The Most (and Least) Accurate Polls

As we head into the final stretch of the campaign, we're likely to be inundated by a spate of poll results from all kinds of different organizations, and if, well, everything is any indication, then different pollsters are likely to show very different things about the state of the race in key swing states. There may be no point in trying to make sense of mind-bogglingly contradictory survey results, but in case anyone is interested, I took a look at the polling done in the run-up to the 2010 midterms and constructed a little table on which pollsters came the closest to foretelling the election results in states that are now considered "battleground" territory and which missed the mark.

I measured polling accuracy by looking at the results of the last public poll of each race by each pollster and comparing the margin between the two highest-polling candidates (sometimes the only two candidates) to the eventual winner’s margin of victory over the second-place finisher. If there was more than one poll tied for most or least accurate by this metric, then I compared each candidate’s individual poll position in the survey results with their eventual share of the final vote. This is why PPP is listed as having the most accurate poll of the Florida governor’s race despite the fact that their final pre election poll had Alex Sink (D) leading Rick Scott (R), the eventual victor, 48% to 47%. This was rated "most accurate" notwithstanding Scott’s victory because it underestimated Scott’s eventual share of the vote (48.9%) by only 1.9 percentage points, while overestimating Sink’s vote share (47.7%) by only 0.3 points. The next-most accurate survey of this race was conducted by Rasmussen Reports, which had Scott leading by three points heading into Election Day.

Race	Most Accurate Poll	Least Accurate Poll
Ohio Gov.	Columbus Dispatch*	CNN/Time
Pennsylvania Gov.	Rasmussen Reports	Franklin & Marshall
Florida Gov.	PPP (D)	Suffolk University
Colorado Gov.	CNN/Time	Magellan Strategies (R)
Colorado Sen.	Denver Post/SurveyUSA	McClatchy/Marist
Florida Sen.	Rasmussen Reports	TCPalm.com/Zogby
Iowa Sen.	Des Moines Register	KCCI/Research 2000
Nevada Sen.	Suffolk University	Rasmussen Reports
New Hampshire Sen.	WMUR/UNH	ARG
North Carolina Sen.	PPP (D)	SurveyUSA
Ohio Sen.	PPP (D)*	Suffolk University
Pennsylvania Sen.	Susquehanna	McClatchy/Marist
Wisconsin Sen.	McClatchy/Marist Rasmussen Reports	PPP (D)
Wisconsin Gov.	McClatchy/Marist	Rasmussen Reports Reuters/Ipsos
Nevada Gov.	PPP (D) Suffolk University	CNN/Time
Iowa Gov.	KCCI/Research 2000	Rasmussen Reports
New Mexico Gov.	Daily Kos/PPP (D)	SurveyUSA
Michigan Gov.	EPIC-MRA	Rossman/Team TelCom
New Hampshire Gov.	WMUR/UNH	ARG

*Nailed it! (Poll results conformed exactly to election results.)

Of course, not all these races were as close as the presidential race is this year. The Senate races in Florida, Iowa and New Hampshire, and the gubernatorial elections in Colorado and Michigan, were all blowouts. Take that however you like. Also notice that the above table does not include any data from Virginia, which had neither a U.S. Senate race nor a gubernatorial election in 2010. In the 2009 race for governor there, SurveyUSA was right on the money; its final poll before Election Day showed Bob McDonnell (R) leading his Democratic opponent, Creigh Deeds, by eighteen points 58% to 40%. McDonnell won that election with 59% of the vote to 41% for Deeds. That was three years ago, however, so it may not be worth much.

Finally, there are a handful of states that are assuredly not up for grabs in the presidential election but are experiencing heated U.S. Senate and/or gubernatorial races. For those who are interested, here are my findings on polls of 2010 races in those states.

Race	Most Accurate Poll	Least Accurate Poll
Connecticut Governor	CT Capitol Report/MRG	Suffolk University*
U.S. Senate – Connecticut	PPP (D)	Rasmussen Reports
Maine­ Governor	PineTreePolitics/MECPO	Critical Insights
Massachusetts Governor	Suffolk/7News	Rasmussen Reports
U.S. Senate – Washington	FOX News/POR-Rasmussen	PPP (D)

Pennsylvania in Play for 2012?

Confusing Poll Numbers Released

Yesterday, political junkies were beset by a flood of polls in presidential and down-ballot elections, primary and general, and to say the results weren't consistent with one another would be a supreme understatement. I'll start with the big kahuna: the presidential race. Barack Obama and Mitt Romney have been locked in a dead heat in national polls for some time now, but we all know the popular vote doesn't matter. An NBC News/Marist poll found the president holding a narrow lead over Romney in the key swing states of Virginia, Ohio and Florida. In each state, 48% of respondents said they plan to vote for Obama; Romney stood at 44% in Virginia and Florida and 42% in Ohio. The Ohio results were confusing for a number of reasons; for one, 44% of respondents identified themselves as "conservative" or "very conservative", but only 21% said they considered themselves "liberal" or "very liberal". This is inconsistent not only with the presidential survey results but also Marist's poll of the Senate race in Ohio, which found leftist Sen. Sherrod Brown (D) leading State Treasurer Josh Mandel (R) by fourteen points, 51%-37%. That also clashes with other polls of the same Senate race, which show Mandel within single points of the incumbent.

