Education Commons^™

The Implications Of Halo Effects And Item Dependencies For Objective Measurement, T Mcnamara, Ray Adams

Prof Ray Adams

Analysis of data from performance-based language tests presents the problem of dealing appropriately with bundles of items that are linked to particular tasks. The authors use recent developments in Rasch modelling which allow analysis of bundles of items (or sets of ratings) to examine the extent of dependence between items and ratings, and they explore the implications of those dependences on the assessments. Data from two performance assessments are used in two studies of this problem.

Interpreting The Parameters Of A Multidimensional Rasch Model, W Wang, M Wilson, Ray Adams

Prof Ray Adams

This chapter introduces the multidimensional random coefficient multinomial logit model (MRCML) which can be applied to multidimensional polytomous test items. Two examples of real data analyses are followed to demonstrates the applications of the MRCML. A detailed discussion is made of interpretation of item parameters, in comparision to those derived from their corresponding unidimensional Rasch measurement models. Finally, it provides some guidelines for constructing scoring matrices and the corresponding models to aid parameter interpretation.

Measuring Student Knowledge And Skills : The Pisa 2000 Assessment Of Reading, Mathematical And Scientific Literacy, Andreas Schieicher, Claudia Tamassia, Geoff Masters, Ray Adams

Prof Ray Adams