Faure and de Neuville XXXXXXXXXXpresented data on the effectiveness of French efforts to reduce accidents in towns where major highways run through the ancient and crowded city centers. They presented...

Faure and de Neuville (1992) presented data on the effectiveness of French efforts to reduce accidents in towns where major highways run through the ancient and crowded city centers. They presented number of accidents before (yi1) and after (yi2) the road safety improvements, as shown in Table 13.16. Varying amounts of years of data were available from each city. They are shown in parentheses after the count of accidents. Adapt the ideas of a paired t test by fitting a Poisson regression model where

mean in town i before improvement μ_i1
= λ_is_i1

and mean in town iafter improvement μ_i2
= ð(λ_i
)s_i2:

You will need to define a set of dummy variables for Town, and one for Time (before/after). The offset variables s_i1
and s_i2
are the years of data available for town i before and after the improvement, shown in parentheses

(a) Has there been a significant decrease in the accident rate? If so, give a 95% confidence interval for the reduction.

(b) Upon carefully rereading the research article, you see that the table is entitled “Some examples of accident reduction.” How does that change your conclusion?