Major inconsistency between frequentist and bayesian non parametric paired t-test

[ArCourtin]

• JASP version:0.13
• OS name and version: Windows 10
• Analysis: (bayesian) paired t-test
• Bug description: The W value returned by the Bayesian and parametric paired t-test are inconsistent. BF and p-values interpretation lead to drastic differences of decision on H0 rejection.
• Expected behaviour: I'm unusure wether the W statistic are supposed to be exactly equal between the 2 tests but I assume they should not differ to such extent

Steps to reproduce:

1. Go to "T-tests>Paired samples t-test" and "T-tests>Paired samples t-test"
2. Tick the "Wilcoxon signed rank" box in the "Test" menus
3. Compare W statistics obtained for the same dataset.

[EJWagenmakers]

@JohnnyDoorn

[ArCourtin]

The inconsistency in W is also present in JASP v0.12.2 but the corresponding BF and p-value seemed to be in agreement. Same dataset in JASP V0.12.2: