Presidential Politics: Possible Flaws in Public Opinion Poll
Although Rasmussen has defended its use of automated and online polling, citing a positive track record, some polling analysts are critical. Nate Silver of the New York Times found Rasmussen and Marist to be the most inaccurate polls of the 2010 gubernatorial and Senate elections, missing the final margins by 3.9 and 4 points, respectively. He identified 55 2010 Rasmussen polls in which margins were overestimated in favor of Republicans, and 12 in which the error favored Democrats. Silver referred to Rasmussen’s attitude toward polling convention as “cavalier,” citing details such as short interview windows and computer scripting as contributing to low response rates and inaccuracy.
Rasmussen praises its record, citing accurate predictions in the 2004 and 2008 Presidential elections. Frequent polling and screening likely voters are factors that Rasmussen claims differentiates it from other pollsters, since it can closely monitor trends among populations that are likely to turn out at the polls. Elections such as the 2008 presidential primaries, the 2010 special Senate election won by Scott Brown, and the 2010 New Jersey gubenatorial race are cited as being predicted correctly by Rasmussen before other pollsters. Rasmussen’s polling results have also been praised by several former presidential pollsters, and by political analyst Larry Sabato, director of the Center for Politics at the University of Virginia.
Romney Beating Obama? Interpreting the Poll Results
In December, 13 major polls featured a hypothetical matchup between Romney and Obama. Romney won 4 of these, 3 of which were Rasmussen polls. While this may indicate bias, it should also be noted that the other poll giving the win to Romney, delivered by Public Policy Polling, was released in the same time frame as an ABC poll that had the two candidates tied. Also, the most recent Rasmussen poll giving Romney a 6 point edge is the only poll so far to have been released following the holidays. Some observers may argue that this factor makes the poll less reliable, since respondent availability is often limited between Christmas and New Years. Others speculate that the end of the payroll tax debate may have halted a temporary approval boost for Obama, and that Romney’s much-publicized campaigning in Iowa may have benefited him. No matter the interpretation, the shift is difficult to compare with aggregate results, since other pollsters have not released data since December 18.
New York Times. Rasmussen Polls Were Biased and Inaccurate; Quinnipiac, SurveyUSA Performed Strongly. (2010). Accessed December 30, 2011.
RealClearPolitics. General Election: Romney vs. Obama. Accessed December 30, 2011.
Decoding Science. One article at a time.