Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Inconsistent results in bayesian t-tests using residuals of co-variates

Dear JASP community,

I would like to ask your help with the following problem for my analysis in JASP.

I performed Informed Independent Bayesian Student t-tests using residuals of a linear regression to control for co-variates (age, sex, education) and set the location of the prior at -0.7 and scale at 0.707 to test my Hp that group 1< group 2.

I obtained the same Bayes factor number for all the variables (Screenshot 1), despite these variables had different distributions (Screenshot 2).

When I performed a Mann-Whitney t-test for not normally distributed data, I got then different Bayes factor numbers for each variable of interest (Screenshot 3).

The fact that my data are not normally distributed may be the reason why I get these strange results with Student t-test? Should I then take for reliable the results obtained from the Mann-Whitney t-test?

I would be super grateful if someone could help me understand the issue.

Thank you in advance.

Best regards,

Alice

Screenshot 1

Screenshot 2

Screenshot 3


Comments

  • This just looks like a bug. If you enter the variables individually you should see different results. I will try to see if I can reproduce this...no, it works fine for me, I get different results. Please post the issue on our GitHub page, and outline a clear reproducible path (for details see https://jasp-stats.org/2018/03/29/request-feature-report-bug-jasp/). This would help us a lot.

Sign In or Register to comment.