predProb: Predicted probabilities and confidence intervals from logit...
In jgabry/QMSS_package: Quantitative Methods in the Social Sciences

Description Usage Arguments Value Author(s) Examples

Predicted probabilities and confidence intervals from logit or probit model

1	predProb(model, predData, ci = TRUE, level = 0.95)

`model`	A `glm` model fit with `binomial` family and either a `logit` or `probit` link function.
`predData`	A data frame to pass to `predict.glm` in which to look for variables with which to predict.
`ci`	Logical value indicating whether to compute confidence intervals.
`level`	The confidence level to use if `ci` is `TRUE`.

A data frame with predData and the associated predicted probabilities. Confidence intervals are included if argument ci is TRUE.

Jonah Gabry <jsg2201@columbia.edu>

GSS_2010$Y <- with(GSS_2010,
                   cut(realinc,
                   breaks=c(-Inf, median(realinc, na.rm = T), Inf),
                   labels=c("Low", "High")))
logitmodel <- glm(Y ~ age + educ, data = GSS_2010, family = binomial)
probitmodel <- glm(Y ~ age + educ, data = GSS_2010, family = binomial(link = "probit"))
predData <- data.frame(age = 20, educ = 15)
predProb(logitmodel, predData, ci = F)
predProb(probitmodel, predData, ci= F)
predData <- expand.grid(age = c(20, 50, 80), educ = c(5, 10, 15))
predProb(logitmodel, predData, ci = T)
predProb(probitmodel, predData, ci= T)