On Fri, 09 Feb 2007 16:21:58 -0000 (GMT), (Ted Harding) wrote:

> 
>   "Let i indexes  a subject and Y_i = (Y_i1, Y_i2,...,Y_iT)
>    be a vector of binomial variables for subject i  such that
>    Y_it ~ Bin(n,p_t) with t = 1,2, ....T.
>    A simple correlation I would like to have is :
>    corr(Y_ij, Y_ik) = c for all (j,k)"

I have only seen only a portion of this discussion, so I hope that
the following is not unrelated to the question.

Would it be Ok to generate a sequence p_1,...,p_t, ...p_T
from a Beta stationary process (which exists, actual more
than one sort), and then sample from Bin(n,p_t)?

Alternatively, some 20 years ago, there was a stream of literature 
for discrete sttionary processes, based on the idea of "thinning"
a discrete variable. There are for sure stationary processes
with Negative binomial distribution of this sort. I have
not followed that literature, but it is conceivable that
the similar process for the binomial case has been considered.

Best wishes,

Adelchi Azzalini

-- 
Adelchi Azzalini  <[EMAIL PROTECTED]>
Dipart.Scienze Statistiche, Università di Padova, Italia
tel. +39 049 8274147,  http://azzalini.stat.unipd.it/

______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Reply via email to