Surmise Relations between Tests


Non-numerical test theory denotes a research program of current interest in behavioral and social sciences for which some components already exist. One example is the Knowledge Space Theory originally developed by Doignon and Falmagne. One advantage of the non-numerical test theory approach is the low requirement of scale level combined with the possibility of empirical validation. This corresponds closely to the status of theories and research methods in many fields of psychology. A second advantage is the interpretability of the relationships between test items by surmise and prerequisite relations which are non-bijective. For application’s in different fields like cognitive, developmental, educational, and personality psychology, a generalization of this approach is needed where the tests and not only the items are the basic elements. Then the prerequisite relationships between different cognitive functions, developmental stages, educational stages, and personality traits, respectively, can be investigated on the basis of non-numerical tests, Thus, we describe a project during which a theoretical model for relations between non-numerical tests will be developed, Based on this modeling and theoretical analysis, we will implement a software system for generating and evaluating hypotheses on non-bijective relations between non-numerical tests, for partitioning and combining tests, and analyzing data. Besides simulation studies a first application of model and software to empirical data will be performed.


  • Dietrich Albert
  • Silke Brandt
  • Cord Hockemeyer
  • Susane Poetzi
  • Ali Ünlü
  • Gudrun Wesiak

University: Department of Psychology, University of Graz, Austria/Europe

Duration: 01.03.1998 – 30.09.2000 (31 months)

Funding: A project supported by the Austrian Science Fund (FWF). Grant No.: P12726