ABSTRACT
In general, Boolean analysis is presented as a method for 0/1-data analysis that has a greater scope than Guttman analysis (Flament, 1976; Theuns, 1989, 1992, 1994; Van Buggenhaut & Degreef, 1987). The output of a Boolean analysis is an implication scheme; this is a graph consisting of responses to binary items (vertices) and implication relations between these responses (edges). Until now approaches inspired by Flament (1976) resulted in one or more implication schemes that comply with some optimality criterion. Restrictions to implication schemes with regard to theoretical constraints on implications have not been dealt with systematically. Several approaches to approximation of implication schemes concentrated on the global fit of implication schemes to the data (Flament, 1976; Theuns, 1992). Methods have been introduced that optimize the trade-off between the simplicity of implication schemes and their quality of fit (Theuns, 1989, 1992, 1994; Van Buggenhaut, 1987; Van Buggenhaut & Degreef, 1987).
