A New Derivation of the Rasch Model

摘要: Abstract The Rasch model is usually derived from statistical requirements of estimation its parameters. principle 'specific objectivity' or, equivalently, 'sample independence' requires that subject parameters can be estimated independent the item and vice versa. This in turn sufficient statistics exist for parameters, which follows. Rather than using a argument, present paper presents derivation based on requirement probability inferential ordering subjects (or items, resp.) sample independent. parallels Ducamp & Falmagne's axiomatization Guttman scale (composite measurement); it first given two items or subjects, subsequently generalized to multiple ordering. core equation is, course, not different derivation.