This uses the recidivism data recid.sas7bdat. Make sure you change the value of time for the censored cases in week 52.
- Use the life table method to estimate the survival experience of the inmates staying not arrested. Use interval width of 5 weeks and pay attention to the end point of the last interval. Show your code and the results. Based on the results, what is the probability of getting arrested in the first 15 weeks?
- Use another PROC LifeTest step, create the plot for the hazard curve using the default KM method, and then plot hazard curve using the Life Table method. Which method will produce the smoothed hazard curve?
- Compare the estimated hazard curve and the smoothed hazard curve. What do you find?
- Use a PROC LifeTest step to plot the smoothed hazard curve with bandwidth of 10, then another PROC LifeTest step to plot with bandwidth of 5. What’s the difference? Which bandwidth is a better choice for this datset?
- Get the LS and LLS plots from PROC LifeTest for this data. By comparing the shape of these plots, can you decide what distribution the event time T has? Why or why not?
