Conduct the analyses of the effect of Family Structure on students’ Substance Use as outlined in this chapter using the NELS data. This is one of the more complex exercises you will do, because it requires the creation of several new variables. It is also probably one of the more realistic examples. I suggest you team up with a classmate as you work on it.

a. Create the Family Structure and Substance Use variables (see note 7). Examine descriptive statistics for each variable, and compute means and standard deviations of Substance Use by Family Structure.

b. Create dummy variables contrasting students from two-parent families with those from parent–guardian families and those from single-parent families. Regress Substance Use on these dummy variables. Interpret the overall regression. Use the table of coefficients to conduct post hoc testing. Make sure you compare singleparent families to parent–guardian families.

c. Create effect variables with two-parent families as the group coded –1 on all variables.

Regress Substance use on these effect variables and interpret the regression results.

d. Convert the Family Structure variable into a single criterion scaled variable and conduct the MR using it. Correct the ANOVA table from the MR for the correct degrees of freedom and compare the results with the other analyses of the same data.