home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!ECN.NL!VANARKEL
- X-Envelope-to: stat-l@MCGILL1.BITNET
- X-VMS-To: in%"stat-l@mcgill1.bitnet"
- Content-transfer-encoding: 7BIT
- Message-ID: <01GSP5TFNVFM00062J@ECN.NL>
- Newsgroups: bit.listserv.stat-l
- Date: Thu, 24 Dec 1992 15:20:00 +0100
- Sender: STATISTICAL CONSULTING <STAT-L@MCGILL1.BITNET>
- Comments: Sent using PMDF-822 , routing is done by ECNICE
- From: vanarkel@ECN.NL
- Subject: SPSS-PROBIT
- Lines: 18
-
- I ask your attention for a relatively simple, yet persistent problem with
- SPSS/PC+. I'm trying to estimate a probit model using the PROBIT-routine in
- the Advanced Statistics module, release 4.01. This procedure requires the
- use of aggregated data (or repeated observations), which I do not have at
- my disposal. From the way in which the PROBIT-procedure has been construc-
- ted (from a user's point of view) and described in the SPSS/PC+ manual, I
- suspected that this procedure used an estimated generalized least squares
- routine to compute it's estimates, which would mean that I could not use my
- case-by-case data in the estimation of the probit-model. The SPSS-helpdesk
- however assured me that the PROBIT-procedure in SPSS used 'a' maximum-li-
- kelihood-algorithm, though they couldn't tell me which one right away. If
- SPSS indeed uses maximum-likelihood, my problem is solved at once. But I
- still have my doubts about the way SPSS computes probit-estimates, so my
- question is: does anyone know if the SPSS PROBIT-routine uses EGLS or ML,
- and in case of ML which particular maximum- likelihood-algorithm is being
- used in computations?
-
- Ron Linssen, Netherlands Energy Research Foundation - Energystudies
-