West Long Branch, NJ
The Monmouth University Poll accurately described the potential outcome in the Alabama Senate race, both in terms of the margin of victory and in the level of turnout. Monmouth’s midpoint model showed a razor thin race that Democrat Doug Jones eventually won by 1.5 percentage points.
This unique special election involved a high degree of uncertainty and Monmouth used this opportunity to provide a realistic range of outcomes. Different turnout models were based both on individual voting history as recorded in the voter rolls and self-reported interest and enthusiasm in this election. Monmouth’s high turnout model (about 55-60% of registered voters) with a light screen based on presidential-electorate demographics showed Jones leading Republican Roy Moore by 3 points. A lower turnout model (about 30-35%) based on typical midterm demographics, including only voters who participated in at least two recent elections or expressed a very high level of interest, had Moore up by 4 points.
Monmouth also created an adjusted midterm model based on patterns seen in recent special elections as well as last month’s Virginia gubernatorial contest. This model projected a slight increase in typical midterm turnout (about 35-40%) driven by Democratic voters in Democratic areas of the state.
This model assumed that, regardless of overall turnout, Democratic strongholds would command a larger than normal share of the electorate. For example, in last month’s Virginia election, the region Monmouth defined as Northern Virginia accounted for 31% of the total vote whereas this area would normally contribute about 28-29% of the final tally, with nearly all that increase coming from Democratic voters. The model based on this turnout pattern produced a tied outcome for the Alabama race.
In the actual results, overall turnout came in at about 45% of registered voters, with relatively higher turnout among Democratic voters in Democratic parts of the state. For instance, Jefferson County – home to Birmingham, the state’s largest city – comprised 16% of the final electorate whereas it usually contributes 14% of the total vote. This result put the actual turnout somewhere between Monmouth’s adjusted midterm model and high turnout model. The final margin of victory – Jones by 1.5 points – was also midway between the estimates provided by these two models.
“The 2016 presidential contest as well as the Virginia gubernatorial race last month showed that slight deviations from typical turnout can have a huge impact on election outcomes,” said Patrick Murray, director of the independent Monmouth University Polling Institute. “I don’t think pollsters should present every possible model under the sun, but the current era of electoral instability suggests it may be a good idea to show a realistic range of outcomes in states where pollsters have little track record or where the nature of the campaign itself invites uncertainty.”
Monmouth’s only other polls in Alabama were conducted during the 2016 presidential primaries. Monmouth’s Republican poll showed Donald Trump with a 23 point lead over his nearest opponent – a race he won by 22 points. That poll was within one percentage point of the actual vote share for 4 of the 5 candidates on that ballot, underestimating only Ted Cruz’s total by 5 points. Monmouth also showed Hillary Clinton ahead by 48 points in a Democratic primary race that other polls suggested would be much tighter. She won that contest by 59 points.
The Monmouth University Polling Institute was established in 2005 to be a leading center for the study of public opinion on critical national and state issues. The Polling Institute’s mission is to foster greater public accountability by ensuring that the public’s voice is heard in the policy discourse. The Monmouth University Poll, which is conducted nationally and in 27 states, received an A+ rating from the polling website FiveThirtyEight.com.
For more information: www.monmouth.edu/polling