R Tutorial - #11 - Factorial ANOVA

Sections:
0:00 Introduction
1:08 Import the data set (Alcohol and Self-control Study) and create a data frame
1:41 Type of sums of squares in R: A caution!
8:02 Why you need to manually set orthogonal contrasts for lm() objects
9:22 R's orthogonal contrast functions
10:18 How to manually set orthogonal contrasts for lm() objects
10:53 Two-way factorial ANOVA using the afex package
18:44 Interaction plot using ggplot
23:20 Extract the residuals/predicted values
24:04 Testing parametric test assumptions
26:42 Two-way factorial ANOVA using the lm() function
34:12 Extract the residuals/predicted values
35:02 Testing parametric test assumptions

I begin by describing important information about types of sums of squares (SS) and balanced vs. unbalanced data, and why you need to be mindful of them when running factorial ANOVAs in R. I then highlight another important issue you need to know about: by default, R doesn't set orthogonal contrasts, which will mess up your results for factorial ANOVAs using Type III SS with the lm() function. I then explain how you can manually set orthogonal contrasts for lm() objects to avoid this problem. The rest of the tutorial is divided into two parts. First, I explain how to run a two-way factorial ANOVA using the aov_4() function from the afex package. Second, I explain how to do the same analysis using the lm() function.

To freely download the R code and data sets used in this tutorial series, visit http://github.com/Statistics4Doctors. This tutorial analyzes a data set modified from data built by Dr. Andy Field (https://www.youtube.com/user/ProfAndyField) for his textbooks on statistics: Discovering Statistics using R (https://tinyurl.com/y6r6wsot) and SPSS (https://edge.sagepub.com/field5e).