Dear all,
I would like to ask the advice of the community on ordinal probit models.
We want to fit a very simple regression model for an ordinal variable named
ttai_cgi coded 1, 2, and 3.
We specifically want to use a probit link to estimate this model, ie, an
ordinal probit model.
We have written a model in Winbugs (see below, we use the function phi as
the inverse of the probit function to write the model), but find rather
different results than in an ordinal probit model in SAS.
In the end, we are not sure that our model in Winbugs is a correct ordinal
probit model.
Would someone have experience in terms of simple ordinal probit models, and
could help us check our model or provide appropriate code of ordinal probit
model for comparison?
Moreover, we observe that the coefficients predicts the odds of lower values
rather than higher values in the ordinal variable, as we initially expected.
Thank you all!
Basile
model {
for (i in 1:6514) {
ystar.mu3[i]<- alpha3 + b3_1*Age[i] + b3_2[homme[i]] + s3[codiris[i]]
ystar3[i] ~ dnorm(ystar.mu3[i], tau3)I(-5,5)
Q3[i,1] <-phi( -ystar3[i])
Q3[i,2] <- phi(gamma3 - ystar3[i])
p3[i,1] <- Q3[i,1]
p3[i,2] <- Q3[i,2] - Q3[i,1]
p3[i,3] <- 1 - Q3[i,2]
ttai_cgi[i] ~ dcat(p3[i,])
}
tau3 ~ dgamma(1,1)
alpha3 ~ dflat()
b3_1 ~ dnorm(0,0.00001)
b3_2[1]<-0
b3_2[2] ~ dnorm(0,0.00001)
for (j in 1 : 1883)
{ s3[j] ~ dnorm(B3, TB3) }
B3 <- 0
TB3 ~ dgamma(0.1,0.1)
a3~dnorm(1,1)
gamma3 <-exp(a3)
}
-------------------------------------------------------------------
This list is for discussion of modelling issues and the BUGS software.
For help with crashes and error messages, first mail [log in to unmask]
To mail the BUGS list, mail to [log in to unmask]
Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html
Please do not mail attachments to the list.
To leave the BUGS list, send LEAVE BUGS to [log in to unmask]
If this fails, mail [log in to unmask], NOT the whole list
|