The best-fitting advertising response function in Example 7.5 fits the observed data. This is because we rigged the observed data to fall close to a curve of the form in Equation (7.4). See what happens when one of the observed points is an outlier—that is, it doesn’t fit the pattern of the others.

a. Specifically, change the number of exposures corresponding to 50 ads from 90.3 to 125, and then rerun Solver. Do you get essentially the same response function as before, or does this one outlier exert a large influence on the estimated response function?

b. Repeat part a, but now change the number of exposures corresponding to 50 ads from 90.3 to 55.