interval: Prediction and Confidence Intervals for glm Objects

In HH: Statistical Analysis and Data Display: Heiberger and Holland

Prediction and Confidence Intervals for glm Objects

Description

Prediction and Confidence Intervals for glm Objects

Usage

interval(glm.object, ...)
## S3 method for class 'glm'
interval(glm.object, linkfit.object,
         type = c("link", "response"),
         conf.level = 0.95, ...)

Arguments

glm.object	result from a call to the glm function.
linkfit.object	result from a call to the predict function for the glm.object with type="link", se.fit=TRUE.
type	Either "link" or "response". See predict.glm for details.
conf.level	Confidence level, for example .95 for 95%.
...	Other arguments to be passed to predict.glm.

Value

Matrix with five columns: fit, ci.low, ci.hi, pi.low, pi.hi and as many rows as predict.glm returns.

Author(s)

Richard M. Heiberger <[e-mail]>

Examples

data(spacshu)
spacshu.bin.glm <- glm(damage ~ tempF, data=spacshu, family=binomial)

## observed data
spacshu.interval          <- interval(spacshu.bin.glm)

## new data, link
spacshu.interval.link     <- interval(spacshu.bin.glm, newdata=data.frame(tempF=30:85))

## new data, response
spacshu.interval.response <- interval(spacshu.bin.glm, newdata=data.frame(tempF=30:85),
                                      type="response")

HH documentation built on May 29, 2024, 6:24 a.m.