Skip to contents

jlst

This package is used to fit joint mean (location) and variance (scale) tests, i.e. joint location-and-scale tests. The package also has functions to perform variability tests using the Breusch-Pagan and Brown-Forsythe methods.

Functions

  • jlssc: joint location-and-scale score test.
  • jlsp: joint location-and-scale test using Fisher’s method.
  • vartest: variability tests with Breusch-Pagan or Brown-Forsythe methods.

Installation 

install.packages("remotes")
remotes::install_github("jrs95/jlst")

Example 

# Libraries
library(jlst)

# Data  
x <- rbinom(1000, 1, 0.5)
y <- 0.5 + 0.025 * x + rnorm(1000, 0, sqrt(0.005 * x)) + rnorm(1000, 0, 0.1)

# Variance test  
vartest(y, x = as.factor(x), type = 1) # Breusch-Pagan test
vartest(y, x = as.factor(x), type = 2) # Brown-Forsythe test

# Joint location-and-scale test using Fisher's method
jlsp(y, x = as.factor(x), var.type = 1) # Breusch-Pagan variance test
jlsp(y, x = as.factor(x), var.type = 2) # Brown-Forsythe variance test

# Joint location-and-scale score test   
jlssc(y, x = as.factor(x), type = 1) # Breusch-Pagan (BP) variance test
jlssc(y, x = as.factor(x), type = 2) # Brown-Forsythe (BF) variance test
jlssc(y, x = as.factor(x), type = 3) # Method of moments version of the test with the BP variance test
jlssc(y, x = as.factor(x), type = 4) # Method of moments version of the test with the BF variance test

Citation

Staley JR, Windmeijer F, et al. A robust mean and variance test with application to high-dimensional phenotypes. Eur J Epidemiol 2022;37(4):377-387.