Hi all#
have had a query this week from someone doing chi square tests in the
context of a cluster trial. They wanted to know how to adjust for the
clustering. I suggested that they (also) look into reporting a cluster
adjusted odds ratio, but meantime given that
for any cluster; ({ = sum of)
{(O-E)^2/E^2 ~ chi sq (1) (assuming a 2x2 table)
is it legitimate to claim that
{X^2(1) (j) ~X^2(N of clusters)
and is this a sufficient adjustment?
thanks
Andy
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|