ordprobit.univar {mprobit}R Documentation

Maximum Likelihood for Ordinal Probit Model

Description

Maximum Likelihood for Ordinal Probit: Newton-Raphson minimization of negative log-likelihood.

Usage

ordprobit.univar(x,y,iprint=0,maxiter=20,toler=1.e-6)

Arguments

x vector or matrix of explanatory variables. Each row corresponds to an observation and each column to a variable. The number of rows of x should equal the number of data values in y, and there should be fewer columns than rows. Missing values are not allowed.
y numeric vector containing the ordinal response. The values must be in the range 1,2,..., number of categories. Missing values are not allowed.
iprint logical indicator, default is FALSE, for whether the iterations for numerical maximum likelihood should be printed.
maxiter maximum number of Newton-Raphson iterations, default = 20.
toler tolerance for convergence in Newton-Raphson iterations, default = 1.e-6.

Details

If ordprobit for repeated measures ordinal probit fails to converge from the simple starting point in that function, this function ordprobit.univar should provide a better starting point. It is also equivalent to ordprobit with an identity latent correlation matrix.

The ordinal probit model is similar to the ordinal logit model (proportion odds logistic regression : polr in library MASS), The parameter estimate of ordinal logit are roughly 1.8 to 2 times those of ordinal probit (the signs of the parameters in polr may be different, as this function may be using a different orientation for the latent variable.

Value

list of MLE of parameters and their associated standard errors, in the order cutpt1,...,cutpt(number of categ-1),b1,...b(number of covariates).

negloglik value of negative log-likelihood, evaluated at MLE
cutpts MLE of ordered cutpoint parameters
beta MLE of regression parameters
cov estimated covariance matrix of the parameters

References

Anderson, J.A. and Pemberton, J.D. (1985). The grouped continuous model for multivariate ordered categorical variables and covariate adjustment. Biometrics, 41, 875-885.

Examples

data(ordinalex)
x=as.vector(ordinalex$x)
y=ordinalex$y
ord.univar = ordprobit.univar(x,y)
print(ord.univar)
startp=c(ord.univar$cutpts,ord.univar$beta,0.5)
ord.exch <- ordprobit.exch(x,y,ordinalex$id,iprint=0,startpar=startp)
print(ord.exch)

[Package mprobit version 0.9-3 Index]