While not all the final numbers are in as this gets written, it looks like Cory Booker has won the Senate race by between 11 and 12 points. Our final poll, which ended last Sunday, had him up 22 points in our likely voter sample. That’s pretty far off. Monmouth and Quinnipiac were essentially right on the mark.
We’ll try to figure out why we missed it this far over the next weeks. After all, while we’d like to get it “right” we always stand by two points: 1. Polling is NOT a prediction, no matter how much we sometimes act and talk as if it is. It provides an estimate at the point in which the poll is done. It does not foretell the future. 2. We are an academic operation, and even when we get what seem to be out-of-the-norm results, we look for the learning opportunities for ourselves and for others in what we find.
What’s interesting to us is that our other numbers in the same poll make “sense” – for example, our favorability ratings for Booker are in the same range as other polls that have reported them. And as our releases over the next week will show, other numbers from the same sample also seem reasonable. So something is odd specifically in what we recorded for the Booker-Lonegan vote question. It doesn’t look like a technical problem; we’ve checked that. We recorded what people told us. And the sample doesn’t seem demographically odd as a whole. Given these two points, no matter how we thought about likely voters in our modeling, we kept showing Booker +20 or so.
On obvious problem, then, is that we failed to get a good “likely voter” screen as much as we tried. Our likely voter calculation did screen out more unlikely Booker supporters than it did unlikely Lonegan supporters, which suggested Lonegan’s supporters would turn out at a higher ratio than would Booker’s.
And in a very quick look, that is be exactly what happened, but even more than we expected.
Last year Republican Joe Kyrillos lost to Sen. Bob Menendez by a 17-point margin. Kyrillos won 7 counties, Menendez won 14 counties. This time around, Lonegan appears to have won the same 7 counties plus another two. More interestingly, Lonegan overperformed Kyrillos in 18 of 21 counties – that is, he did better as a percentage of the vote than did last year’s GOP candidate throughout the state.
Booker overperformed in exactly zero counties as this is written, instead getting the same percentage in 3 counties that Menendez did last year. So it does look like Lonegan supporters were much more likely to turn out than we were seeing in our data.
Another indicator is that in the 7 counties Kyrillos won last year, turnout was just under half of what it was in 2012 (about 46% of last year’s total at the moment).
But in the counties Menendez won last year (and Booker won all but two this year) turnout appears to only be about 40% of the number that voted last year. Again, Booker not only underperformed Menendez, but also had lower turnout in his winning counties.
Finally, there are real regional differences: Lonegan’s “best” over-performance came in:
Cumberland +13 points over Kyrillos’ results
Cape May +7
Except for Sussex and Warren, these are southern and shore counties.
Booker’s least worst counties were:
Essex +0 points versus Menendez’s results
Nothing South Jersey in this list, though Booker held even in a couple Republican counties.
This leads us to wonder if the fact that Booker is African-American played any role. If it did for us, why not for other pollsters? We have some thoughts on this point we will be investigating, and we’ll come back and talk about that once we’ve done so.
These are not final numbers at this point, so there may be a percentage point shift here and there. But one part of the story seems clear. We assumed better turnout in Democratic counties than actually occurred and we underestimated Republican turnout. And, Booker underperformed compared to Lonegan all over the state, if we use 2012 as a baseline, which we did not see coming in what people were telling us last week.
It might be worth noting that in most places a 11-12 point win is pretty much a landslide. But given the low initial expectations for Steve Lonegan, it probably seems disappointing to Booker forces. But give Lonegan lots of credit. he ran an impressive attention-getting campaign, for better or worse, and probably can count this as a moral victory if nothing else.
As for us, well we do have another election on which to focus!