[Bug]: Bayesian equivalence test with Normal prior fails if interval starts with minus infinity

[henkkiers]

JASP Version

0.19.1

Commit ID

No response

JASP Module

Equivalence T-Tests

What analysis are you seeing the problem on?

Two sample t test

What OS are you seeing the problem on?

Windows 10

Bug Description

Using the equivalence t test module, I used the Normal (rather than default) prior, and encountered that the analysis led to infinity and 0 values when I entered as interval "minus infinity to -0.1", while "-0.1 to infinity" it gave decent values. I varied the interval, and it actually appears that this always happens when using the normal prior, and the second option for intervals, that is the one starting with -Inf. I show the output in a word document. I also uploaded a zipped version of my JASP file, and a screenshot of what happened.
Screenshot of bug jpg
Bug report on Equivalence t test.docx
st2shortBugReport.zip

It's a very annoying bug, because I use the equivalence t test to learn students how to interpret probabilities under posterior distributions, so I hope it can be solved soon.

Expected Behaviour

I actually explained what should happen, namely, I should get the complements to the results for the complement of the interval, which you get by choosing the interval from -0.1 to infinity.

Steps to Reproduce

I simply get it all the time if I choose the "Equivalence region" minus infinity to a particular value, when using the normal prior (not when using Cauchy prior or t prior)

Log (if any)

No response

More Debug Information

No response

Final Checklist

• I have included a screenshot showcasing the issue, if possible.
• I have included a JASP file (zipped) or data file that causes the crash/bug, if applicable.
• I have accurately described the bug, and steps to reproduce it.

[FBartos]

seems like to be an issue when copying results of the analysis when the "Informed-Normal" prior distribution is selected @boutinb

[henkkiers]

Hi Frantisek, I'm not sure whether this was a kind of question to me, but it surely not only happens when coping analyses It also happens with a brand new analysis. best regards henk

…

On Mon, Oct 28, 2024 at 2:30 PM František Bartoš ***@***.***> wrote: seems like to be an issue when copying results of the analysis when the "Informed-Normal" prior distribution is selected @boutinb <https://github.com/boutinb> — Reply to this email directly, view it on GitHub <#2984 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AXIY4XO3H6BAIQOEBMTZ3ODZ5YU4ZAVCNFSM6AAAAABQQQX6POVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINBRGU4TMMRTGI> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- Henk A.L. Kiers Department of Psychology University of Groningen Grote Kruisstraat 2/1 9712 TS Groningen The Netherlands tel. +31631984821 email: ***@***.*** http://kiers.webhosting.rug.nl

[FBartos]

Hi Henk, the message was for other people working on the issue. I see now that I'm also reproduce the issue even without copying the analysis.

[henkkiers]

Hi Frantisek, Interestingly, just this morning I did encounter a problem in the same module, which did seem to be caused by copying parts of results where it shouldn't. You easily can see what went wrong in the attached.pdf with four screenshots. I think it is important that you know about these errors, and probably best while you're working at this module anyway.. best regards Henk

…

On Mon, Oct 28, 2024 at 2:55 PM František Bartoš ***@***.***> wrote: Hi Henk, the message was for other people working on the issue. I see now that I'm also reproduce the issue even without copying the analysis. — Reply to this email directly, view it on GitHub <#2984 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AXIY4XJATSNEYBHIYIO2X7DZ5YX3LAVCNFSM6AAAAABQQQX6POVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINBRGY2TSNRZGY> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- Henk A.L. Kiers Department of Psychology University of Groningen Grote Kruisstraat 2/1 9712 TS Groningen The Netherlands tel. +31631984821 email: ***@***.*** http://kiers.webhosting.rug.nl

[EJWagenmakers]

Thanks Frantisek!!