Refer back to the experiment in Exercise 21. Instead of mpg redo the analysis using log (mpg) as the response. Do your conclusions change? How? Are the assumptions of the model better satisfied?
In Exercise 21
An experiment to test a new gasoline additive, Gasplus, was performed on three different cars: a sports car, a minivan, and a hybrid. Each car was tested with both Gasplus and regular gas on 10 different occasions and their gas mileage was recorded. Here are the partial box plots:

A two-way ANOVA with interaction model was run, and the following ANOVA table resulted:

A plot of the residuals vs. predicted values showed:

Transcribed Image Text:

50.0 37.5 25.0 - 12.5 32 E 28 24 20 Ct Hybrid Minivan Sports Type Gasplus Regular Additive Sum of Mean Source DF Squas Square F-ratio P-value Type Additive 1 92.1568 92.1568 21.57 <0.0001 Type × 2 51.8976 25.9488 6.0736 0.0042 223175.4 11587.7 2712.2 <0.0001 Additive Error Total 54 230.711 59 23550.2 4.27242 25 2.5 -5.0 12.5 25.0 37.5 50.0 Predicted Values (mpg) キ 丰+丰 (Bdu) senpisau