Premium
Digital ITEMS Module 03: Nonparametric Item Response Theory
Author(s) -
Wind Stefanie A.
Publication year - 2018
Publication title -
educational measurement: issues and practice
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.158
H-Index - 52
eISSN - 1745-3992
pISSN - 0731-1745
DOI - 10.1111/emip.12239
Subject(s) - nonparametric statistics , parametric statistics , item response theory , computer science , key (lock) , visualization , scaling , data mining , econometrics , theoretical computer science , mathematics , statistics , psychometrics , geometry , computer security
In this digital ITEMS module, we introduce the framework of nonparametric item response theory (IRT), in particular Mokken scaling, which can be used to evaluate fundamental measurement properties with less strict assumptions than parametric IRT models. We walk through the key distinction between parametric and nonparametric models, introduce the two key nonparametric models under Mokken scaling – the monotone homogeneity and double monotonicity model – and discuss modern extensions of the basic models. We also describe how researchers and practitioners can use key nonparametric statistics and graphical visualization tools to evaluate the fundamental measurement properties of an assessment from a nonparametric perspective. Finally, we illustrate the key reasoning steps and associated best practices using video‐based worked examples completed with the mokken package in R .