Capitol Offense

Hardly anyone gave Donna Howard much of a chance. After all, she was a Democrat running in Texas House District 48—an area that hasn’t been friendly to Democrats in recent elections. The district covers a wealthy, largely Anglo swath of West Austin, stretching into the city’s burgeoning suburbs. The residents live in large houses, drive large cars, and, by a roughly 57 percent majority, vote Republican. Todd Baxter, a GOP backbencher with few legislative accomplishments, managed to beat well-funded Democratic candidates in 2002 and 2004. So when Baxter suddenly resigned from the Legislature in fall 2005, many Texas political observers expected the GOP to retain the seat in a special election. Howard, a progressive Democrat, former nurse, and public school activist, entered the special election as a first-time candidate with scant campaign money and even less name ID. Facing another Democrat and an experienced Republican in a GOP district, Howard was a major underdog.

Yet on Election Day, January 17, 2006, Howard trounced the field, winning 49.5 percent in the four-way race. Alarmed by the result, the Texas GOP establishment mobilized for the runoff election a month later. Checks rolled in to Republican Ben Bentzin’s campaign from the biggest conservative donors in Texas. He nearly doubled Howard’s fundraising total. Yet in the runoff, Howard thumped Bentzin again, earning a remarkable 58 percent of the vote in the GOP-leaning district.

How did she do it? The Texas political chattering class attributed Howard’s surprising success to anti-Republican, anti-incumbent anger among voters. But Howard’s upset may contain another important lesson for Democrats in Texas and around the nation about how modern campaigns must be waged. Howard didn’t just craft a winning message; her campaign identified and targeted exactly which voters would be most receptive to her appeals. To do that, the Howard team tapped into the latest high-tech tool in political campaigns: voter targeting, also known as data mining.

The premise is simple: In modern politics, candidates for public office are products that must be sold to voters. As in any corporate marketing campaign, it helps immensely to know who the consumers are. In data mining, a campaign adopts the techniques of corporate marketing to compile as much personal and demographic information about as many voters as possible. Voter targeting takes into account everything from which primaries you’ve voted in, to your marriage status, to how long you commute to work, to how often you use FedEx. Each bit of information yields a clue about a voter and voting tendencies. The campaign runs this information through a series of complex, computer-driven statistical analyses. The end result is a surprisingly accurate portrait of who lives in the district, which voters the campaign should pitch to, and how to reach them in a media-saturated culture.

Republicans have used these techniques in recent election cycles to successfully target and turn out conservative religious voters. Democrats in Texas, however, have been slow to catch on. Yet the Democratic Party’s biggest successes in Statehouse races the past two years—Mark Strama in Austin, Hubert Vo in Houston, David Leibowitz in San Antonio, and Howard—all used data mining to some degree.

Not everyone is enamored with the technique. Some of the top Democratic consultants and campaign contributors in Texas believe the effects of voter targeting are overblown, just hype generated by a handful of self-promoters that distracts from more important elements of campaigning. A debate has begun within the party about the effectiveness of targeting voters in small legislative races. In particular, the influential Texas Trial Lawyers Association, the deepest well for Democratic campaign money, has generally refused to provide major support to campaigns that use data mining. Howard, for instance, received hardly any money from trial lawyers in her race. The result is that voter targeting has been slow to gain traction among Democratic candidates. Only about a dozen around the state are using targeting this election cycle.

“We make decisions on funding campaigns based on their viability—the candidate, the district, the message, and the competence of the campaign team,” says Russ Tidwell, political director for the Texas Trial Lawyers Association. “There’s a danger of candidates being sold snake oil—they think something is a magic elixir, and it’s nothing of the sort. And it gets them off the other campaign fundamentals. That’s why we’re wary of things like this. You need a good candidate, with a good message, talking to a lot of voters.” Tidwell and other critics in the Democratic Party argue that voter targeting segments the electorate too much. They charge that tailoring specific messages for certain groups of voters denigrates the debate in campaigns and risks over-reliance on one piece of technology. “One new, sophisticated tool is not a panacea,” Tidwell says.

Even the biggest proponents of voter targeting concede it wasn’t the exclusive reason for Howard’s victory—her policy ideas, a message that appealed to a frustrated electorate, and mistakes by Bentzin were all important factors. But could Howard have won without voter targeting? “No,” says her campaign consultant, Kelly Fero, “because we would have had to campaign in a more general sense instead of specifically targeting the voters we needed to get to the polls to win this thing.” Fero contends that Americans are inundated with advertising and informational messages each day on radio, television, and the Internet. “Political candidates are just another product in that sense, trying to drive a message through that fog of advertising,” he says. “So the message has to be good. But you also have to know who you’re trying to reach with that message. That’s why the targeting is so important.”

