[Openspace] Spatial Lag Model in GeoDa versus SAS

[Julia Koschinsky]

Krista,

 

GeoDa implements maximum likelihood estimation for both its lag and error
models. Make sure your weights are row-standardized – that has accounted for
small differences between SAS and GeoDa results before. Here is the
reference for the technical details of the models implemented in GeoDa:

 

Smirnov, O. and Anselin, L. (2001). Fast maximum likelihood estimation of
very large spatial autoregressive models: A characteristic polynomial
approach. Computational Statistics and Data Analysis, 35:301-319.

 

Julia

 

  _____  

From: openspace-bounces at sal.uiuc.edu [mailto:openspace-bounces at sal.uiuc.edu]
On Behalf Of Krista.Collins at statcan.ca
Sent: Tuesday, September 20, 2005 2:50 PM
To: openspace at sal.uiuc.edu
Subject: [Openspace] Spatial Lag Model in GeoDa versus SAS

 

Hi 
  I am working at Statistics Canada on a spatial project predicting crime
rate based on various socio-economic variables. We know the data is
spatially autocorrelated and are attempting to fit the spatial lag model to
the data.  I am able to produce the spatial lag model for my data in GeoDa,
but I am trying to also do the analysis in SAS.  From my SAS model I get
similar values for the regression coefficients and for Moran's I statistic,
but they are not identical.  In SAS, my model essentially just adds the
lagged dependent variable as a predictor in the model and does not change
the covariance structure of the error terms.  Would this account for the
difference? How does GeoDa change the covariance of the error terms? What
type of model does it use?

Any help you can offer would be greatly appreciated. 

Thank-you, Krista 

Krista Collins 
613-951-0667 | facsimile / télécopieur 613-951-3100 
Krista.Collins at statcan.ca 
Statistics Canada | 120 Parkdale Avenue Ottawa ON K1A 0T6 
Statistique Canada | 120 avenue Parkdale Ottawa ON K1A 0T6 
Government of Canada | Gouvernement du Canada 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://sal.uiuc.edu/pipermail/openspace/attachments/20050920/80b93a7f/attachment.html