Hello all,

After a rather exhaustive traversing of stats journals, I have yet to find a
satisfactory answer to this problem.  I have a Bayesian linear model that
takes the following form (this is actually the reduced form from a larger
problem):

Y_jk = a_k + b_k *gamma_j x_j  .

The parameters *a *and *b *are just slopes/intercepts, x_j and Y_jk are
data.  The *gamma_j *are bounded parameters - by definition, they are
constrained to the until interval (i.e. gamma_j in [0,1]).  I should note
that this model looks eerily similar to the standard ideal point problem,
though it isn't.

In writing a gibbs sampler for this problem, I tried to use the technique
that Simon Jackman used in his 2000 AJPS for the AR-1 correlation
parameter.  Namely, I gave normal priors to the gammas and simly "tossed
out" posterior draws that lied outside the necessary interval.  This is
turning out to be a huge problem, though.  My sampler gets stuck *very
*easily.
I also dabbled in truncated priors but these were not any good either --
draws would often ram against the boundaries due to far-away means and/or
tight variances.

Any helpful suggestions are welcome.

Regards,

Adam

-- 
Adam Ramey
Ph.D. Candidate
Department of Political Science
Harkness Hall 338
University of Rochester
Rochester NY 14627-0146
Phone: 585-273-1678
Fax: 585-271-1616
E-mail: [log in to unmask]
Website: http://mail.rochester.edu/~aramey

**********************************************************
             Political Methodology E-Mail List
   Editors: Melanie Goodrich, <[log in to unmask]>
            Delia Bailey, <[log in to unmask]>
**********************************************************
        Send messages to [log in to unmask]
  To join the list, cancel your subscription, or modify
           your subscription settings visit:

          http://polmeth.wustl.edu/polmeth.php

**********************************************************