search

jasp-stats / jasp-issues

This repository is solely meant for reporting of bugs, feature requests and other issues in JASP.
58 stars 29 forks source link

[Feature]: 95% HDI's from ANOVA regression coefficients #2119

Open Wouterwolf opened 1 year ago

Wouterwolf commented 1 year ago

JASP Version

0.17.1

Commit ID

No response

JASP Module

ANOVA

What analysis are you seeing the problem on?

AN(C)OVA

What OS are you seeing the problem on?

Windows 11

Bug Description

In the newer versions of JASP, when I conduct a Bayesian ANOVA and check the descriptives, the 95% credible intervals of the means for the groups no longer show up (the table stops after the coefficient of variation).

Screenshot 2023-05-03 183003 Similarly (although I am not sure this next one used to be the case), when I do a Bayesian ANCOVA with 1 fixed factor and 1 covariate, I can make a descriptive plot of the regression lines per factor, but I cannot actually find a way for JASP to give me the regression coefficient (or their HDI's) of these regression lines, even though they are apparently being computed for the creation of the graph.

Expected Behaviour

At least for the ANOVA descriptives, I expected two more columns with an 95% credible interval min and max for the means for each group.

Steps to Reproduce

  1. Open JASP with a dataset
  2. Start an ANOVA (or ANCOVA)
  3. Insert a factor (and a coveriate for the ANCOVA)
  4. Check the descriptives box (and make sure the credible interval box is set to 95%)
  5. run the analysis ...

Log (if any)

No response

Final Checklist

tomtomme commented 8 months ago

@Wouterwolf At least the CIs for the BANOVAs descriptives table are restored with jasp 0.18.3 But the missing regression coefficient (or their HDI's) in BANCOVA are still a thing, it seems.

Can you confirm? I would however consider this as a feature request, not a bug. Would be cool if you could change the title accordingly. Thx

Wouterwolf commented 8 months ago

Yes, correct. The group means were a bug because they were there before, but the regression coefficients have never been, but would be a really nice feature, especially given that the lines themselves (+ 95% HDI's) are already visually obtainable in a graph)

tomtomme commented 8 months ago

image The graph that shows the (B)ANCOVAs slopes and HDIs/CIs Wouterwolf would like to have their values in table form