CLICK HERE FOR THE TEXT OF THE 10/14 BOOKER-LONEGAN RELEASE FROM THE RUTGERS-EAGLETON POLL. Alternatively, the full text of the release follows a somewhat lengthy speculation on why this poll differs from a brand new Monmouth poll showing a much smaller margin for Booker.
The following analysis is by Rutgers-Eagleton Poll Director David Redlawsk [EDITS as noted 6:40pm EDT, 10/14/13].
Well, these are the times that pollsters hate. I think I know how the Gallup people must have felt when their 2012 vote models showed Romney ahead in the waning days of last year’s election, something that was clearly not true.
Why, you ask? Because today’s Rutgers-Eagleton Poll puts Cory Booker up 22-points on Steve Lonegan, a far larger margin than any other recent poll, and, in particular well above Monmouth’s 10-point lead for Booker, also out today. Monmouth’s polling dates fall within the dates we polled as well – though we were in the field longer. However, looking at our daily numbers we don’t think the longer fielding time is the issue, though we do see movement more toward than away from Booker over the last week, especially on our last day of polling, Oct 13, the day after Monmouth ended their poll. In fact, of our week of polling, the two days with the best Lonegan performance are the 10th and 12th, two of the three days Monmouth polled.
Still, I am puzzled, to be quite honest. I understood our large margin for Booker right after the primary – at that time Lonegan had gotten nothing but bad press and Booker had just come off a huge primary victory. But here it is harder to see what’s happening.
What’s especially odd from our perspective is that numbers other than the vote – for example, favorability – are very similar between the two polls.
So, anyway, which is it: Booker up 10, or Booker up 22?
We’ve looked at our data over and over. We’ve looked at other indicators besides the Booker-Lonegan race and those numbers seem quite reasonable, as we’ll report later. We looked at our weighted partisan breakdown for likely voters, and that’s not substantially different either. What is different between us and Monmouth today? We see a few key areas:
No, we don’t mean the Lonegan campaign’s attacks on Booker’s masculinity. That’s irrelevant. But the difference between how men and women plan to vote does matter.
Monmouth shows essentially NO gender gap – in their poll 53% of men and 51% of women support Booker. We show a huge gender gap: 51% of men support Booker (basically the same as Monmouth) but we have 67% of women on his side (versus only 27% voting for Lonegan.) This alone could account for much of the difference. Interestingly, other polls have also shown a gender gap, most notably a recent Quinnipiac poll which had women going for Booker 62%-31%, although they showed men supporting Lonegan 51%-44% .
We have independent voters supporting Booker over Lonegan by a margin of 49% to 41%. But Monmouth’s poll has it the opposite: independents are 48% Lonegan and 43% Booker. we also have more Democrats (96%) supporting Booker and fewer Republicans (74%) supporting Lonegan. For Monmouth the number are 90% and 86% respectively.
Is it the Old Folks?
In our poll, voters 65 and over favor Booker 53%-43% over Lonegan. Monmouth does not report an age crosstabulation so we do not know how different we are. However, we do know that Monmouth has a lot more senior citizens in its Likely Voter sample, at 36% of all voters, where we have just 30% of likely voters 65 and over. Since older people are more likely to support Lonegan, that could be a factor.
METHODOLOGY! [EDIT, 6:40PM, 10/13: Huffpost Pollster talks about this point this evening…]
So here’s an interesting issue – Monmouth is using IVR for part of their polling. IVR stands for Interactive Voice Response – or in other words, robo-polling. In this method, computers dial landline numbers and respondents answer computer generated questions by keying responses on their phone keypad. With IVR the pollster has no idea who they are actually talking to – it could be anyone who picks up the phone. There is no human interaction. For the most part people who respond to IVR polls are often quite different from those who respond to live callers. Monmouth knows this so they supplement the IVR with live caller cellphones and some live landlines. But more than half their sample is from IVR. It would be very interesting to see what the numbers say just for the IVR sample, compared to their live callers. However, they do not report this. Maybe they see no difference. [EDIT: Patrick Murray at Monmouth tells Huffpost that they did in fact see no difference. But since that was NOT disclosed in their press release or their methodology statement, it was impossible to know that when writing this post. I will note we don’t actually know what “no difference” means. Is in no difference in the marginals? Or no difference in the makeup of the samples? Pre or post-weighting?]
