CLRtools (version 0.1.0)

confidence.interval: Compute Wald-Based Confidence Intervals for Logit and Predicted Probability

Description

Calculates Wald-based confidence intervals for the predicted logit values and estimated probabilities for new data points, based on a fitted logistic regression model.

Usage

confidence.interval(model, data, confidence.level = 0.95)

Value

A list with two elements:

logit

A data frame with predicted logit values and their lower and upper confidence limits.

estimated.prob

A data frame with predicted probabilities and their lower and upper confidence limits.

Arguments

model

A fitted logistic regression model from glm() with family = binomial.

data

A data frame containing the new observations for which predictions and confidence intervals are needed.

confidence.level

Confidence level for the interval estimates. Defaults to 0.95.

Examples

# Example from Hosmer et al., 2013
# Applied Logistic Regression (3rd ed.), Chapter 2

# Fit logistic regression model with selected predictors, Table 2.3
mod2.3 <- glm(
  fracture ~ age + priorfrac + raterisk,
  family = binomial,
  data = glow500
)

# Create a new data point for prediction
new.x <- data.frame(
  age = 65,
  priorfrac = "Yes",
  raterisk = "Same"
)

# Compute the 95% confidence interval for the predicted probability
confidence.interval(mod2.3, data = new.x, confidence.level = 0.95)