Date & Time: Wednesday, July 21 at 9:00AM EST
I will review the way psychometrics has addressed reliability of measurement through the years, and notice that with respect to reliability psychometrics has great difficulty to sail a straight course. The topics I briefly discuss are:
- Definition of reliability through the years
- Methods of estimating reliability
- Different kinds of reliability
- Reliability in different statistical models
- Does knowledge development sail a straight course?
- What do consumers of reliability need?
I will draw a conclusion of where we are with respect to reliability and provide recommendations for future research.
About the Speaker
Klaas Sijtsma received his PhD in psychology at the University of Groningen in 1988. His topic was Mokken scale analysis. He has taught psychometrics and applied statistics at several Dutch universities and has been affiliated with Tilburg University since 1997. His research focusses on measurement topics with respect to psychological attributes, such as item response models, in particular nonparametric models, person-fit analysis, missing data and outliers, cognitive diagnostic models, and many other topics, including classical test theory and an occasional digression to research integrity issues. He published many papers on these topics, and three books on test theory and measurement. The most recent book is Measurement Models for Psychological Attributes, co-authored with Andries van der Ark and published by Chapman & Hall / CRC in 2021. This book provides an anthology of the prevailing measurement models in psychology, education, health, and other human sciences.
Klaas’ invited paper is a sequel to his 2009 Psychometrika paper on coefficient alpha. It will appear later this year in Psychometrika together with three discussion papers and a rejoinder. He co-authored the paper with Julius Pfadt from Ulm University in Germany.