CLRtools (version 0.1.0)

delta.coefficient: Delta-beta hat percentage: Change in Coefficients when Adding a Variable

Description

This function checks the percentage change in the coefficient of a variable when it is added to a model. The comparison is made between two models: model1 (with the new variable included) and model0 (without the new variable).

Usage

delta.coefficient(model1, model0, variable = NULL)

Value

A vector of percentage changes in the coefficients between model1 and model0.

Arguments

model1

A fitted model (e.g., glm, lm) with the new variable included.

model0

A fitted model (e.g., glm, lm) without the new variable.

variable

A vector of variable names to check. If NULL (default), checks all coefficients.

Details

This function compares the coefficients of variables between two logistic regression models or conditional logistic regressions: one that includes a variable of interest (model1) and one that excludes it (model0). It calculates how much the coefficients of other variables change, expressed as a percentage. The change is calculated as ((beta0 - beta1) / beta1) * 100, where beta0 is the coefficient from model0 and beta1 is the coefficient from model1.

Examples

# Example from Hosmer et al., 2013
# Applied Logistic Regression (3rd ed.), Chapter 3, Table 3.10

mod3.1 <- glm(fracture ~ priorfrac, family = binomial, data = glow500)
mod3.2 <- glm(fracture ~ priorfrac + height, family = binomial, data = glow500)

# delta-beta-coefficient
delta.coefficient(model1 = mod3.2, model0 = mod3.1, variable = 'priorfrac')