Of all the epithets that one can throw at a politician, “poll-driven” and “focus-group tested” might be the meanest. It’s a little more polite than calling him a flip-flopping, finger-in-the-wind opportunist, but it amounts to the same thing. Why, we voters plead, can’t candidates just say what they believe rather than what they think we want to hear?

Count on Mark Mellman ’78 to defend his profession. Mellman is a pollster and focus-group user, one of the most successful political consultants in the business, and he suggests, only half in jest, that whoever first articulated that line of political attack probably relied on polls and focus groups to determine that it worked. Kidding aside, he thinks the tricks of his trade are not tricks at all, but tools for helping candidates on a limited budget reach voters with a limited attention span. 

“Our role is to say, who is this candidate and what do they believe?” he explains. “They are a lot of things, and they believe a lot of things. Our job is to figure out which are the one or two things that, if we repeat them over and over again, are most likely to get them the most votes. Our job is not to create candidates but to take candidates who are multi-faceted and narrow their focus.” 

More to the point, he adds, pollsters “are reflecting what people are telling us. And I think it’s valuable for politicians to know what people really think.”

Mellman is president and CEO of the Washington, D.C.-based Mellman Group, which offers its clients “research-based strategy.” When it comes to electoral politics, Mellman works only for Democrats, unions, and liberal interest groups, though he has plenty of counterparts across the aisle. (The firm also does market research for corporate clients.) “Partisan” is another dirty word in politics, but Mellman defends that, too. 

“People are in our business because we believe in something,” he insists. “I believe in the ideals and ideas of the Democratic Party, and there are people who are my friends who believe just as strongly in the ideals of the Republican Party. The way in which those agendas get enacted or not in our system depends on who gets elected. So our goal is to help people get elected because we care about where this country goes.”

Over his 30 years in the business, Mellman’s record for getting people elected is very good. His client list includes 21 past or present senators, 26 House members, eight governors, and dozens of other state and local officials. In his most recent race this fall he helped Scott Stringer thwart Elliot Spitzer ’81’s comeback bid in the Democratic primary for New York City comptroller. Alumni solidarity, if there is such a thing, ends at the ballot box. 

Mellman has received a lot of credit for the Democrats holding their Senate majority in 2010 (when he helped Harry Reid to a come-from-behind victory in Nevada) and 2012 (when he helped Heidi Heitkamp pull off one of the biggest upsets of the night in North Dakota). A current client, Alison Lundergan Grimes, is challenging GOP Minority Leader Mitch McConnell in Kentucky, making it likely that Mellman will have a lot to say about whether Democrats hold the Senate in 2014, as well.

Statistician and writer Nate Silver cited Mellman, along with CNN and Grove Insight, as the most accurate in predicting last year’s presidential race. Last November, the Doonesbury comic strip called him one of the “princes of polling” as it poked fun at a “gang of angry low-information voters.” The American Association of Political Consultants recently named Mellman Pollster of the Year, for the third time. The AAPC also honored Mellman in 2013 for best international campaign (in Israel) and best campaign for a ballot initiative (to expand casino gambling in Maryland, which passed). 

As those awards suggest, Mellman branches far beyond the marquee races. He increasingly takes on work overseas. Mellman has worked on legislative campaigns in more than half a dozen countries, including Russia, Ukraine, Spain, and Uruguay, and counts the mayor of Tirana, Albania, and the presidents of Colombia and Costa Rica as satisfied clients. Last January, he helped the Yesh Atid party gain 19 seats in the Israeli parliamentary elections and become the second-largest party in the Knesset by focusing on issues of concern to the middle class. One Israeli website described him as “the kingmaker’s kingmaker.”

Mark Mellman ’78 sometimes works while walking on a treadmill. A colleague explains: “Mark can’t always get to the gym, so we brought the gym to him.”
Mark Mellman ’78 sometimes works while walking on a treadmill. A colleague explains: “Mark can’t always get to the gym, so we brought the gym to him.”
Jeffrey MacMillan p’14

In person, Mellman might be described as rumpled and avuncular. For exercise, he sometimes works at a high desk while walking on a treadmill. He’s a hard man to catch up with, as likely to be off meeting with potential new clients in places like the Philippines as in the office. As an Orthodox Jew, he does not work on the Sabbath, even in the hectic final weeks of a campaign. His clients learn to work around that schedule. 

