?
It depends, and as we get closer to the election, we increase the sample size to make it more accurate. A year before the election, does it really matter if you are up by a point or two here or there? When you get closer to the election, our sample size increases, and this last [Albuquerque] Journal poll, which will hit the Sunday before the election, will be an 800 statewide sample.
How do you get a wide demographic range?
For each county, we set quotas so each county will get its right mix of Democrats and Republicans. Democrats tend to be undercounted if you leave it to pure randomness, because Democrats also correlate with income and mobility and things like that. So we set quotas to get the right mix of Democrats and Republicans not at the state level, but at the county level. We also set minimum quotas so we don't undercount Hispanics, so we don't overcount women, and so we don't overcount seniors. Or you can put it another way and say we set quotas for Hispanics, ethnicity, gender and age. Bottom line is women are more likely to answer the phone than men, seniors are more likely to be home than young people. So we set quotas to make sure we don't overcount certain groups.
How representative are polls in determining the outcome of an election?
If you generate a random sample, and if the sample is representative of the defined universe—in this case, the voting population—then the results can be generalized to the larger population. In general elections, we have a 97 percent accuracy rate. Basically, [in] 59 out of 61 races we polled for the Journal in the last 20 years, we had the right person on top. The first time we blew it, it was only by 366 votes that Gore beat Bush in 2000. We had Bush up by one point and he lost by .06 of a percent. So thanks to a snowstorm on the Eastside, or a lot of stay-at-homes, we lost our perfect number. So I blame the weather (laughs). In this case, we were still in the right margin of error.
In polling, do you try to determine where people get their information from? For instance, does the Christian Right tend to get a lot of their information from church? Or do liberals get their information from a primary source?
That kind of information is sought out by the pollsters who do the work for the campaigns, because they're the political consultants who want to help the campaigns win. Since we do the polls for the media, our primary objective is not to find out where they getting their information, but how they're going to vote so that we can predict the outcome. But what you're suggesting does happen, it just happens at a different type of poll.
We'll ask age, income level, education level, ethnicity, residency, gender, and we'll cross tabulate and cross reference the results of all the questions against the demographics, and we'll be able to say, “Did you notice that Heather Wilson is slipping among the Hispanics? [In] the first Journal poll she had 35 percent with Hispanics then 29 percent in the next poll, and now she's down to 20 percent, and that explains why she's slipping.” We'll look for explanations for why candidates may be on the rise or may be on the decline by looking at demographics.
In the last poll I saw, it had Madrid ahead of Wilson by three points. Do you think there's been much change since the other night’s debate? How close do you think the race is?
It's hard to say. We had the debates the other night, and everyone's spinning them in their way. I have a feeling about how many people watched the debate—it was sandwiched between “Dancing with the Stars” and the World Series (laughs). Most people get their information about what happened at the debates from what we call public opinion leaders, so most people hear about the debate filtered through the eyes of others.
In my opinion, Heather Wilson was a more skilled debater and came off as more at ease. And whether that will help her or not will remain to be seen at the next poll. The way that Heather came out of that debate, guns a-blazing, tells me she knew she had to be tough. You don't act that way if you're up 10 points. The trend of the public polls from August to mid-October has been in Madrid's favor. But that's why it's called a horse race—the only thing that matters is what happens at the finish line. So we're seven furlongs in and Madrid pulled ahead, but they haven't hit the finish line to see if there's going to be some surge at the end.
There's a lot of history here. This seat was first created in 1968, roughly, and for nearly 40 years this seat has been 100 percent of the time in Republican hands. There have only been three incumbents during that whole time: Manuel Lujan, Steve Schiff and Heather Wilson. If you look at most of the other races in Bernalillo County—and Bernalillo County is 95 percent of the Congressional district—Democrats have a tremendous edge and win most of the races in the county. But for whatever reason this seat has always gone Republican.
I read that you said Wilson has never been tied or behind in a Journal poll before.
Never been behind. Never had a Democrat ahead. In the 20 years I've been polling this district. So this is historic.
