Welcome!

Sign in with your CogSci, Facebook, Google, or Twitter account.

Or register to create a new account.

We'll use your information only for signing in to this forum.

Supported by

Bayesian paired t-test gone weird?

Hello everyone!

I am currently trying to run Bayesian paired t-tests in JASP. I am comparing the average hours of daylight (hours between sunrise and sunset) across the days of July in two different places, Hornsund and Edinburgh. In Hornsund, the average hours of daylight is always 24 because the sun doesn't set. In Edinburgh, the average is roughly 15hrs.

For some reason, when I entered the data with Hornsund first and then Edinburgh into the paired t-tests, it told me that it couldn't calculate this because there was a "TRUE/FALSE value missing". I went into my data set and discovered that indeed, I had failed to enter the 24 for July 31. I fixed this in Excel and went back to JASP. I re-ran the test but by mistake entered Edinburgh as the first variable, then Hornsund. Now it computed the test results. So I was happy. However, because for all the other t-tests (sunlight in Hornsund/Edinburgh in different months) I had entered Hornsund first I wanted to be a bit OCD and swapped the variables around in the July analysis. This got me the same error message as before. Changing them to Edinburgh-Hornsund again gave me results...

What am I doing wrong here...?

Comments

  • EJEJ Posts: 247

    This is difficult to tell without the data. Maybe a screenshot helps.
    E.J.

  • eniseg2eniseg2 Posts: 6

    Hi E.J.,
    I've just uploaded the file into the OSF: https://osf.io/zxukj/

    If you scroll all the way down in the file, you can see two t-tests with the identical variables but in different order. I've annotated them, too.

    Many thanks!
    Anna

  • EJEJ Posts: 247

    Hi Anna,
    I think you have to make this public, or else invite me (EJ.Wagenmakers@gmail.com)
    Cheers,
    E.J.

  • eniseg2eniseg2 Posts: 6

    Oh dear! I'm so new to the forum, haha. I've added you as a reading contributor, I hope that worked. :-)

    Meanwhile, I have a slightly different question. I remember you saying that there "is such a thing as too many pizza plots", hence JASP doesn't provide them for Bayesian ANOVA. In the example calculations paper for JASP there's a formula of how to calculate the pizza plots' odd ratios. BF/BF+1=odds ratio=area in the pizza plot that is for H1.
    Is it okay to use these calculations to create pizza plots by myself for my ANOVA Bayes Factors?

    Thanks a lot!
    Anna

  • EJEJ Posts: 247

    Hi Anna,

    Thanks for sending the file. What has happened is that you conducted a one-sided analysis for the July test, probably by mistake. The first one made sense (Hornsund > Edinburgh), but the second one (Edinburgh > Hornsund) is so strongly contraindicated by the data that the result cannot be computed.

    And of course, you are free to compute as many pizza plots as you like!
    Cheers,
    E.J.

Sign In or Register to comment.