Cox Proportional Hazards Redux 1 of 5

1
Cox Proportional HazardsRedux (1 of 5)

• No knowledge of the structure for duration
  dependence
• Handling right censorship
• Regression oriented
• Data Structure
• Tied data
• Hazard rate ratios

2
Cox Proportional HazardsRedux (2 of 5)

• HRR
• Beware of the small HRR and variable scaling
• Dummy variable coding
• Chi-square comparisons
• Relation between b and HRR

3
Cox Proportional HazardsRedux (3 of 5)

• Main Effects and Interactions
• What is an interaction?
• Interpreting main effects and interactions
• Centering
• Efficient ways to center
• Dummy variables
• Effects on main effects vs. interaction effects
• Collinearity Diagnostics
• How to do this
• 5-50 rule
• Structural vs. sample-based collinearity
• Centering to the rescue?

4
Cox Proportional HazardsRedux (4 of 5)

• Proportionality
• What is it?
• Graphical methods for assessment
• Allow for non-proportionality
• Interaction between duration and a covariate
• Exception to centering
• Interpreting the non-proportionality effect
• Continuous time or discrete time?

5
Cox Proportional HazardsRedux (5 of 5)

• Is the model any good?
• Goodness of fit
• -2LogL
• Nesting
• Special case for dummy variables
• Avoid R2 envy
• Using SAS to do the testing
• Plotting
• Simple plots in SAS Analyst, PROC Lifetest/Phreg
• Hazard and survival
• Watch your intervals affects the look of the
  plot
• Adjusted plots (later)

6
P 164 Kleinbaum and KleinClinic
7
P 164 Kleinbaum and Klein Prison
8
P 164 Kleinbaum and Klein Dose