Tutorial in Latent Class Analysis

Course Objective

Students who have completed this tutorial will understand the principles and the uses of Latent Class Analysis and they will be able to apply LCA in their own research question.

Course Content

Latent Class Analysis (LCA) is a person-centered probabilistic statistical method that is extensively used in Social Sciences. It is mainly (but far not exclusively) used as model-based clustering method where individuals (or other research entities) are grouped into clusters with similar characteristics according to some observed indicators. The fact that it is a model-based approach allows us to apply it not only for explorative purposes (i.e. to identify groups of individuals) but also for causal inference. In this way, we can predict group (i.e. latent class) membership by covariates or predict distal outcomes using the groups (i.e. latent classes).
This tutorial begins with an introduction to logistic regression which is necessary to proceed. Relations between variables in LCA is always modelled as a multinomial logistic regression. Then we proceed by introducing LCA and its main uses in Social Sciences. We end by discussing the application of LCA to
longitudinal data. The latter application is named hidden Markov model.
The practical work consists of meeting-specific exercises with applications of the material that is discussed in class. Students get real data from the European Values Survey (EVS) to practice with R or Latent Gold. Finally, students are asked to work on the data and the topic of a published paper.

Teaching Methods

The course include four sessions on location and one session with student presentations. The sessions combine theory with exercises. Attendance is required for all sessions. Students are also expected to study chapters of the designated books. After the second session, students can start with the final assignment

Method of Assessment

The assessment of the course includes 3 parts:
1. A participation grade (20% of the final grade). Students are expected to be present and actively participate in all meeting. Failing to do so in a meeting (without a reason) will result in the deduction of 1/8 of this part of the final grade.
2. Completing an individual assignment on latent class analysis (70% of the final grade). This assignment will be given in week 3 after the second session and should be submitted via Canvas.
3. Presenting the individual assignment in the last session (10%).

Literature

Agresti, A. (2018) An Introduction to Categorical Data Analysis, 3rd Edition, John Wiley & Sons, Inc, Hoboken, New Jersey.
Collins, L. and Lanza S.T. (2010) Latent Class and Latent Transition Analysis: With Applications in the Social, Behavioral, and Health Sciences, John Wiley & Sons, Inc, Hoboken, New Jersey

Target Audience

Research Master in Societal Resilience

Recommended background knowledge

Basic statistics, regression analysis