#### Howdy, Stranger!

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

Supported by

# BF package: null interval

Hi!

I have a directional hypothesis I want to test with ttestBF (R-BayesFactor-package) using a half-Cauchy prior. According to the documentation the null interval can be used to test a directional hypothesis and should contain the lower and upper bounds of the interval hypothesis to test. The null interval should be in standardized units but I'm not quite sure what is meant by standardization here.

For example, I have a dataset in which the values can theoretically be between 0 and 1. Most of the values are, however, around chance which is 0.25 or above. The hypothesis I want to test is which values are larger than chance (I don't want to test whether anything is smaller than chance). So the null value of the mean is 0.25. Should my nullInterval be [0.25, 1] or something else (because of the standardization)?

Any help very much appreciated.

Cheers,

Lina

• The interval is on the scale of Cohen's d - the standardized difference.

• Great, thanks, that was helpful! I think I just misunderstood what the term "null interval" means and how it relates to the units of the data. I now calculated the effect size from baseline data (mean/SD) and set this as the lower bound of the null interval. So now it is [0.4, Inf] and everything makes sense :)

Cheers,

Lina

• It sounds like you are post-hoc setting a prior (the null interval is a type of prior, on the null). This is ill-advised.

If you don't specify the null internal, a point null of 0 is used.