Date: Tue, 2 Feb 1999 01:04:20 -0600
Reply-To: "Frank H. Jurden" <fhj@QNI.COM>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: "Frank H. Jurden" <fhj@QNI.COM>
Organization: Suba Communications, http://www.suba.com
Subject: Re: Correlated sampling?
Content-Type: text/plain; charset=us-ascii
a good place to start is the lisrel/prelis software manual. it includes a data
generator which can reproduce mulitvariate normal and non-normal data sets
according to the means/sds/and correlation matrices you desire.
hope this helps.
> I have a question about Monte Carlo simulation, which I haven't found an
> to. Suppose one wants to simulate together two random variables, both of
> are Normal(0,1). If their correlation is -1, 0, or 1, it is trival. But how
> should one proceed if the correlation is not one of these? Linear treatment
> seems plausible, but a co-worker believes that isn't right, but can't back it
> up. In other words, if X1 and X2 are the independent random draws, how does
> get the corresponding correlated random variables?
> Any help? Any references to book or articles would be much appreciated, too.