In Rkabacoff/oneway: One-way Analysis of Variance

knitr::opts_chunk$set(
  collapse = TRUE, message=FALSE, warning=FALSE,
  comment = "#>"
)

Introduction

The *oneway package provides a simple interface for completing a one-way Analysis of Variance (ANOVA). This is useful when evaluating mean differences between groups on a quantitative variable.

Setup and results table

First the model is fit. In this example, we want to know if there is a difference in fuel efficiency based on the number of cylinders in a car engine. The data come from the mtcars dataset.

Summary statistics and the ANOVA results are printed.

library(oneway)
fit <- oneway(mpg ~ cyl, data = mtcars)
print(fit)

These results can be summarized using the summary generic function.

summary(fit)

Plotting the results

Group differences can be visualized using three types of plot - boxplot, violin plot, and ridges plots.

plot(fit, plot="boxplot")

plot(fit, plot="violin")

plot(fit, plot="density_ridges")

Test Assumptions

The results assume normality of the outcome variable (mpg) and equal variances on this variable for the three groups. We can test this with the assumptions function.

assumptions(fit)

Here, we can see the the normality assumption is met, but 4 cylinder cars have greater variance.

Post hoc comparisons

The ANOVA test indicates that the groups are different, but not where the differences lie. The posthoc function performs pairwise comparisons, controlling for the number of tests performed.