Kurse: Kursinformationen

  Analysis of Total, Direct, and Indirect Effects in Experiments and Quasi-Experiments

Kursleitung: Prof. Dr. Rolf Steyer

Sommersemester 2010, Workshop, Kurslänge: 11.75 Stunden, Sprache: Englisch, Thema: Analysis of causal effects

This course is an introduction to the stochastic theory of causality (Steyer et al. Springer: 2010), which is a generalization of the theory of causal effects in the tradition of J. Neyman and D. B. Rubin. All designs and models for the analysis are developed for the purpose of learning about conditional and/or average total, direct, and indirect effects. Unlike other courses on the analysis of treatment effects, it uses structural equation modeling (with or without latent variables) instead of analysis of variance techniques, the General Linear Model or related techniques.

This course is a synthesis of different traditions in methodology: Rubin's approach to causality, the Campbellian tradition of quasi-experimentation and internal validity, and structural equation modeling.

Although this workshop does not require experience and knowledge in structural equation modelling (SEM), we do not recommend this workshop as a first introduction to SEM, if the motivation is to have an introduction into SEM. For this purpose, I rather suggest our course "Introduction to latent variable modeling with Structural Equation Models using Mplus" held in April 2009 at the University of Jena. This and other courses are still available in the internet and on DVDs at: http://www.metheval.uni-jena.de/courses.

The course, "Analysis of Total, Direct, and Indirect Effects in Experiments and Quasi-Experiments" aims at those who are interested in data analysis in experimental and quasi-experimental studies involving covariates such as one or several pretests, a discrete treatment variable, and one or several outcome variables.

Data analysis can be done with EffectLite, which is a program developed by Rolf Steyer and Ivailo Partchev, which will be provided to all participants. It analyzes a generalized multivariate analysis of variance and covariance. It creates LISREL and/or Mplus input files, reads and interprets the results, computes some statistics, and produces an output file containing the results that are most important for the analysis of total, direct, and indirect treatment effects. In the univariate case, EffectLite does not assume homogeneity of variances of the outcome variable between groups. In the multivariate case with two or more outcome variables it does not assume homogeneity of covariance matrices of the outcome variables between groups. Furthermore, it allows analyzing mean differences and adjusted means differences - aimed at estimating causal effects - between groups with respect to

• several manifest covariates and/or outcome variables
• one or more latent covariates and/or outcome variables, and
• a mixture of the two kinds of variables.

The covariate(s) may also be qualitative (blocking factors). In this case, we estimate and test average effects for non-orthogonal analysis of variance designs, provided that the covariates are specified as qualitative indicator variables. If the covariates fulfil certain assumptions, the program estimates and tests the conditional and average causal effects.

In the course we will

• present the theory of stochastic causality with an emphasis on causal effects
• show how to use EffectLite, LISREL, and Mplus for the analysis of total, direct, and indirect effects Theory
• Motivation: Simpson's paradox, non-orthogonal ANOVA
• The scope of the theory: random experiments
• The mathematical structure of causal models: causality spaces
• True outcome variables, average and conditional causal effects
• Prima facie effects
• Sufficient conditions for unbiasedness
• The role of randomization and other design techniques and strategies of data analysis Applications
• Estimating and testing average and conditional total, direct, and indirect effects via structural equation modeling (Applications using EffectLite, LISREL and Mplus)

Um die Videos anzuschauen, benötigen Sie den VLC media player, den Sie hier herunterladen können. Die Videos werden bereitgestellt über die Digitale Bibliothek Thüringen (dbt). Die grün verlinkten Videos und Materialien sind kostenfrei abrufbar. Klicken Sie auf den grünen Link, um das Video zu sehen oder die Datei herunterzuladen! Um alle Materialien und Videos abrufen zu können, müssen Sie sich einloggen.

Videos

Material

• Handout-version of the slides for the workshop

• Steyer, R., Partchev, I., Kroehne, U., Nagengast, B., & Fiege, C. (2010). Probability and Causality: Theory. Manuscript in preparation.Heidelberg: Springer.
• Steyer, R., Partchev, I., Kroehne, U., Nagengast, B., & Fiege, C. (2010). Probability and Causality: Theory. Manuscript in preparation.Heidelberg: Springer. Glossaries.
• Steyer, R., Fiege, C. & Rose, N. (2010). Analyzing Total, Direct and indirect Causal Effects in Intervention Studies. ISSBD Bulletin, 57, 10-13
• Steyer, R., Nachtigall, C., Wüthrich-Martone, O. & Kraus, K. (2002). Causal regression models III: covariates, conditional and unconditional average causal effects. Methods of Psychological Research-Online, 7, 1, 41-68.
• Steyer, R., Gabler, S., von Davier, A. & Nachtigall, C. (2000b) Causal regression models II: unconfoundedness and causal unbiasedness. Methods of Psychological Research-Online, 5, 3, 55-86.
• Steyer, R., Gabler, S., von Davier, A., Nachtigall, C. & Buhl, T. (2000a) Causal regression models I: individual and average causal effects. Methods of Psychological Research-Online, 5, 2, 39-71.

• Steyer, R. (2005). Analyzing Individual and Average Causal Effects via Structural Equation Models. In: Methodology European Journal of Research Methods for the Behavioral and Social Sciences, 1, 39-54.
• Pohl, S., Steyer, R. & Kraus, K. (2008). Modelling method effects as individual causal effects. Journal of the Royal Statistical Society. Series A, 171, 41--63.
• Pohl, S. & Steyer, R. (2010). Modelling common traits and method effects in multitrait-multimethod analysis. Multivariate Behavioral Research, 45, 1-28.
  
  A short excursus concerning true change models
  • Steyer, R., Eid, M. und Schwenkmezger, P. (1997). Modeling true intraindividual change: True change as a latent variable. Methods of Psychological Research-Online, 2, 21-33.
  • Steyer, R., Partchev, I. & Shanahan, M. (2000). Modeling True Intra-Individual Change in Structural Equation Models: The Case of Poverty and Children's Psychosocial Adjustment. In: Little, T. D., Schnabel, K. U., & Baumert, J. (Eds.), Modeling longitudinal and multiple-group data: Practical issues, applied approaches, and specific examples (pp. 109-126). Hillsdale, NJ: Erlbaum

• SEM and latent variable modelling with Mplus
• LISREL Reference Sheet

• EffectLite for LISREL
• Manual for EffectLite for LISREL

• EffectLite for Mplus
• Manual for EffectLite for Mplus

• Causal Effects Explorer

In Lisrel-versions which had compiled before August, 29, 2007 there were a bug in the FIML option for multiple group analyses. Here you can get an updated version of the executable for the student edition of Lisrel 8.8 without this bug. If you wanted to have the updated executable for the full version of Lisrel, please contact the technical support from Scientific Software International, Inc.

• Lisrel88s.exe without the bug in the FIML option for multiple group analyses

• Simulated data to the corresponding tables of Steyer et al. (2010).
• Klauer data set
• kirchmann_data.zip, Kirchmann data set