Howdy, Stranger!

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

Supported by

Trend analysis (NHST) in JASP

Hi,

I was wondering about how JASP calculates the trends (linear, quadratic, etc.) when asking for polynomial contrasts in an ANOVA (NHST). Is there any reason why it reports the results in a t test rather than in an F test (like SPSS). I guess ultimately the result should be identical when taking the square root of values from the F test, but I do wonder whether there are some crucial differences. I assume that they still are basically simple regression analyses? In this regard, it would also be handy to not only see t and p values but also the dfs for that analysis.
More generally, is there somewhere a documentation on the tests that JASP uses and whether certain corrections have been applied (e.g. I am thinking about the multiple corrections of Cousineau and Morey when it comes to the calculation of within-subject confidence intervals)?

Thanks,
Michif

Comments

  • Dear Michif,

    I'll ask Johnny to clarify this. We probably call an existing R package.

    Cheers,
    E.J.

  • Hi Michif,
    The polynomial contrasts in JASP are calculated with the "contr.poly()" function in R, the result of which is then supplied to the "contrast()" function of the emmeans package. This is also the function that gives the t value for the model comparison. For the non-RM ANOVA, we do provide the df's - I will look into supplying these for the RM ANOVA contrasts as well!
    As for your documentation suggestion - that is an excellent idea! Especially for the confidence interval scenario you describe - which seems to be quite a subjective field (how to compute the SE's and whether to apply additional corrections etc). Currently, we apply the Morey and Cousineau correction to the confidence intervals for within-subject factors and we let the user decide whether to used the pooled SE for those factors or not.
    We are currently in the process of rewriting a lot of our R-code, which will greatly enhance the readability of the code - hopefully this will also make it more inviting for people to go and look at what's going on under the hood. It would be cool to have an additional table with those details though.
    Cheers,
    Johnny

Sign In or Register to comment.