This series attempts to explain in simple terms what Bayes factors do, assume, mean and require people to be OK with if they want to use them. I (Uri) do not believe that many social scientists would embrace Bayes factors, if they understood them, and this is my attempt to convey that message.
Post 1. DataColada[78a] – Milton and Minimum Wage
The first post uses an example, where Milton predicts an effect between 1% and 10%, and upon seeing 1%, the Bayes factor deems this effect, which was predicted by Milton, as contradicting Milton's prediction. The example is used to convey the intuition of how Bayes factors assess if data are consistent with a theory, and contrasts it to how researchers do.
Post 2. DataColada[78b] – Hyp-Chart: the missing link between p-values and Bayes factors
This posts introduced Hyp-Chart, a plot that shows how consistent the data are with every possible hypothesis, compared to the null hypothesis. The Bayes factor is but a (bad) summary of Hyp-Chart.
Post 3. DataColada[78c] – Looking at 10 papers in Psych Science that report Bayes factors
In three Psych Science papers obtaining a non-significant effect (p>.05), the Bayes factor is shown to be non-diagnostic of whether the data do or do not support the null.
