|
|
Einrichtungen >> Fakultät Sozial- und Wirtschaftswissenschaften >> Institut für Soziologie >> Lehrstuhl für Soziologie, insbes. Methoden der empirischen Sozialforschung >>
|
Fortgeschrittene Verfahren der Längsschnittanalyse: Applied Panel Data Analysis- Dozentinnen/Dozenten
- Prof. Dr. Michael Gebel, Chen-Hao Hsu, M.A.
- Angaben
- Seminar
Rein Präsenz
4 SWS, Unterrichtssprache Deutsch
Zeit und Ort: Mo 8:00 - 12:00, RZ/00.06
- Voraussetzungen / Organisatorisches
- Requirements:
Students are expected to be familiar with the content of the compulsory lecture 'Research Design' and with multiple linear and binary logistic regression analysis. Students are also expected to be familiar with the statistical package Stata. These skills can be acquired or refreshed either through self-study or through an online tutorial. Link to Stata tutorial
Registration: Please register in the VC (Link to VC ) of the event until 12.10.2023. Only those registered in the VC will receive all information about the course.
Type of instruction:
On-site teaching
Language of instruction:
English
Module exam:
Portfolio in English (time: 3 months)
Barrier-free participation: Please contact the lecturer in advance of the course if you have any needs with regard to barrier-free participation in the course.
- Inhalt
- Learning targets:
The central aim of this course is to enable participants to critically discuss the basic concepts and assumptions of various panel regression estimators for linear and binary dependent variables, to conduct theory-driven empirical research, to select and specify appropriate regression models according to the ideas of modern causal analysis, to independently carry out panel data analyses using the statistical package Stata and data from the German Socio-Economic Panel (SOEP), to correctly interpret the results and to clearly present the results of regression analyses in tables and graphs.
Course contents:
The course begins by explaining the basic logic of panel design compared to cross-sectional design, and by practicing the preparation and descriptive analysis of panel data in Stata. In the next step, pooled ordinary least squares (POLS), fixed effect (FE), first difference (FD), random effect (RE) and hybrid models for linear dependent variables are presented and explained. The different assumptions for the consistency of the estimators (=causal inference) are compared: the exogeneity assumptions are illustrated using causal graphs, and the implications of the full rank assumption for the inclusion of time-constant regressors and the need for sufficiently time-varying regressors are discussed. In addition, the assumptions for the efficiency of the estimators (= statistical inference) are presented, appropriate statistical tests (e.g. tests for serial correlation of error terms, Hausman test) are practised, and implications in the form of the use of panel-robust standard errors are explained. The different models are compared and practical recommendations for model selection are given. Based on the principles of theory-driven empirical research and modern causal analysis, the choice of time-constant and time-varying control variables is explained. In terms of model specification, the role of lagged dependent variables (LDV) and lagged explanatory variables (LEV) as control and causal variables (keywords: dynamic panel models, impact function, anticipation effects) is also discussed. Problems and possible solutions of bidirectional causality in the form of feedback loops and simultaneity are discussed. Aspects of modelling age, period and cohort effects and growth processes are discussed. Panel data models for binary dependent variables are presented. Finally, the issue of missing data in panel models is addressed.
In addition to theoretical introductions to the models, the logic of the various panel data estimators is illustrated using very simple game data. In laboratory sessions, participants will learn how to implement panel data analysis in practice using the statistical package Stata. The lab sessions and the seminar papers will be based exclusively on current sociological issues in life course research (consequences of life course events) and data from the German Socio-Economic Panel (SOEP). Specifically, the course offers an applied introduction and hands-on experience in the complex preparation of panel data for statistical analyses during the lab sessions.
- Englischsprachige Informationen:
- Title:
- Applied Panel Data Analysis
- Zusätzliche Informationen
- Erwartete Teilnehmerzahl: 30
- Institution: Lehrstuhl für Soziologie, insbes. Methoden der empirischen Sozialforschung
Hinweis für Web-Redakteure:
Wenn Sie auf Ihren Webseiten einen Link zu dieser Lehrveranstaltung setzen möchten, verwenden Sie bitte einen der folgenden Links:Link zur eigenständigen Verwendung Link zur Verwendung in Typo3
|
 |
|
|
UnivIS ist ein Produkt der Config eG, Buckenhof |
|
|