Our approach is to compute a quantity of interest, its fully sampling (or
posterior) distribution, and to possibly summarize this with a mean and a
confidence interval if necessary. At that point, you've provided all
information there is, and hypothesis testing will not help teach us
anything relevant. There is of course a one-to-one correspondence between
CI's and hypothesis tests, but thinking about the quantity of interest
normally makes a lot more sense. That's why we don't compute any of the
hypothesis tests that one might. However, by counting the proportion of
simulations that fall in a specified region, you can calculate the exact
probability (p-value) that the truth lies in that region (to put it in the
Bayesian way).
Gary King
: Gary King, King(a)Harvard.Edu
http://GKing.Harvard.Edu :
: Center for Basic Research Direct (617) 495-2027 :
: in the Social Sciences Assistant (617) 495-9271 :
: 34 Kirkland Street, Rm. 2 HU-MIT DC (617) 495-4734 :
: Harvard U, Cambridge, MA 02138 eFax (617) 812-8581 :
On Wed, 31 Mar 2004, yumin sheng wrote:
Dear all,
I used Clarify to obtain the predicted probabilities
for the four categories of a dependent variable and
the "first difference" effects of independent
variables of interest in an ordered logit model. The
output generated provides statistics on "mean values
for quantity of interest," "standard errors" and the
"confidence intervals." However, in trying to
calculate the p-values and determine the exact "level
of significance," I am not quite sure on how to
proceed.
Page 27 of the Manual says that "simqi" for ordered
logit is drawn from a multinomial distribution with n
(degree of freedom?)=1. If so, how could I find the
exact cumulative densities with Stata? Or, did I
misunderstand the distribution underlying the reported
statistics by Clarify so that I could simply get
t-values or z-values from the means and standard
errors?
Any help will be tremendously appreciated. Many
thanks in advance.
Best wishes,
Yumin
=====
Yumin Sheng
Ph.D. Candidate
Dept. of Political Science
Yale University
__________________________________
Do you Yahoo!?
Yahoo! Finance Tax Center - File online. File on time.
http://taxes.yahoo.com/filing.html
--
Clarify mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive:
http://lists.hmdc.harvard.edu/?info=clarify
--
Clarify mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive:
http://lists.hmdc.harvard.edu/?info=clarify