Johnny

[Johnny @JohnnyBvanDoorn]

@DrShariEllen @JASPStats This is normal behavior, as the test-value of the signed rank statistic is equal to n*(n+1)*0.5. So any deviation from this value (lower or higher) will go towards evidence against H0, so the p-values you report seem to make sense =)

[Johnny @JohnnyBvanDoorn]

@ZacHyams @JASPStats @t_palma @zacHyams I am not sure if this feature was already there in 10.12, so just to be sure, I would recommend an update. Also, this option is only available under RM ANOVA, not AN(C)OVA.

[Johnny @JohnnyBvanDoorn]

@AlxEtz @pedromsmoreira Hi Pedro, we use the following formula:
W / ((n*(n+1))/2) * 2 - 1
where W is the Wilcoxon statistic and n the sample size. You can read more about this approach here: https://t.co/rHaFKH9l7m

[Johnny @JohnnyBvanDoorn]

@MMarsiske @JASPStats Hi Michael, that's a great suggestion! We will have a marginal means table for the Bayesian ANOVA (including these credible intervals) in the next release or the one after that!