social interactions using spatial analysis
Julia Koschinsky
Hello,
I am interested in examining social interactions using spatial analysis. However, I am relative newcomer to this field. My current understanding is that if a spatial lag model provides a better fit than a spatial error model this is indicative of interaction*. I also understand that there is a suggested specification search procedure whereby the significance of the LM test statistic provides a strong indication of whether spatial lag or spatial error is the proper alternative. My query is that my dataset has high levels of heteroskesdasticity. I have used GeodaSpace to correct for this, running the spatial lag and error models and it appears that the
latter is a better fit. I just want to check whether this is a sensible approach given my interest in social interactions. I am also wondering if I should also try an alternative approach - correcting the data for heteroskedasticity and then run it in a Geoda regression? (using the specification approach). If so, advice on how best to correct for heteroskedasticity in this context would be appreciated. I have a macro for HCSE (Hayes and Cai, 2007) installed on SPSS but as far as I am aware the outputs are not appropriate to use in a Geoda regression........
Many, many thanks
Alison
*Note I am aware of Moran's I and LISA, and the dataset suggests spatial autocorrelation. I am now accounting for the impact of co- variables.
The University of Aberdeen is a charity registered in Scotland, No SC013683.
************************
Julia Koschinsky, Ph.D.
Research Director
Assistant Research Professor
Arizona State University
School of Geographical Sciences and Urban Planning
GeoDa Center for Geospatial Analysis and Computation
julia.ko...@asu.edu
http://geodacenter.asu.edu
http://www.facebook.com/geodacenter
http://twitter.com/GeoDaCenter
Pedro Amaral
I'd recommend you to check GeoDa Space. The alpha release can be found here: http://geodacenter.asu.edu/geodaspace_alpha . The software provides LM/robust LM tests to help you to decide between error / lag models, HAC/White corrections for heteroskedasticity, along with the estimator proposed by Arraiz et al (2010) to model heteroskedasticity. In sum, you should be able to take care of everything using just that software.
I hope it helps,
Pedro.
---
Pedro Amaral
PhD Candidate in Land Economy - University of Cambridge
Visiting Student - GeoDa Center for Geospatial Analysis and Computation
--
You received this message because you are subscribed to the Google Groups "Openspace List" group.
To post to this group, send email to openspa...@googlegroups.com.
To unsubscribe from this group, send email to openspace-lis...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/openspace-list?hl=en.
Frank Howell
rhos...@u.washington.edu
Dick Hoskins
UW, Seattle
On Fri, 10 Feb 2012, Frank Howell wrote:
> Alison,
>
> You might also read Stewart Fotheringham's work on spatial interaction models....especially on handling origin
> vs destination characteristics, etc.
>
> Good luck and let us hear about your progress!
>
> Frank
>
> On Feb 10, 2012, at 9:45 AM, Pedro Amaral <pvm...@cam.ac.uk> wrote:
>
> Dear Alison,
>
> I'd recommend you to check GeoDa Space. The alpha release can be found here:
> http://geodacenter.asu.edu/geodaspace_alpha . The software provides LM/robust LM tests to help you
> to decide between error / lag models, HAC/White corrections for heteroskedasticity, along with the
> estimator proposed by Arraiz et al (2010) to model heteroskedasticity. In sum, you should be able
> to take care of everything using just that software.
>
> I hope it helps,
>
> Pedro.
>
> ---
> Pedro Amaral
> PhD Candidate in Land Economy - University of Cambridge
> Visiting Student - GeoDa Center for Geospatial Analysis and Computation
>
>
>
> On Thu, Feb 9, 2012 at 8:59 AM, Julia Koschinsky <jkos...@asu.edu> wrote:
> on behalf ofï¿œAlison Pridmore (alison....@abdn.ac.uk).
> Julia
>
>
>
> Hello,
>
> I am interested in examining social interactions using spatial analysis. However, I am relative
> newcomer to this field. ï¿œMy current understanding is that if a spatial lag model provides a better
> fit than a spatial error model this is indicative of interaction*. ï¿œI also understand that there
> is a suggested specification search procedure whereby the significance of the LM test statistic
> provides a strong indication of whether spatial lag or spatial error is the proper alternative.
