Date of Award
Open Access Thesis
Master of Public Health (MPH)
School of Public Health
Peter M. Aronow
Background: Redelmeier and Tibshirani conducted a study, indicating polling hours of presidential election days created an increased risk of driving fatalities . In another paper, they suggested a negative association between the 4 hours after polling and risk of driving fatalities . Such a strong negative relationship cast doubt on the methodology used in the original paper . We were concerned that the approach for hypothesis testing was underconservative, and that the confidence intervals presented lacked proper coverage.
Methods: We performed analysis on time-based relative risk of driving fatalities and fatal crashes for US population on election hours, non-election hours, and full 24 hours of election days with other comparable days with updated dataset to 2012. We also calculated the relative risk and constructed the distribution of 100 Tuesdays and 6 days before and after election days to evaluate the influence of election days in a large scope.
Results: The risk of driving fatalities was 1.19 (95% CI: 1.11-1.27) during election hours, 0.85 (95% CI: 0.78-0.93) during non-election, 1.09 (95% CI: 1.01-1.12) during full 24 hours on election days. 7% of the Tuesdays and 8.3% of the 12 days had risk of driving fatalities at least as extreme as election days during election hour. For the risk of fatal crashes, the risk was 1.15 (95% CI: 1.03-1.27) during election hours. 7.5% of the Tuesdays and 8.3% of the 12 days had risk of driving fatalities at least as extreme as election days during election hours.
Conclusions: Our results suggested that presidential elections do not appear to be particularly unusual relative to other comparable days in four-year election cycles.
Zhang, Fan, "Driving Fatalities On Us Presidential Election Days: A Reanalysis" (2015). Public Health Theses. 1343.