help for starjas
-------------------------------------------------------------------------------

Title

    starjas --     Arjas plot to check proportional hazards assumption


Syntax

        starjas varname [if exp] [in range] [, adjust(varlist) levelplot(#
               [#] ..) atobs(#) rrglance(#) twoway_options ]


    options                        Description
    -------------------------------------------------------------------------
    Options
      adjust                       specify the variables fitted in the Cox
                                     model before checking if varname is
                                     proportional
      levelplot                    specify the levels of varname to be
                                     displayed in the plot
      atobs                        restrict the plot to the first # events
      rrglance                     draw a line approximately corresponding in
                                     the case of binary covariate to a #
                                     relative risk

    Y-Axis, X-Axis, Caption, Legend
      twoway_options               some of the options documented in
                                     [G] twoway_options
    Add plot
      plot(plot)                   add other plots to generated graph
    -------------------------------------------------------------------------

    You must stset your data before using starjas; see stset.


Description

starjas graphs Arjas plot by one categorical variable declared in varname.

Arjas plots allow to check if a covariate should be included in a proportional
hazards model and if it has proportional hazards effect.
If a covariate does not need to be in the model, then for each level a roughly
45° line through origin should be plotted.  Otherwise Arjas plot will give
curves which are approximately linear but with slope differing from 1.
If a covariate has a nonproportional effect on the hazard rate, then the curves
will differ nonlinearly from 45° line. Assume an indicator variable having only
two values, the curve for the level 1 should be concave if the hazard ratio is
increasing in t whereas the curve for the level 0 should be convex.



Examples

            use "C:\Data\BMT", clear
            starjas MTX, legend(pos(5) ring(0)) lw(medthick)
            starjas Z7cat, adjust(MTX Z1-Z6) legend(pos(5) ring(0)) lc(green
                    navy red brown)

Downloading ancillary files in one of your `"`c(adopath)'"' directory you can
run this example.

          (click to run)


Also see

 Manual:  [S] st stcox

 Online:  stcox postestimation, stcox diagnostics



References

Klein JP, and Moeschberger ML: Survival Analysis: techniques for Censored and
Truncated Data - 2nd ed. (2003), p. 369 - 374.  Springer


Authors

Enzo Coviello (enzo.coviello@alice.it)