5.1 See page 85

5.7 part a: In SAS, you could have a program like this:

    proc reg data = one ;
    model y = x2 x3 ;
    output out = r1  r = resy ;
    run ;

    proc reg data = r1 ;
    model x1 = x2 x3 ;
    output out = r2  r = resx1 ;
    run ;

    proc reg data = r2 ;
    model resy = resx1 ;
    run ;

    part a: In R, you could have a program like this:

    CanPrestige1.lmYX <- lm(prestige ~ income +women, data=CanPrestige1)

    CanPrestige1.lmXX <- lm(education ~ income +women, data=CanPrestige1)

    lm(CanPrestige1.lmYX$residuals ~ CanPrestige1.lmXX$residuals)

    part b: See page 85

5.9 Use the definition of a standardized regression coefficient on page 102

6.5 You can review the concept of maximum likelihood estimation in the online supplement at: 
    http://socserv.socsci.mcmaster.ca/jfox/Books/Applied-Regression-3E/Appendices.pdf starting on
    page 92. The problem gives the probability density function (pdf) of the normal distribution for a 
    single observation. You will need to use the pdf of the sample of n observations, take logs,
    and then take three partial derivatives, then set them equal to 0 and solve.

6.6 Similar approach for parts a and b. If X' = 10(X - 1), then first show that the mean of X' 
    (X'bar) = 10(Xbar - 1), where Xbar is the mean of X. Continuing in this fashion, you can show what 
    Sx, Sy, and Sxy are in terms of X', which will lead to finding B' and SE(B').

6.13  A computational hint is included with the homework listing