The Victory Lab (8 page)

Barabba imagined it should be possible to augment the precinct-level political histories with demographic profiles of voters derived from U.S. Census data about race, ethnicity, age, and family type. It was good for a Republican candidate to know that a neighborhood had broken in favor of his party in the last four statewide elections, Barabba thought, but that information became all the more useful if coupled with the insight that its bungalows were packed with married white retirees. Barabba went back to UCLA and proposed such an analysis as his dissertation for a business degree, but professors rejected it because it did not pursue a “new theoretical approach.” So he quit school and reached out to an old boss, Stu Spencer, who had directed Rockefeller’s California campaign. Together they started a business called Datamatics, and took Ronald Reagan’s 1966 gubernatorial campaign as one of their first clients.

Barabba was fascinated by the work of two California sociologists who had developed a model they called Social Area Analysis to profile the socioeconomic character of the state’s urban neighborhoods. Based on 1960 Census data, they categorized tracts along two axes, one reflecting class and the other family composition. Barabba wanted to map these onto precincts, where he could merge them with political information. He hired a programmer to write a computer script and rented time at a local insurance company’s IBM 1401 when the computer wasn’t in use. Late at night, the two would feed it punch cards that merged these two levels of data to create what Datamatics marketed as the Precinct Index Priority System
(PIPS). In 1965, Barabba’s numbers helped two candidates get elected to the Los Angeles City Council, and Datamatics looked for new clients.

The next year, Barabba was summoned to Flint, Michigan, to meet a Harvard Business School student with big plans. Don Riegle had worked at IBM as an analyst before heading to Cambridge and was now intent on returning to his hometown to run for Congress there as a Republican. Barabba met Riegle in the lobby of the Durant Hotel, in a downtown still vibrant with General Motors’ Buick City, and detailed how PIPS worked. The district was a traditionally Democratic one, and Flint’s rich union tradition usually scared off Republicans. “There were enough Republicans to get you close, but if you got every Republican out there you could never win,” says Barabba. “So you had to get swing voters.”

Barabba thought Riegle’s campaign would be a perfect test case for a demographically driven approach to political organizing. The scale of the challenge was clear: the first poll, commissioned in May, showed incumbent John C. Mackie leading Riegle, 63 percent to 26 percent. Riegle’s advisers believed that their candidate, a smart young moderate technocrat, was the campaign’s greatest asset and that the rising specter of inflation was a boon to any challenger. Barabba and Spencer had bought thirty minutes of television time on a Sunday afternoon in an effort to have him directly address local elites. At one point during the broadcast, Riegle dramatically took a cleaver to a piece of steak and threw one-third of it onto the trash—a demonstration, he said, of what inflation did to workers’ buying power.

The performance helped establish Riegle as a serious challenger, and Barabba knew if he could get his candidate in front of the right voters he would be able to pick up support. Polling showed that, despite their historically Democratic allegiances, union members would be persuadable by Riegle’s appeals. Barabba put his Precinct Index Priority System to work finding neighborhoods whose demographics reflected a more conservative profile even if they were not traditionally thought hospitable to Republicans—the white, working-class union members who would later become known as Reagan Democrats. “Everyone had classified union
workers all the same,” says Barabba. Now he was going to try to capitalize on that mistake.

In September, the gap stood at 20 points, with Mackie still comfortably at 51 percent. Barabba doled out the candidate’s time in the places where PIPS told him it would be most valuable. The Riegle for Congress Volunteer Committee canvassed the most promising areas by phone to identify individual voters, while the candidate went to knock on doors in the precincts where Barabba thought his physical presence could have the biggest influence. On election night, Barabba was at the Biltmore Hotel in Los Angeles, at what became Ronald Reagan’s victory party in his first run for office. Around the same time that Reagan was declared California’s governor, Barabba got a telephone call from Michigan. It was Riegle reporting that he had beaten Mackie by eight points. The man who emerged from the Biltmore would become the greatest broadcast performer ever in American politics, but Barabba believed that what happened on the streets of Flint ought to resonate as widely. “At that point we were in an era of mass communication, and everybody thought that that would be the way to go,” says Barabba. “We kept questioning that.”

Barabba’s success with Riegle, and four other congressional candidates for whom Datamatics played less significant strategic roles in 1966, caught the eye of the American Medical Association, whose political action committee had emerged during the 1960s as one of the business world’s leading political players. Imagining itself as a counterweight to the campaign clout of organized labor, AMPAC participated in congressional races nationwide, almost always to boost Republicans. “The guys who were heading the PAC at the AMA were disappointed at how much money was spent on political campaigns. They wanted to make sure that when the doctors got involved, their money was spent wisely,” says Barabba. During Barry Goldwater Jr.’s 1969 congressional campaign, Barabba’s analysis enabled the AMA to target telegrams emphasizing different issues by precinct in the Los Angeles area district. “We thought we could do a better job by focusing on smaller areas.” But after Watergate, as campaign finance
laws curtailed the ability of outside groups to spend freely on behalf of campaigns,
the AMA struggled to sustain its influence on elections.

Across Washington, one of the AMA’s nemeses confronted a similar identity crisis and headed in an altogether different direction. The National Committee for an Effective Congress had been founded in 1948 by Eleanor Roosevelt as a backer of liberal congressional campaigns nationwide. Its financial primacy may have been threatened by post-Watergate reforms, but NCEC was intent on remaining a central player in stacking Capitol Hill with allies. What if instead of merely giving money to campaigns, committee strategists wondered, they developed common resources that Democratic candidates and party committees could use to plot strategies and tactics for districts nationwide?

