Editor’s Note: The most watched election in America in 2006 was the Senate race in Connecticut, where veteran Democratic incumbent Joe Lieberman was defeated in a primary by a political unknown but came back in the general election to win easily. The Lieberman race offered pollsters unique and demanding challenges, which the Quinnipiac University Polling Institute met with a total of twelve polls during the election year—a record for one election.
There is no more difficult job in political polling than trying to determine who will win a primary election. Nearly as difficult is polling a three-way general election race with the incumbent running as an independent against Republican and Democratic challengers. Throw in an unpopular war in Iraq and the fact that Senator Joe Lieberman was not your run-of-the-mill incumbent, but a Democrat once nominated by his party to run for vice president, and you have a polling nightmare.
On the eve of Election Day 2006, Lieberman and both his Democratic and Republican rivals claimed that the final Quinnipiac Poll which showed Lieberman ahead with 50 percent, followed by Lamont with 38 percent and Schlesinger with 8 percent, was wrong. The next day the voters proved the poll was right.
When it comes to using polls to predict primary outcomes, most pollsters don’t, and for good reason. Voter turnouts are always small, candidates are often unknowns, and it is impossible to know what party or political organization or candidate has the better get-out-the-vote operation when the goal is not to get out a majority to vote, but maybe just 20 percent or so. In the best general elections, screens to determine likely voters still produce unrealistic turnout predictions. In a primary such screens are even more risky predictors. Even a history of voting in past primaries can’t be trusted because voters who came out in the last primary where, say, abortion was the big issue probably won’t come out the next time if the issue is the Iraq War.
Connecticut presented a particularly difficult venue for primary polling, as the state has an August primary, when many voters (but who can tell which ones?) are on vacation; and the fact that Lieberman was challenged by a young—but wealthy and attractive—Ned Lamont meant that there would be a lot of first-time voters casting ballots (but who could tell how many?)
What was known was that the tremendous interest in the race promised a turnout larger than the usual 20 percent, which helped because it increased the chances that those Democrats polled were likely voters. Normally only two in ten polled would show up, but with the 43 percent turnout in this primary, the numbers more than doubled.
In the event, the primary and the general election that followed produced some twists and turns that will keep pollsters and pundits analyzing them for years to come.
A first rule of primaries violated by Connecticut voters was that incumbent senators aren't supposed to lose them. Only four in the past twenty-five years had done so; and Lieberman was a popular three-term U.S. Senator and former vice presidential nominee, liberal and Jewish in the bluest of blue states. But his strong backing for an unpopular war had cost him support among his own party members. As early as January, the Quinnipiac Poll found a surprising danger sign for Lieberman: Only 52 percent of Democrats wanted the party to nominate him.
Despite some grumbling from antiwar voters in his own party at that time, however, Lieberman was considered safe because no potential candidates were on the horizon with the stature to pose a serious threat. There was some buzz that perhaps Lowell Weicker, the former Connecticut senator and governor, would take on Lieberman in a fascinating rematch of their 1988 battle, but Weicker declined, and instead encouraged his friend Ned Lamont to be the candidate.
Challenging the old political saw that you can’t beat a somebody with a nobody, Lamont went on to take a page out of Senator Eugene McCarthy’s political handbook for antiwar candidates and demonstrate that with the right message and lots of money, even a political novice can defeat a well-known incumbent, at least in a primary.
When Lamont jumped into the race he was fifty-five points behind Lieberman. But the same polls that found Connecticut voters unhappy about Lieberman’s status as President George W. Bush’s favorite Democrat in the Senate also showed that Connecticut was one of the earliest and strongest antiwar states in the nation. The Lieberman camp was slow to react to the forewarnings, but young Democratic leaders around the state sensed that in Lamont they had a candidate with a message that would appeal to the growing poll of young voters who had never cast a primary ballot in their lives.
Several steps taken by the Quinnipiac Poll put us on the right track in successfully predicting the results of the primary. First, while critics say polls often miss the surge in new voters, our likely voter model tried to find them with questions seeking intention to vote, interest in the election, and interest in politics—and not, significantly, past voting.
Second, we sought to disprove the common misconception that a poll cannot predict a primary if it does not have an accurate prediction of the turnout. Since we knew that it would be impossible to predict turnout (an assumption justified when the old turnout record was shattered by eighteen points—something nobody predicted), instead we made it our goal to predict the primary. We had Lamont by six points the day before the election; he won by four.
Third, we were determined not to be one of the polls that end up being wrong because they stop too early and miss the last-minute shifts that almost always occur in the weekend before voting. As it happened, in this race movement occurred in the final week as Lieberman cut Lamont's lead from thirteen points to six in the Quinnipiac survey. This shift was confirmed by the CBS News exit poll, which showed that the late deciders broke significantly for Lieberman, reversing Lamont’s momentum.
Immediately after Lieberman's loss in the primary, Quinnipiac conducted a poll whose results may have influenced the general election campaign that followed. While pundits around the country were writing the senator’s political obituary, the poll showed that Lieberman, running as an independent, still had a comfortable lead over Lamont. This finding was a warning to Democratic leaders in Washington that while they were obligated to endorse Lamont as the party’s candidate, they should not campaign strongly against Lieberman. Bill and Hillary Clinton, for example, never came to Connecticut to campaign with Lamont. The Democrats wanted to win control of the Senate, and they weren’t about to alienate Lieberman enough that he might consider switching parties once he made it back there.
