Blinded by History
How political journalists relied too heavily on past patterns and conventional wisdom, missing what was really going on in the 2002 midterm election
By Rachel Smolkin
In the end, the zealous burrito lovers ranked among the elite prognosticators of Election 2002.
At California Tortilla, a Bethesda, Maryland-based Mexican restaurant chain, patrons indicated preferences in two closely watched races by selecting burritos named after the candidates. Their tastes correctly predicted the underdog victories of Republican gubernatorial candidate Robert L. Ehrlich Jr., represented by a hickory chicken and horseradish-mashed potato burrito, and Democratic House candidate Christopher Van Hollen, symbolized by a mesquite chicken, portobello mushroom, onion and pepper concoction.
"We knew right away Van Hollen was going to win," says California Tortilla co-owner Pam Felix. "His supporters were so into it. They had such an energy and had so much fun with the contest." Ditto for Ehrlich's supporters, who purchased burritos with the confidence and zest of future victors.
The highly unscientific burrito poll tapped into a fervor among some voters that much of the media missed. Most reporters carefully emphasized that many races were too close to call in the weeks and days before November 5. But few pollsters, pundits or political analysts foresaw the breadth of Republican success on Election Day and the ensuing power shift in Washington.
Guided by historical precedent, the press largely missed an opportunity to delve into the unusual circumstances surrounding Election 2002. Some reporters even dubbed it the "Seinfeld election," calling it a campaign about nothing. But beneath the facade of nothing lay something. It was the first election since President Bush's disputed ascension to the White House, the first chance for Republican voters to validate his presidency and an opportunity for party faithful to rally behind their commander in chief. Additional ingredients flavored the political stew: an impending war with Iraq, a Democratic Party with no clear leader and no clear message and a Republican Party with a clear leader, loads of money and a get-out-the-vote effort modeled on past successes of Democrats and labor unions.
Bush and White House political guru Karl Rove understood all this, even if many reporters and analysts did not. While polls vacillated and reporters explained that history prescribes losses for the president's party during midterm elections, Bush defied conventional wisdom, dashed around the country and fired up Republicans.
The media's performance was better than in 1994, when the press underestimated the currents that swept Republicans, led by Newt Gingrich and his "Contract with America," to dominance in the House for the first time in 40 years. "There are no real issues at all this year, at least in the traditional sense," Richard L. Berke of the New York Times wrote on October 23, 1994.
This time, the largely unforeseen Republican Senate takeover occurred because a small percentage of voters ultimately favored the president's party. "Bush was quick to order a ban on post-election gloating," Newsweek's Howard Fineman wrote on November 18, "and his aides were equally quick to downplay the magnitude of the victory, noting (as did their critics) that a switch of 150,000 votes would have kept the Senate in Democratic hands."
A narrow victory, perhaps, but one with huge repercussions. Much of the news media's pre-election take recalled the sniper profilers who had saturated the airwaves just a few weeks before. Most of them had suggested, based on history and precedent, that the suspect terrorizing the Washington, D.C., region would turn out to be a white, middle-aged man. Most of them were wrong. (See "Off Target," December.)
Likewise, political analysts and pundits predicted, based on history and precedent, that 2002 would not be a great year for Republicans in Congress. The president's party had gained House seats in only two midterm elections in the last century: in 1934, during the Great Depression, and in 1998, during the battle over whether to impeach President Bill Clinton. The president's party had picked up Senate seats in five of 22 midterm elections since 1914. So the odds supported reporters and analysts who forecast Republican setbacks. But they also were wrong.
"Everyone was blinded by history," says Darrell West, a Brown University political science professor who runs a politics Web site (www. InsidePolitics.org). "History does not always predict the future. The current times are more extraordinary than people have realized. Just over the last four or five years, unusual things have been happening: the Clinton impeachment, the Election 2000 fiasco, September 11. All these things are combining to produce more volatile politics not guided by past precedent. It's an extraordinary time period that we're living through, and we probably don't even fully grasp that."
