Skip to main content

Latent Variable Analysis (LAVA) Lab

Virtually all measurements in psychology and the social sciences are affected by effects of persons, situations, methods or observers, and random measurement error. Statistical models with latent variables allow separating various systematic and unsystematic sources of individual differences in psychological measurements and thus help investigators learn more about these effects and obtain less biased results in empirical studies. In my lab, we develop, evaluate, and apply latent variable models for a variety of purposes, including models for measuring variability and change over time and models for analyzing multimethod/multirater data.

A particularly important question in research on latent variable methodology concerns the actual meaning of latent variables. What are these variables and how do we interpret them correctly? In my lab, we follow Rolf Steyer's approach of defining latent variables "constructively", based on the well-defined concepts of classical psychometric measurement theory and latent state-trait theory. This ensures that all latent variables have an unambiguous meaning and interpretation.

Substantive research interests pursued in my lab concern individual differences in spatial abilities and how they can be explained.



Christian GeiserChristian Geiser

Biographical Sketch

I am currently a professor and the director of the Quantitative Psychology Specialization at USU. I studied psychology in Magdeburg (Germany), Geneva (Switzerland), and Berlin (Germany).  I obtained my PhD in psychology from Free University Berlin in 2008.  Before coming to Utah State University, I was an Assistant Professor in Arizona State University's Quantitative Psychology PhD program.

Research Interests

My research interests are in quantitative psychology and particularly in latent variable models and psychological measurement.  My main areas of expertise are in multitrait-multimethod modeling, longitudinal data analysis and latent state-trait analysis.

I also do research on individual differences in spatial abilities.  My thinking about latent variable methodology and the utility of latent variables in psychological research is heavily influenced by Rolf Steyer's and Michael Eid's framework of defining latent variables.  According to this approach, latent variables are not simply "out there", but have to be "constructed" by the investigator.  That is, latent variables should be explicitly and clearly defined based on concepts that are themselves well defined in psychometric theory, such as true score variables.  For example, how are "method factors" or "growth factors" in complex structural equation models defined and how should we interpret them?  In my lab, we emphasize the need for models in which laent variables are constructively defined, making the resulting statistical models more useful for testing theories in psychology


I teach the following courses at USU: PSY 3010 "Psychological Statistics", PSY 5330/6330 "Principles of Psychological Measurement and Test Theory", PSY 6600 "Statistical Foundations", PSY 7070, "Measurement and Psychometrics", PSY 7610 "Regression Analysis", PSY 7770, "Longitudinal Data Analysis".

Editorial Activities

Associate editor, New Directions for Child and Adolescent Development. Editorial board member: Journal of Personality, Personality Science, Frontiers in Psychology