So, starting with the day he was defeated in the primary, Lieberman set out to be an exception to the old rule that there are no second acts in politics. His primary loss should have been devastating, with Lieberman down for the count. But despite Lamont’s pouring $15 million into his own campaign on a wall-to-wall television blitz, Lieberman's numbers didn't drop by as much as some expected them to, and he retained a double-digit lead that he would never relinquish. And while political pundits said that third-party candidates don't win election to the Senate, past experience had shown differently. Back in 1970, Jim Buckley, running as an independent in New York, won as a conservative, and Senator Harry Byrd, who had been a Democrat, won as an independent in Virginia. And, of course, it hadn't been that long since Connecticut elected an independent to statewide office: In 1990, Connecticut elected Lowell Weicker governor as an independent.
The analysts who said Lamont's primary victory would be tantamount to victory in the general election were also overlooking one important number: Only about 7 percent of all Connecticut voters cast ballots for Lamont in the Democratic primary. In addition, the Democratic primary electorate was more liberal than the entire statewide electorate, which includes Republicans and independents. Democratic primary voters were more likely to disagree with Lieberman on Iraq and more likely to vote against him just on that issue than general election voters. For most general election voters, Lieberman's experience outweighed Lamont's position on the war.
As a matter of fact, Lamont’s victory in the Connecticut primary would come to symbolize what our polls later found out—that voters were increasingly worried about whether he had the experience to be their senator. On primary night, voters statewide had an opportunity to see the candidates speaking on television, unfiltered by the media. For Lamont it was an opportunity to move from the left to become the candidate of the middle ground; but he became too wrapped up in his own moment of glory, and it cost him dearly. As he spoke, the divisive Al Sharpton stood right behind him. That night Lamont failed to make a good first impression with many independents and Republicans, and his momentum was lost. Lieberman, on the other hand, used his speech to go on the offensive and turn the race around.
After the primary, skeptics once again doubted that the polls were accurately reflecting what was going on. Lieberman's lead was remarkably stable throughout the fall campaign, hovering around 50 percent, while Lamont had peaked at 40 percent, and Alan Schlesinger, the Republican, was stuck in single digits. Some argued that the race would be much closer in the end because Lieberman would be hurt by his low position on the ballot. The theory was that with popular Republican governor Jodi Rell at the top of the ticket, some voters would simply forget to look for Lieberman and just vote for the name next to Rell’s, Alan Schlesinger. And, besides, a Republican candidate, regardless of how unknown and how poor a campaigner, would surely get more than 10 percent of the vote when running against two Democrats. Once again, the pundits were wrong. Quinnipiac final poll had Lieberman at 50 percent, Lamont at 38 percent, and Schlesinger at 8 percent; the final result was Lieberman 50 percent, Lamont 40 percent, and Schlesinger 10 percent.
Again dismissing conventional wisdom, the voters had disproved that there is a generic Republican vote. Once Schlesinger was viewed as a weak candidate with no chance to win, thousands of Republicans realized that party loyalty would mean helping Ned Lamont beat Lieberman. In the end, 70 percent of Republicans, as the polls predicted, voted for Lieberman, and Connecticut voters (as they had in the Weicker race) again displayed their already solid reputation for ticket-splitting. At the top of the ticket Republican Rell won the governorship overwhelmingly, while Lieberman took a majority of the votes running as an independent and Democrats won all the other statewide constitutional offices.
It should be noted that Schlesinger contributed considerably to his own terrible poll numbers. The first thing many voters learned about the GOP candidate was that he had gambled in the state’s Indian-run casinos under a false name—not an illegal act, but not one to endear an unknown candidate to the voters. It just got worse when Rell suggested he should reconsider his candidacy; then President Bush wouldn’t endorse him; and, finally, it was disclosed that he had run up large gambling debts.
Our preelection polls also showed what the exit poll confirmed: that the war in Iraq was not as big an issue in the general election as in the primary. Only 35 percent said it was the most important issue to their vote in the general election. And while Lamont won those voters, he lost those who said other issues were most important to their votes, from the economy to terrorism to health care.
Two other points were also confirmed—first, while most agreed with Lamont that the war was a mistake, most didn't agree that the United States should withdraw troops from Iraq right away; and, second, voters’ concern about the war was outweighed by their concerns about Lamont's experience. A majority said he didn't have the right kind of experience to be a U.S. Senator.
When all three candidates on election night pronounced the final Quinnipiac Poll wrong, they might have been motivated more by politics than any sure feeling about the numbers. All three said the race was closer than we were predicting, Lamont and Schlesinger probably because they wished it were true, and Lieberman because he was afraid his “big mo” would stall short of victory as his overconfident supporters stayed home on election day.
The election also gave pollsters and analysts a couple of lessons that need to be remembered in future Connecticut elections. First, Connecticut is a deep blue state when it comes to national politics. Nutmeggers are reliably Democratic in presidential elections but not so in state elections, and moderate Republicans can be very successful. Governor Jodi Rell followed the Arnold Schwarzenegger strategy of appealing to the center by supporting abortion rights, civil unions, stem cell research, and campaign finance reform. She was reelected by twenty-eight points. Connecticut hasn't elected a Democrat for governor since Bill O'Neill in 1986.
Second, although Lamont lost, many viewed his primary victory as helping energize antiwar voters, which ultimately helped the Democrats take back the House of Representatives. But while there is no doubt that the war in Iraq hurt Republicans, there was no surge in Democratic turnout or a significant increase in the percentage of Democrats voting Democratic, according to the exit polls. The reason the Democrats won House seats was because of the independents who voted overwhelmingly for them.
Douglas Schwartz is director of the Quinnipiac University Poll.