Political operatives have long thought of winning a vote as a three-step process. First a voter needs to be registered. Then comes “persuasion”: the challenge of emerging as the preferred choice among two or more candidates. Finally, once a person has been registered and persuaded, the campaign has to convert that support into a vote: mobilizing him or her to the polls through get-out-the-vote operations, often known simply as GOTV, that can include a battery of last-minute reminders by phone or mail or election day visits offering a ride to the polls.

Every voter is different and has varying degrees of openness to a candidate’s arguments or a need to be pushed to the polls. Campaigns use canvassers (either volunteers or paid workers) to touch these voters one by one, through a phone call or a doorstep visit, and gauge their support and likelihood of voting. Canvassers are typically given a script—do you expect to vote? which candidate do you support? do you think you could change your mind?—and then charged with plotting responses on a five-point scale: a 1 is a certain supporter, 5 a firm backer of the opposition. But in the 1970s few campaigns had the time or resources to tailor a distinctive strategy for following up with each of them. As a result, operatives often had little choice but to blanket an entire category of people with arguments they thought would be persuasive—or, in the scramble of election day, to
blindly pull them out of their homes and to the polls. As a result, the parties conducted direct contact almost exclusively on turf they deemed safe.

Improving their knowledge of each patch of turf was of special value to Democrats. The party’s strongholds, usually in cities, tended to be packed full of precincts that turned out for the party in such overwhelming numbers that Republicans reflexively accused their opposition of vote fraud when they saw the lopsided election returns. (There has indeed been vote fraud, but also many precincts that Democrats legitimately carry with 95 percent of the vote.) Meanwhile, Republicans were spread more widely, between cities, suburbs, and rural areas and taking on a slightly different cast in each. There were few precincts anywhere in the country that consistently voted 70 percent Republican.

These differences in political geography shaped the way each party practiced politics. Democratic ward leaders in Chicago and Boston and Philadelphia rarely employed the vote-counting rigor and discipline that Lincoln advised. They would rely on TV and radio ads blanketing the city to do the work of persuasion, air cover for a ruthless ground war below. Party bosses would take areas known to be more than 65 percent Democratic and flood them with manpower on election day, hitting the door of every voter they could find without bothering to check their party identification or candidate preference. Operatives had different terms for this get-out-the-vote practice, but each reflected a relative lack of nuance. In Philadelphia they dubbed it “knock and drag”; in the black counties of rural Virginia it was “hauling and calling.” Both represented a fast-food strategy, identifying reliable margins and counting on volume for victory.

Because of these tactics, pinpointing the difference between 70 percent support in a precinct and 60 percent was very valuable to Democrats. In large swaths of the country that job was getting more difficult. What mattered wasn’t just how people registered but the way districts voted, and the partisan realignment that began during the civil rights era confused the map. Through the end of the twentieth century, Dixie was filled with registered Democrats, but—at least when it came to presidential contests
at the top of the ticket—few Democratic voters. (Alabama voted for only one Democratic presidential candidate after 1964, but its state legislature didn’t go Republican until 2010.) Working-class whites in Milwaukee and Cleveland may, too, be self-identified Democrats and even union members but increasingly they were open to voting for Richard Nixon and, later, Reagan. Flushing out a precinct’s worth of votes on historically Democratic turf could mean producing margins for the other side.

As part of its post-Watergate reinvention, the National Committee for an Effective Congress decided in 1974 that it would take on the challenge of mapping political geography so Democratic candidates could intelligently target every precinct in the country. Nearly all the data necessary to accomplish that was publicly available but prohibitively fragmented. Some states maintained voter files—which usually listed an individual’s name, age, and gender—but assembling precinct vote data was a laborious process. NCEC hired teams of researchers whose days were spent calling county election boards for past vote returns and maps that showed where precinct lines were drawn. It would all arrive on paper and need to be inputted manually to computers. Then analysts added neighborhood-level Census data to enliven the portrait of residents, by categories like race, ethnicity, and household type. (In states covered under the 1964 Civil Rights Act, officials were required to keep individual racial markers in their voter file. Ethnic name dictionaries helped flag those likely to be Jewish, Hispanic, or Asian.) Of course, NCEC analysts knew that every ten years, districts at all levels would be redrawn to match new Census figures—and many precinct lines would change even more frequently—so they had to not only keep tallies of votes but also translate them over time to align with changing boundaries.

NCEC analysts boiled this stew of data down to three key formulas they could apply to any precinct in the country. The first was a Democratic-performance index, an estimate of how an average Democratic candidate would fare on the ballot. It was a useful indicator of which party controlled the turf: anything over 50 percent was a Democrat-friendly area,
while beyond 65 percent represented the party’s base. Then NCEC calculated a “persuasion percent,” which took the measure of how much an area swung between parties due to crossover votes. (Some people thought it made more sense as a “volatility index,” measuring how much voters were willing to swing between the parties.) A persuasion percent of 30 indicated an area where a campaign could expect nearly a third of voters to be open to appeals from both sides, like a Denver suburb where partisan identity is weak and voters hopscotch between Democrats and Republicans as they make their way down a ballot. A persuasion percent of zero reflected a place where party loyalties are firm and voting patterns predictable regardless of which candidate is on the ballot—whether that means Democrats win with 70 percent of the vote each year or always lose with 30 percent, or whether races come down to a 50–50 split. A third formula, the “GOTV percent,” measured the volatility of turnout: how much did the number of people who actually voted shift from year to year?

When combined, these calculations made it possible for Democratic electoral strategists to meticulously separate their voter contact resources between persuasion and turnout and spread them around to areas where each could have the most impact. By the mid-1980s, NCEC had become a crucial utility in Democratic politics, contracted by the DNC and other party organs to crunch precinct-level election data for them. In essence, NCEC had determined the odds that a given voter in any precinct in America had Reese’s green nose or purple ears. A campaign no longer needed the resources to count every single person; it could just play the averages.