Data set

9. cross section 11. length and width are indeterminate; we’d expect the coefficient on weight to be negative and that on the Japan dummy to be positive 13. mileage = 44 – 0.004Xlength – 0.106Xwidth – 0.004Xweight – 1.323Xjapan as expected the coefficient on weight is negative; the coefficient on Japan contradicts our expectations. In these data, Japanese cars have lower mileage than American cars. 15. at 10% significance, weight is the only significant x variable 17. there is clearly a significant relation; p-value for the F test = 1.4X10; the equation explains 67% of the variation in gas mileage. 19. using the rule of thumb that any |r| > 0.5 is worrisome, there is multicollinearity among length, width and weight with r = 0.72, 0.75 and 0.74. It would be best to drop width and length and re-run the equation.