[R-sig-Geo] unbalanced panel in spatial panel data models

[Patrick Meyfroidt]

Dear Francis,

thank you very much.
No in fact I did not consider multiple imputation, and have to admit that
I don't know what it is.

Now, I see that I can calculate these multiple imputations using Amelia's
package, right? I suppose I'll figure out how.

After that, I'm not entirely sure how I have to proceed.
Do I have to run my analyses multiple times using the multiple imputed
datasets, and then combine the results? But I don't know how to combine
the results?
Do you have any resource that would help me do that?

Thanks,

patrick


> Hi Patrick,
>
> I have no experience with unbalanced spatial panels, but an alternative to
> doing a complete case analysis might be multiple imputation, which has
> been
> adapted for longitudinal designs by the biostatistics community.
>
> You probably already have already thought of this, but I thought I'd raise
> the option just in case.
>
> Cheers,
>
> Francis
>
> On 19 September 2012 22:50, Patrick Meyfroidt <
> patrick.meyfroidt at uclouvain.be> wrote:
>
>> Hello,
>>
>> I have a spatial panel database (regions over years), over which I would
>> like to run linear and most probably logistic regressions also.
>>
>> I started to use the package splm (Giovanni Millo, Millo and Piras
>> 2012),
>> calculating the spatial weights with spdep.
>> But it appears that splm only deals with balanced panels.
>>
>> Is there another package or way to make spatial panel regressions, or
>> another way to deal with possible spatial autocorrelation effects, with
>> unbalanced panel data in R?
>>
>> I know that I can manually remove incomplete regions or years, but given
>> the data I have this will be very costly (in terms of number of
>> remaining
>> observations).
>>
>> Thank you very much for considering the message, and for any help.
>>
>> Patrick Meyfroidt
>> ------
>> F.R.S. - FNRS and
>> Université catholique de Louvain - UCLouvain
>> Earth and Life Institute
>> Georges Lemaître Centre for Earth and Climate Research (TECLIM)
>> Place Pasteur 3, bte L4.03.08
>> 1348 Louvain-la-Neuve
>> Belgium
>> http://www.uclouvain.be/teclim
>> http://www.redd-alert.eu/
>>
>> _______________________________________________
>> R-sig-Geo mailing list
>> R-sig-Geo at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>>
>