Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Fwd: st: logit fe with svy panel data.

From (Jeff Pitblado, StataCorp LP)
Subject   Re: Fwd: st: logit fe with svy panel data.
Date   Tue, 20 Dec 2011 12:09:01 -0600

Sumedha Gupta <> sent me a follow-up question, which
I will share here:

> Thank you so much for your reply. I need a clarification though. What does
> 'levels of variable 'hhidpn' are
> > nested within the final stage sampling units' mean? hhidpn is the
> > individual level id number and I want to run an individual level fixed
> > effects model. Would that work?

Sumedha didn't show us how the data is -svyset-, so I felt I should mention
that -svy: clogit- puts certain restrictions on how the data is -svyset-; most
other -svy- estimation commands do not have this type of restriction.

My comment was a warning to Sumedha that if there is a variable that
identifies the sampling units in the stage just before the 'hhidpn' units are
selected, then that variable must be -svyset-. Also, if individuals were
sampled in the final sampling stage, then 'hhidpn' needs to be -svyset- in
the final stage.

For example, suppose Sumedha's data was a simple random sample of individuals,
and 'sampwgt' is the name of the variable containing the sampling weights.
The proper -svyset- for this setup is

	. svyset hhidpn [pw=sampwgt]

Failure to specify 'hhidpn' as the PSU variable will cause -svy: clogit-
report the error

	the group() variable is not nested within the final stage
	sampling units

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index