As Washington Post media reporter Howard Kurtz noted in a November 7 article headlined "Media Flunk the Midterms," many newspapers predicted in the final weekend before the election that Democrats would cling to Senate control. "Republicans face stiff odds in their bid to reclaim a Senate majority, but Democrats have even a tougher climb to take back the House," Washington Post reporters David S. Broder and Dan Balz wrote on November 3. Ronald Brownstein of the Los Angeles Times agreed: "Analysts in both parties consider Republicans the favorite to maintain--and possibly even expand--their slender six-seat majority in the House. And Democrats seem positioned to maintain--or enlarge--their one-seat Senate majority. Enough races remained within reach, however, to sustain Republican hopes of a takeover."
On ABC's "This Week with George Stephanopoulos" on November 3, the network's political director, Mark Halperin, said, "We're making ourselves a little bit more optimistic in terms of Democrats for the Senate, up one we think now, and obviously, that allows them to keep control."
And on Tuesday, November 5, John J. Miller of the National Review observed, "Republicans may be disappointed if they don't capture the Senate tonight, but they should put things in perspective: Midterm elections usually lead to big-time losses for the party that controls the White House. At most, it appears the GOP will lose a seat or two. Not bad. By my count, the two parties will trade four seats--and the Senate's balance of power won't change."
Once the votes were counted, the Republicans had gained two seats in the Senate and six in the House.
The press always grapples with how much horse-race coverage and how many political predictions are helpful, but in this case, efforts to anticipate the balance of power perhaps shortchanged other aspects of the election.
"We sometimes tend to miss the forest, focusing not only on the trees but the twigs," says Fineman, Newsweek's chief political correspondent and an NBC analyst. "The forest was Bush's popularity in a new, somewhat undefined war period. Absent some other compelling thing, I think people are going to vote to support the president."
The press also largely overlooked the importance of Election 2002 in bolstering Bush's presidency. Some news accounts, including an October 18 story in the Washington Post, suggested that Florida Democrats, angered by the outcome of the presidential election, might rally against presidential brother Gov. Jeb Bush and flock to the polls to support his Democratic rival, Bill McBride. Democratic leaders "said the constant presidential attention to Florida is just firing up their voters by reviving memories of the disputed election of 2000, and some Republican strategists agree," the Post's Mike Allen wrote. But few press accounts explored the other side of that equation. Republicans, too, had lingering feelings about Election 2000, and they also could cast ballots to reflect those feelings. "It's almost as if they viewed this as overtime in a football game, and they were going to play like hell," Fineman says. "Too few of us, including myself, understood the psychology of that."
Tucker Carlson, conservative cohost of CNN's "Crossfire," won the Washington Post's 2002 Crystal Ball contest by correctly predicting GOP gains in the House and a takeover in the Senate. Like California Tortilla's customers, Carlson also rightly foresaw that Van Hollen, one of the few Democrats to pick up a House seat, would end Republican Connie Morella's 16-year tenure in the House.
"The press made the mistakes it always makes," Carlson says, deriding the media's "addiction to conventional stupid ideas." Politics, he observes, is ruled by clichés: endorsements never help (except Bush's endorsements clearly did) and historical models are worthwhile guides (except this time they clearly weren't).
When the country is so evenly divided, discerning a national trend is no easy task. But Carlson says the press could have paid more attention to certain aspects of the campaign season. "Nobody noticed, for some reason, that the Democrats weren't really running on anything," he says. "I do think that you kind of need to run on something. It can be something stupid or unfair or misleading, but in the end it has to be something.... The other thing that people missed is that [the White House] employed an incredibly bold strategy."
White House tentacles reached into every facet of the midterm elections, from candidate recruitment to fundraising to campaign appearances. While reporters cited Bush's vigorous campaigning and prodigious fundraising, few grasped the impact of that strategy.