Odd-numbered years are called off-years in American politics, but there is little downtime for pollsters, who are already gearing up for midterm congressional races. Each race and election cycle is different, so Mellman begins work, even for repeat customers, by reading as much information as possible and talking at length to the candidate and local party officials. Next, Mellman might convene a series of focus groups so he can bore into the minds of a small group of representative voters. He might, for example, show them video of a candidate with no sound and ask for their impressions. Although participants often complain that they can’t form an opinion without hearing the words, they quickly develop elaborate beliefs based on nothing more than posture and body language. And Mellman, who still does much of this work himself, notes them carefully.

Focus-group sessions help Mellman develop broader opinion surveys in which he can test as many as 20 or 30 variations on a theme or message. Polling is an art as well as a science; the order in which questions are asked and the way they are phrased can influence the results enormously. To give one example, Mellman says that people are much more likely to agree if asked whether something “should not be allowed” than if they are asked whether it should be “prohibited.” 

In the same way, candidates can look very different with only small changes in presentation. Do voters in a particular race, for example, react more positively to someone who comes across as a strong leader or as a competent fiscal manager? No candidate is a blank slate; each has a voting record to defend or previous statements to explain, so there are limits to what Mellman can — or wants to — do. Polls and focus groups help determine which attributes to trumpet and which to explain.

Another Mellman tenet, though, is that voters often don’t know what they think or — more to the point — they don’t always think what they think they think. They can be very poor reporters about what is important to them and why, which means that simply asking them questions and tabulating the responses is not enough. Mellman applies various regression analyses and mathematical models to test the strengths of voter opinions. 

“Part of our job,” he explains, “is to be able to say to politicians, you may think these are the concerns of voters, but we really understand exactly what those concerns are and sometimes they are quite different [from what you think they are].”

Unlike many pollsters, who only survey likely voters, Mellman tries to construct a model of what he calls the likely electorate, which can be quite different. Some who describe themselves as likely voters stay home, while others, who are considered unlikely voters, end up voting. Mellman believes his analytics can tell which are which.

All this information tells campaigns how to spend their money and their time. Although Mellman does not write the ads they build from the data he gives them, he works closely with their media teams and continues to advise the campaigns, often doing follow-up media testing, through Election Day. He doesn’t talk about how much all this costs, but says that the sort of work he does generally accounts for between 3 and 10 percent of a campaign’s budget, much less than the estimated 70 percent it spends buying TV airtime.

In some cases, Mellman continues to advise clients after their elections, and his most useful service can be telling them what not to do. Jennifer Granholm recalls that when she was governor of Michigan, she wanted to create a fund to promote clean-technology jobs in the state. Certain that the Republican-dominated legislature never would approve the idea, she planned to go over legislators’ heads and place a funding initiative on the ballot. Mellman’s polling persuaded Granholm that there was little public support for the proposal. Rather than fight a losing battle, Granholm scrapped the initiative and funded her plan by tapping the state’s tobacco settlement instead.

Analysis rooted in academic research, higher mathematics, and even psychology is what set Mellman apart when he first entered the business. A self-described political junkie who passed out leaflets for Democratic candidates when he was still a high school student in Columbus, Ohio, Mellman joined Princeton’s Committee for the Impeachment of the President soon after he arrived on campus in 1973 (he entered with the Class of 1977) and traveled to Washington to line up press coverage of students lobbying for Nixon’s removal from office. He took on his first client in 1982 when he was a graduate student in politics at Yale and an unknown congressional candidate, Bruce Morrison, came looking for help in an uphill race. A Yale professor Morrison had hoped to retain was unavailable, but the professor referred him to Mellman.

Mellman was called a “prince of polling” in G.B. Trudeau’s “Doonesbury” comic strip Nov. 23, 2012.
Mellman was called a “prince of polling” in G.B. Trudeau’s “Doonesbury” comic strip Nov. 23, 2012.
DOONESBURY (c) 2012 G.B. Trudeau. Used by permission of Universal Uclick. All rights reserved.

Applying some of the data analytics he was learning in class, Mellman determined that Social Security was the issue voters in the district cared about most, and persuaded Morrison to devote his entire media budget to hammering that issue. Although Congressional Quarterly had not listed Morrison’s race as competitive, he won narrowly. Mellman has been busy ever since.

Campaign consulting has changed considerably since 1982, when fax machines were little used and data had to be entered on punch cards. Technology has compressed a pollster’s schedule, which used to be governed by the Federal Express pickup deadline. Now, Mellman says, “if someone calls you at 8 o’clock, they expect [data] at 8:01.”

