Post hoc tests for robust mixed design ANOVA using R

Question

Is it possible to compute the function mcp2atm with unequal sample size? I ran the robust Mixed Design ANOVA by using tsplit and want to do post hoc tests now.

Answer 1

The robust mixed-design ANOVA function is now called rmanova instead tsplit and unfortunately the WRS2 package does not yet contain a corresponding post-hoc test for it. (mcp2atm is solely for t2way). In case you need to conduct a post-hoc test, you must use the original WRS package for now.

This package contains the functions bwmcp for multiple comparisons of variable A (between), B (within) and the interaction between these two. If you are interested just in one of these you can use bwamcp for A, bwbmcp for B respectively bwimcp for the interaction. The usage of these function is not as straightforward:

# load the latest WRS package
> source("https://dornsife.usc.edu/assets/sites/239/docs/Rallfun-v38.txt")

# convert your data for the function from long format into list format. 

> data <- bw2list(yourdf, 4, c(7,8))

# In this example "4" stands for the column where your independent groups are located.
# c(7,8) indicates the position of the columns with the within groups data

Now that we have the data prepared suppose that our between subject variable has 3 levels (groups) and our within-subject variable (as seen above) has 2 levels. The function would then look like this:

> bwmcp(3,2, data, tr = 0.2) # uses by default 20% trimmed means

Answer 2

You can use the WRS2 package and run the following posthoc tests: sppbb (within effects p), sppba (between effects), and sppbi (interaction effects).

You can read more about it here: https://link.springer.com/article/10.3758/s13428-019-01246-w