[Openspace] Geoda Vs Spatial Econometrics Toolbox

[Mihai Nica]

I worked with conty level shapefiles and most of the time they have missing 
observations. If the two software handle missing observations differently 
(and they do many times) the results are of course different. I would run a 
simple OLS diagnostic regression and compare all results (including the 
degrees of freedom) and then go further... but then again, I might be wrong 
:-)

Mihai Nica, ABD
Jackson State University
170 East Griffith Street G5
Jackson, MS 39201
601 914 0361





>From: Anil Rupasingha <arupasingha at aus.edu>
>To: openspace at sal.uiuc.edu
>Subject: [Openspace] Geoda Vs  Spatial Econometrics Toolbox
>Date: Thu, 22 Sep 2005 08:34:25 +0400
>
>Dear All:
>I have a question for those who are familiar with Geoda and Lesage's
>Spatial Econometrics Toolbox in Matlab.  I am a learner in both but have
>used Toolbox more often.  I have a dataset on USA Counties that has
>about 3041 observations.  I have run spatial models first using Toolbox
>programs.  I used make_nnw() a nearest neighbor spatial weight matrix
>and xy2cont() creates a contiguity matrix alternatively.  All
>specifications returned the results indicating that there are both forms
>of spatial effects in data (lag and error) in the data.  I also ran the
>likelihood ratio test (to test for spatial error) and it was highly
>significant.  Then I used Geoda and ran (using several forms of W)
>robust LM test to separately identify the extent of the error vs. lag
>effects and results were not significant.  The univariate Moran I ran
>using my dependent variable has an I = .04.  How do you explain these
>differences?  Thanks in advance for any clarification.
>
>Anil Rupasingha
>Department of Economics
>American University of Sharjah
>Sharjah 26666
>U.A.E.
>www.aus.edu
>
>


>_______________________________________________
>Openspace mailing list
>Openspace at sal.uiuc.edu
>http://sal.uiuc.edu/mailman/listinfo/openspace