03/27/2014 05:33 pm ET Updated Mar 27, 2014

HUFFPOLLSTER: One Pollster Reviews Why It Missed The Mark In 2013


A polling misfire in New Jersey renews a lesson from Pre-Election Polling 101. We're thrilled to announce the latest addition to the HuffPost Pollster team. And there's a side to "lefty" voting bias you've probably never considered. This is HuffPollster for Thursday, March 27, 2014.

WHY RUTGERS-EAGLETON GOT IT WRONG - Matt Friedman: "The Eagleton Center for Public Interest Polling at Rutgers University — whose polls on the 2013 race for governor and special election for U.S. Senate were far off the mark — today released a study it commissioned to find out why its surveys were so wrong. 'We have a special obligation to take our educational mission seriously, which includes informing the public as well as learning from our own errors,' said David Redlawsk, director of the center.....According to the report, written by Langer Research Associates of New York, Rutgers-Eagleton flubbed by asking respondents a series of questions about the candidates before finally asking who they would vote for. In effect, that 'primed' respondents to think more positively about candidates, and made them more likely to choose them when they finally arrived at the 'head-to-head' question….Going forward, Redlawsk said, his organization will either stop asking the head-to-head questions or ask them in separate polls. Simply reversing the questions' order, he said, could bias the responses to the other questions." [Newark Star-Ledger]

Questions 'primed' support - Gary Langer and Greg Holyk on the questions that appeared to "prime" support for Christie and Lonegan: "In a departure from best practices, vote-preference questions were preceded by questions on the leading candidates’ personal favorability and 'feeling thermometer' ratings, and, in the case of Christie, job performance ratings." These included specific issue ratings for Christie. Perhaps most problematic, these items included Christie's Hurricane Sandy Recovery job rating and a "most important problem" question, that prompted for eight specific issues including, again, Hurricane Sandy Recovery. For the other polls, these kinds of questions came after the vote. The result? Langer & Holyk: "Results by partisan groups indicate that the November Rutgers-Eagleton poll overestimated Christie’s support, and underestimated Buono’s, among Democrats and independents, compared with the Quinnipiac, Monmouth and FDU polls. The polls were more closely aligned on Christie’s near-unanimous support among Republicans. [Langer report]

Why didn't anyone notice? - Except for favorable ratings of Christie and Buono, the Rutgers November 4 release made no mention of 16 relevant questions that preceded the gubernatorial vote question. The questions that Rutgers did choose to list in the "Questions and Tables" section of their release omitted question numbering, concealing the existence of these key questions. [Rutgers release from Nov. 4, 2013]

VOTERS UNHAPPY WITH UKRAINE RESPONSE - Dana Blanton: "By a widening margin, more voters think the United States is weaker since Barack Obama became president. And the highest number in a decade feels the country is less safe than it was before 9/11, according to a Fox News poll released Wednesday. The poll also finds most voters think President Obama has not been tough enough on Russia, although a majority still says the U.S. should stay out of the situation in Ukraine….So far the U.S. has responded to Russia’s annexation of Crimea with economic sanctions. Voters are divided on this action: while 41 percent think it’s 'too weak,' almost as many -- 40 percent -- say it’s 'about right.' Just seven percent see it as 'too strong.' While 35 percent of voters say the U.S. should be more involved in Ukraine, a 53-percent majority says the U.S. shouldn’t. At the same time, many more think military force will be required to stop Russia from taking control of Crimea (50 percent), than think diplomacy and sanctions alone will work (30 percent)." [Fox]

WARNER LEADS GILLESPIE IN VIRGINIA SENATE RACE - Quinnipiac: "Virginia, considered a swing state, is swinging blue for U.S. Sen. Mark Warner in his reelection drive this year and for former Secretary of State Hillary Clinton if she runs for the White House in 2016, according to a Quinnipiac University poll released today. Sen. Warner tops former White House aide Ed Gillespie, a possible Republican challenger, 46 - 31 percent, with 6 percent for Libertarian candidate Robert Sarvis, the independent Quinnipiac University poll finds. Virginia voters say 49 - 36 percent that Warner deserves reelection. They approve 55 - 33 percent of the job he is doing and have a 49 - 30 percent favorable opinion of him. For the other candidates, 64 percent don't know enough about Gillespie to form an opinion and 80 percent don't know enough about Sarvis." [Quinnipiac]

Gillespie pollster Glen Bolger (R) responds [POS]

ANNOUNCING A NEW MEMBER OF THE POLLSTER TEAM! - Natalie Jackson will be joining the Huffington Post as our Senior Data Scientist. In that position she will work with the Pollster team to refine and expand our poll tracking models and political campaign forecasts. Natalie received a PhD in political science from the University of Oklahoma and has advanced training in statistical methods. She comes to us after working as a Senior Analyst for the Marist Institute, where she conducted research on polling and survey methods and helped oversee Marist's methodological processes. Prior to that, she did postdoctoral work at Duke University as part of their Initiative on Survey Methodology. She also taught statistics and data analysis to students at Marist College and at the summer program of the Inter-University Consortium for Political and Social Research (ICPSR). Natalie's full-time presence on the Pollster team will help us to improve and expand HuffPost's polling aggregation and modeling. At a time when statistical models used to forecast elections have gained great prominence, she will help Pollster leverage our unique knowledge of and experience with pre-election polls.

HUFFPOLLSTER VIA EMAIL! - You can receive this daily update every weekday via email! Just click here, enter your email address, and and click "sign up." That's all there is to it (and you can unsubscribe anytime).

THURSDAY'S 'OUTLIERS' - Links to the best of news at the intersection of polling, politics and political data:

-Russia's unfavorable rating soars with Americans. [Gallup]

-Rasmussen finds Iowa's Bruce Braley (D) leading his GOP opponents for Senate in a poll taken mostly before a controversial recording of Braley was released. [Rasmussen]

-An online St. Leo University survey finds Florida Gov. Rick Scott (R) closing the gap against challenger Charlie Crist. [Miami Herald]

-A California poll finds Gov. Jerry Brown (D) the favorite for reelection. [PPIC]

-Most New Mexicans like Susana Martinez but don't think she should run for president. [PPP (D)]

-A GOP poll finds David Perdue leading the Republican Senate field in Georgia. [Atlanta Journal-Constitution]

-Obamacare will never be popular, Jonathan Bernstein says. [Bloomberg]

-Geoffrey Skelley predicts Democrats' courting of young voters could cause them turnout problems in the short term. [Sabato's Crystal Ball]

-Americans think the Cold War may be returning. [Gallup]

-Celinda Lake (D) takes on Fox News' Steve Doocy. [MediaMatters]

-Paul Krugman offers praise for Charles Gaba's ACASignups.net (and more criticism of Nate Silver's FiveThirtyEight). [NYTimes]

-Nate Silver slings parody at Paul Krugman. [538]

-Kieran Healy finds some variables omitted from Silver's Krugman model. [kieranhealy.org]

-Neil Paine explains why random chance makes the perfect baseball forecast impossible. [538]

-Most Americans think George Washington could beat up Abraham Lincoln. [YouGov]

-Left handed people are more likely to vote for candidates on the left-hand side of the ballot. [National Journal]