What makes it different this year?
What makes it different this year is the same thing that is making it different in 35 other Congressional districts around the nation. The 32 most competitive House seats going on right now in the nation all have Republican incumbents. Think about that. And of the top 50 of those competitive seats, 45 have Republican incumbents. So what we're seeing is larger than two local personalities running for office.
If everything were focused on local issues, then you would think that for every Republican in trouble, there would be a Democrat in trouble around the nation. But the fact that of the top 32 competitive Congressional seats, all of them have Republican incumbents, tells you that old historical phenomenon—the president's party tends to lose seats during the midterm elections when the job approval is low. In 1994, Clinton's first term, Democrats lost over 50 seats, and the Republicans took over Congress. Clinton was seen as too liberal at the time. Ironically, the Democrats didn't lose seats midterm during his second term, during the impeachment. They thought the impeachment was overblown. In his first term, the Democrats got their butts kicked, but now we're seeing the shoe on the other foot. The Republicans are on the defense on these Congressional seats; same thing on the Senate seats but not quite as dramatic.
What's the president's approval rating in Albuquerque?
It's 38 percent in Albuquerque.
How does that compare to the nation?
It's 37 percent nationwide, and they always mirror each other.
Do you think the Iraq war is affecting the elections this year?
Absolutely. If it weren't for the Iraq war and the president's 38 percent approval rating, Heather Wilson would be leading this race comfortably just like she did two years ago and four years ago. She won by nine points two years ago and 10 points the election before that. It got close there for a while, and the issues were the same. But now we're two years deeper into the issues of Iraq, two years deeper into the president's approval rating.
What the Democrats are trying to do is to nationalize the race. They're trying to take Heather Wilson and nationalize her and tie her to the Republican leadership in Washington, the White House and Congress. Tie her to the president, tie her to Iraq. Those are all national issues.
The Republicans are trying to localize the issues. Trying to tie Patricia Madrid to being soft on public corruption, looking the other way on the Robert Vigil stuff. Trying to tie Patricia Madrid to ethical concerns regarding campaign contributions she received from people. The Republicans are localizing it, trying to make it between Patricia Madrid and Heather Wilson. Two people, two different personalities. Looking for public corruption. Looking for ethics. The Democrats are trying to nationalize it. “Look at how Heather Wilson is buddies with the president, and the president is raising money for her. Look how she supports the president on ‘stay the course’ on Iraq.” One's going national, and the other local. That's happening in all of these 30 some odd districts nationwide.
Do you think the Democrats have a chance of taking back the House?
Yes. For the Senate, it looks tougher, but for the House, it looks like they just have to pick up 15 seats and, as I said, there are over 30 that are in play.
Do you know how many New Mexicans support the war?
In my last poll, 52 percent said the war was a mistake. That was two weeks ago.
Do you know how many are Republicans and Democrats?
Yes. We asked, “Looking back, do you think that the United States did the right thing in taking military action against Iraq, or should the U.S. have stayed out of Iraq?” Forty-two percent said “did the right thing,” 52 percent said “should have stayed out.” Among Democrats, 18 percent “did the right thing,” 73 percent “should have stayed out.” Among Republicans, 78 percent “did the right thing,” 17 percent “should have stayed out.” So they're opposites. So party affiliation is a big predictor of attitudes toward Iraq.
How many people do you think will turn out to vote for this election?
Turnout is a lot harder to predict than how people are going to vote. But, quite frankly, I don't see a reason for high turnout. We've got a boring governor's race; we've got a boring U.S. Senate race. We have an interesting CD1 race that we're talking about now, but in the other two Congressional districts in the north and the south, they're boring. So what do we have?
We've got an interesting Land Commissioner race, between you and me. We have a great Congressional race here in Albuquerque that only represents 1/3 of the state. Then we have these stupid paper ballots that I've been reading about in the paper. There are long lines for early voting—not because there are long lines for voting, but because it just takes a long time to fill out the ballot, which might scare some people away. So what do I think? When I start to throw out percentages, then those are the percentages used for the next two weeks around the state. It makes me nervous, but usually I have pretty good instincts on that. It was 75 percent last time in the presidential year; I'll be happy if it's 50 percent this year.
