look to the table2.20 and 2.21. Are there any obvious scenarios where one might be more useful that the other?

Figure 2.20: A contingency table with row proportions for the app-type and homeownership variables. The row total is off by 0.001 for the joint row due to a rounding error. Figure 2.21: A contingency table with column proportions for the app_type and homeownership variables. The total for the last column is off by 0.001 due to a rounding error