survival was used as the dependent variable to normalize the
distribution of values on that variable.
Inferential study
The inferential study of these variables consisted of a
series of regression analyses. In the first series of
regression analyses two linear models were tested for each of
the three cancer sites. These analyses were exploratory in
nature. The levels of statistical significance of the
partial regression coefficients of the independent variables
in these analyses were used as a criteria for the development
of equations for prediction.
The first of these two linear models included only the
main effects of the independent variables. The second linear
model included main effects and selected interaction effects.
Table 4 summarizes the variables used in the regression
analysis using the main effects only and the analysis using
main and interaction effects.
The regression analyses using the variables listed in
the table were carried out in an identical manner for each of
the three cancer sites with one exception. Since all cases
in the breast cancer group were female, the variable sex was
dropped as an independent variable for that site.
The second series of regression analyses involved tests
of regression equations for each cancer site that included
combinations of the independent variables that appeared to be