Use software with the Crime2 data file at the text website, excluding the observation for D.C. Let y = murder rate. For the five explanatory variables in that data file

(excluding violent crime rate), with α = 0.10 in tests,

(a) Use backward elimination to select a model. Interpret the result.

(b) Use forward selection to select a model. Interpret the result.

(c) Use stepwise regression. Interpret the result.

(d) Compare results of the three selection procedures.
How is it possible that a variable (percentage with a high school education) can be the first variable dropped in

(a) yet the second added in (b)?

(e) Now include theD.C. observation.Repeat

(a) and (b), and compare to results excludingD.C.What does this suggest about the influence outliers can have on automatic selection procedures?