0
$\begingroup$

I am examining the difference in illness models (by asking several questions: DV) for refugees and non-refugees (IV). In addition, we asked about several confounding variables / covariates (demographic data).

Some items of my dataset do not fulfil the requirement of homogeneity of regression slops. I was wandering how I can best describe the relationship of the covariate and DV for those items. I am aware that if I use a MANCOVA/ANCOVA regardless of this assumption violation, this will increase the TYPE II Error.

I have read about Johnson Neyman procure. It seems complicated to me.

Can I calculate a regulare MANOVA/ANOVA for thes items, that do not fulfil the assumption of homogeneity of regression slopes (not enter the covariate)? And then check the way the covariate influences the DV by doing a regression?

I am not aware of a valid method. Does anyone have experience with heterogeneity of regression slopes?

Thank you for your help

$\endgroup$

1 Answer 1

0
$\begingroup$

If you center your variables before creating cross products you can interpret the model with the cross product in it. The main effects will be tested at the mean values of the other variables. Keep in mind that different slopes can sometimes be interesting theoretically.

$\endgroup$
1
  • 1
    $\begingroup$ Hallo David, thank you for your fast as lightneing ;) help. I will have to read up on your suggestion a bit as I am more or less a novice/ insecure around statistics and will get back (to you) as soon as possible. $\endgroup$
    – T.H.
    Mar 17, 2017 at 17:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.