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

I need to check a cubic trend (polynomial contrasts) using Bayesian analysis. can i use JASP for that?

We don't have Bayesian polynomial contrasts in there just yet, but it is on our list. You can push us to do this by posting the issue on our GitHub page (for details see https://jasp-stats.org/2018/03/29/request-feature-report-bug-jasp/).

E.J.

Thank you for your quick response.

Is it still possible to do Bayesian polynomial contrasts in your program using the following transformation?

The goal is to check the cubic contrast for 7 levels.

In this case, the cubic contrast is -1 1 1 0 -1 -1 1.

Can I

*multiply Level1 by -1 Level2 by 1, Level 3 by 1 and so on…

*Then add level 1,2 &3 to a new variable, for example, var 1

*Then add levels 5,6 &7 to a new variable, for example, var 2 (absolute values)

*In the next step, can I use a T test to check the differences between the two means

and then look at the Bayesian value to give me an indication if there is or isn’t a cubic trend?

This seems a little fishy. I am always wary of these kinds of shortcuts, or at the very least I'd like to see a comparison to a more principled way of doing things. We're going to implement contrasts at some point in the future though.

## Comments

We don't have Bayesian polynomial contrasts in there just yet, but it is on our list. You can push us to do this by posting the issue on our GitHub page (for details see https://jasp-stats.org/2018/03/29/request-feature-report-bug-jasp/).

E.J.

Thank you for your quick response.

Is it still possible to do Bayesian polynomial contrasts in your program using the following transformation?

The goal is to check the cubic contrast for 7 levels.

In this case, the cubic contrast is -1 1 1 0 -1 -1 1.

Can I

*multiply Level1 by -1 Level2 by 1, Level 3 by 1 and so on…

*Then add level 1,2 &3 to a new variable, for example, var 1

*Then add levels 5,6 &7 to a new variable, for example, var 2 (absolute values)

*In the next step, can I use a T test to check the differences between the two means

and then look at the Bayesian value to give me an indication if there is or isn’t a cubic trend?

This seems a little fishy. I am always wary of these kinds of shortcuts, or at the very least I'd like to see a comparison to a more principled way of doing things. We're going to implement contrasts at some point in the future though.

E.J.