GESIS Training Courses

Wiss. Koordination

Dr. Nora Müller
Tel: +49 621 1246277

Administrative Koordination

Claudia O'Donovan-Bellante
Tel: +49 621 1246-221

Regression Modelling Strategies to test Research Hypotheses: Theory and Applications with Stata

Prof. Moris Triventi, PhD

Datum: 27.06 - 29.06.2018 ics-Datei

Veranstaltungsort: Mannheim, B2, 8 / Kurssprache: Englisch

Referenteninformationen - Prof. Moris Triventi, PhD


Regression analysis, in its various forms, is a key analytical tool for every social scientist interested in using quantitative methods to investigate social phenomena. Most of the students and scholars have attended courses introducing the basics of regression analysis or discussing the most sophisticated details regarding its assumptions and statistical properties. Often, less attention is devoted to connecting regression analysis to, on one side, the more specific research questions a researcher has in mind and, on the other, on the practical decisions that have to be made to apply regression models to concrete datasets and actual research problems. This course aims to fill this gap, by providing a comprehensive discussion of regression modelling strategies as useful tools to test research hypotheses.
I will show the various purposes of regression analysis and the importance of clarifying the specific research aim before deciding crucial aspects such as choice of the variables and model specification. I will show that the common practices of introducing a lot of covariates and commenting the associated regression coefficients suffers from many flaws and I will discuss viable alternatives.
An extensive part of the course will be focused on discussing common mistakes that should be avoided when using regression analysis. I will also provide a comprehensive overview of what can be done and what cannot be done when using nonlinear regression analysis. Can we compare coefficients from different model specifications or from different groups? At which conditions? In which case can we attribute the status of causal effects to our estimates?
The last part of the course is focused on how to obtain the appropriate quantities of interest from regression models, and how to present and comment effectively the results from regression analysis. We will pay attention to substantial significance and effect size compared to mere statistical significance, and the use of graphical representations along with tables to illustrate intuitively the main findings in relation to the research hypotheses guiding the empirical investigation.
The course will be useful not only for beginners of regression analysis, but also for those applying their research more advanced statistical techniques (e.g. panel data analysis, multilevel models), since most of the issues covered during the course can be applied in such more sophisticated kinds of analysis.  



Master and PhD Students, as well as post-doc researchers doing quantitative research in the fields of social and political sciences, in particular involved in the analysis of large scale datasets


  •         Be aware and avoid common mistakes when applying regression analysis
  •         Select the best analytical design to test specific kinds of research hypotheses
  •         Interpret in the correct way the results of various kinds of regression analysis
  •         Choose the appropriate and best ways to present and communicate results from regression models


Participants should have previous experience of secondary data analysis and basic knowledge of the statistical software Stata.


Weitere Informationen