Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: RE: clustering option after ivprobit


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE: st: RE: clustering option after ivprobit
Date   Wed, 3 Sep 2008 15:51:09 +0100

Your code is not what Austin recommended. 

As I understand it, bootstrapping panel data must produce samples that
no longer look like panels, as sampling with replacement naturally
implies that some observations will occur twice or more. Unless there is
some extra white magic, such samples will be indigestible to commands
expecting panel data. 

Nick 
n.j.cox@durham.ac.uk 

Choudry T Shehzad

Thanks Austin:

Yes, lag operator might have been a cause and seems intuitive as well
but  problem still remains the same.  Same error message again with
bootstrap

ivprobit dependent x y l.z (zz = mz pz), twostep
vce(bootstrap,cluster(country))

repeated time values within panel
the most likely cause for this error is misspecifying the
cluster(),idcluster(), or group() option
r(451);

The only thing that comes to my mind is that data structure is defined
as panel before ivprobit but even in that case why it would produce
cluster-robust vce but not cluster-robust bootstrapped vce.


Austin Nichols wrote:

> When you resample clusters, you can no longer use lag operators since
> -tsset- would return an error with multiple instances of each time
> period with the same cluster id.  You could write your own program to
> bootstrap that handles redoing the -tsset- with a new id, or simply
> generate the lag before doing the bootstrap: try
>
> g lagz=l.z
> ivprobit dependent x y lagz (zz = mz pz), vce(cluster) twostep first
>
> On Wed, Sep 3, 2008 at 6:19 AM, Choudry T Shehzad <C.T.Shehzad@rug.nl>
wrote:
>   
>> Brian P. Poi wrote:
>>     
>>> On Tue, 2 Sep 2008, Choudry T Shehzad wrote:
>>>
>>>       
>>>> Thanks Nick:
>>>> In help file for ivprobit, they do ask for cluster variable but
even
>>>> without that stata says "vcetype 'cluster' not allowed"
>>>>
>>>> ivprobit dependent x y l.z (zz = mz pz), vce(cluster) twostep first
>>>> vcetype 'cluster' not allowed
>>>> r(198);
>>>>
>>>> ivprobit dependent x y l.z (zz = mz pz), vce(cluster country )
twostep
>>>> first
>>>> vcetype 'cluster' not allowed
>>>> r(198);

*
*   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   |   What's new   |   Site index