mixequiv

Equivalence testing of (generalized) linear mixed model fixed effects parameters

Installation

devtools::install_github("w-decker/mixequiv")

Usage

Some libraries we will use

library(mixequiv)
library(lme4, quietly = T)
library(lmerTest, quietly = T) # MUST LOAD `lmerTest` to estimate DoF!
library(tidyverse, quietly = T)

Load some data

data(iris) 

df <- iris %>%
  select(Sepal.Width, Petal.Width, Species) %>%
  mutate(
    Species = as.factor(Species)
  )

Define linear model

model <- lmer(Sepal.Width ~ Petal.Width + (1 | Species), 
              data = df, REML = F, control = lmerControl(optimizer = "bobyqa"))

Define minimal meaningful difference (MMD)

mmd <- 0.3

Test equivalence

results <- test_equiv(model, term = "Petal.Width", method = "TOST", mmd = mmd)

See results

summary(results)

Equivalence Test Summary
-------------------------
  Estimate: 0.741
  Degrees of Freedom: 147.399
  Standard Error: 0.119

TOST results:
  t1: 8.779, p1: 0.000
  t2: 3.717, p2: 1.000

Alpha level: 0.05 
Result: Not Equivalent