glm.cis: Compute confidence intervals for predictions.

In GLMpack: Data and Code to Accompany Generalized Linear Models, 2nd Edition

Description

Apply an exponential transformation to the confidence intervals and predictions from binomial and Poisson models.

Usage

glm.cis(preds, ses, alpha, df)

Arguments

preds	The predictions based on the additive linear component of the model.
ses	The standard error(s) of the prediction.
alpha	The desired confidence level.
df	The desired degrees of freedom.

Value

The output is a matrix.

Examples

data(campaign)
attach(campaign)
cmpgn.out <- glm(TOTCONTR ~ CANDGENDER + PARTY + INCUMCHALL + HISPPCT,
                 family=Gamma(link = 'log'), data=campaign)
newdat_gender <- data.frame(CANDGENDER = c('F','M'), PARTY= rep('Democrat',2),
                            INCUMCHALL=rep("C", 2), HISPPCT=rep(mean(campaign$HISPPCT),2))
preds_gender <- predict(cmpgn.out, newdata = newdat_gender, se.fit = TRUE)
glm.cis(preds_gender$fit, preds_gender$se.fit, 0.95,cmpgn.out$df.residual)

GLMpack documentation built on July 19, 2019, 5:05 p.m.