I have dataset of household dietary consumption (based on food weighing) and each food items were classified into 12 food groups from 9,668 household. I've conducted factor analysis and generated 5 dietary patterns (which was the factors/components).Now, I conducted a MANOVA with 5 dependent variables (Dietary Pattern I, Dietary Pattern II, Dietary Pattern III, Dietary Pattern IV, and Dietary Pattern V) and my factor variable was Food Insecurity Level (i.e. Secure, Mild Insecure, Moderate Insecure, and Severe Insecure) adjusted for some household characteristics and my result was significant.

However, multivariate normality and homogeneity of variance-covariance matrices was violated. Also, dietary patterns were not correlated to each other since values were generated factor loading.

I wish to look at how the independent variable predicts or influences the multiple dependent variables. What statistical method should you recommend?

Thanks in advance!

Marvin