Prediction Intervals for Log-Link and Logit-Link Generalised Linear Models

Teaching helper for predictions from log-link and logit-link generalised linear models. It wraps predict.glm with standard errors and returns fitted values with confidence limits on either the link scale or the response scale.

Usage

predictGLM(object, newdata, type = "link", cilevel = 0.95, quasit = FALSE, ...)

Arguments

object: a glm object, i.e. the output from glm.
newdata: prediction data frame.
type: "link" (default) or "response" for estimates and confidence intervals on the linear predictor or response scale.
cilevel: confidence level for the intervals.
quasit: if TRUE, use a t multiplier rather than a normal multiplier for confidence intervals when object is a quasi model.
...: optional arguments that are passed to predict.glm.

Value

A data frame with columns fit, lwr, and upr containing fitted values and confidence limits on the requested scale.

Details

New work should use predict20x, which handles lm and glm fits with one function and gives results on the response scale by default. predictGLM() keeps its own defaults and column names for existing teaching material.

Note: newdata must include all first-order terms used in the fitted model. This simplified requirement reflects the teaching-wrapper interface and is not a complete reproduction of predict.glm().

Note

Deprecated in s20x 3.4.0. Use predict20x instead.