Log in
RegisterDropping subjects to be balanced? [General Statistics]
Dear Ken,
you really want discard 3 subjects with full data because an other one dropped out?
IMHO that's not a good idea. I guess that no regulatory body will accept such an approach, random choice of the subjects to be dropped or not.
Moreover there is no need to act this way. AFAIK the study is evaluable even if the balancedness of the original design is not maintained by drop-outs. SAS Proc GLM f.i. does the job without any problem.
It may be even worth considering to retain the data of the Drop-out to the extent they are available if the evaluation is done by the EMA recommended approach "use only the data relevant for the comparison under consideration". Or if you use mixed model analysis in the evaluation (not EMA!).
you really want discard 3 subjects with full data because an other one dropped out?
IMHO that's not a good idea. I guess that no regulatory body will accept such an approach, random choice of the subjects to be dropped or not.
Moreover there is no need to act this way. AFAIK the study is evaluable even if the balancedness of the original design is not maintained by drop-outs. SAS Proc GLM f.i. does the job without any problem.
It may be even worth considering to retain the data of the Drop-out to the extent they are available if the evaluation is done by the EMA recommended approach "use only the data relevant for the comparison under consideration". Or if you use mixed model analysis in the evaluation (not EMA!).
—
Regards,
Detlew
Regards,
Detlew
Complete thread:
- Randomization before or after sample analysis Ken Peh 2014-05-05 18:44
![[Mix]](https://static.bebac.at/img/mix.png "open in Mix view")
- Dropping subjects to be balanced?d_labes 2014-05-06 08:09
- Randomization before or after sample analysis Tushar.g 2014-05-06 13:41
Ing. Helmut Schütz