Applied Analysis of Variance
and Linear Modelling
Prof Dr. Edgar Erdfelder,
Professor at the Department of Psychology, University of Mannheim, Germany
University of Göttingen, Germany (1974 - 1980)
University of Trier, Germany (Research Assistant and Lecturer, 1981-1984); University of Bonn, Germany (Senior Lecturer, 1984-2001); University of Gießen, Germany (Professor of Psychological Methodology, 2001-2002); University of Mannheim, Germany (Professor of Experimental Psychology, 2002-)
- Teaching experience:
Experimental Psychology, Experimental Design, Statistical Methods and Data Analysis, Methodology of Psychology
- Research interests:
Experimental Cognitive Psychology (especially memory); Quantitative research methods in the social sciences; Statistical power analysis; Psychological measurement
- Editorial board memberships:
"Experimental Psychology" (2002-), "Journal of Experimental Psychology: Learning, Memory, and Cognition" (2000-2002), "Methods of Psychological Research - online" (1996-), "Psychology Science" (2003-).
For more details and publications
The workshop will cover the analysis of experimental and quasi-experimental designs with continuous dependent variables from an applied perspective. Among the topics are:
One- and multi-factorial analysis of variance with fixed effects (ANOVA)
- Post-hoc comparisons: to use or not to use?
- Planned comparisons and "tailor-made hypothesis tests"
- Analysis of covariance (ANCOVA) and alternatives
- Random and mixed effects ANOVAs: to use or not to use?
- Repeated-measures ANOVA
- Multivariate analysis of variance (MANOVA)
- Statistical power analyses for (M)ANOVAs, ANCOVAs, and planned comparisons
- What to do when the distributional assumptions are not met?
- Hierarchical Linear Models (Multilevel Modelling)
We will study and discuss these topics using concrete research problems and both real and simulated data. We will apply most of these methods using the SPSS and the GPOWER computer programs.
Relevant background knowledge
Hays, W.L. (1994). Statistics (5th ed.). Fort Worth: Harcourt Brace College
Material that will be covered during the course:
- Cohen, J., Cohen, P., & West, S. G. (2003) Applied multiple regression/correlation analysis for the behavioral sciences. Mahwah, NJ: Lawrence Erlbaum Associates
- Edwards, L.K. (Eds.). (1993). Applied analysis of variance in behavioral science. New York, NY, US: Marcel Dekker, Inc.
- Erdfelder, E., Faul, F. & Buchner, A. (1996). GPOWER: A general power
analysis program. Behavior Research Methods, Instruments, & Computers, 28,
- Myers, J. L. & Well, A. D. (2003). Research design and statistical analysis (2nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates.
- Sahai, H. & Ageel, M.I. (1999). The analysis of variance. Fixed, random,
and mixed models. Birkhäuser Verlag.
- Snijders, T. A. & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage
- Stevens, J. (2002). Applied multivariate statistics for the social sciences (4th ed.). Mahwah, NJ: Lawrence Erlbaum Associates
You should have some background knowledge in experimental design and applied statistics as covered, for example, in the first one or two years of psychology studies (see, e.g., Hays, 1994; Myers & Well, 2003).
You should be familiar with the SPSS handling (i.e., data input, variable
and value labels, data transformations, merging and splitting data files,
and the SPSS statistics menu).
In addition, you should familiarize yourself with the GPOWER power analysis
program (Erdfelder, Faul, & Buchner, 1996). GPOWER is freeware. The program
may be obtained from the web site
A manual is available at:
Bring your own data
I urge you to bring your own data to the workshop so that we can aim at
solving the data analysis problems you are faced with. If you are planning
to collect data that will not yet be available in August 2001, think about
preparing a "simulated data set". In any case, if you will bring your own
(real or simulated) data to the workshop, please contact me in advance
(preferably by email:
email@example.com), and describe both your data set and the underlying
research questions. Please do so as early as possible but not later than
August 18, 2003.