Dear List,
Some of you already heard a bit of my data. and thanks for all the
comments.
I do still have slights problems though. i end up having large CIs with
big ORs.
for a summary:
I have a dataset of 1000 controls, 330 patients which are matched by age
and sex; end up with 162 pairs.
i have 683 variables.
i took out the variables that has a prevalance of less than 5%.
and tried to do mcNemar, and corrected with FDR. (adjusted p-values)
Q1) is it usual when one takes the significance level of p <.001( since i
have so many variables)
Q2) from my clinical experience i know that it is true to have these high!
ORs but what does it tell statisticaly?
many thanks,
H
=====================
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
