The 2004 Democratic nomination was decided in Iowa. John Kerry's decision to focus his efforts in the Hawkeye State, with the support of a $6.4 million loan secured by his home, paid off handsomely: Kerry rallied in the final two weeks to upset Howard Dean, and surged to an easy win in New Hampshire just eight days later. The rest of the states then fell like dominoes.
"Inside the Dean Campaign" (April 8, 2004)
Howard Dean's political pollster talks about the campaign's extraordinary rise and crashing fall.
Dean and his chief strategists—Joe Trippi, the campaign manager; the media consultants Steve McMahon and Mark Squier; and I—were not surprised: winning Iowa had been the heart of our own victory plan. As was the case with so many other parts of our campaign, somebody stole our hopes along the way. This is the story of how it happened.
Polling in Iowa is both important and impossible. The vagaries of the caucus process make it very difficult to predict who might actually turn out on caucus night and to target those voters in a poll sample. The problems are compounded because Iowa voters have become "professional," and they know it is to their advantage to delay making up their minds.
Our first stab-in-the-dark-attempt was in late April of 2003, nine months before the caucuses. It confirmed some of Howard Dean's most fervent hopes about the state: he was actually in the lead among Iowans who had attended a caucus in the past and said they would definitely go this time. But it exposed a serious weakness: he was virtually unknown to many Iowans, particularly those inclined to support the 1988 Iowa winner, Richard Gephardt, of Missouri. The data also hinted that John Kerry could become a formidable candidate and that John Edwards was a direct threat to Dean. The overall results were these: Gephardt, 34 percent; Kerry, 16 percent; Joseph Lieberman, 15 percent; Dean, 10 percent; Edwards, six percent; others, four percent; undecided, 15 percent.
Some revealing additional findings came out of that first poll. One concerned the war in Iraq. Dean's support was 24 percent among those who strongly opposed the war; 13 percent among those who somewhat opposed it; and only six percent among those who favored it. Nearly three quarters of our sample opposed the war. The lesson for our planning purposes was that antiwar messages were among our least effective means of increasing Dean's support. We already had most of the core antiwar vote.
The second finding involved the dynamics among the candidates. It was clear even then that Kerry and Edwards posed a threat to us; support for the three tended to overlap. As I wrote in a memo just after the April poll, "In some ways Dean and Edwards are battling to be the 'New Face' ... just as Dean and Kerry have been battling over who is the 'True Democrat.'" The scenario was very different when it came to Joe Lieberman. We thought Lieberman had probably peaked in this first poll, where he was a strong third, and that his center-right run would fail badly among Iowa's liberal voters. Still, the votes Lieberman got would come mainly at the expense of candidates other than Howard Dean. We needed him to stay in the race in Iowa to sap others' support; unfortunately, he pulled out of the state in the fall.
The third finding had to do with biography and record. More than any other, this one showed how John Kerry might actually win. Not counting Lieberman, there were two Washington-establishment candidates in the race—Gephardt and Kerry. And there were two non-establishment, non-Washington candidates—Dean and Edwards (who is serving his first term in the Senate). Voters did not indicate a clear preference for one kind of candidate over the other. In the April poll we found that 36 percent of voters would support a candidate "whose main experience is in Washington, D.C., as a member of Congress," and 32 percent would support a candidate "whose main experience is outside of Washington, D.C., as a governor." But when voters were asked to make a choice based on job description and without knowing the candidates' names, the winner was a "lawyer/ congressional leader" (Gephardt) over a "medical doctor/governor" (Dean), by 49 to 18 percent. "Vietnam vet/ senator" also won over "medical doctor/governor," by 49 to 17 percent. In the actual Iowa-caucus vote, nine months later, Kerry and Gephardt together received 49 percent, and Dean 18 percent.
Fortunately from our point of view, when voters heard briefly about the qualifications of each candidate by name, including Dean's record in Vermont (providing health care, balancing budgets, and so on), Dean was the only candidate to gain support; he moved ahead of Lieberman and Kerry into a strong second against Gephardt. We knew we could probably never win an outright battle of the bios, but a fight about the candidates' achievements as elected officials looked promising. When we met to discuss these results, we debated the odds of defeating Gephardt in Iowa, and whether a second-place finish would be acceptable for us. It was a short debate; we decided to go for the win.
Since the campaign-finance reforms of the 1970s, primary campaigns have lived with the reality of what's known as "the cap." To be eligible for federal matching funds, candidates have to keep their spending below per-state limits that are set by a complex formula. In practice the cap has really mattered only in the first two states, Iowa and New Hampshire—and in 2004 it would matter in a new way, because the McCain-Feingold law stiffened the penalties for exceeding the cap. Late in 2003 the Dean campaign decided to opt out of the public-financing system, which meant it could spend as much money as it wished. But before that decision, in the early summer of 2003, it had to calculate when it could start running expensive TV ads without using up cap money too soon.