"With a president at a 63 percent approval rating, the last blitz that he made in those five days, we shouldn't have underestimated the impact that would have," says Gloria Borger, a columnist at U.S. News & World Report and CBS News special correspondent. Borger notes that Bush willingly risked his popularity by campaigning in states with very tight races. His South Dakota Senate candidate, John Thune, ultimately lost by 524 votes. But the president's candidates triumphed in Georgia, Missouri, New Hampshire and Minnesota, where he also campaigned in the final days. "The lesson is that a president with a 63 percent approval rating has to really be taken into account in these kinds of close races," Borger says.
Borger also was intrigued by the shifting political dynamics of the homeland security debate. When Sen. Joe Lieberman, D-Conn., initially proposed a Department of Homeland Security in October 2001, the White House opposed the idea. But when congressional hearings exposed intelligence lapses before the September 11 attacks, Bush seized the idea and successfully turned it against its creators. "It became a very potent issue and backfired on the Democrats," Borger says. "It's fascinating to go back and reconstruct it."
In a November 25 column, she did just that. Borger wrote that the issue has "morphed into a wildly successful Republican campaign mantra, as in: Our homeland security bill shows that we are serious about fighting terror and the Democrats are not.... For the Republicans, it was the perfect political storm."
She says the homeland security debate and election results taught her a "journalistic lesson" that will influence her political perceptions in the future. "Foreign policy has become domestic policy now," she says. "National security is personal security....We always say people vote on domestic policy, but [national security] is now domestic policy. That's where things have changed for me in a very interesting way."
New Republic Editor Peter Beinart describes the campaign coverage as "a bit disappointing" because "more thought could have been put into this. A lot of people relied on this sort of soundbite stuff about how presidents don't win in midterm elections."
Beinart says he would have preferred deeper analysis, including, for example, observations that the president's party often loses seats in midterm elections because weaker House candidates sweep into office on the president's coattails and do not survive re-election. That occurred in 1980 during Ronald Reagan's landslide, but not in 2000, when Bush lost the popular vote. "There wasn't a lot of low-hanging fruit for the Democrats to take," Beinart says.
He adds that the media is too wedded to the "stock model" of campaign coverage. "You send someone out to spend a day with the candidates; they watch the [campaign] commercials and call a couple analysts in that state," Beinart says. "The press is still catching up to the fact that political campaigns have changed."
National reporters could learn more, Beinart suggests, by paying more attention to campaign mechanics such as direct mail and turnout operations, and by spending more time immersed in communities.
In Georgia, for example, Republican gubernatorial challenger Sonny Perdue staged one of the year's biggest upsets by becoming the state's first GOP governor in 130 years. He had pledged to let voters decide whether to revive the state's old flag, dominated by the Confederate battle emblem. The defeated governor, Democrat Roy Barnes, had successfully pushed for legislation to minimize the emblem. "Sonny Perdue used the issue of the flag to great effect under the radar screen," Beinart says. Reporters "might have picked that up from direct mail or by spending more time on the ground."
While the Washington Post covered the Maryland governor's race extensively, Beinart adds, other national reporters could have spent some time in Prince George's County to illustrate the lack of enthusiasm for Democratic candidate Kathleen Kennedy Townsend. Frustration and disinterest among black voters helped a Republican capture the Maryland governor's mansion for the first time in more than three decades.
But Beinart concedes his own election analysis was less than perfect. "I, like most people, thought the Democrats would hold on to the Senate," he says. "I didn't pick up on the turnout, the stronger than anticipated Republican turnout, and the Bush effect in the final days."
Beinart had plenty of company. Most reporters and pollsters missed the late Republican surge. On the weekend before the election, several high-profile polls conflicted, illustrating the limitations of polling in very tight races and the difficulties of predicting which voters will cast ballots.
"There was a lot of movement at the end," says Thomas Patterson, a professor of government and the press at Harvard University's Kennedy School of Government. "What's happening in part is we've got an electorate that's a little less grounded than once upon a time. It's more movable than in the past. People psychologically are just kind of kicking in at a later point."
