Problem 15 (KNN 8.39 p. 341, not in NKNW) does not work as stated because X3 (NE region indicator) is collinear with other variables in the model. You have two options:
1. Do the problem after replacing X1 (total population) by its logarithm. In SYSTAT you would create a variable, say lx1, as
let lx1=l10(x1)
Then the regression
model y = constant + lx1 + x2 + x3 + x4 + x5
(where all variables have been renamed from the original names in the file) works as intended.
For Part b do not use a confidence interval; instead test directly the hypothesis that beta 3 = beta 4.
2. Or, do Problem 8.41 p. 342 (using the SENIC data set, that will be posted on the web) instead of 8.39 p. 341. For Part c, do not use the Bonferroni procedure; instead test directly the hypothesis that the coefficients of all 3 regional indicators are the same.
30 Mar 2004