> ï¿œMy query is that my dataset has high levels of heteroskesdasticity. ï¿œI have used GeodaSpace to
> correct for this, running the spatial lag and error models and it appears that the
> latter is a better fit. ï¿œI just want to check whether this is a sensible approach given my
> interest in social interactions. ï¿œI am also wondering if I should also try an alternative approach
> - correcting the data for heteroskedasticity and then run it in a Geoda regression? (using the
> specification approach). ï¿œIf so, advice on how best to correct for heteroskedasticity in this
> context would be appreciated. ￜ ￜI have a macro for HCSE (Hayes and Cai, 2007) installed on SPSS
> but as far as I am aware the outputs are not appropriate to use in a Geoda regression........
>
> Many, many thanks
>
> Alison
>
> *Note I am aware of Moran's I and LISA, and the dataset suggests spatial autocorrelation. ï¿œI am
Julia Koschinsky
The alpha GeoDASpace seems very nice. I managed to get something to run without any documentation which bodes well for the design in my opinion.
Dick Hoskins
UW, Seattle
On Fri, 10 Feb 2012, Frank Howell wrote:
Alison,
You might also read Stewart Fotheringham's work on spatial interaction models....especially on handling origin
vs destination characteristics, etc.
Good luck and let us hear about your progress!
Frank
On Feb 10, 2012, at 9:45 AM, Pedro Amaral <pvm...@cam.ac.uk> wrote:
Dear Alison,
I'd recommend you to check GeoDa Space. The alpha release can be found here:
http://geodacenter.asu.edu/geodaspace_alpha . The software provides LM/robust LM tests to help you
to decide between error / lag models, HAC/White corrections for heteroskedasticity, along with the
estimator proposed by Arraiz et al (2010) to model heteroskedasticity. In sum, you should be able
to take care of everything using just that software.
I hope it helps,
Pedro.
---
Pedro Amaral
PhD Candidate in Land Economy - University of Cambridge
Visiting Student - GeoDa Center for Geospatial Analysis and Computation
On Thu, Feb 9, 2012 at 8:59 AM, Julia Koschinsky <jkos...@asu.edu> wrote:
on behalf of Alison Pridmore (alison....@abdn.ac.uk).
Julia
Hello,
I am interested in examining social interactions using spatial analysis. However, I am relative
newcomer to this field. My current understanding is that if a spatial lag model provides a better
fit than a spatial error model this is indicative of interaction*. I also understand that there
is a suggested specification search procedure whereby the significance of the LM test statistic
provides a strong indication of whether spatial lag or spatial error is the proper alternative.
My query is that my dataset has high levels of heteroskesdasticity. I have used GeodaSpace to
correct for this, running the spatial lag and error models and it appears that the
latter is a better fit. I just want to check whether this is a sensible approach given my
interest in social interactions. I am also wondering if I should also try an alternative approach
- correcting the data for heteroskedasticity and then run it in a Geoda regression? (using the
specification approach). If so, advice on how best to correct for heteroskedasticity in this
context would be appreciated. I have a macro for HCSE (Hayes and Cai, 2007) installed on SPSS
but as far as I am aware the outputs are not appropriate to use in a Geoda regression........
Many, many thanks
Alison
*Note I am aware of Moran's I and LISA, and the dataset suggests spatial autocorrelation. I am
now accounting for the impact of co- variables.
The University of Aberdeen is a charity registered in Scotland, No SC013683.
--
************************
Julia Koschinsky, Ph.D.
Research Director
Assistant Research Professor
Arizona State University
School of Geographical Sciences and Urban Planning
GeoDa Center for Geospatial Analysis and Computation
julia.ko...@asu.edu
http://geodacenter.asu.edu
http://www.facebook.com/geodacenter
http://twitter.com/GeoDaCenter
--
You received this message because you are subscribed to the Google Groups "Openspace List" group.
To post to this group, send email to openspace-list@googlegroups.com.
To unsubscribe from this group, send email to openspace-list+unsubscribe@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/openspace-list?hl=en.
--
You received this message because you are subscribed to the Google Groups "Openspace List" group.
To post to this group, send email to openspace-list@googlegroups.com.
To unsubscribe from this group, send email to openspace-list+unsubscribe@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/openspace-list?hl=en.
--
You received this message because you are subscribed to the Google Groups "Openspace List" group.
To post to this group, send email to openspace-list@googlegroups.com.
To unsubscribe from this group, send email to openspace-list+unsubscribe@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/openspace-list?hl=en.
--
You received this message because you are subscribed to the Google Groups "Openspace List" group.
To post to this group, send email to openspace-list@googlegroups.com.
To unsubscribe from this group, send email to openspace-list+unsubscribe@googlegroups.com.