boolprob: Calculate predicted probabilities

In boolean3: Boolean Binary Response Models

Description

This function calculates predicted probabilities for the selected covariate profiles.

Usage

boolprob(obj, vars = NULL, newdata = NULL, k = 50, conf.int = FALSE,
  n = 100, as.table = TRUE, scales = list(x = list(relation = "free")),
  between = list(x = 1, y = 1), xlab = "x",
  ylab = "Predicted probability", ...)

Arguments

obj	object of boolean-class containing a fit boolean model.
vars	vector selecting a set of covariates from the fitted model. This can be a character vector of covariate names (as output from summary(obj)), or a numeric vector indexing the desired covariates.
newdata	data.frame with the same structure as model.matrix(boolean).
k	integer indicating the number of points at which the predicted probability should be calculated.
conf.int	logical; should confidence intervals be simulated.
n	number of draws to take from the estimated parameter space.
as.table	logical (default TRUE), to be passed to xyplot.
scales	list of settings for the scales argument passed to xyplot.
between	numeric specifying the space between panels.
xlab	string, the x-axis label.
ylab	string, the y-axis label.
...	Additional arguments to pass to xyplot. See that documentation for details.

Value

Returns an object of boolprob-class, the default action being to present the default plot.

Author(s)

Jason W. Morgan ([e-mail])

Examples

## Not run: 

## Note: This example assumes a boolean model has already been fit.

## Plot predicted probabilities for a fitted model. Either vars or
## newdata *must* be specified.
boolprob(fit, vars = c("x1_a", "x4_b"))
boolprob(fit, vars = c(2, 3, 4, 6))

## Specifying conf.int = TRUE produces simulated confidence intervals.
## The number of samples to pull from the distribution of the estimated
## coefficients is controlled by n; n=100 is default. This can take a
## while.
(prob <- boolprob(fit, vars = c(2, 3, 4, 6), n = 1000, conf.int = TRUE))

## End(Not run)

boolean3 documentation built on May 30, 2017, 12:30 a.m.