I propose a framework in which psychometrics items such as response to questionnaries or clinical symptoms are grouped together in terms of their informational content, which can be classified as synergistic or redundant.

This approach will extend and complement the emerging field of network psychometrics, providing advances and solutions that are both statistical and conceptual. The implemented tools will allow researchers in psychometrics to focus on the higher order behavior of the data, moving beyond a pairwise network representation.

I will reanalyze datasets which have been represented as pairwise networks, and provide a reinterpretation of the results, in a complementary way with respect to the usual interpretation in terms of latent variables and factors. The concept of synergy will furthermore be used as a warning sign of impending transitions on longitudinal data with the focus on prevention of mental health issues.