Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Re: Capabilities of svymlogit

From   "R.E. De Hoyos" <[email protected]>
To   <[email protected]>
Subject   st: Re: Capabilities of svymlogit
Date   Wed, 13 Jul 2005 15:27:06 +0300


In -mlogit- (or its survey version) the dependent variable is discrete and takes more than two values. If I understood you right, your data is set in a way that you observe several rows for the same individual each of them accounting for the different "participations". Therefore each column representing a different "participation" will only take values 0 and 1. If every individual "participates" only in one option then you can -reshape wide- your data to obtain a single variable (column) taking different values for the different chosen participations (say if you have 5 possible "participations" your new variable will take values 1 to 5).

help reshape

Once you have done that you can either -svyset- your data and run -svymlogit- or else use -mlogit- with [pw] and the cluster() option.

Alternatively you can use -clogit- with the present format of your data (-mlogit- and -clogit- are mathematically the same). Unfortunately -clogit- does not have a survey version, all you can do is to include the cluster() option.

clogit Y vars, id(personid)

I hope this helps,

R.E. De Hoyos
Faculty of Economics
University of Cambridge

----- Original Message ----- From: "Jens Clausen" <[email protected]>
To: <[email protected]>
Sent: Wednesday, July 13, 2005 1:55 PM
Subject: st: Capabilities of svymlogit

Dear All,

I am new to Stata, and trying to use it to analyze a survey dataset
where some respondents have several positive responses (ie: "Have you
participated in A?" -Yes, "Have you participated in B?" -Yes, etc.) and
therefore have several lines in the dataset.
Now my question is whether SVYMLOGIT is capable of handling this, and
whether I am going about it the right way. I am using:

svyset [pweight=WGT], psu(PERSONID)
svymlogit Y <vars>, base(0)

which produce estimates identical to

mlogit Y <vars>, base(0)

except for the standard errors. Is this correct?

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index