The results out of Florida were also inconsistent with other recent polls; a Quinnipiac poll released Wednesday shows Romney with a six-point lead (47%-41%) over Obama in the Sunshine State. Given this disparity, it should not come as a surprise that Quinnipiac found Rep. Connie Mack IV (R) locked in a statistical dead heat with incumbent Sen. Bill Nelson (D), who garnered 41% support among the respondents to Mack's 42%, while Nelson led Mack, the prospective Republican nominee, by four points (46%-42%) in the Marist poll.

Perhaps the biggest single outlier in the NBC News/Marist survey concerned the Virginia Senate race, where NBC News finds former Gov. and DNC Chair Tim Kaine (D) holding a six-point lead over former Sen. George Allen (R), who held the seat from 2001 to 2007. Kaine polled at 49% in the Marist survey, higher than he's polled in any poll of this race I've seen. Earlier this month, a Washington Post poll showed Kaine and Allen tied at 46%, and a survey conducted last month by the Democrat-run Public Policy Polling (PPP) had the Democrat leading by just one point, 46%-45%. (It's worth pointing out that Marist's last survey of the race, which gave Kaine a nine-point lead over Allen, was an outlier too.) 

Staying on Senate races, Democrats may have reason to worry in yet another state that was supposed to be safe: Rasmussen Reports provided the latest news out of arguably the most undercovered Senate race this cycle. Sen. Bob Casey, Jr. (D-PA), who defeated Rick Santorum in 2006 in what must fairly be called a rout, was not supposed to face a tough fight for reelection. No Republican heavyweight stepped into the race to challenge him, but that didn't deprive the GOP of a spirited primary. Steve Welch, a dashingly handsome businessman who had switched his party registration to Democrat in 2008 and then switched back, was what you might call the establishment candidate; he had the endorsements of the state Republican Party, Gov. Tom Corbett and the state's major newspapers. The "TEA Party candidate", former State Rep. Sam Rohrer, had lost his last race (against Corbett in the 2010 gubernatorial primary), but this time he appeared to be the frontrunner, supported by multiple grassroots organizations and former presidential contenders Michele Bachmann and Herman Cain. But on election day, Pennsylvania Republicans picked neither Welch nor Rohrer, opting instead for Tom Smith, a wealthy businessman with a high school education who worked as a coal miner and farmer in western PA and spent much of his life as a Democrat. (His record in local government, supporting multiple tax increases, provided much fodder for his opponents.) Perhaps even more surprising than Smith's victory was the margin; he took nearly 40% of the vote in a five-way primary, nearly twice what the second-place finisher got. Then yesterday, Rasmussen published the results of a poll showing Smith within striking distance of Casey; he trails the incumbent by just seven points, 48%-41%. By way of comparison, a PPP survey released earlier this week had Casey leading Smith by sixteen points, 49%-33%. (Interestingly, every poll of a head-to-head matchup between Casey and Smith, including those conducted before the primary, has the Democrat polling at just under 50%; this kind of consistency is rare, even for a well-known incumbent. Smith's numbers have, understandably, been all over the place.)

Finally, there's the marquee matchup in my neck of the woods: the race to succeed Kay Bailey Hutchison in the U.S. Senate. Winning the Republican primary is tantamount to victory in the general election, and in my opinion, this race has been severely underpolled. The only outfit I know to be tracking this primary is PPP, and their latest poll raised some eyebrows. Lt. Gov. David Dewhurst (R) was widely regarded as the inevitable candidate, and big-name challengers who had initially thrown their hats into the ring later backed off. Still, many Republicans weren't content to just accept Dewhurst's coronation; former Dallas Mayor Tom Leppert and some jock named Craig James are also vying for the GOP nod next Tuesday. Dewhurst's main competition, though, has come from north of the border. Ted Cruz, a Canadian-born Cuban-American (not a typo) who served as state solicitor general from 2003 to 2008, has slowly but surely cultivated an enthusiastic base of support and amassed a vast campaign war chest. PPP has shown him gaining support with each successive poll, moving from 18% in January to 26% in April to 29% this week. Dewhurst's support, however, has increased as well: he's now at 46%, his best showing to date. If the lieutenant governor receives more than 50% of the primary vote, then he'll avoid a runoff.