Higher measurement quality of tests and questionnaires by means of more powerful statistics

Hendrik Straat (PhD student)
MTO, Tilburg School of Social and Behavioral Sciences, Tilburg University

Project: Project financed by Tilburg University

Project running from:1 September 2008 – 1 September 2012

Promotores: Prof. dr K. Sijtsma (Tilburg University), prof. dr B.W. Junker (Carnegie Mellon University)

Summary:
Tests or questionnaires are often used to measure personality traits, attitudes, opinions, skills, and abilities. A measurement model transforms the respondents’ item scores into a meaningful measurement value. Using a measurement model that does not fit the data may lead to incorrect conclusions with possibly severe consequences: e.g., a wrong diagnosis of a mental patient or an incorrect educational placement. For nonparametric item response theory models—a very general class of measurement models—the available methods to assess fit are insufficient to allow good test construction. In this project better methods are developed that have more power.

Date of defence:   23 November 2012

Title of thesis:   Using scalability coefficients and conditional association to assess monotone homogeneity. Ridderkerk: Ridderprint BV. ISBN / EAN: 978-90-5335-598-5