Hi all,
sorry for the weird subject line, I didn't know how to express the
problem briefly ... which is: I am working on a sample including
teachers rating how well they know a child. My goal was to standardize
the values classwise (M=0, SD=1) to make classes comparable. However,
many teachers indicated they know all children equally well. Due to
zero variance, no standardized values are computed, thus drastically
reducing my sample to about two thirds. My first idea was to set all
these constants to zero (the mean of the subsample value thus
equalling the mean of the standardized value; it's a linear
transformation, after all). What would you say? I feel a little silly
asking this question yet hope someone can answer it :)
Regards
Tanja
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
