Date:         Tue, 27 Sep 2005 21:07:57 -0400
Reply-To:     Hans Chen <spsshanschen@GMAIL.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Hans Chen <spsshanschen@GMAIL.COM>
Subject:      Missing Data and Probit Model
Content-Type: text/plain; charset=ISO-8859-1

Dear SPSS Listers

My friend has met some problems to handle in his study. Could you please
kindly give some suggestion for him?

He is dealing a panel consisting of 10,000 enrollees of a public health
insurance plan for 10 years. He wants to use this panel to measure the
effect of co-payment on patients¡¯ choice of outpatient services on
different levels of hospitals.  The dependent variable is defined as the
choice of three levels of hospitals, namely 1 2 3. His questions is that
each enrollee will encounter different times of choices among 1 2 3
hospitals respectively, i.e. they share different risks of being sick. So
if He wants to use a probit model to analyze the choice structure, there
must be some ¡°missing values¡± in the error matrix (It is not a repeated
choice procedure!). Could you please offer some suggestions on how to deal
with this situation in SPSS?



Many Thanks!