Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: intraclass correlation for Categorical data

From   John Litfiba <>
To   Dirk Enzmann <>
Subject   Re: st: intraclass correlation for Categorical data
Date   Fri, 13 May 2011 00:49:27 +0200

Many thanks Dirk,

But this doesnt work, as I have more than 2 categories ...
To be more precise, I have a panel dataset which is too large for
using random or fixed LOGIT models...
Therefore I am constrained to fit a multinomial logit (which works
fine even with millions of observations), and I was trying to figure
out if the "correlation" between the categorical observations within
my clusters (the individuals of my panel dataset) show a high or low
level of correlation... I hope of course to find a low "correlation"
(can we say "correlation" for nominal variables? not sure) in order to
justify the independence assumption of the multinomial logit ;-)
Maybe there is no short answer to this problem... ;)


On 13 May 2011 00:08, Dirk Enzmann <> wrote:
> With respect to dichotomous data, see -xtmrho- on ssc.
> Dirk
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index