Advanced longitudinal modeling in Mplus

Description

Many researchers in the social and behavioral sciences are interested in using structural equation modeling (SEM) to analyze longitudinal data and investigate their theories. In this five-day course, SEM experts will teach you about the fundamentals of SEM and various longitudinal data analysis techniques for both panel data and intensive longitudinal data, including growth curve analysis, (random intercept) cross-lagged panel models, and dynamic structural equation modeling (DSEM). The course consists of in-depth lectures, and computer lab meetings in which you learn how to implement these techniques in the powerful and flexible software package Mplus.

On the first day, the focus is on the formulas behind SEM, calculating the number of parameters and degrees of freedom by hand as a way of model checking, interpreting the TECH1 output, when to worry about the default settings in Mplus, model parameterization, and model fit. In the following four days, diverse longitudinal models are discussed, including recent development with respect to cross-lagged panel models, latent growth curve models, latent class growth models, latent transition analysis, and DSEM. In addition, the way in which diverse panel data models relate to causal inference is discussed.

On each day there are, besides lectures, one or more computer labs in which participants obtain hands-on Mplus experience with the discussed topics and models.

Prerequisites 
For this course, we assume participants have Mplus experience and basic SEM knowledge (but you do not need to know matrix algebra, calculus, or likelihood theory).

Recommended combination

A good preparation for this course is our summer course 'Introduction to Structural Equation Modeling using Mplus' (S20), or possibly our summer course Structural Equation Modeling in R using lavaan (S65); an e-learning course which uses lavaan instead of Mplus).

We also offer tailor-made M&S courses and in-house M&S training. If you want to check out the possibilities, please contact us at ms.summerschool@uu.nl

More information

You will receive a certificate upon course completion. Please be aware that this course does not include graded activities, and therefore we cannot provide a transcript of grades.

Participants are requested to bring their own laptop. If needed, software will be available online (please make sure to install this before the start of the course, you will receive an e-mail with information about this).

Lecturers

Prof.dr. Ellen Hamaker, dr. Beth Grandfield,  dr. Jeroen Mulder

Target audience

(Research) master students or (post-graduate) researchers who already use Mplus and/or followed a multiple-day introduction into Mplus, for example the USS course ‘Introduction to Structural Equation Modeling using Mplus’. Participants from a variety of fields, including sociology, psychology, education, human development, marketing, business, biology, medicine, political science, and communication sciences, may benefit from the course.

The selection will be done on a first-come-first-served basis. Please note that there is always a one-day delay between the registration system and the information on the website about the availability of places.

For an overview of all our summer courses offered by the Department of Methodology and Statistics please click here.

Aim of the course

The main objective of the course is to acquire an expert understanding of longitudinal models as applied in the social and behavioural sciences are analyzed using Mplus.

Learning goals per day:

Day 1 

• General overview of SEM as a way to model mean and covariance structures. 
• Gain an insight into the LISREL model and its importance for understanding SEM and what happens behind the scenes in software. 
• Learn path analysis and latent variable models intertwine to answer complex research questions. 
• Gain an understanding of model estimating and fit. 
• Understand how Mplus implements the LISREL model to help understand the software Tech output. 
• Learn about multiple group models in Mplus. 
• Learn how SEM can help evaluate if a questionnaire is biased for different groups by examining measurement invariance and how to implement this in Mplus 

 Day 2 

• Learn about Latent growth curve modeling (LGCM) and how these models can be used to describe change over time. 
• Understand the traditional pieces of a simple LGCM along with the options of how to run these models in Mplus and interpret the output.  
• Gain an introductory knowledge of some of the many extensions of LGCM including non-linear change, adding covariates, and multiple indicator models. 
• Discuss additional interests such as individually varying times of observations and overfitting. 
• Apply your knowledge by analyzing example data and practicing interpreting the results.

Day 3

• To get acquainted with categorical latent variables, specifically in the context of latent class analysis (LCA).
• To understand how to assess model fit for LCA models, how to do model comparisons, and how to interpret its results.
• To learn how to extend the LCA model to longitudinal settings, in particular latent class growth analysis, and growth mixture modeling, and get hands-on experience with fitting these models.

Day 4

• To learn about various popular longitudinal SEM models for investigating cross-lagged relationships.
• To understand the defining differences between these models, and how they impact the interpretation of model parameters.
• Understand some of the critiques on the use of these models for causal inference, and discuss current research that aims to address these critiques.

Day 5

• Learn about dynamic structural equation models.
• Apply your knowledge by analyzing example data and practicing interpreting the results.

Study load

Five full days.

Days 1 and 5: The morning session consists of lectures. In the afternoon session, participants obtain hands-on Mplus experience with the discussed topics and models during a computer lab.

Days 2-4: Both the morning and afternoon sessions consist of lectures followed by a computer lab where everyone can practice working with Mplus.

Tags

• dynamic structural equation modeling (DSEM)
• growth mixture model
• latent class growth model (LCGM)
• latent growth model (LGM)
• M&S
• Mplus
• panel data models
• random intercept cross-lagged panel model (RI-CLPM)
• structural equation modelling (SEM)
• SEM models