In the monograph is represented the method of multiple comparison for a data set, divided into groups on external criteria (one or several). On the basis of an author's method of multiple comparison is designed the new approach of analysis of different form of statistical dependence in psychological research, thus the strength of connection and form of dependence are determined (the linear dependences are determined as a particular case). The examples of non-linear psychological systems are considered: Five-factor personality model (Big Five); Research of stress; Cross-cultural research of relations to vital difficulties and problems in various ethnic groups. Methodology of correlative relativism in psychology is shown on the example of investigation of dependences between psychological parameters in five systems of the report (data sets of five high schools) in one task.