Patterson and his colleagues at the Joan Shorenstein Center on the Press, Politics and Public Policy conducted a Lexis-Nexis search about 10 days before the election and concluded that media coverage had dropped about 15 percent from the same point in 1998. "The media are kicking in later, the voters are kicking in a bit later, and so a lot of decisions that might have been made a week out, two weeks out, a month from the election are being made later," Patterson says.
In the turbulent Minnesota Senate race, where Democrat Walter Mondale replaced the late Sen. Paul Wellstone less than a week before the election, two newspaper polls conducted from October 30 through November 1 showed opposite results. The St. Paul Pioneer Press/Minnesota Public Radio poll showed Republican Norm Coleman ahead, 47 percent to 41 percent, while Minneapolis' Star Tribune had Mondale leading, 46 percent to 41 percent.
"The difference between the polls suggests an electorate in tremendous flux but also could be the result of different polling methods," the Star Tribune's Patricia Lopez wrote on November 3. Coleman won, 49.5 percent to 47.3 percent.
In the Colorado Senate race, two newspaper polls put Democratic challenger Tom Strickland ahead but well within the margin of error. A Denver Post/9News/KOA poll conducted October 29 through November 1 showed Strickland at 42 percent to Republican Sen. Wayne Allard's 41 percent, and a Rocky Mountain News/News4 tracking poll around the same time had Strickland over Allard 42 percent to 38 percent. But Allard prevailed with 50.9 percent.
Mark Mellman, a Democratic pollster in Washington, says polls across the nation were "unusually bad this year. Usually most polls most of the time pick up the results very closely, and this year a lot of the polls seemed to be very far off."
In many states, pollsters were handicapped by newspaper deadlines, CNN Polling Director Keating Holland says. He has observed that editors who want to include a final poll in their Sunday papers often request results by Thursday to allow sufficient time for graphics. That means pollsters have to complete survey interviews a week before the election.
Editors are "asking pollsters to do the impossible: to predict the election more than a week in advance," Holland says. "It's the same as telling the sports reporter in a football game to leave the game in the third quarter and write the results."
But some national pollsters encountered difficulties even with late canvasses. Pollster John Zogby gained fame in 1996 by coming within one-tenth of 1 percent of the presidential race tally and further enhanced his reputation in 2000 by showing Al Gore pulling ahead in the popular vote. But this time his numbers seemed less prescient. As late as November 4, Zogby's polls correctly showed Republican Jim Talent ahead of Democrat Jean Carnahan in Missouri. But Zogby also showed leads for Democrats Mondale, Strickland and Max Cleland in Georgia, all of whom lost. Pollsters have to estimate turnout, and Zogby based his turnout models on historical voting patterns in each state.
"Some of us screwed up in some places, to be sure, but it was always too close to call until the very end what the Senate would look like," Zogby says. "I know that Gallup looked at a possible Republican surge a few days before, and my hat's off to them for that. I didn't see it. I had no evidence at all, to be honest with you."
Gallup did indeed detect a Republican upswing just before the election. "Late shift appears to favor GOP," declared a page-one USA Today headline on November 4. "Key Senate races in Tuesday's congressional elections are too close to call, but Republicans appear to have gained strength in the final weekend as they fight to retain and perhaps add to their thin House majority," the paper's Richard Benedetto wrote in a story that included the findings of a USA Today/CNN/Gallup Poll.
A New York Times/CBS News poll also detected the Republican advantage on the "generic ballot question," which asks voters which party's candidate they'll support. But the newspaper's handling of its findings highlights the challenges of assessing late polls, even if they ultimately prove to be correct.
"IN POLL, AMERICANS SAY BOTH PARTIES LACK CLEAR VISION," declared the Times on its November 3 front page.
