Citation

Material Information

Title:
Predictors of Traffic Fatalities
Series Title:
Math 7393 Bayesian Statistics
Creator:
Lancaster, Heath

Subjects

Genre:
Course Material ( sobekcm )

Notes

Acquisition:
Collected for Auraria Institutional Repository by the Self-Submittal tool. Submitted by Heath Lancaster.
Publication Status:
Unpublished
General Note:
I think the abstract is already on the website. The uploaded powerpoint should play the presentation once you click 'start slide show'

Record Information

Source Institution:
Auraria Institutional Repository
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.

Auraria Membership

Aggregations:
Auraria Library
Added automatically

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

Predictors of Traffic Fatalities Heath Lancaster

PAGE 2

Introduction Traffic accidents result in between 35,000 40,000 deaths each year Engineering and stricter laws have caused declines, but distracted driving remains a persistent problem in the U.S Understanding economic and demographic relationships can help direct policy improvements

PAGE 3

Background Fatality Analysis Reporting System (FARS) Small Area Estimates (SAIPE and SAHIE) US Census Bureau Local Area Estimates Bureau of Labor Statistics Economists have looked at connection between the Patient Protection and Affordable Care Act and ambulance response times

PAGE 4

Research Question Using available data on age, sex, economic, and race are most correlated with traffic fatalities. This analysis will use a Bayesian framework and will be conducted in the statistical software R and Stan.

PAGE 5

Methods Data is aggregated from all 50 states over the period 2010 2018. Variables included in the analysis cover economic, race, age, and sex variables. Because the data is grouped by states the demographic variables represent the share of the population in that state , of that demographic Women is percent of total population that are women.

PAGE 6

Methods be a glm . Negative Binomial Multiple model fitted using different combinations of economic and demographic variables . Final model was chosen by WAIC and LOO.

PAGE 7

Results Final model is comprised of economic and racial variables Economic includes median household income, percent of population with health insurance and the average unemployment rate over the period in each state. Racial variables include the percent of the population in each state that is Black, the percent of the population that is Asian, and the percent of the population that is Hispanic.

PAGE 8

Results

PAGE 9

Results

PAGE 10

Conclusion Considering economic, racial, sex, and age demographics this analysis has show that economic and racial variables best fit the data to explain traffic fatalities. There could be many explanations, but one that seems likely is that safer cars are more expensive and newer. Therefore a straightforward policy recommendation would be subsidizing safety innovations and improve streets and highways.