[Bug]: The p-value of the Shapiro–Wilk test is inconsistent.

[akwan168]

JASP Version

0.95.0

Commit ID

No response

JASP Module

Distributions

What analysis are you seeing the problem on?

Normal

What OS are you seeing the problem on?

Windows 10

Bug Description

In the distribution module, the p-value of the Shapiro–Wilk test is inconsistent with the data in the descriptive statistics table.

Expected Behaviour

In the distribution module, the p-value of the Shapiro–Wilk test is inconsistent with the data in the descriptive statistics table.

Steps to Reproduce

1. download the data file
   https://drive.google.com/file/d/10Si1MmXTnBXBSLLLKbZ_-gQ7-jdug1wA/view?usp=sharing

2. Please watch the video.
   https://youtu.be/OCmo23VwdDQ

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.

[github-actions]

@akwan168, thanks for taking the time to create this issue. If possible (and applicable), please upload to the issue website (#3590, attaching to an email does not work) a screenshot showcasing the problem, and/or a compressed (zipped) .jasp file or the data file that causes the issue. If you would prefer not to make your data publicly available, you can send your file(s) directly to us, issues@jasp-stats.org

[tomtomme]

@JohnnyDoorn
can you comment on this one? thx

[JohnnyDoorn]

Hi @akwan168,

It seems that when the sample size is below a certain threshold, a different algorithm is used to derive the p-value corresponding to the test statistic (which is the same across the descriptives and distributions modules). Apparently deriving this p-value is surprisingly complex, which is why different methods can be employed. For low sample sizes, nortest::sf.test is used, while for larger ones, stats::shapiro.test is used. I am not sure what this behavior is based on - perhaps @Kucharssim knows more? I do like the simplicity of always using the default R-function (which also seems consistent with other software).

Cheers
Johnny