Hi Han,
You still need to simulate betas and gammas from the multivariate normal
(becase we need to integrate over uncertainty in the parameters), but once
you have these you plug into the formulae for pi and lambda, and directly
determine the expected outcome using the formula given, rather than
simulating from the zero-inflated poisson and averaging over simulations a
bunch of times.
Iain
On Wed, Apr 14, 2010 at 11:16 PM, Han He <han.he at college.harvard.edu> wrote:
I thought that by stating that it is unnecessary to
simulate from the
zero-inflated poisson distribution that we should not simulate. I may be
wrong though and your method does make sense in finding errors. Anyone else
care to comment?
Best,
Han
Harvard College Class of 2013
614-329-1324
On Wed, Apr 14, 2010 at 11:12 PM, Michael Barnett <mlbarnett at gmail.com>wrote:
I figured that we need to simulate 1,000 or
10,000 sets of parameters
from multivariate normal and then get the first difference and the
standard error from the set of simulated first differences. But we
don't need to use the zero inflated poisson, like the problem set
says, just need the multivariate normal for the parameters. Just like
in problem set 5.
-Michael
On Apr 14, 2010, at 11:07 PM, Han He wrote:
Hey,
I am not particularly sure if I did the first difference right but I
took the median values and changed a married variable, ran it
through the expected value equation and took the difference. My
question is how do we then get the standard error. Anyone have any
idea/hints?
Best,
Han
Harvard College Class of 2013
614-329-1324
_______________________________________________
gov2001-l mailing list
gov2001-l at
lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l at
lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l at
lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l