Inverse distance weights and spatial error model

neigh.dist <- nbdists(neigh.nb, coords, longlat=TRUE)
inverse <- lapply(neigh.dist, function(x) (1/(x^2)))

neigh.listw <- nb2listw(neigh.dist, glist=inverse,  
style="W",zero.policy=TRUE)

Hi Everyone,
I'm new to R, so thank you in advance for your patience.  I'm using  
US census block groups in Buffalo New York to examine how  
neighborhood characteristics affect crime.  I would like to use an  
inverse distance weights (distance decay) of block group centroids  
that is banded at 1 kilometer.  In other words, I would like to  
create a one kilometer buffer around each of the centroids, then use  
row standardized inverse distance weights.  Finally, I would like to  
run a spatial error model using these weights.

Here is my code:

library(foreign)
library(spdep)

buffalo <- read.dta("buffalo_blkgrps.dta")
attach(buffalo)
names(buffalo)

[1] "bgidfp00"     "gage2900"     "gavghhinc00"  "gblack00"      
"gcrwd00"
[6] "gethhet00"    "ghhincsdln00" "glatino00"    "gocc00"        
"gowner00"
[11] "gpop00"       "gpov00"       "assaul"       "robber"        
"burglr"
[16] "motveh"       "murder"       "larcen"       "x"            "y"

#distance based neighbors - making a neighbor list - bounded to 1  
kilometer
coords <-(cbind(x,y))
neigh.nb <- dnearneigh(coords, 0, 1, longlat=TRUE)
summary(neigh.nb)

Neighbour list object:
Number of regions: 409
Number of nonzero links: 5298
Percentage nonzero weights: 3.167126
Average number of links: 12.95355
1 region with no links:
12
Link number distribution:

0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22  
23 24 25 26
1  1  3  3  4  5 14 25 28 23 19 26 32 44 29 24 29 25 24 13 15 12  3   
4  1  1  1
1 least connected region:
3 with 1 link
1 most connected region:
382 with 26 links

#making inverse distance weights
neigh.dist <- nbdists(neigh.nb, coords, longlat=TRUE)
inverse <- lapply(neigh.dist, function(x) (1/(x^2)))

#creating row standardized spatial weights from list
neigh.listw <- nb2listw(inverse, style="W",zero.policy=TRUE)

Error in nb2listw(inverse, style = "W", zero.policy = TRUE) :
 Not a neighbours list

traceback()

2: stop("Not a neighbours list")
1: nb2listw(inverse, style = "W", zero.policy = TRUE)

assault <- errorsarlm(assaul_r ~ gage2900 + gblack00 + gcrwd00 +  
gethhet00 + ghhincsdln00 + glatino00 + gocc00+ gowner00 + gpov00,

+ data=buffalo,listw=neigh.listw, zero.policy=TRUE)
Error in eval(expr, envir, enclos) : object 'assaul_r' not found

The model will run when I do not include the code with the inverse  
weights and only interpoint distances, but it's unclear to me why I  
can't include the inverse distances when using nb2listw.  Any ideas  
on why this is occurring and/or help to alleviate this issue would  
be much appreciated.  The spatial error model from errorsarlm  
(package=spdep) will also run fine when I don't include the inverse  
weights.  Is there a better way to go about running this spatial  
error model using row standardized inverse distance weights?   Would  
spautolm from the spdep package be better suited for these data?   
Any comments or suggestions are welcome.
Thank you for your time!

Adam

Adam Boessen
Doctoral Student
Department of Criminology, Law and Society
University of California, Irvine
aboessen at uci.edu

p.s. If it's helpful, here is my version of R:

version

              _
platform       i386-apple-darwin9.8.0
arch           i386
os             darwin9.8.0
system         i386, darwin9.8.0
status
major          2
minor          10.1
year           2009
month          12
day            14
svn rev        50720
language       R
version.string R version 2.10.1 (2009-12-14)




	[[alternative HTML version deleted]]