Dear listmembers,
suppose I have a series of chi^2 test statistics that are dependent and
that the nullhypothesis under study can be rejected if any of the test
statistics yields a significant outcome. Would it be appropriate to use the
maximum of these test statistics to determine whether or not the
nullhypothesis holds? And if so, can this maximum still be compared to a
chi^2 square distribution?
Best regards,
Jerry
|