What’s the trend in voter turnout?
The trend in voter turnout over the last few decades is downward, percentage-wise, with the exception of ’04. ’04 was record highs, but everything was leading to record highs: Bush versus Kerry, different philosophies and styles, we’re at war, tons of money was being spent by 527 organizations trying to get out the vote. If we didn’t have high turnout in ’04, we’d never have it. But we did have it, it was an exception; we’re going to go back now to lower turnouts.
Millions of dollars are spent from both sides in negative campaign ads every election cycle. How do you see that working out in terms of benefiting one side versus another?
Ultimately, the Madrid and Wilson negative ads will cancel each other out. They have basically made both of these people have low approval ratings and low honesty ratings. People’s images are tarnished by being the target of the ad, but the voters are also mad at the perpetrators of the ad.
I really like this campaign reform law that passed that makes the candidates at the end say, “I approve this message. I authorized this ad. I paid for it.” I think that’s really good, because that way the voters can see which ads are the candidate-based ads and which are ads that are being paid for by third-party organizations. Either way, they’re pathetic; the ads all have a grain of truth and a pound of lies. And it’s sad, and the voters aren’t sure if they’re truthful or not, but because they’ve been going on since July, they’re sick and tired of them.
But they have their effect. You know, they neutralize each other or cancel each other out in a race like Wilson versus Madrid, but look at a low-profile race like Attorney General: Gary King, partially known guy—his daddy was governor three times, his legislature lot—runs against Jim Bibb, who’s a good-looking young man and Republican, has no experience in politics. And so Gary’s going to have a 30-point lead because people know his name and he’s likable and there’s no reason to vote against him and they’ve never heard of his opponent. And his opponent spends a quarter-million dollars running some negative ads and narrows the gap from 30 points to probably 15 points in three weeks. I mean, the power of negative ads is amazing. So now Gary King has to go out there and say, “Bibb’s a fibber,” and try to build his lead back up again. And so people say they hate the ads but sometimes they can be effective.
Do you think they ever keep people from voting? That they just get so turned of that they don’t vote?
Well, in the Journal polls we’re consistently getting four percent in the Congressional race who are saying, “Yeah, I’m voting, but I’m not voting for either of the Congressional candidates.” Four percent are saying they’re not gonna vote for either one of them, then you’ve got the people who don’t vote at all, whether they’re using the negative ads as an excuse—“I’m not gonna vote because I hate all politicians and I hate the negative ads”—or whether they sincerely are staying home as a result of the negative ads. It’s always hard to know—probably a combination of the two.
You mentioned earlier that in a overwhelmingly Democratic area, Heather Wilson has and other Republicans have always maintained that seat. Why do you think that is?
Well, as I said, she won her last two elections by very comfortable margins. I think she might not be in everyone’s hearts, but she’s in their minds. You know, when you’re in the House, you run for election every one of two years. That means one out of every two years you’re being called a dog by your opponent. And so imagine having to run for Congress, and every blink of the eye there’s another negative campaign running against you for six months. So she’s subjected to that. Whereas, if you’re a U.S. Senator you’ve got a six-year term—you at least have a five-year breather before they pound on you again.
Heather Wilson, many people see her as honest, she has impressive credentials in the area of intelligence and she’s a female veteran, so she’s got an impressive résumé. Some people see her as aloof, too cold or stilted, not quite warm, but no one’s ever accused her of not being smart. But every two years there’s someone else pounding at her, as is the case with any person who wants to subject themselves to being a congressperson. So it’s a tough place to campaign from; that’s why I’m sure she wouldn’t mind having Pete Domenici’s Senate seat one day.
You think she’s going to run for that?
Sure. But now Pete’s thinking about running again. He’s only been there 32 years; I guess that’s not long enough.
