Felix Clouth

Methodology and Statisticsphoto
Tilburg School of Social and Behavioral Sciences
Tilburg University

Academic webpage Felix Clouth

On 1 March 2024 Felix Clouth defended his thesis: Latent Class Models for Causal Inference at the university of Tilburg.

Summary
Causal inference methods such as inverse propensity weighting (IPW) or the g-formula
allow for the estimation of causal effects on non-randomized observational data. In
contrast to, e.g., structural equation modeling, modern causal inference separates the
estimation of a causal effect from its definition and identification. For instance, a causal
effect might be defined as the difference between two potential outcomes, identified
using assumptions such as exchangeability, consistency, and positivity, and estimated
using IPW. While this view on causality is widely adopted in epidemiology, public health
research, and economics, applications in the social and behavioral sciences are few.
One reason for this might be the high prevalence of measurement error in the data
requiring the use of measurement models such as latent class analysis (LCA). LCA is a
modelling based unsupervised clustering technique that is used to identify patterns in
the data and cluster individuals into unobserved classes based on the similarity in their
response patterns. For instance, LCA is frequently used for analyzing patient reported
outcome measures data as such data often measure an underlying construct that is not
directly observable.
In this thesis, we developed new methodologies to combine LCA with modern causal
inference techniques. Using a combination of LCA and IPW or the g-formula, new
methods were presented for estimating causal effects of an exposure on an
unobservable outcome measured through several indicators. These methods were
extended to deal with measurement non-invariance, to estimate the causal effect of an
unobservable exposure on observed outcomes, and to estimate the causal effect of a
time-varying exposure on dynamic unobservable outcomes.

Supervisors
Prof. dr. Jeroen Vermunt, prof. dr. Steffen Pauws

Financed by
NWO (project: DATA2PERSON)

Project period
1 September 2018 – 1 March 2024