Dear all,
I'm dealing with SPARSE TABLES (the dimension is >2: 3 or K-ways table): I need
to calculate a statistic for assessing the indepence/non indepedence of the
marginal distributions of 3+ categorical variables involved in the table.
I'm expecting to have a lot of cells with a very small number of subjects (say
1-3), some with 0, and few cells with more than 10 subjects.
The classical chi-square test would be useful, but non robust in this case;
Fisher's exact test is too much expensive in terms of computer calculation
time... I'm searching for something else, without recurring to loglinear models.
Any suggestion?
Thanks!
Cristian
===========================================
Cristian Pattaro
===========================================
Unit of Epidemiology & Medical Statistics
Department of Medicine and Public Health
University of Verona
[log in to unmask]
http://biometria.univr.it
===========================================
-------------------------------------------------
Biometria - biometria.univr.it
|