Aside from the Wilson-Madrid race, what do you think is one of the more interesting races in the state right now that people aren’t talking about?
Oh, the Land Commission race. You know, most people don’t know much about the Commissioner of Public Lands, but because we have two well-known people who are vying for that seat right now—Jim Baca, the former land commissioner, former mayor, and Pat Lyons, the current land commissioner—because they’re well-known personalities, this race has never received as much visibility as it does this go-around. Plus, lines are drawn with their philosophies and their style.
Pat Lyons is sort of a go-along, get-along, friendly kind of [guy]. You know, customer-oriented, works with the gas and oil industry, very business-like. And Jim Baca’s more of an environmentalist, doesn’t hesitate to take a controversial stand or come up with ideas for constitutional amendments to change the way things are or reform things. So you’ve got a status-quo kind of guy who’s just proud of being a good manager against a guy who’s willing to shake things up and be a reformer and environmentalist. So you’ve got two really differing philosophies. Pat Lyons has a lot more money, but Jim Baca has narrowed the gap a little there, so they’re on TV and radio and newspaper, and that’ll be an interesting race to watch.
Where are they in the polls right now?
In my three polls Baca has gone from up five to up four to up two. And so you see where the trend’s going.
A reader wrote in to us recently wondering about the polling process, and they had heard that the Albuquerque Journal owners also own Research & Polling and wanted to know how there’s a separation between the interests of the companies. So I figured now would probably be a good time to ask …
That’s a good question. Basically, the shares of Research & Polling are owned by myself and the person who owns the company that owns the newspaper—so, basically, Tom Lang, the publisher of the paper. My relationship with the newspaper is as a client. In other words, my business relationship is with the person who owns the company that owns the newspaper, and my relationship with the newspaper itself is as a client. I charge them fair market prices, I haggle with them over the questions, I have no undue influence over them and they have no undue influence over me; we keep a very independent, hands-off relationship.
I don’t want people to think they can come to us to get a poll and get a front-page story in the newspaper, nor do I want people to think that the newspaper somehow could influence us. It’s totally hands-off. The people over there don’t even know who our other clients are. We have an agreement of total independence. Basically, 98 percent of our work has nothing to do with politics. We do patient satisfaction and we do corporate image work, and the politics is the fun part. It’s where my area of expertise is, but most of our work is done for the Intels of the world or the Presbyterian Health Care Services. Those are who our clients are. The Journal is about one-half of one percent of our business. But, you know, they’ll do a poll and drag it out for a week, so it might seem like a lot, but it’s actually less than one percent of our business.
Because you’ve covered so many elections in New Mexico, what’s the most exciting state election you remember?
Well, the most exciting one was the 2000 presidential race between Gore and Bush, where Gore won by 366 votes out of 570,000. I mean, you can’t get any more exciting than that. So in terms of a race being close and exciting, we didn’t know who won that one for a week after the election. In terms of earth-shaking, in 1982, Jeff Bingaman knocked off a Republican Senatorial incumbent; he knocked off Harrison Schmitt, a Republican who had walked on the moon. It was the only Republican U.S. Senate candidate who lost in the nation. So Jeff Bingaman, a Democrat, knocked off a Republican incumbent for U.S. Senate in 1982 and Toney Anaya, a real long-shot, became the governor. And so that was an exciting year as well.
You probably won’t be able to answer this, but who are you voting for?
Plenty of people ask me that question and what I tell people jokingly, and it gets me a good laugh during my speeches, is I vote for whoever’s ahead in my final Journal poll so the results will be correct (laughs). Although it may not always be true, it always gets a laugh.
I have my views like anybody else, and I go to great lengths to keep them to myself, and I never let them get in the way of my polls because, ultimately, what drives me is my selfishness to be as accurate as I can be and to have people recognize they can count on us and rely on us. But there have been times when I could have gone either way and I will vote for whoever’s ahead in my poll (laughs). There are a few times when I can’t stomach myself to do that, though. So that’s what I say to get laughs at the speeches, and there’s only marginal truth to it.