[Feature Request]: Option to model hypothesis with uniform distributions

[fschuman]

Description

Uniform priors / models

Purpose

Test hypothesis that are best modelled by uniform distributions

Use-case

Any situation where a uniform distribution is more appropriate than a normal or Cauchy distribution

Is your feature request related to a problem?

The best model for H1 in my current data is a uniform model, not a Cauchy or normal distribution

Is your feature request related to a JASP module?

Unrelated, ANOVA, T-Tests

Describe the solution you would like

Option to specify a hypothesis with a uniform prior in t-test, ANOVA etc.

Describe alternatives that you have considered

Uniform distributions can used here to to model the hypothesis under test: https://harry-tattan-birch.shinyapps.io/bayes-factor-calculator/

Additional context

No response

[tomtomme]

@fschuman This is done for ANOVA. Maybe for some time now, I don't know exactly. You can find it under "additional options". Since you can to a t-Test within the ANOVA module, is this then sufficient to close this request?

[fschuman]

Thanks for letting me know! This helps !

Since you can to a t-Test within the ANOVA module, is this then sufficient to close this request?

Here I'd say it depends - from a usability perspective, making uniform priors a regular option for t tests as well seems still very useful. Otherwise the software nudges users into using the easily available (non-uniform) priors without reflecting on whether they fit to the question to be tested at hand. But non-uniform priors are not always the best representation for a model. If uniform priors were as easily available among the choices, it would more likely prompt a reflection as to which model is most appropriate in the given case. That would be my thoughts about it.

[tomtomme]

closing for now as duplicate of https://github.com/jasp-stats/jasp-issues/issues/785