munroe.9 at osu.edu
Wed Apr 23 12:11:42 CDT 2008
See the GeoDa manual - Dr. Anselin recommends the use of a permutation
approach (rather than comparing against a z-distribution) in order to
determine significance. Right-click on your scatterplot, and select
"Randomization". The default is 99 permutations, but I'd choose 199 or 499.
The p-value is partly a function of the number of permutations. You can see
the histogram to show where your observed distribution of values is (yellow)
relative to a histogram of random permutations. This is more robust than a
z-stat, because you don't have the distributional assumptions of normality
do not have to hold.
The permutations should also give you an expected value (bottom of the
window), so if you really want a z-test, you can use that expected value to
calculate the z-stat yourself (not too awfully hard).
- Darla Munroe
From: openspace-bounces at sal.uiuc.edu [mailto:openspace-bounces at sal.uiuc.edu]
On Behalf Of Ola Claësson
Sent: Wednesday, April 23, 2008 1:04 PM
To: openspace at sal.uiuc.edu
Subject: [Openspace] z-score
Hi there everyone!
does anyone of you know how to attain the z-score in geoda. And note that I
do not want the z-score of the residuals from a regression but the z score
from an ordinary calculation of Morans I. Please help!
Ola Claësson, phd student, Lund, Sweden
Openspace mailing list
Openspace at sal.uiuc.edu
More information about the Openspace