Return to ChangeLog CVS log

Up to [Local Repository] / imach / src

*** empty log message ***

    1: 2003-03-28    <brouard@BROUARD>
    2: 
    3: 	* imach.c (Module): In version up to 0.92 likelihood was computed
    4: 	as if date of death was unknown. Death was treated as any other
    5: 	health state: the date of the interview describes the actual state
    6: 	and not the date of a change in health state. The former idea was
    7: 	to consider that at each interview the state was recorded
    8: 	(healthy, disable or death) and IMaCh was corrected; but when we
    9: 	introduced the exact date of death then we should have modified
   10: 	the contribution of an exact death to the likelihood. This new
   11: 	contribution is smaller and very dependent of the step unit
   12: 	stepm. It is no more the probability to die between last interview
   13: 	and month of death but the probability to survive from last
   14: 	interview up to one month before death multiplied by the
   15: 	probability to die within a month. Thanks to Chris
   16: 	Jackson for correcting this bug.  Former versions increased
   17: 	mortality artificially. The bad side is that we add another loop
   18: 	which slows down the processing. The difference can be up to 10%
   19: 	lower mortality.
   20: 
   21: 2002-05-30    <brouard@BROUARD>
   22: 
   23: 	* 
   24: 
   25: 	* imach.c (Module): Add correlation matrix of one-step
   26: 	probabilities and covariance matrix
   27: