Part 1

The data set Fatalities looks at car crash fatalities from 48 US states from 1982-1988. Data is collected at the state year level and includes population, alcohol consumption, miles driven, percent of young drives, and other relevant state level data.

Car Fatalities by Year
year avgRate avgRate15 avgRate18 avgRate21
2 1983 -2.2127069 1.1512829 -8.921645 -2.2796617
3 1984 2.3005811 3.0342559 3.215174 3.1900455
4 1985 -0.6822277 4.0467586 -6.882254 0.7986467
5 1986 5.7468966 21.1298062 12.531081 1.6022218
6 1987 0.7514006 0.9962656 -5.154774 -1.5827526
7 1988 2.0316392 -1.8995045 8.587338 0.0034473

The table above shows average rate of fatalaties by year across all states, and average rate of fatalities by age group, 15-17, 18-20, and 21-24. The table doesn’t indicate a time trend or that fatalities have dropped over the 7 years. Young drives have seen a positive increase in fatalaty rates with 1986 having a 21% increase in car crash fatalities over 1985. This large increase is also observed in 18-20 year old drives with a 12.5% increase in fatalities in one year.

The map above shows the average rate number of car related fatalities per capita by state not including Hawaii and Alaska. The next map below looks at alcohol consumption per capita by state.

The consumption of alcohol seems to be relatively uncorrelated with the per capita fatalities. This likely due to the high rate of car accidents unassociated with alcohol consumption and that many people will consume alcohol without driving.

Univariate Regression Results
Spirits P-value
-7.7e-06 0.5193822

As seen in the simple regression above, the correlation between average per capita alcohol consumption and per capita fatalitie from 1982 to 1988 is near zero with a large p-value (.5). However since the data is aggregated we lose a lot of data points and time trends. Below is a regression testing whether or not alcohol consumption impacts using state fixed effects and time trends

Regression Results
Spirits P-value
0.0001062 0
Alcohol Consumption and Car Fatalities
Avg_Fatal Marginal_Effect_Alcohol Percent_Increase
0.000204 0.0001062 52.03525

Using a regression that controls for time trends and state fixed effects we find that the impact of spirit consumption on per capita car fatalities is 0.00016. It is important to look at the mean of the outcome variable before assuming that the effect is small. As seen in the table above titled “Alcohol Consumption and Fatalities” we see that the effects are quite large compared to how small the average per capita fatality is. Therefore, we can conclude that alcohol consumption has a significant and relevant impact on car fatalities in the contiguous United States