This paper argues that qualitative theories may have interesting statistical applications. More precisely, it is shown that Q-theories, i.e., qualitative theories stated in monadic languages with two or more families of predicates, can be used in describing the statistical structure of cross classified populations and that their adequacy in this task can be evaluated by measuring their ‘statistical verisimilitude’. Firstly, a short survey of some applications of the post-Popperian notions of verisimilitude to the analysis of statistical inferences is provided. Secondly, a new class of measures for the verisimilitude of the universal generalizations stated in monadic languages is introduced. Finally, such measures are applied to Q-theories and used as the starting point for defining appropriate measures of the statistical verisimilitude of Q-theories.
