However, we encountered several problems during modelling: variables that based on the chi-square test clearly significantly influence preferences for personalized nutrition fell from the model (this was presumably due to multicollinearity on the one hand, which is also evidenced by the high VIF index, and to the effort to maximize the explanatory power of the model in multivariate modelling on the other); the combined explanatory power of the model can be regarded as so low (Naglerke R 2 = 0.30) that the interpretation of the results and the determination of the actual correlations became questionable.
1
—
—