The man behind Donna Howard’s data was Leland Beatty. A longtime Democratic operative, Beatty is an unreconstructed number-cruncher with graying hair, a neat mustache, and a pair of prim glasses. He served as an aide to Jim Hightower in the Texas Department of Agriculture in the early 1980s, and he still salts his talk of regression analysis and error rates with the kind of folksy, populist phrases he once wrote into Hightower’s speeches.

Beatty was with Hightower during the 1982 campaign—the high-water mark for liberal Texas Democrats. Ann Richards (treasurer), Garry Mauro (land commissioner), Jim Mattox (attorney general), and Hightower were all elected to statewide office. One of the reasons for that success was the Texas Democratic Party’s turnout effort. The party took lists of voters who had participated in previous Democratic primaries and divided that list of likely Democrats by their voting precincts. The party and campaign staffs then focused on turning out the voters in the heaviest Democratic precincts. This wasn’t a novel approach, but party officials in 1982 were particularly adept at identifying reliable votes, especially among progressive Democrats. Party leaders have tried to re-create that success ever since.

In the 25 years since the 1982 election, Texas has shifted from a one-party Democratic state to a one-party Republican hegemony. And the precinct-by-precinct model that worked so marvelously in 1982 has delivered diminishing returns. In recent years, it has sputtered badly (see Tony Sanchez campaign for governor, 2002). Because there are simply more Republicans than Democrats in Texas, the old precinct approach has been focusing on an ever-shrinking pool of reliable Democratic voters. The party still ushers a high percentage of hard-core Democrats to the polls. Problem is, Beatty says, the small supply of solid Democrats will rarely—even if they all vote—be enough to win an election. It’s mission impossible.

The answer to this conundrum, in Beatty’s view, is to expand the universe of voters you hope to attract—go after not just the Democratic base, but also what Beatty calls “persuadable” voters. These are the people Democrats in Texas and around the country have failed to attract in recent election cycles. To win over these folks, Beatty contends, you have to know who they are.

Both national parties already have toyed with voter targeting, incorporating demographic and consumer data into their voter files. Thus far, however, Republicans have been far more adept at utilizing the data. Beatty was wowed by the ingenuity of the Bush-Cheney 2004 campaign that presidential adviser Karl Rove devised. Rove recognized that conservative religious voters were severely underrepresented in the 2000 election, and used data mining to identify these voters and win them over in key battleground states, particularly Ohio. As Beatty will tell you, one of the reasons the exit polls were so wrong on election day 2004 was because the Bush-Cheney campaign simply knew who its voters were better than exit-polling agencies did.

Another recent example of GOP data mining prowess was the 2005 vote on a Texas constitutional amendment to ban same-sex marriage. For Beatty, the reason the amendment won 76 percent of the vote—a tally that shocked some progressives—was evident in the data. Conservative religious groups rolled out a strong targeting effort to uncover and attract new voters who would support a ban on gay marriage. “It had the largest turnout in a constitutional amendment election since 1990,” he says. “Twenty-five percent of those people had never voted in a general election in their lives.” Many of those new religious voters look enticing to Gov. Rick Perry’s re-election campaign. Perry was a high-profile supporter of the ban. Thanks to targeting in the constitutional amendment election, the Perry folks now know exactly which voters to go after this fall.

Beatty believes that Democrats must be similarly ambitious. “I’m tired of hearing this thing that, ‘Well, we just have to wait for the census to turn around,'” he says, referring to the conventional wisdom that an emerging Latino majority in Texas eventually will propel Democrats back into power. Voter targeting, he contends, can speed up the process.

On a July afternoon, Beatty settles down in front of his PC to demonstrate how data mining works. His home office occupies a small corner of his South Austin bungalow. A small blizzard of cigarette ash covers his workstation and keyboard. Smoldering Camel in one hand, mouse in the other, Beatty opens a software program called SPSS. Rows of numbers appear on the screen. This is raw voter data for Congressional District 14, which Beatty is analyzing to predict how the Southeast Texas area—currently represented by Congressman Ron Paul—will vote in the upcoming governor’s race.

