[R-sig-Geo] question about nb2listw row standardization

[Roger Bivand]

On Tue, 4 Aug 2015, Mueller,Drew wrote:

> I was hoping for a clarification on the row standardization of nb2listw. 
> I have implemented a nb list based on distance, and a function
>
> dlist <- nbdists(col.gal.nb, coords)
> dlist <- lapply(dlist, function(x) 1/x)
>
> my question is, if I call
>
> Wa <- nb2listw(x, glist = dlist, style = "W")
> Wb <- nb2listw(x, glist = dlist, stlye = "B")
>
> is Wa then row standardized based on inverse distance or does the glist 
> override this and I get actual undstandardized inverse distance weights? 
> Also, would Wb actually be binary, or does the glist argument override 
> this to make the weights 1/x instead of 1's and 0's?  If my goal is to 
> maintain the actual inverse distance weight without standardization (for 
> better economic interpretation), does Wb give me an unstandardized W, or 
> do I need to call a different function to avoid row standardization? 
> If my goal IS to row standardize a weights matrix with weights based on 
> inverse distance, then is Wa the correct function call?  Mainly I was 
> having trouble interpreting the help file for nb2listw as to whether 
> style = "B"  always creates 1's and 0's regardless of what glist is set 
> to.  Thanks

Please do check:

library(spdep)
example(columbus)
coords <- coordinates(columbus)
dlist <- nbdists(col.gal.nb, coords)
dlist <- lapply(dlist, function(x) 1/x)
Wa <- nb2listw(col.gal.nb, glist = dlist, style = "W")
Wb <- nb2listw(col.gal.nb, glist = dlist, style = "B")
summary(sapply(Wa$weights, sum))
summary(sapply(Wb$weights, sum))

"W" row standardises the general weights, "B" passes through general 
weights. IDW are perverse, and without - contrary to your assertion - 
economic meaning, because the scaling depends closely on the metric of 
the coordinates:

summary(unlist(dlist))

are all < 1.

Roger

>
> Drew
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

-- 
Roger Bivand
Department of Economics, Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; fax +47 55 95 91 00
e-mail: Roger.Bivand at nhh.no