Dear All
"Dear Sue
I disagree with your previous correspondents and their
quoted authorities.
We randomise to achieve prior similarity and failure to
achieve this might mean that any observed difference - eg
the treatment effect - is due to an imbalance in one
or more baseline covariates rather than the treatment, per se.
Failure to identify imbalance and to adjust for it
is bad Science. For example, what interpretation will one
place on a significant RCT result which does not survive
adjustment for an observed baseline imbalance.?
The many objections to adjusting for baseline imbalance
are based on particular philosphical statistical arguments
which though sometimes interesting do not transcend the
basic science involved.
Moreover these philosophical objections can be met one by one.
In Medical Statistics one has to guard against un-necessary
and narrow fashion
Best for now
Gilbert"
_____________________________
Prof. Gilbert MacKenzie
Director,
Centre for Medical Statistics
Keele University
Staffordshire ST5 5BG
England
Email: [log in to unmask]
Tel: (01782) 583651
Fax: (01782) 584268
______________________________
|