Modern Psychometric Theory and Methods

PSY 2070 Subject & Catalog Number

Description

By its classical definition, Psychometrics is concerned with the theory and techniques of psychological measurement. In this class we will cover a wide variety of modern psychometric methods; a big portion of them going beyond the classical psychometrics definition boundaries.

The first half of the class covers a variety of exploratory scaling (unsupervised learning) methods. The aim is to scale and visualize association patterns in complex, multivariate datasets. Such techniques include principal components analysis (PCA), correspondence analysis (CA), Gifi methods, multidimensional scaling (MDS), and (social) networks.

The second half of the class deals with parametric psychometric methods.

We start with basic elaborations on measurement and reliability, before moving on to latent variable models. Within this context we cover exploratory and confirmatory factor analysis, and structural equation models which allow us to model complex relationships among (latent) variables. Finally we introduce item response theory, a measurement framework for categorical data. One overarching goal of these latent variable units is to replace a naive sum score by something more sophisticated.

The last unit (psychometric theory) will be held by Richard McNally where he will talk about validity, and will cover theories of intelligence, personality, and behavior genetics fundamentals.

All topics covered will be supported by corresponding computations and illustrations in R, and supported by lab sections.