We understand by data analysis the full set of methods which are used to point out the structural relations between certain objects and the characteristics observed in them. In the case where these characteristics admit a process of dichotomy, the relations can be expressed in terms of propositional logic.In the present paper, we endeavour to make clear the opposition between the problems of propositional logic and those of certain model utilised in data analysis: in propositional logic, we look for a (...) solution verified by a given set of propositions; as long as we are concerned with data analysis, we look for a solution in terms of propositions which include as proper parts the observed objects.Finally, we draw the main lines of two very used methods which can be relevant with the aid of computer: the factor analysis and the automatic classification (cluster analysis). (shrink)