The graph below suggests that the New Mexico highway fatality rate is highly
related to the U.S. fatality rate. Other aspects need to be considered, however.
Among these are the lagged New Mexico rate and the lagged U.S. rate, and 
perhaps time itself. Whatever model is finally adopted, the residuals from the 
model should be independent of time (as indicated by some test such as the
Durbin-Watson statistic). Use stepwise regression to sort out the variables 
significantly related to the New Mexico fatality rate, and indicate what a test of
the serial residuals tells you.