"The battle for control of Congress moved into its final stretch with Americans unsettled about conditions at home and threats from abroad, but saying that Democrats and Republicans have failed to offer a clear vision about how they would lead the nation, the latest New York Times/CBS News Poll shows," wrote Adam Nagourney and Janet Elder.
Their approach drew a rebuke from Michael Barone, a senior writer at U.S. News & World Report, a contributor to Fox News Channel and coauthor of "The Almanac of American Politics." Barone faults the Times for stating that neither party had a clear plan for the nation, when the paper's poll numbers showed that 42 percent of respondents said Republicans had a clear plan, compared with 31 percent who said Democrats had a clear plan. The margin of error on that question was + or - 3 percentage points.
"The Republicans were significantly more likely than the Democrats to be seen as having a clear plan," says Barone, who correctly predicted Republicans would recapture the Senate. "They missed the story from their own poll. That's astonishingly bad journalism."
But Nagourney, national political reporter for the Times, says he dislikes horse-race coverage and wanted to approach the story cautiously, especially the weekend before the election. What if he led with the apparent Republican advantage, and then the Republicans lost? "This story gave readers a clear indication without overstating it," Nagourney says. "Any fair reader of that story would understand that the Democrats were facing some difficulties here." He says he believes that he handled the story correctly and fairly.
Other reporters and political observers also object to horse-race coverage. "The obsession with pre-election polling is obscene and tantamount to a direct threat to democracy," says Kellyanne Conway, a Republican pollster in Washington. "Clients' polls are meant to be used for strategic internal insight from which you figure out how to win, not whether you are going to win. The media's focus on horse-race coverage compels everyone to look at who will win instead of who should win. Public polls are what are sucking the lifeblood out of our democracy, because they're thwarting turnout."
Mark A. Schulman, president of the American Association of Public Opinion Research and a New York-based pollster, adds, "People tend to look at the importance of the polls in terms of the horse race, who's ahead and who's behind. But a lot of what we pollsters do, and probably our more important function, is to explain why. What are the issues in the public's mind? Why are they gaining or losing? A lot of what we do is try to understand the electoral dynamics."
Brownstein, the Los Angeles Times' national political correspondent and a CNN political analyst, says reporters can't ignore horse-race polls, because they offer insight into the "state of play" in an election. But he says of predicting results: "I hate it. Most of these things are not knowable.... By and large the political press corps would be better off with less prediction and more analysis of what already happened. I would be perfectly happy if my editor said to me, 'Don't call any more races.' "
Noting that small political shifts have substantial impact in an electorate split so evenly, Brownstein says he isn't "really that hard on the press" for its predictions and coverage of the midterm elections. "It's not like 1994 happened, and everybody missed it," he says.
But he does question why the press spent so little time digesting the results. "The electorate always surprises you," he says. "The more valuable part [of political reporting] is understanding what's already occurred. I've just been struck by how fast we've moved off it, of trying to understand it."
Comprehending the election results has been more difficult than usual because of the election night meltdown by Voter News Service, a consortium whose members include ABC, CBS, NBC, CNN, Fox and the Associated Press. A computer snafu, which VNS is investigating, devastated the service's ability to provide exit polls that not only help networks project winners but also provide valuable information about who voted and what issues motivated them. (See Free Press, December.)
Perhaps VNS will successfully revamp itself during the next two years. And maybe during the next election the press can curb its enthusiasm for horse-race coverage and historical precedent, instead savoring the singularity of Election 2004. It will be the first presidential race since the election mess of 2000, the first since the terrorist attacks and the first since Bush achieved historic gains in the House and Senate.
Will the Democratic candidate have a clear message and that "vision thing" once alluded to by the president's father? Will Democrats reassert their savvy in propelling voters to the polls? Will the Supreme Court sit the next one out? Will Americans care? And will hope or fear spur them to participate?
And if reporters do hunger for on-target surveys in 2004, perhaps we can all turn to California Tortilla. Owner Felix promises her customers will sink their teeth into the presidential contest with gusto.###