To get the data, open R and type >install.packages("alr4"), require(alr4), the datafile name is stopping

7.6 (Data file: stopping) The (hypothetical) data in the file give automobile stopping Distance in feet and speed in mph for n = 62 trials of various automobiles (Ezekiel and Fox, 1959) 7.6.1 Draw a scatterplot of Distance versus Speed. Explain why this graph supports fitting a quadratic regression model. 7.6.2 Fit the quadratic model but with constant variance. Compute the score test for nonconstant variance for the alternatives that (a) variance depends on the mean; (b) variance depends on Speed; and (c) variance depends on Speed and Speed?. Is adding Speed helpful? 7.6.3 Refit the quadratic regression model assuming Var(Distancel Speed-Speed . Compare the estimates and their standard errors with the unweighted case. 7.6.4 Based on the unweighted model, use a sandwich estimator of vari- ance to correct for nonconstant variance. Compare with the results of the last subproblem. 7.6.5 Fit the unweighted quadratic model, but use a case resampling bootstrap to estimate standard errors, and compare with the previ- ous methods. 7.6 (Data file: stopping) The (hypothetical) data in the file give automobile stopping Distance in feet and speed in mph for n = 62 trials of various automobiles (Ezekiel and Fox, 1959) 7.6.1 Draw a scatterplot of Distance versus Speed. Explain why this graph supports fitting a quadratic regression model. 7.6.2 Fit the quadratic model but with constant variance. Compute the score test for nonconstant variance for the alternatives that (a) variance depends on the mean; (b) variance depends on Speed; and (c) variance depends on Speed and Speed?. Is adding Speed helpful? 7.6.3 Refit the quadratic regression model assuming Var(Distancel Speed-Speed . Compare the estimates and their standard errors with the unweighted case. 7.6.4 Based on the unweighted model, use a sandwich estimator of vari- ance to correct for nonconstant variance. Compare with the results of the last subproblem. 7.6.5 Fit the unweighted quadratic model, but use a case resampling bootstrap to estimate standard errors, and compare with the previ- ous methods