By Kent R. Kroeger (Source: NuQum.com, December 12, 2017)
How do we reconcile the final three polls for the Alabama U.S. Senate race? One, conducted by Fox News, shows Democrat Doug Jones ahead by 10 percentage points (50 to 40 percent). Another, by Emerson College, shows Republican Roy Moore ahead by 9 points (53 to 44 percent). Finally, a poll by Monmouth University shows the race to be a virtual tie (both candidates at 46 percent).
They can’t all be correct.
All were conducted at about the same point in time (between Dec. 6th and Dec. 10th). So, we can’t attribute the differences to the times in which the polls were taken.
As to the questionnaires, the questions themselves are similar enough to rule out format differences.
We do see, however, major differences in the methodologies used by the three polls: particularly with respect to the mode of data collection and the sampling frames.
The Fox News poll was conducted by the research firms, Anderson Robbins Research and Shaw & Company Research. They used live telephone interviews from a random sample of 1,408 Alabama registered voters. (You can find their methodology here.)
In general, a registered voter population skews more Democratic when compared to ‘likely voters.’
The Emerson College poll used a more complicated (and less expensive) data collection method. They employed an Interactive Voice Response (IVR) system of landlines only and an online panel provided by Survey Sampling International (SSI). Their target population was of registered voters that were ‘very likely’ to vote. (You can find their methodology here.)
Based on NuQum.com’s experience with IVR (landline-only) samples, there tends to be a strong bias towards older, more conservative voters. In contrast, online panels tend to be younger and more liberal. In the Emerson College poll, the net bias of both data collection methods should presumably cancel out.
Finally, the Monmouth University poll was conducted by telephone from December 6th to 9th with 546 Alabama residents likely to vote in the December 2017 special election. Their sample was drawn from a list of registered voters who voted in at least one of the last four general or primary elections or have registered to vote since January 2016, and indicate they are likely to vote in the upcoming election. (You can find their methodology here.)
Unlike the Fox News poll, Monmouth filters their sample frame down to likely voters. Furthermore, by using a telephone (live interview) methodology, Monmouth does not contaminate its sample with the bias inherent in IVR (landline-only) samples.
So who do we believe?
Based on NuQum.com’s experience, the Monmouth Poll will come the closest to predicting tonight’s results in the Alabama Senatorial election.
It is going to be tight race.
We will see.
About the author: Kent Kroeger is a writer and statistical consultant with over 30 -years experience measuring and analyzing public opinion for public and private sector clients. He also spent ten years working for the U.S. Department of Defense’s Office of the Under Secretary of Defense for Personnel and Readiness and the Defense Intelligence Agency. He holds a B.S. degree in Journalism/Political Science from The University of Iowa, and an M.A. in Quantitative Methods from Columbia University (New York, NY). He lives in Ewing, New Jersey with his wife and son.