My guess is that IVR brings the older sample, and also results in the lack of a gender gap. But that’s only a guess right now. [EDIT, 6:40PM, 10/14: And apparently Patrick Murray suggests the same gender result in both IVR and live landlines, but again he isn’t completely clear on this.]
This leads to one other interesting possibility. With IVR respondents don’t have to talk to anyone. This lessens what we call “social desireabilty” which has been shown to matter when race and ethnicity are a factor in preferences. People know it is “wrong” to express overtly racist attitudes. Likewise, there has been evidence in the past that voters may over report support for a non-white candidate in a live-caller poll. IVR does not have this problem – people can be honest without anyone knowing except the computer. Is it a factor here? I’m really not sure how deeply race comes into this election. If it does, then we may well get more supporters for Booker in our live calling than he will get in the privacy of the voting booth.
Finally, it’s also about likely voters and sampling
Both Rutgers-Eagleton and Monmouth try to identify likely voters. From what Monmouth says in their release, they used a listed registered voter sample and consider people who voted in two of the last four general elections to be likely voters. If they asked any other questions to determine this, they did not say in their release so we assume they did not. [EDIT 6:40PM, 10/14: Apparently, according to Huffpost, Monmouth did ask two additional screening questions. But again this was not made clear in the Monmouth release or methodology statement.]
We use Random Digit Dialing (RDD), not a listed sample. The disadvantage is we do not know voting history, and we have to screen for registered voters – and people may lie about this. The advantage is that we can hear people talk, and can get to the right person. We also theoretically have better coverage in that many people do not provide phone numbers when they register or provide incorrect numbers. With RDD theoretically anyone can get a call.
We ask a series of questions to determine likely voters – including awareness of the election date, attention to the campaign, a direct “will you vote question”, and a vote history question. Combining these gives us a model for likely voters that has fewer older people, and slightly more white people, but about the same share of each party as Monmouth. So the likely voter screen does not seem to be a huge source of the difference, at least in terms of partisan breakdown.
[EDIT 6:40PM, 10/14: We also checked different levels of screening that were more or less strict, based on different cutoffs on our questions. The results did not vary by more than a couple points.]
The upshot is that we report numbers quite different from other recent polls. Are we certain about them? Of course not – there are many reasons we could be way off. But we could also be within the ballpark. In the end, every poll is an estimate and some will be on the mark and some will be off. Take each one with a grain of salt. Looking across all polls, if I had to guess (as opposed to poll), I think Booker will win with a margin in the mid-teens.
Still, a lot depends on who remembers to vote on Wednesday, October 16 and how effective the campaigns are a getting out their supporters. If we’re wrong, we’ll try to figure out what happened. If we’re right, well, you heard it here first!
Finally, we could have simply not released this poll, but we might as well put it out there and see if others can see a problem that we are missing.
RELEASE TEXT FOLLOWS:
RUTGERS-EAGLETON POLL: BOOKER HOLDS LARGER THAN EXPECTED LEAD OVER LONEGAN ON EVE OF SPECIAL SENATE ELECTION
NEW BRUNSWICK, N.J – With just two days to go until the Oct. 16 special U.S. Senate election, Newark Mayor Cory Booker holds a 58 percent to 36 percent lead over former Republican Bogota Mayor Steve Lonegan among likely voters, a new Rutgers-Eagleton Poll finds. Another 3 percent prefer someone else, and about 3 percent remain undecided. While the Democrat’s lead is less than he enjoyed following the August primary, compared to other polls this latest Rutgers-Eagleton poll suggests Booker has recently gained in his bid to succeed the late Frank Lautenberg.
The debates between Booker and Lonegan, which highlighted the stark policy differences between them, may have played an important role in Booker’s current performance. While 43 percent of all likely voters paid some attention to a debate, independent voters who followed the debates strongly favor Booker, 59 percent to 37 percent. But independents ignoring the debates prefer Lonegan by a 45 percent to 42 percent margin.
For partisans, attention to the debate reinforced support for their own party’s candidate. At this point, Booker does much better than Lonegan within his party: 96 percent of likely Democrats are in Booker’s corner, while Lonegan wins only 74 percent of Republicans. GOP backers are less likely to have followed the debates: 36 percent compared to 49 percent of Democrats and 43 percent of likely independent voters.
