Swiss Summer School 2002

Causal Models and Structural Equations for panel data

Peter Schmidt

Peter Schmidt is Professor for Social Research at the University of Giessen. Presently, Program Director at ZUMA (Centre for Survey Research and Methodology, Mannheim, Germany), responsible for Social Monitoring, (General Social Survey, Social indicators, and Microcensus Data). His research interests are the foundations and applications of structural equation models, analysis of panel data, and empirical testing of rational choice theory. Applications include national identity and environmental behavior, topics on which he has published several books and papers.

Workshop contents

Part 1: Confirmatory Factor Analysis

Day 1

Overview of the whole course. Different model specifications, causality and empirical research. Process and strategy of theory testing. Notation for Structural Modelling. Use of SEMNET and AMOS manual. Theoretical exercise 1. Description of the data-sets and overview of the input files.

Practical Session: AMOS and the logic of its' use. Confirmatory Factor Analysis (CFA) with one measurement model. Preparation of Example 1a (input file: corger.sav) "discrimination against foreigners" with four indicators. Computation of model 1. Output interpretation. Example 1b "Factor Models with two panel waves without autocorrelation". Example 1c "Factor Model with two panel waves with autocorrelation".

Foundation of Confirmatory Factor Analysis (CFA): Assumptions, model specification and identification. Types of restrictions. Typology of model testing: parallel, tau-equivalent and congeneric models. Theoretical exercise 2. Discussion of example 1.

Practical Session: Factor Models for longitudinal data Preparation of ExampleS 2a, b, c. Preparation of tables with detailled and global modal fit with two panel waves. Estimation and identification in CFA. Model modification and the strategy of theory testing: New factors, new factor loadings or residual correlations. The structure of autocorrelations and multiple time points. Preparation of ExampleS 3a and 3b: Authoritarianism, discrimination against foreigners with three time points. Examination of detailed and global model fit.

Day 2

Reliability and validity estimates in CFA with multiple time points. Variance decomposition. Multiple Groups. Missing Values. Theoretical exercise 4. Discussion of examples 3a and 3b.

Practical Session: SCFA. Preparation of multiple group comparisons due to a) East/West Germany, b) education and c) age. Examples 4a preparation of tables.

CFA with observed and latent means and multiple time points. Higher order factor models and MTMM-Models (Multi Trait Multi Method). Equivalent models. Theoretical exercise 5. Discussion of examples 4a, b, c. PRACTICAL SESSION: CFA with means: Subgroups East/West Germany, education, and age. Preparation of Example 5a. (higher factor vs. residual correlation); example 6: Missing Values

Part 2: Structural Equation Models

Day 3

Structural Equation Models (SEM) with latent variables and multiple indicators for panel data: Specification, identification and estimation. Cross lagged panel models. Causality and equivalent models. Typology of model testing. The "two step strategy". Theoretical exercise 6. Discussion of example 5a.

Practical session: Preparation of SEM for ethnocentrism with three constructs and multiple indicators. Preparation of examples 7a, b.

Model testing and model modification in panel models. Detailled and global fit measures. Interpretation of parameters. Decomposition of effects. Theoretical exercise 7. Discussion of examples 7a,b.

Practical session: SEM with multiple groups. Model specification and estimation. Theoretical Exercice 8. Discussion of examples 8a, 8b.

Day 4

SEM with multiple groups and multiple time points: Model specification and estimation. Types of restrictions. Growth Curves 1. Theoretical exercise 8. Discussion of examples 8a,b.

Practical Session: SEM, multiple groups with latent means: Preparation of ExampleS 9a,b,c. Model Competition

Growth Curves 2: Specification and test. Typology of model testing revised. Theoretical exercise 9. Discussion of examples 9a,b,c. Practical Session: Preparation of ExampleS 10a,b,c.

Day 5

Presentation of participantīs models. Strategy of model testing. Hot topics. Open Questions.


The objective of this course is to show how structural equation modeling can be used to develop and/or test both measurement models and causal theories between latent variables. A further important aim is to familiarize participants with the AMOS program. The program will be run by graphical input via path diagrams (AMOS Graphics) and by command language (AMOS Text).

A student version of AMOS is available via internet: Smallwaters


Basic text/overview

Remedial Reading