Cellphones and voice mail have made voters much harder to reach and polls more expensive — and more tricky —to conduct. Mellman believes that the reliance on so-called robo calls, which go only to landlines, explains why so many Nevada polls predicted that Harry Reid would go down to defeat in 2010, while his own poll showed Reid ahead.

Still, not every candidate wins. In 2004, Mellman was the lead pollster for John Kerry’s unsuccessful presidential campaign. Reviewing the campaign in December 2004 in a discussion at Harvard, Mellman explained the loss this way: “The reality is the country was not feeling the level of pain that was required to oust an incumbent president. ... As an incumbent president in times that weren’t that bad, they [the campaign of President Bush] had some real advantages and they used those advantages extremely well.” Two years earlier, he was criticized in left-leaning publications for advising Democratic congressional candidates to sidestep the war on terror, which didn’t poll well for the party, and focus their messages on education and expanding prescription-drug benefits — a failed strategy, according to the critics. Mellman still insists that his polling was accurate and that those clients who followed his advice did quite well, even if the party as a whole underperformed.

In any campaign, the question of whether or not to go negative is a tricky one: One person’s negative ad can be another person’s truth telling. In general, Mellman believes that going negative is both effective and proper. Because negative ads often refer to specific votes or positions the other candidate has taken, he says, they usually are more rigorously fact-checked than positive ads. Because voters process negative information more quickly and retain it longer than positive information, negativity also gives candidates a bigger bang for their advertising buck. “That’s a psychological fact of life, and until someone abolishes it, we’re going to have negative campaigns,” Mellman says. Each campaign, though, is unique. In 2010, he advised Illinois Gov. Pat Quinn, who was running for his first full term, that he needed a positive message to introduce himself to independents while also attacking his Republican opponent. (Quinn won.)

Mellman believes that the Supreme Court’s 2011 decision in the Citizens United case, which enabled the emergence of independent super PACs, may have had the unintended consequence of increasing the number of negative ads. Because those super PACs cannot coordinate their message with the campaign or use the candidate’s image, he believes it is easier for them simply to attack the other side, whether the candidate wants them to or not. 

Mellman has confessed that he would like “a direct pipeline into the brain of each voter,” something social media soon may offer. Last year, he partnered with Twitter, the search engine Topsy, and Republican pollster Jon McHenry to form the so-called Twitter Index, or Twindex, which assessed more than 200 million Twitter feeds each day during the presidential campaign. Though Twindex is still somewhat experimental, Mellman believes it will offer increasingly sophisticated ways of assessing what people are talking about and how strongly they feel. 

Does the pollster/consultant class help our democracy, or is it another pox on our politics? Politicians certainly think pollsters are essential. “For a politician to eschew careful polling is like a soldier favoring a blunderbuss over a scoped M4,” former Rep. Jim Marshall ’72, a Vietnam vet and ex-Mellman client, writes in an email. “[I]t is also a tired truism that the majority of voters haven’t the time, inclination, and/or capacity to sort out politics. So polling arguably provides a public service by focusing campaign advertising in competitive races on a few issues most important to the voting public, lessening the likelihood of election outcomes determined merely by a strong chin or big checkbook.” 

Nolan McCarty, a Princeton professor of politics and public affairs, describes the work Mellman and other pollsters do as “marketing, in the best sense of the word.” Although some might denounce it as manipulative, McCarty, like Marshall, thinks it helps candidates understand what voters think so they can present their best arguments. 

A year away from the midterm elections, Mellman is cautiously optimistic about the Democrats’ chances of holding the Senate, less so about their chances of retaking the House. He attributes Republican strength in the lower chamber less to gerrymandering and more to the fact that Democratic strength is concentrated in relatively few areas, mostly cities, making it easier for Republicans to draw congressional districts that lean toward the GOP.

After President Barack Obama’s re-election, some suggested that superior data analysis and strength with growing demographic groups gave the Democrats a built-in edge, but Mellman warns them against complacency. “We do have certain advantages going forward,” he says, “but demography is no longer destiny.” Independents remain up for grabs, and even voters who have turned out reliably for one party can still be persuaded to change sides or stay home if they don’t like what they hear. If they couldn’t, Heidi Heitkamp could not have been elected in deep-red North Dakota or Republican Scott Brown in liberal Massachusetts. The right poll-driven, focus-group-tested message can make a difference — but even Mellman concedes there is only so much he or anyone can do to turn a sow’s ear into a senator.

“By and large,” he says, “if you don’t have a good candidate, you have a really, really hard job.” 

Mark F. Bernstein ’83 is PAW’s senior writer.