Howdy, Stranger!

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

Sign In Register

Quick Links

Categories

Supported by
Bestel Een wereld vol denkers door Sebastiaan Mathôt via bol.com

Help for covariate analysis with Mixed ANOVA

kev_baguekev_bague
in JASP & BayesFactor

Hi,


I need help to explain my data.


I'm conducting a mixed ANOVA with one within subject factor (A) and one between subject factor (B). Results show a main effect of A, and an interaction AxB.

However I would like to test anxiety as a covariate assessed by STAI.


First, what is the best way to compute this ? Should I use ANCOVA or just add STAI in covariate box in my ANOVA ?


Secondly, there are two component in STAI, should I compute my analysis for each component individually, or add both in covariate box ?


Finally, I tried to compute this analysis in my ANOVA directly, I added STAI-Y1 in covariate box. So I don't know how explain results. They showed that with covariate main effect of A disappeared, but there was not a main effect of my covariate. What that is mean ?


I did not find articles, tutorials for this design so if you have these tell me.


Have a nice Sunday.


Kevin

Comments

  • PM_StatPM_Stat

    Hi,


    Could you add a screen of your JASP output?


    PM

  • EJEJ

    I've asked our ANOVA expert to take a look at this as well

  • JohnnyBJohnnyB

    Hi Kev_bague,


    1. There is no covariate box in the ANOVA, only in the ANCOVA - which is the difference between the two analyses: in the underlying code, the ANOVA just calls the code of ANCOVA but supplies a model without covariates (this is not JASP specific).
    2. I would advise to add the two covariates simultaneously, as they probably have a lot of overlap in how well they predict your dependent variable (i.e., I would expect a high correlation between the two components) and adding them both would take that into account.
    3. This is probably because the effect of your independent variable (A) was confounded by the effect of the covariate. Without including the covariate, it seems that A is a good predictor of your dependent variable. However, it could be the case that Anxiety is the best predictor of your dependent variable and that Anxiety differs for each group of A. Adding Anxiety to your model is a way to check whether A still has any explanatory value (i.e., does A explain patterns in your data that are not already explained by Anxiety?). In your case it seems that Anxiety does have some explanatory power: enough that it spuriously amplifies the main effect of A (when not including Anxiety in the model), but not enough to warrant concluding that there is a main effect.

    Pease let me know if things are still unclear =)

    Cheers,

    Johnny

  • kev_baguekev_bague

    Hello @JohnnyB


    Thank you very much for your answer very helpful !

    We're very lucky to have this software as well as your tips !


    Thank you !


    Cheers,


    KB

  • JohnnyBJohnnyB

    Thanks KB, that's great to hear!

Sign In or Register to comment.
Sebastiaan Mathôt » v2.0.18 » Copyright 2011 — 2019 » Vanilla flavored » License