An interesting experiment took place beginning in April 1979 in Albuquerque, New Mexico. The local police department

Question:

An interesting experiment took place beginning in April 1979 in Albuquerque, New Mexico. The local police department tried a procedure they thought might have the effect of reducing driving-while-intoxicated (DWI) related accidents. The procedure was quite simple. A squad of police officers used a special van that housed a blood alcohol testing (BAT) device; the van became known as the "Batmobile."
In the quarterly data set below is information on the following variables:
ACC = Injuries and fatalities from Wednesday through Saturday nighttime accidents
FUEL = Fuel consumption (millions of gallons) in Albuquerque

The first 29 observations in the data set are a control period before the implementation of the Batmobile program. The following 23 quarterly observations are for the experimental period.
Your job is to explain statistically using standard forecasting procedures whether the Batmobile program was effective.
a. Using the "fuel" variable as a proxy for the amount of driving in any given period, calculate the average injuries per gallon of fuel for both the pre-Batmobile period and for the treatment period. Do the results lead to the inference that the Batmobile was effective?
b. The data appear to have some seasonality. Construct a multiple-regression model using the fuel variable, seasonal dummy variables, and a separate dummy variable for the Batmobile program to explain the injuries and fatalities variable. Explain the diagnostic statistics and present your reasoned opinion on the efficacy of the Batmobile program.
c. There is some evidence that programs like the Batmobile program take some time to "catch on." If this is actually the case, the dummy variable representing the existence of the Batmobile program can be modified to examine whether there is a "cascading" effect to the program. The data set also contains a "ramped" version of the Batmobile dummy in which the variable is zero in the pretreatment period but takes on a value of one in the first treatment period and increases in value by one for each subsequent treatment period. In a sense, this treatment mimics a ramping up of the effectiveness of the Batmobile effort. Run such a model and examine the relevant summary statistics. Is there evidence of a ramping effect? What degree of confidence do you have in your answer?