Date: Thu, 2 Jul 1998 16:33:07 -0500
Reply-To: "Nichols, David" <nichols@SPSS.COM>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: "Nichols, David" <nichols@SPSS.COM>
Subject: Re: Var-Cov Matrix, MV Normal Dist
If you have Release 7.5 or above, you can save the coefficient statistics
in a linear regression in REGRESSION to a file. This file will include the
coefficients, their standard errors, significances (p-values), error
degrees of freedom, and the covariance matrix of the estimates. This can be
done through the dialog boxes.
The following will (with appropriate numeric substitutions for ncases and
nvariables) create the desired number of cases with the desired number of
standard normal variates. I know of no way to directly generate data from a
multivariate normal distribution with SPSS (you can only generate variables
one at a time and I don't know of a way to enforce the multivariate
aspect).
input program.
loop #i=1 to ncases.
vector v(nvariables).
loop #j=1 to nvariables.
compute v(#j)=rv.normal(0,1).
end loop.
end case.
end loop.
end file.
end input program.
execute.
David Nichols
Principal Support Statistician and
Manager of Statistical Support
SPSS Inc.
----------
From: Jason Wittenberg [SMTP:jwittenb@LATTE.HARVARD.EDU]
Sent: Wednesday, June 17, 1998 9:35 AM
To: SPSSX-L@UGA.CC.UGA.EDU
Subject: Var-Cov Matrix, MV Normal Dist
I have two questions. First, has anyone figured out a way to save, either
as variables or matrices, the vector of estimated parameters and the
variance-covariance matrix of these parameters that are generated after an
estimation? I know it is possible in some cases to print them out, but I
would like to save them for future manipulation.
Second, I would like to take random draws from a multivariate normal
distribution. Has anyone tried to write the code to do this?
Thank you very much for any help you can provide.
Sincerely,
Jason Wittenberg
Massachusetts Institute of Technology
| 
