In the final Emerson College poll before the election, GOP candidate Roy Moore now leads the Democratic candidate Doug Jones 53% to 44%, a six point bump from last week’s Emerson poll. The poll suggests Moore has weathered the storm of alleged sexual misconduct; The survey has a sample size of 600 very likely voters, and has a margin of error of +/-3.9 percentage points. Since the Emerson Poll of Nov. 12, a few days after allegations of sexual misconduct, Moore’s lead dropped from 10 points, to 6 points Nov. 26, and to 3 points last week on Dec. 3. A major event that might have contributed to Moore’s improved poll numbers is his endorsement by President Trump this past weekend. The President is more popular than either candidate with a 55%/40% favorable/unfavorable rating; Moore is at 45%/45% and Jones at 43%/45%.

The Alabama Special Election Poll is the third in a series of mixed mode experiments using online panels and automated polls (IVR) to test the reliability and validity of different forms of data collection, by comparing the poll’s margin of victory with actual election results. The previous two studies yielded mixed results on which form of data collection is the most reliable and valid. In the New Jersey Governor’s race, the data favored IVR polling to be the most valid and reliable, while in the Virginia Governor race all three modes of data collection performed equally as well. The full results of the three part study will be released after the Alabama Special Election vote total is certified by the State of Alabama.

The Emerson College Polling Society is a charter member of the Transparency Initiative of the American Association of Public Opinion Research (AAPOR).

The Alabama Emerson College poll was conducted December 7-9, 2017 under the Supervision of Professor Spencer Kimball. The sample consisted of only very likely registered voters, n=600, with a margin of error (MOE) of +/- 3.9 percentage points. The Alabama data was weighted by region, 2016 vote, and mode of data collection. It is important to remember that subsets based on gender, age, party affiliation, and education carry with them higher margins of error, as the sample size is reduced. Data was collected using both an Interactive Voice Response (IVR) system of landlines only and an online panel provided by Survey Sampling International (SSI).

Transparency Initiative

Results