Sufficient statistics and latent trait models

Abstract

logistic latent trait, minimal sufficient statistic, Rasch model, equidistant scoring, sufficiency,

Development and validity of a work functioning impairment scale based on the Rasch model among Japanese workers

“…The model also assumes that the probability of a particular person interaction (with regard to rating high or low) is determined only by how difficult the item is, and the subject's ability. Second, the degree to which the trait has been mastered is represented by the summed rating of the attributes, since for Rasch measurement, the raw score is considered a "sufficient statistic" 40) . The statistical sufficiency of raw scores means that the person total score includes the complete information available about the individual in the specified context with regard to the relevant latent trait, regardless of response pattern.…”

Towards an alignment of engineering and psychometric approaches to uncertainty in measurement: Consequences for the future

“…Statistical sufficiency formulated at the individual level, per Rasch [20,21], is conceptually identical with invariance in measurement [22,23].…”

“…It seems to me that Andrich's account is likely to turn out to be an image of the deep structure of how certain kinds of noise-induced order [33,34] come about. To what extent are the "sufficient conditions for a system to exhibit stochastic resonance" [35] the same as or different from the conditions sufficient for fit to a stochastic Rasch measurement model [15,16,20,21,23]? How similar or different from the True Score Theory-Rasch or Guttman-Rasch contrasts is "the transition from chaotic behavior to ordered behavior induced by external noise…observed in a certain class of one-dimensional mappings" [33]?…”

Élaboration de l’Échelle du sentiment d’efficacité personnelle des enseignants en gestion de classe (ÉSEPGC)

“…les données au prétest ont été analysées par dimension à l'aide du rating scale model (andersen, 1977 ;andrich, 1978 ;Wright & Masters, 1982) à partir de la version 3.81 du logiciel WinstePs (linacre & Wright, 1999). ce logiciel est utilisé pour estimer les statistiques d'ajustement et le niveau d'endossement des énoncés (logit) aux prescriptions du modèle.…”