Thank you, I have a question nobody around me can answer so I'm à bit stuck : I did a mancova in spss (with a 3-group indépendant variable, 6 dépendant variables and two covariates), asking for estimated marginal means as well as saving the predicted (unstandardized) scores for all my individuals in the data file. From what I understand, these scores are calculated based on the marginal means (taking into account the covariates). And indeed, they are quite different from the initial scores. But what I don't understand is that, when I then do a simple manova using these new scores (and no covariates, just a manova), not only the results (F) are not the same as the mancova's corrected model, but the marginal means are also different. Would you know where I am mistaken please? Sorry for the length of this question (and probably my English since I'm French !)
Levenes test indicates a problem with heteroscedasticity (or lack of constant residual variances) when significant. It is not used to test for heterogeneity of regression slopes. Heterogeneity of regression slopes is tested by including a group x covariate interaction term in your model. If that term is significant, it is an indication that the linear relationship between covariate and dv varies across groups.
Pretty good explanation ❤️ Thanks 🙏
Thank you, I have a question nobody around me can answer so I'm à bit stuck : I did a mancova in spss (with a 3-group indépendant variable, 6 dépendant variables and two covariates), asking for estimated marginal means as well as saving the predicted (unstandardized) scores for all my individuals in the data file. From what I understand, these scores are calculated based on the marginal means (taking into account the covariates). And indeed, they are quite different from the initial scores. But what I don't understand is that, when I then do a simple manova using these new scores (and no covariates, just a manova), not only the results (F) are not the same as the mancova's corrected model, but the marginal means are also different. Would you know where I am mistaken please? Sorry for the length of this question (and probably my English since I'm French !)
Could you make a video about two-way ANCOVA testing for assumptions?
I was under the impression Levine's test indicated caution when the result WAS significant. Wrong?
Levenes test indicates a problem with heteroscedasticity (or lack of constant residual variances) when significant. It is not used to test for heterogeneity of regression slopes. Heterogeneity of regression slopes is tested by including a group x covariate interaction term in your model. If that term is significant, it is an indication that the linear relationship between covariate and dv varies across groups.
Sir, where we find adjusted post test mean in this ... Kindly highlight that
Thanks