There seems a belief in some quarters that Republican Steve Lonegan has gained momentum against Democrat Cory Booker in the US Senate special election in New Jersey this Wednesday. Beyond the fact that a Tea Party candidate stands little chance of defeating a Democrat in a state that went for President Obama by nearly 18pt in the last election, the numbers really don’t suggest any trouble for Booker.
Booker holds a 12pt lead per the Real Clear Politics average, with a little over two days to go. I can’t think of a single campaign polled this extensively in the past decade, in a non-primary, major statewide election, that had a greater than 12pt error. That’s hundreds and hundreds of races. But this is a special election, you might say, where turnout is going to be low.
We can look back to the 2013 Senate special election race in Massachusetts, just a few months ago, for a similar example. Turnout in Massachusetts was low. The polls, however, remained accurate. Democrat Ed Markey led by 12.3pt in the Real Clear Politics average and went on to win by a little over 10pt.
It’s not that low turnout doesn’t increase the chance of a polling error; it’s that any error is not likely to be large enough to allow a Lonegan victory.
The reason is that any good pollster (pdf) is already accounting for the low turnout typical of an election taking place on a Wednesday in the middle of October. They are projecting who is going to vote and who isn’t. That’s potentially a part of the reason why Booker is leading by 12pt, instead of 15 or 20pt.
Isn’t there a chance the pollsters are way-off? But even when pollsters don’t do a good job modeling the electorate, 12pt is still too big a hill to climb. Consider the 2010 Nevada Senate race, where it was clear pollsters simply couldn’t figure out what was going on with Latino voters. The final polls had Harry Reid down by a little less than 3pt. He won by 5.5pt – an 8pt polling error. That’s still far short of 12pt.
What about pollster accuracy when a candidate is cutting the lead, as Lonegan has (from 16pt to 13pt, to 10pt, in the last three Monmouth surveys)? We can look to the case of Scott Brown in 2010. That was also a special election taking place at an odd time (the middle of January) and in which the Republican candidate in a blue state came up dramatically from behind.
The problem for Lonegan is that the late movement for Scott Brown had already occurred by this point. Every poll conducted by a legitimate pollster in the final week of that campaign had Brown leading. There hasn’t been a single poll in the 2013 New Jersey Senate race that has had anything but a Booker margin of at least 10pt.
The biggest obstacle for the Lonegan comeback story, though, is that any “momentum” he has gained isn’t coming at the expense of support for Booker. The last four Monmouth surveys have Booker holding steady at 53%, 54%, 53%, and 52% respectively. There is no statistically significant difference between these percentages. The last two Quinnipiac polls of likely voters also have Booker at 53%.
The only percentage that has changed is Lonegan’s, which is nice and all for him, but it doesn’t cut it when your opponent is over 50%. Even if Lonegan picked up every undecided voter (and my guess is many won’t vote), he would lose by half-a-dozen points. Chances are, however, that Steve Lonegan is not going to pick up every undecided voter.
The smart bet here is to average the Monmouth and Quinnipiac surveys to project an 11pt Booker win. That’s certainly disappointing to some Booker supporters, as is a campaign that has revealed Booker as more neoliberal and less accomplished than some of his supporters like to believe.
Yet, a win is a win. And Cory Booker is poised to win a spot as the next United States senator from New Jersey.