Two polls released over the weekend suggest that Gov. Scott Walker of Wisconsin, a Republican, remains the clear favorite to win Tuesday’s recall election.
Although the contest is fairly close, polls of gubernatorial races are ordinarily quite reliable in the late stages of a race. We have not officially released a forecast for the race, but Mr. Walker’s lead of about six points would translate into almost a 95 percent chance of victory if we used the same formula we did to evaluate gubernatorial races in 2010, which derives its estimates from the historical accuracy of gubernatorial polls over the past 15 years.
One of the new polls over the weekend, from Public Policy Polling, which conducts polling on behalf of Democratic clients as well as publishes its own polls independently, showed a somewhat tighter race, with Mr. Walker’s Democratic opponent, Mayor Tom Barrett of Milwaukee, having closed his deficit to three percentage points. However, the firm has showed somewhat more favorable results for Mr. Barrett than other polling firms, and this reflected a relatively minor change from the firm’s previous poll, which had Mr. Walker ahead by five percentage points.
At the same time, the Public Policy Polling survey had Mr. Walker at 50 percent of the vote and had very few undecided voters. The presence of undecided voters tends to correlate with higher unpredictability on Election Day, while the absence of them, as in this case, means that even a small lead is more likely to hold up.
Another poll from We Ask America, which is a subsidiary of the Illinois Manufacturers’ Association and often shows Republican-leaning results, had a larger lead, 12 points, for Mr. Walker. There had been no change from the firm’s prior poll, which also had Mr. Walker 12 points ahead.
Democrats have made various claims that Mr. Barrett might overperform his polls – citing, for instance, what they say is a superior turnout operation.
These claims need to be weighed against the long actuarial odds of a candidate who is down by this amount in the consensus of polls coming back to win at the last minute.
Campaigns that are down by about this margin in the polls often say that there is some dynamic that the polls are not capturing. Sometimes they are making reasonable arguments, and sometimes they are just spinning. But either way, these factors are rarely enough to allow the candidate to overcome the deficit. The exceptional cases are often remembered precisely because they are rare events.
It could be that Mr. Barrett does overperform his polls, but not by enough to win. A benchmark for a superior turnout operation is that it might typically be worth an additional two or three points – fewer than the six points by which he now trails Mr. Walker in the average of surveys.
Nor do the polls seem to be suggesting that Mr. Walker will win because of lopsided turnout. Instead, the same polls that show Mr. Walker with a six-point lead also show President Obama ahead by about the same margin in his matchup against Mitt Romney.
This suggests that it would be dubious to come to too many conclusions about what Tuesday’s outcome could mean for November. A fair number of Mr. Walker’s voters also seem prepared to vote for Mr. Obama. But this also implies that the polling results are not simply an artifact of skewed sample designs. Swing voters in Wisconsin seem to have made an affirmative choice on Mr. Walker’s behalf.
From a macroscopic view, the mechanics of why Mr. Walker is likely to prevail are not that hard to discern. The results of another recall election last August, in which Democrats succeeded in recalling two Wisconsin state senators but failed in efforts to oust four others, had served as something of a referendum on Mr. Walker. My interpretation of the results was that they implied that opinion in the state was about evenly divided on Mr. Walker at the time in terms of how it translated into actual votes.
Since then, however, Mr. Walker’s performance ratings have improved, with his approval rating exceeding his disapproval rating in most surveys. It is difficult for an incumbent to lose with a net-positive approval rating under any circumstances, and it is probably more so in the case of a recall election, when some voters might give Mr. Walker the benefit of the doubt to allow him to serve out his term. (Mr. Walker, if he wins on Tuesday, would be up for a vote again in 2014 when his original term expires.)
With that said, recall elections are rare events, and it is plausible that the true margin of error in polls of recall elections is intrinsically higher than in regular contests. The results are worth watching, but it would be a true upset if Mr. Barrett were to prevail.