Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


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

Re: st: comparing policies across countries: multilevel estimation?


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: comparing policies across countries: multilevel estimation?
Date   Wed, 15 Dec 2010 19:23:47 -0500

On Wed, Dec 15, 2010 at 6:48 AM, Laura R. <laura.roh@googlemail.com> wrote:
> Dear Stata users,
>
> my question is whether I have to use multilevel models or not for my
> estimations.
>
> I have panel data with individual data, and I added some variables on
> the country level, e.g. the unemployment rate and some dummies for
> different policies. Several countries may have the same policies or
> not.
>
> As my aim is the estimation of the impact of these policies on the
> dependent variable, I wonder if I have to estimate multilevel models
> or not. As I have understood it, in multilevel models you cluster
> groups of individuals, e.g. by country.

Not quite. As Austin said, you need to clarify what you mean by
"multilevel models", as these terms might mean different things to
different people. What is the discipline that you work in? Are there
publications in your discipline that use multilevel models? In what
way do they use them? What have you read on multilevel models?

> But, as my aim is to find out
> about differences on the policies on the country level, wouldn't the
> clustering by country somehow cancel out (i.e. abolish) differences in
> the policies between the countries?

Depends on what exactly you will end up doing. I wouldn't really
expect this to happen.

> If I had to use multilevel models, are they also appropriate for a
> two-stage selection process? If so, I would have to use -xtmelogit- in
> the first step and -xtmixed- in the second step, wouldn't I?

What is your selection based on? If you really want to model a
multilevel selection process (which frankly I have never seen done,
although I can imagine in might make sense in some applications), you
would probably have to use -gllamm- for simultaneous estimation of
both the selection and the main equation at multiple levels.

-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index