Ok, it's helpful to know that I need to zoom in on those three things.

I created the Random Effects Matrix by hand in Excel, so I read that into R

and put the matrix into the format recommended here:

https://cran.r-project.org/web/packages/HSAR/vignettes/hsar.htmlBelow, I show how I made W, M, and Delta.mat, before I try to estimate the

model. Hopefully this helps.

constit<- readShapeSpatial("Population

Weighted/Constituencies_2008/20170209_Constit")

constit.nb<- poly2nb(constit, row.names = constit$X20160526_5)

ghana.constit.weights.binary<- nb2listw(constit.nb, style="B", zero.policy

= TRUE)

W.constit<- listw2mat(ghana.constit.weights.binary)

W.constit <- W.constit / rowSums(W.constit)

W.constit <- as(W.constit,"dgCMatrix")

dist2008<- readShapeSpatial("Population Weighted/Districts_2008/Volta

Variable/20170226_Districts")

dist2008.nb<- poly2nb(dist2008, row.names = dist2008$DIST_2008)

ghana.dist2008.weights.binary<- nb2listw(dist2008.nb, style="B",

zero.policy=T)

W.dist<- listw2mat(ghana.dist2008.weights.binary)

W.dist <- W.dist / rowSums(W.dist)

W.dist <- as(W.dist,"dgCMatrix")

Delta<- read.csv("Random Effects Matrix_Ghana.csv",

header = T, row.names = 1)

Delta.mat<- as.matrix(Delta)

Delta.mat <- as(Delta.mat,"dgCMatrix")

> HSAR.model1<- hsar(Count_ ~ ndc_pres_3

+ + volatility + turnout_21

+ + volatili_1 + X20160526_6

+ + DENSITY_RD + Count_3

+ + MEAN + pov_p_2008

+ + gini_2008 + ferat_2008

+ + Count_4 + literacy

+ + grid_perCa, data=constit, W=W.constit,

+ M=W.dist, Delta = Delta.mat,

+ burnin = 5000, Nsim = 10000,

+ thinning = 1, parameters.start = NULL)

Error in hsar(Count_ ~ ndc_pres_3 + volatility + turnout_21 + volatili_1 +

:

not an S4 object

On Thu, Sep 27, 2018 at 3:57 PM Roger Bivand <

[hidden email]> wrote:

> This code tells nothing, the problem is in your construction of W, M

> and/or Delta. Pleaseng show this code too, best as a reproducible example.

> Tip: sometimes running traceback() after an error shows where it happens.

>

> Roger Bivand

> Norwegian School of Economics

> Bergen, Norway

>

> Fra: Justin Schon

> Sendt: torsdag 27. september, 21.36

> Emne: [R-sig-Geo] Question about HSAR package

> Til:

[hidden email]
>

>

> Dear all, I am receiving the error "not an S4 object" when I attempt to

> estimate the hierarchal spatial auto-regressive model from the HSAR

> package. I have attempted several ways of creating the lower level matrix

> and higher level matrix. Rather than asking if members of this list can

> help with the code, I am first wondering if anyone can explain why this

> error would appear. I am including the code that estimates the model, as

> well as the error, below: > HSAR.model1

>

--

Justin Schon

Post-Doctoral Researcher on Environmental Change and Migration

MURI Migration Research Team:

http://murimigration.org/University of Florida

Fellow, Initiative for Sustainable Energy Policy (ISEP)

[[alternative HTML version deleted]]

_______________________________________________

R-sig-Geo mailing list

[hidden email]
https://stat.ethz.ch/mailman/listinfo/r-sig-geo