Möbus, Claus (1978) Zur Fairness psychologischer Intelligenztests: Ein unlösbares Trilemma zwischen den Zielen von Gruppen, Individuen und Institutionen. Diagnostica, 24 (3). pp. 191-234. ISSN 0012-1924

Preview	- Published Version Volltext (9Mb)
	- Published Version Volltext (8Mb)

Abstract

The purpose of this article is to discuss the non-bayesian definitions of testbias. It can be shown that most contradictory issues are due to unrealistic error assumptions in the underlying regression models. It is necessary to introduce equation ard measurement errors. By leaving classical test theory and using instead latent trait theory, it is further shown that (with the help of the Rasch.model) it is possible to avoid item- and testbias. In the last part of the paper we discuss random and nonrandom admission strategies in the context of testbias. It is argued that nonrandom strategies which are based on cutoff-orientated decisions are antagonisiic to the idea of man as a learning system.

Item Type:	Article
Uncontrolled Keywords:	testbias, itembias, Raschmodel, regression model, equation error, measurement error, error in variables
Subjects:	Philosophy and psychology > Psychology
Divisions:	School of Computing Science, Business Administration, Economics and Law > Department of Computing Science
Date Deposited:	09 Oct 2025 09:26
Last Modified:	09 Oct 2025 09:26
URI:	https://oops.uni-oldenburg.de/id/eprint/7246
URN:	urn:nbn:de:gbv:715-oops-73278
DOI:
Nutzungslizenz:

Actions (login required)

View Item