1. Spatial Distribution of Alcohol Involved Crashes in Louisiana
Jeff Dickey, Ph.D.
LSU Highway Safety Research Group
2017 ATSIP Traffic Records Forum

2. Highway Safety Research Group
Funded by Louisiana Department of Transportation and Development
Research group within the Stephenson Department of Entrepreneurship & Information Systems (SDEIS) at LSU
Developed LaCrash application to electronically capture crash reporting information
Collect, correct and analyze crash data for the state and other stakeholders

3. Research and Support Seat belt use Drugged driving Roadway departure
Crash map spotting
Crash hot spot mapping for regional safety coalitions
In depth hot spot analysis

4. Alcohol Involved Crashes
Alcohol-impaired driving accounts for 31% of all traffic-related deaths in the USA
– CDC
Half of Youth Car Crash Deaths Involve Alcohol
-medpagetoday.com/pediatrics February 13, 2017
Every 2 minutes, a person is injured in a drunk driving crash
– MADD
In Louisiana, 296 of the 758 total fatalities in 2016 were alcohol- involved crashes (39.05%)
-LSU HSRG

5. Spatial Analysis of Alcohol Involved Crashes
Use driver’s home zip code
Drinking and driving as a risky behavior
Cultural norms “Laissez les bon temps rouler”
“Left of boom”

6. Data for Spatial Analysis
All moderate injury, severe injury, and fatal crashes on local and state roads for
100,868 Louisiana drivers in alcohol involved crashes
82,827 records with valid driver zip code
4,991 recorded alcohol – 6.0%
13,204 predicted alcohol – 15.7%

7. Data for Spatial Analysis
Import crash records into ArcMap
Create points with ZCTA latitude and longitude
Spatial join to ZCTA to sum recorded and predicted alcohol
Merge ZCTA with < 10 crashes into surrounding or adjacent ZCTA
Calculate recorded alcohol crash rate and predicted alcohol crash rate

8. Tools for Spatial Analysis
ESRI ArcMap 10.4
Spatial Statistics Tools
Mapping Clusters
Optimized Hot Spot Analysis (Getis-Ord Gi*)

9. Recorded Alcohol Crash Rate

10. Correlation of Recorded Alcohol Crash Rate

11. Predicted Alcohol Crash – hour, day of week
Driver – alcohol, condition, age, sex, protection system, violations
Pedestrian – alcohol, condition
Vehicle type, number of vehicles, parish code

12. Predicted Alcohol Crash Rate

13. Correlation of Predicted Alcohol Crash Rate

14. Correlation of Difference Predicted minus Recorded Alcohol Crash Rate

15. Interesting Findings
Predicted alcohol generally higher across the state
Highest recorded and predicted alcohol in the Houma area
Alcohol involvement appears to be under reported in north Louisiana

16. Future Research
Incorporate socioeconomic data from American Community Survey
Administer survey of behavior and perceptions of risk
Examine assumptions underlying the predicted alcohol algorithm

17. Questions & Discussion
Thank you!
Jeff Dickey, Ph.D.