[S] Common starting value in Random Slopes model

Robert Kinley (KINLEY_ROBERT@lilly.com)
Thu, 06 Aug 1998 10:17:25 +0100


I often encounter the situation where there are several sets of X,Y data
for samples from the same population, kept under different storage
conditions.

Typically X is time in months, Y is the feature of interest, and there is
one set
of X,Y data for each storage condition.

Unfortunately, all the X,Y sets have the SAME Y at X=0 .

The intention is to apply a random-slopes model and interest lies in
(a) are the slopes are sig diff,
(b) if not sig diff, getting a confidence region for a pooled slope.

The common 1st value messes this up - treating the datasets as
independent
is clearly wrong because the degrees of freedom won't be right. Leaving out
the
initial point means you can apply the model OK, but I don't like excluding
data.

I generally place the problem of finding a correct solution using all the
data
in the 'too-hard' basket -

any suggestions as to how I could get it into the 'can-do' basket ?

Thanks ...... Bob Kinley

-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news