Search
Search type

Cathie Marsh Institute for Social Research

Longitudinal Structural Equation Modelling

Please note that this course has unfortunately been postponed.  A new date will be announced in due course.

Date: 5 December 2016
Duration: 1 day (9.45am - 4.30pm)
Instructor:  Dr Alexandru Cernat
Level:
 Advanced
Course fee: £195 (£140 for those from educational, government and charitable institutions).  The Cathie Marsh Institute (CMIST) offers five free places to research staff and students within the Faculty of Humanities at The University of Manchester and the North West Doctoral Training Centre. Postgraduate students requesting a free place will be required to provide a letter of support from their supervisor.

Outline

Structural Equation Modelling (SEM) has been growing in popularity due to its ability to estimate complex relationships between variables, the possibility to control for measurement error and the comprehensive model fit indicators. This approach is also very well suited to deal with longitudinal data (i.e., data collected repeatedly from the same unit). In this context, it offers a number of unique modelling opportunities. Thus, using SEM with longitudinal data it is possible to investigate how the change in one variable is related to the change in another. For example, it can estimate how the change in physical health is linked to changes in happiness. Similarly, the SEM framework makes it possible to create typologies based on patterns of change in time. Thus, it is possible to see what are the typical patterns of change in cognitive ability or party support and who are the people that manifest them.

This one-day course introduces two of the main models used to analyse longitudinal data using SEM: cross-lagged models and latent growth models (LGM). The cross-lagged model is typically used to better understand the causal relationships in longitudinal studies. On the other hand, LGM models are used to estimate individual level change. Both help answer essential questions related to how individuals/organisations/countries change in time.

The course will be a mix of short conceptual presentations and hands on applications using Mplus (which will be available in the computer cluster). No prior knowledge of Mplus is required.

Objectives

  • Introduce simplex models for change;
  • Introduce cross-lagged models in longitudinal studies;
  • Introduce/refresh model fit evaluation in SEM;
  • Introduce Latent Growth Models (LGM) with practical examples;
  • Short presentation of more advanced longitudinal SEM models: second order LGM and growth mixture models.

Pre-requisites

Participants should be familiar with statistical modelling using linear regression and with confirmatory factor analysis. 

Recommended reading

  • Little, T. D. (2013). Longitudinal Structural Equation Modeling. Guilford Press.
  • Newsom, J. T. (2015). Longitudinal Structural Equation Modeling: A Comprehensive Introduction. Routledge.

The presenter

Dr Alexandru Cernat is a lecturer in the Social Statistics Discipline Area at The University of Manchester. Before he was a Research Associate in the National Centre for Research Methods, based at the University of Manchester. He has received a PhD in survey methodology from the University of Essex working on the topic of mixed modes designs in longitudinal studies.

His research interests are in: latent variable modelling, measurement error, longitudinal surveys/analysis, missing data and survey methodology.