Does Poll Sample Selection Make a Difference?
Gallup and Rasmussen both use a 3-day rolling average based on daily results collected from 1500 respondents, contacted from April 20, 2012 to April 22, 2012. All of Gallup’s respondents were contacted via live telephone calls to a pre-determined ratio of landlines and cell phones. 1,350 of the Rasmussen poll respondents were contacted via telephone; the other 150 were polled online. Although they were invited to participate in the poll at random, they had, at some point, volunteered to be part of an independent polling panel. Rasmussen also did not contact respondents via cell phone, citing its online polling as a way to reach households that do not use landlines. Although internet polling offers greater access to polls, it is sometimes criticized when used in major political polls, since there is some degree of voluntary selection in poll participation, even though respondents may not know which specific pollster is gathering results. Rasmussen also excludes exclusive cell phone users who do not sign up for online polling panels.
Obama Approval Poll Weighting Differences
Rasmussen weights its results for political party affiliation, while Gallup and most other major pollsters weight only for demographics. This practice is also debated, with some analysts arguing that it skews results based on a variable that is subject to frequent changes, and others arguing it can correct inadvertent oversampling of specific groups.
Comparing Polls: What do the Trends Say?
While several arguments can be made to favor Rasmussen’s method, and others can be made to favor Gallup’s method, the clearest way to determine polling accuracy is by looking at overall trends. Both results seem to stand out from other recent surveys, as most other polls conducted since mid-April have had very low spreads between approval and disapproval, ranging from -1 to 3. There was one other poll conducted recently that seems to back up Gallup’s results: a CBS News/New York Times poll conducted from April 13 to April 17 indicated a 6 point spread between “approve” and “disapprove,” the same spread as the recent Gallup poll.
Evaluating Political Polls
Although the Rasmussen poll isn’t consistent with other current polls, a greater number of recent polls are required to reach any conclusion with a level of certainty. One of the biggest challenges in political polling is that voter sentiment tends to shift faster than the amount of time required for several similar polls to be completed.
Gallup. Gallup Daily: Obama Job Approval. Accessed April 24, 2012.
Rasmussen Reports. Daily Presidential Tracking Poll. Accessed April 24, 2012.
RealClearPolitics. President Obama Job Approval. Accessed April 24, 2012.
Correction: A previously published version of this article stated that 1000 of the 1500 respondents who participated in Rasmussen’s Daily Tracking poll were contacted via internet polling. This was incorrect: 150 respondents were contacted via online polling. Information has also been updated regarding how the online polling panel from which Rasmussen surveyed was compiled.
Decoding Science. One article at a time.