This paper regards considerations on methodological consequences of using a realistic approach to social research. There are three main argumentations. (1) There is no realistic method that could establish itself as a possible alternative to scientific method. (2) The realistic concept of social research refers to a perspective of unmetaphysical realism. (3) The operationalization of sociological concepts related to latent dimensions could benefit by a reduction procedure less oriented towards a lazarsfeldian latent structure analysis and more attentive to the suggestions of the «circumstantial paradigm», as originally formulated by Carlo Ginzburg.