TRENTON A respected polling institute offered an explanation Wednesday for why it was way off the mark predicting the margin of victory for two key New Jersey elections last fall: Chris Christie for governor and Cory Booker for U.S. Senate.
Survey participants were influenced by the order in which questions were asked, according to a report released Wednesday by Langer Research Associates, which was commissioned by the Eagleton Institute of Politics at Rutgers University to study the error of its ways.
In its final pre-election surveys, the institute had Christie winning by 36 percentage points (he actually won by 22) and Booker beating his opponent by 22 points (he won by 11).
In both races, Rutgers-Eagleton asked respondents questions about personal favorability and other subjects before asking whom they planned to vote for.
The psychological effect, known as priming, suggests "some respondents who reported favorable views of Christie and Booker were influenced by those positive attitudes when subsequently asked their vote preference, thereby overstating their inclination to support these candidates," the report says.
Likewise, if poll respondents reported unfavorable views of the opposing candidates, they may have understated their support, according to the report.
In addition, the report says, the theory of cognitive consistency suggests respondents who report a favorable view of a candidate are less likely to then report a negative one, such as not voting for him.
Democratic respondents were more likely to be affected by the questionnaire design when asked about Christie, while Republicans were more likely influenced when asked about Booker.
"Asking about these candidates' personal popularity and professional achievements thus may have had a particular priming effect on respondents not affiliated with the candidates' political parties," the report says.
Christie, a Republican who was running for reelection, won 60 percent of the vote, defeating his opponent, Democrat Barbara Buono, who had 38 percent. In its final pre-election poll, Rutgers-Eagleton projected a 36-point victory for Christie - overestimating support for Christie by 6 percentage points and underestimating support for Buono by 8.
In that poll, respondents were asked about Christie's job performance, Hurricane Sandy recovery, taxes, and other topics before vote preference. More Democrats reported support for Christie in that poll (38 percent) than in Quinnipiac (30 percent) and Monmouth University (23 percent) surveys.
Similarly, fewer Democrats voiced support for Buono in the Rutgers-Eagleton survey than they did in other polls.
In the special election for U.S. Senate, Booker, a Democrat and former Newark mayor, defeated Republican Steve Lonegan 55 percent to 44 percent. In its final pre-election poll, Rutgers-Eagleton gave Booker an edge of 58 percent to 36 percent.
That overstated support for Booker by an insignificant 3 percentage points, the report said, and understated support for Lonegan by 8 points.
Voters reported being relatively unfamiliar with Lonegan - 36 percent said they had no opinion of his favorability. Republicans supported Lonegan by 13 fewer percentage points in the Rutgers-Eagleton poll than they did in the Quinnipiac survey, and by 12 points in the Monmouth one.
David Redlawsk, director of Eagleton's Center on Public Interest Polling, said Wednesday that the institute had asked questions about Christie at the beginning of its poll since Christie was first inaugurated. "[W]e had concerns that moving these questions after a head-to-head vote question would bias those results for the same reason we ended up biasing the vote questions," Redlawsk said in a statement.
He said Rutgers-Eagleton should have asked the voter-preference question in a separate survey or focused only on its longer-term work.