What separates Beatty’s work from traditional voter identification techniques is the level of detail. The Democratic Party’s tried-and-true, precinct-based approach ends with a person’s voting history; Beatty’s analysis begins there. He takes the standard voter file, and “I add to that 10 to 15 commercial variables—are they a homeowner or renter? How old are their cars?” For the same reasons that shopping at Banana Republic may mean you receive the Pottery Barn catalog in the mail a few weeks later, political campaigns can obtain reams of consumer and demographic information about potential voters. These data troves—which can list everything from the nonprofits you support to the magazines you receive in the mail—are relatively easy to obtain. If not enough commercial data is available, a campaign can call a household to gain more information. Beatty then adds census information to the mix: How much do they earn? Married or single? Have kids or not? What race are they? Beatty runs the data through a series of complex analytic models on the computer. The software extrapolates from his sample to predict how people might vote. When the predictions prove to be mostly accurate, he groups voters by household, assigning each a house identification number based on address. So-called “householding” of voters is important, Beatty contends, because people’s living environment affects their voting. Is a likely Democratic voter married to a Republican? Is a 20-year-old living at home with two Democratic parents? Is a soft Democratic voter living with a probable GOP voter? All these change the formulas.

With enough data, Beatty can produce for the campaign a detailed demographic portrait of the district. He also calculates two probabilities for most voters—how likely they are to vote in the first place, and how likely they are to vote for his candidate. The campaign can then decide if it wants to target voters who are impulsive (vote 50 percent Democrat and 50 percent Republican), but who definitely will turn out (90 percent probability). Or the campaign may go after people who vote, say, only 55 percent of the time, but when they do, they will almost certainly vote Democratic (90 percent).

The analysis is more precise than traditional methods. Instead of measuring by large voting precincts, Beatty can isolate residential areas, certain blocks, and even specific houses. That gives a campaign more flexibility to target, say, a set of four houses, instead of going after an entire precinct. A campaign can go after persuadable voters even within unfriendly precincts.

For Donna Howard’s race, Beatty’s numbers revealed that District 48 wasn’t quite as Republican as the campaign had originally thought. The numbers also revealed a voting population wealthier than average, more educated than average, more politically astute, and majority female.

Fero, Howard’s consultant, gleaned from this information that he could craft a complex message that highlighted Republican failures on public education. Howard’s initial message—”If you want change, vote for it”—was geared toward astute voters since it assumed knowledge of previous failures on public school finance. Fero and his partner, Jeff Hewitt, then created a media strategy, eschewing campaign mailers and broadcast television commercials for a limited ad buy on cable stations such as Lifetime that attract women viewers. Hewitt and Fero like cable because it’s cheaper and they can specify ad buys by neighborhood. Ads on Austin’s network affiliates, meanwhile, are more expensive, and only a fraction of the viewers live in the district. “My male friends were asking me when we were going up on the air. We had been on the air two weeks already, but they weren’t seeing it on ESPN,” Fero says. He believes that effective targeting helped the Howard campaign erase two of the GOP’s inherent advantages in Texas—a majority of Republican voters in many districts, and a surplus of major campaign donors. The Howard campaign’s targeting was spot on. Turnout in the special election was extremely low—only about 9 percent of eligible voters went to the polls. Usually, the GOP—so adept at turning out its base support—flourishes in low-turnout affairs such as this. But Howard turned the tables, effectively targeting and energizing the district’s Democrats and disenchanted Republicans in a decisive win.

Critics of targeting contend that its effects are over-hyped. For instance, several Democratic consultants argue that targeting had little to do with Howard’s victory. They point out that District 48 has been trending Democratic for several years. They also note that the Travis County Democratic Party sent campaign mailers using traditional voter ID methods supporting Howard that had more impact than her campaign’s limited cable advertising. “Other factors are much more important—the quality of the candidate, the relative win-ability of the district, the message you choose,” says the TTLA’s Tidwell. “Data mining could be most useful in statewide races where resources are relatively scarce. Its value in legislative race
is marginal, and
e’re shying away from it because of the danger of over-targeting and over-reliance. There’s a danger of not communicating with enough voters. You either have a good message, or you don’t. The audience just doesn’t need to be segmented that much.”

Fero responds that data mining doesn’t shrink the electorate at all, but rather—by targeting “persuadable” voters—expands the number of voters a campaign communicates with. He concedes that data mining alone can’t win an election—a campaign still needs a good candidate, a good message, and a good media strategy. “You shouldn’t use it as a crutch,” he says. For proponents of targeting, the debate comes down to whether a party that has been losing elections for years wants to try something new, change its strategy. “It’s an old school-new school deal in Texas Democratic politics,” Fero says.

Howard sees little reason for debate. She says it would have been difficult to win such a high percentage of votes in a Republican-leaning district without effective targeting. “Clearly we were attracting beyond the [Democratic] base,” Howard says. “If [Democrats] are going to turn the tide and start winning elections, you can’t keep doing the same things and expect a different result.”