Several factors with complex interactions influence the physical stability of solid dispersions, thus highlighting the need for efficient experimental design together with robust and simple multivariate model. Design of Experiments together with ANalysis Of VAriance (ANOVA) model is one of the central tools when establishing a design space according to the Quality by Design (QbD) approach. However, higher order interaction terms are often significant in these ANOVA models, making the final model difficult to interpret and understand. As this is ordinarily the purpose of applying ANOVA, it poses an obvious problem. In the current study, the GEneralized Multiplicative ANOVA (GEMANOVA) model is proposed as an alternative for the ANOVA model. Two complex multivariate data sets obtained by monitoring the physical stability of a solid dispersion with image analysis and X-ray powder diffraction (XRPD) as responses were subjected to GEMANOVA analysis. The results showed that the obtained GEMANOVA model was easier to interpret and understand than the additive ANOVA model. Furthermore, the GEMANOVA model has additional advantages such as the possibility of readily including multivariate responses (e.g., an entire spectral data set), model uniqueness, and curve resolution abilities. © 2012 Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci 102:904-914, 2013.