Alex PikeJan 19, 2023
Frederik Aust

In response to helpful feedback from reviewers and readers, we have updated our preprint on #Bayesian repeated-measures ANOVA. Relevant if you have ever conducted one with {BayesFactor}, JASP, or Jamovi.

We hope it's now clear that the issue discussed is one of model specification and comparison—not a Bayes vs. Frequentism one.

We now also discuss a third model specification approach and discuss how to choose between them.

https://psyarxiv.com/fb8zn/

Jan 18, 2023 at 11:13amWeb
Show threadFrederik AustJan 19, 2023

To use our recommended model specification in {BayesFactor} replace

```
library("BayesFactor")
data(puzzles)

anovaBF(
RT ~ shape*color + ID
, data = puzzles
, whichRandom = "ID"
, whichModels = "top"
)
```

with

```
generalTestBF(
RT ~ shape*color*ID - shape:color:ID
, data = puzzles
, whichRandom = "ID"
, neverExclude = "ID"
, whichModels = "top"
)
```

for fully within-subject aggregated data. Expect longer computation times.

Show threadTom FaulkenberryJan 18, 2023
@FrederikAust Great work! The original paper was excellent, and this one is even better. This is such an important discussion.
Show threadFrederik AustJan 18, 2023

@tomfaulkenberry

Thanks, I‘m glad you like it! It was a fun project.