“Other recent polls showed a narrower lead for Booker, but voters we talked to seem to have moved back in his direction,” said David Redlawsk, director of the Rutgers-Eagleton Poll and professor of political science at Rutgers. “The debates presented a stark picture of the differences between the candidates, which appears to have led independents to prefer Booker. While both candidates came out swinging, Lonegan’s attacks bordered on the taboo, with his ‘floating dead bodies’ and ‘big black hole’ comments about Newark – reminiscent of the rhetoric that got him into hot water early in the campaign. Independents in particular dislike these kinds of attack.”
Results are from a sample of 513 likely voters with a margin of error of +/- 4.3 percentage points. These likely voters are part of a sample of 798 registered voters polled statewide using live callers to both landline and cell phone households from Oct. 7-13.
Voters paying some attention
The uniqueness of the special Senate election makes estimates of turnout difficult and suggests that the final result will be unusually dependent on the candidates’ abilities to motivate their supporters. There is some evidence that voters are paying attention. More than half of registered voters know there is an election in October, and 45 percent know it will be held Oct. 16. Just over half are following the election at least fairly closely, with about one in five paying very close attention. Sixty percent of registered voters claim they are very likely to vote Wednesday.
Booker holds a commanding lead with minority voters, urbanites and voters in the South Jersey/Philadelphia region. Lonegan leads among voters in shore counties, 46 percent to 44 percent, while losing northwestern exurban voters, 51 percent to 41 percent, a better performance than his statewide margin.
“Here’s the key: if the Booker campaign can turn out urban voters and Democrats who say they will vote, he will be in the range we estimate,” said Redlawsk. “If Lonegan’s turnout operation is stronger and Booker’s base stays home thinking it is all wrapped up, then all bets are off.”
Booker ratings down slightly on Lonegan attacks
Lonegan’s attacks throughout the campaign appear to have had some impact. While Booker receives positive ratings from 54 percent of likely voters, this is a nine-point drop from an early September Rutgers-Eagleton Poll. Concurrently, his unfavorability has almost doubled to 32 percent. Still, asked to rate the Democrat on a scale of 0 to 100 degrees, voters remain warm toward Booker with an average score of 56 degrees. The overriding emotional response to Booker is “hope,” with 48 percent saying his campaign makes them feel hopeful and 33 percent proud. In contrast, only 12 percent are angered and 19 percent worried by Booker.
Lonegan, however, continues to leave either a negative impression or none at all on most likely voters. While favorable impressions of the Republican have risen eight points to 30 percent, the share of voters who hold unfavorable impressions climbed to 34 percent. Over one third have no opinion on the GOP candidate. In general, likely voters are quite cool toward Lonegan; he averages only 41 degrees on the poll’s thermometer. Voters feel more negative about Lonegan than Booker: 37 percent are worried, and 30 percent angered by his campaign. Thirty percent are hopeful and 16 percent are proud of the candidate.
Many voters see Booker as a self-promoter
Booker is not without weaknesses. A third of likely voters say Booker is more about self-promotion and that life in Newark has improved little under this watch. While Lonegan has gained some support though this line of attack, a plurality (47 percent) continues to believe Booker has made real improvement in Newark.
Booker’s position on issues is seen as “about right” for 42 percent of likely voters, but 44 percent think he is too liberal, even for blue New Jersey. Lonegan, however, is thought too conservative by a large majority of likely voters; 60 percent say he is further to the right than most of the state, while only 18 percent say his views are right on par with the preferences of New Jerseyans.
“By all measures, voters see Booker as somewhat more in tune with New Jersey and see Lonegan as less so,” said Redlawsk. “Although Lonegan has tarnished Booker’s image, Booker remains the clear preference of voters, generating warmer feelings and a strongly positive impression, and taking positions more in line with what voters want.”
The party base, women and independents support Booker
Booker’s advantage over Lonegan is driven by overwhelming support from his own party base and a winning margin with independents. Ninety-six percent of likely Democratic voters back Booker, compared to only 74 percent of likely Republican voters who support Lonegan. Independents also are in Booker’s corner, 49 percent to 41 percent for Lonegan. Newark’s mayor captures 16 percent of the Republican vote, while Lonegan peels off only 2 percent of Democrats.
Booker leads across virtually all demographic groups. While both men and women go for Booker, women are 13 points more likely than men to say they will vote for the Democrat and 18 points less likely than men to say they will vote for his opponent, evidencing a wide gender gap.
Booker also gets support from almost 9 in 10 likely voters who say he has made a real difference in Newark, but among those who think Booker is too much about his own self-promotion, only one in five say they will vote for him.