Skip to content

Spatial Panel Multinomial Probit

2 messages · Michael Ndwiga, Roger Bivand

Original
Roger Bivand
# 
Please do read Sarah Goslee's sensible reply to a different thread:

https://stat.ethz.ch/pipermail/r-sig-geo/2013-October/019577.html

see:

https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20131017/8bacc39c/attachment.pl

because the thread was in HTML.

Although it is flattering that users seem to believe that all new methods 
are implemented first by "somebody" in R and released as packages, this is 
not the case. Many methods are still unfortunately published as working 
papers or published articles with no implementation (so would nevver be 
published in statistics journals). Those who also write and provide code 
are often sceptical when faced with methods proposals with no way of 
checking whether an implementation actually matches the method.

When asking about new and speculative methods, look first in the usual 
search mechanisms, and if asking on the list, do provide clear and 
complete references to papers/articles so that it is clear what you are 
asking about. Showing whether there is anything like what you ask about is 
also sensible, spatial probit is in the McSpatial and spatialprobit 
packages, but they are neither panel nor multinomial. My guess is that you 
are looking at land use change, but there you probably also need to worry 
about the units of observation, and possible non-linear covariates. A 
brief description (one sentence) of why you are asking also helps.

Hope this clarifies,

Roger
On Fri, 18 Oct 2013, Michael Ndwiga wrote: