Contains data from sj_ms2_fig1scen2b.dta
  obs:        20,000                          simulated data for fig 1,
                                                scenario 2, SJ article. see
                                                -notes-
 vars:             3                          4 Jun 2008 12:51
 size:       200,000 (98.7% of memory free)   (_dta has notes)
--------------------------------------------------------------------------------
              storage  display     value
variable name   type   format      label      variable label
--------------------------------------------------------------------------------
d               byte   %8.0g                  outcome
z               byte   %8.0g                  covariate
y               float  %9.0g                  marker
--------------------------------------------------------------------------------
Sorted by:  

*********************
notes:

_dta:
  1.  Simulated data for a continuous marker y, binary outcome d, and binary
      covariate z.  Marker values vary with z, but the discriminatory accuracy
      of the marker (i.e. the ROC curve) is the same in the covariate-specific
      populations.  In scenario 1, there are different proportions of cases
      (d=1) and controls (d=0) in the two covariate groups.  In scenario 2, the
      case-control ratio is constant across the covariate-specific populations.
  2.  for figure 1, scenario 2, SJ article: Janes, Longton, Pepe.  Accommodating
      covariates in ROC analysis. 2008.
  3.  see fig1dat_init.do.