hyreg: Perform hybrid tobit-logit regression

In WannabeSmith/hyreg: Perform Hybrid Tobit-Logit Regression

Description

This function gets estimates of beta, sigma, and theta in the hybrid tobit-logit model. That is, when the logit betas are a scalar multiple of the tobit betas.

Usage

hyreg(formula.tobit, formula.discrete, data.tobit, data.discrete,
  start.beta = NULL, start.sigma = NULL, start.theta = 1,
  start.gamma = NULL, left = -1, method = "Nelder-Mead",
  ch.terms = NULL, ch.link = "log")

Arguments

formula.tobit	a regression formula describing the relationship between the tobit response and the covariates
formula.discrete	a regression formula describing the relationship between the bernoulli response and the covariates
data.tobit	the data.frame containing the tobit responses and covariates
data.discrete	the data.frame containing the bernoulli responses and covariates
start.beta	a numeric vector of starting values for beta. If not specified, start.beta is taken from a non-hybrid tobit model
start.sigma	a numeric starting value for sigma. If not specified, start.sigma is also taken from a non-hybrid tobit model
start.theta	starting value for theta. This must be specified
start.gamma	starting value for gamma. This must be specified.
left	a number specifying where left-censoring occurred
method	a string specifying the optimization routine to be used by optim
ch.terms	a vector of names of variables to be included for conditional heteroscedasticity. An intercept will be included by default
ch.link	one of "log", "quadratic", or "identity" indicating the type of link to be used for conditionally linear heteroscedasticity

Value

a list containing the following parameter estimates (from maximum likelihood):

beta	the regression coefficients
sigma	the standard deviation of the censored normal distribution
theta	the multiplicative factor relating the two sets of regression coefficients

WannabeSmith/hyreg documentation built on July 22, 2019, 5:08 p.m.