McAuliffe remains the heavy favorite to win in Virginia, but don't be shocked if his margin is closer than expected. The new poll tracking model finds two true toss-up races to watch tomorrow night. Shockingly, neither of them involves Chris Christie. This is HuffPollster for Monday, November 4, 2013.
Model is certain that McAuliffe leads... - As of this writing, the HuffPost Pollster chart for the Virginia governor's race, which uses a poll tracking statistical model designed by Stanford professor Simon Jackman to combine the results from all available public polls, currently gives McAuliffe a lead of just over 7 percentage points (45.4 to 38.0 percent) with Sarvis at 8.9 percent. The colored bands around each trend line in the Pollster chart illustrate the potential variation in the estimate of support for each candidate in the public polling. For example, the chart currently shows McAuliffe leading Cuccinelli, 45.2 to 38.2 percent. The model tells us that there is a 95 percent probability that the current polling snapshot falls between 43.5 and 47.4 percent for McAuliffe and between 36.3 and 39.8 percent for Cuccinelli. Accordingly, because these confidence bands do not overlap, we are extremely confident that McAuliffe is currently leading Cuccinelli.
...but how certain that McAuliffe will win? - There's a bit of a catch to the statistical certainty: It rests on the assumption that the public polls in Virginia have been collectively unbiased. In recent off-year elections, however, public polling averages have occasionally produced large errors. In Nevada in 2010, for example, every poll conducted in late October showed Republican Sharron Angle narrowly defeating Democrat Harry Reid, and the Pollster chart gave Angle a nearly 3 percentage point advantage (48.8 to 46.0 percent). Reid had the last laugh, defeating Angle by better than five points (50.3 to 44.6 percent). An error of the same magnitude in Virginia in Terry McAuliffe's favor would be enough to hide a Cuccinelli lead. Of course, the Nevada meltdown the only statewide race of 35 in 2010 or 2006 decided by ten percentage points or less with a polling error that big, so Cuccinelli remains an extreme long shot. At the same time, 8 of those 35 races produced errors four percentage points or better on the margin separating the leading candidates. So while a polling misfire big enough to conceal a hidden Cuccinelli lead is unlikely, a reasonable possibility exists that we could see a much than expected finish. [Nevada 2010 chart]
A photo-finish for Attorney General - The polling is far more uncertain about the race for Attorney General. The final PPP poll gives Democrat Mark Herring a 2-point edge over Republican Mark Obenshain (47 to 45 percent), but other polls released in the last week have produced different leaders, with results ranging from a 6-point Obenshain advantage (Hampton University) to an 11-point Herring landslide (Roanoke College). Our poll tracking model combines these results to show a near tie, Obenshain 43.1 percent, Herring 43.0 percent. The model's simulations tell us that the probability of Obenshain is basically a coin-toss (51.2 percent) -- this race is the definition of a toss-up. At the same time, the model estimates nearly 14 percent still undecided, so the potential for a big late shift in preferences is very real. [Virginia AG Chart]
A looming blow-out for Lt. Governor - On the other hand, the race for Lt. Governor between Democrat Ralph Northam and Republican E.W. Jackson is not at all close. Every poll conducted in October has shown Northam ahead, and the Pollster model gives the Democrat a 12 percentage point lead (48.2 to 36.2 as of this writing). [Virginia LG Chart]
A NAIL-BITER IN BOSTON - We have also applied the poll tracking model to the polls conducted in the Boston mayor's race. As of this writing, the model gives a half a percentage point edge to state Representative Martin Walsh over Councilor at Large John Connolly (44.7 to 44.2 percent), a margin nowhere near enough to be considered statistically significant. As with the Virginia AG's race, the probability of a Walsh lead is barely bigger than a coin flip (56 percent). Under ideal conditions, the model will reduce the "house effects" of organizations that demonstrate consistent deviation from what other pollsters have been showing. Polling has been relatively sparse in Boston, however, and we have included three surveys in the model sponsored by a pro-Connolly organization that have shown him ahead. The last two independent polls have given Walsh a single digit advantage. Our classic chart filtered to remove the polls with partisan sponsorship would give Walsh a slightly bigger advantage (46.4 to 42.7 percent) though it rests mostly on one final Boston Herald/Suffolk University poll. So in this case too, the Walsh edge would not be big enough to be considered statistically significant. [Pollster Boston chart; trend lines filtered to remove partisan polls]
A YAWNER IN NEW JERSEY - With just one day left until the election, the New Jersey governor's race remains stable, with Chris Christie leading Barbara Buono by double digits. At the time of writing, the HuffPollster model puts Christie ahead by about 23 points, 57 percent to 34 percent, with a 100 percent chance that he leads. Final polls, however, differ significantly on the size of Christie's lead, from 18 or 19 points (Kean/Pulse and Fairleigh Dickinson, respectively) to 36 points (Rutgers-Eagleton). [HuffPollster chart]
NJ polling also varies - RealClearPolitics' Sean Trende: "For all the talk about variance in the VA polling (McAuliffe up 15 or 2), the NJ polling is even crazier (Christie up 18 or 36). But all the NJ polls are showing a landslide for Christie, so [no one pays attention]. But [equal] to if we had polls showing Buono up 9 and Christie up 9" [@SeanTrende]
-Fairleigh Dickinson University pollster Dan Cassino: "@SeanTrende I think a lot of polls were spooked by the bizarre turnout patterns in the Booker special, and we're all recalibrating models." [@DanCassino]
-Monmouth University pollster Patrick Murray: "Party ID splits in NJ polls, Qpac=D+3, Eagleton=D+11, Monmouth=D+6, FDU=D+7. [Christie margin: Q=28, Eagl=36, Mon=20, FDU=19] ???" [@PollsterPatrick]
WENDY DAVIS TRAILS BY SINGLE DIGITS IN TEXAS GOVERNOR'S RACE - Ross Ramsey: "Attorney General Greg Abbott, the leading candidate for the Republican nomination for Texas governor, holds a single-digit lead over the likely Democratic nominee, state Sen. Wendy Davis of Fort Worth, according to the latest University of Texas/Texas Tribune Poll. In a head-to-head race, Abbott got 40 percent of registered voters to Davis’ 34 percent, with 25 percent of the voters undecided. In a three-way general election, he would get 40 percent, Davis would get 35 percent and Libertarian Kathie Glass would get 5 percent." [Texas Tribune]
HUFFPOLLSTER VIA EMAIL! - You can receive this daily update every weekday via email! Just enter your email address in the box on the upper right corner of this page, and click "sign up." That's all there is to it (and you can unsubscribe anytime).
MONDAY'S 'OUTLIERS' - Links to more news at the intersection of polling, politics and political data:
-A Battleground poll finds Americans "extraordinarily dispirited." [GWU]
-Older Americans reject raising the retirement age or reducing Social Security cost-of-living increases. [AP]
-Americans today are about as satisfied with the future as they were in the 1960s. [LA Times]
-Harry Enten shows that when Americans decide to "throw the bums" out, it's usually the incumbents of one party or the other that lose, not both. [Guardian]
-Perry Bacon predicts Chris Christie's cross-racial appeal won't translate to 2016. [The Grio]
-Dan Hopkins answers the shot at "The Nate Silver wing of the Internet" by Double Down reviewer Peter Hamby. [WaPost's The Monkey Cage]
-Twitter users are younger, more mobile and better educated than other Americans and their conversations do not necessarily track with public opinion. [Pew Research]