Message-ID: <alpine.LFD.2.21.1907110917120.9807@reclus.nhh.no>
Date: 2019-07-11T07:27:50Z
From: Roger Bivand
Subject: LR1.sarlm specifications
In-Reply-To: <CANcbSEgDVsFjHR8TkzD2FCVab25WX3bTPL8F6YWpeLWzW2z2Zw@mail.gmail.com>
On Thu, 11 Jul 2019, Amitha Puranik wrote:
> Hi Prof. Roger,
>
> I am using the approach proposed by Prof Paul Erhorst for choosing a
> spatial model in his paper '*Applied spatial econometrics: raising the
> bar*' . As per the strategy, one has to check the likelihood ratio test
> for theta (spatial autocorrelation in exogenous (independent) variables)
> and also in theta+rho*beta (spatial autocorrelation in residuals).
> Suppose I fit a spatial durbin model and use the code LR1.sarlm(sp.dm),
> how would I know whether the likelihood ratio test checks for
> autocorrelation in dependent variable or autocorrelation in the
> independent variable?
The spatialreg::LR1.sarlm() test simply between the fitted model and the
same model assuming the spatial coefficients are zero, so it only tests
the possible benefit of including (a) spatial process(es).
spatialreg::LR.sarlm() lets you test between nested models, and works like
lmtest::lrtest(). The models need to be nested, so you can test SDM/SLM,
SDM/SEM (equivalent to a Common Factor test), and so on, but only if the
models nest (not SEM/SLM, because they do not nest).
Hope this helps,
Roger
>
> Thanks in advance.
> Amitha Puranik.
>
> [[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; e-mail: Roger.Bivand at nhh.no
https://orcid.org/0000-0003-2392-6140
https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en