United Kingdom, Coleraine

Analysing Longitudinal Data: Using Latent Variable Models to Assess Change

when 12 September 2019 - 13 September 2019
language English
duration 1 week
credits 10 EC
fee EUR 300

The analysis of change is central to much psy­chological and social research. Latent Growth Models (LGM) are an important class of models for the assessment of change. In essence these describe individuals’ behaviour in terms of an initial starting point (intercept) and their subsequent devel­opmental trajectories (slope). The technique also allows for the introduction of predictors (covariates) of change. These predictors can be both time-invariant and time-varying and the model can be extended to incorporate other ad­vantages of latent variable framework, e.g., the ability to handle missing data, to introduce both direct and indirect effects and correction for measurement error.

In the context of longitudinal data, latent variable modelling facilitates robust estimation of direct and indirect effects, together with controlling for, and assessing the impact of, moderating and mediating variables.

Course leader

Professor Gary Adamson

Target group

It is expected that participants will have some knowledge and understanding of Structural Equation Modelling.

Course aim

This session will introduce some of the recent developments in the area. Furthermore, applications of the Cross-lagged panel model will be explored and extended to include mixture distributions.

Growth mixture models (GMMs) will be introduced. These models enable the researcher to explore longitudinal data for the presence of unobserved or latent subgroups. In GMMs the assumption of a single homogenous population with a single growth trajectory is relaxed. Instead, a latent categorical variable is introduced with the intention of capturing latent subpopulations in the longitudinal data. These subpopulations are not directly observed, but are inferred from the patterns of responses in the data. In sum, the GMM facilitates the exploration of longitudinal data for unobserved subgroups and estimates latent growth parameters for each of the subgroups.

Credits info

10 EC
This course can be taken as part of a Postgraduate Certificate in Quantitative Methods for the Behavioural and Social Sciences (30 ETCs; 60 UK-credits). It can also be taken without credits.

Fee info

EUR 300: Full fee
GBP 200: Discounted fee (students/unwaged)

Scholarships

NA