Practical Predictive Analytics

Improving the education survival curve by adding gender

As we have seen, we cannot conclude that a significant difference exists among the education levels. Often analyzing interaction effects can uncover significance when other covariates are added. If we are interested in seeing if there is a difference in education levels that depends on gender, we can create a new variable that contains dummy variables for each combination of education and gender:

• We will first create this new variable (factorC) by using the interaction() function. Then we will use our new plotsurv() function, to plot the curve based upon all interactions of education and gender.
• Next, we will use the survdiff() function to test for the significance of these effects: ...