Home Studium Workshops / Konferenzen Videos Forschung Dienstleistungen Team
Impressum / Kontakt
deutschsprachige Version 
Besucher seit 27.02.2014:
ganze Website: 293894
aktuelle Seite: 116
Causal Effects EAM Fachgruppe Methoden und Evaluation (DGPs)

Kurse: Kursinformationen

en  Introduction to Survival Analysis and Event History Modeling

Kursleitung: Prof. Dr. Tenko Raykov, Michigan State University, USA

Wintersemester 2009/2010, Workshop, Kurslänge: 6.00 Stunden, Sprache: Englisch

Instructor: Prof. Dr. Tenko Raykov, Michigan State University, USA (Homepage: www.msu.edu/~raykov)

Dr. Raykov is a Professor of Measurement and Quantitative Methods at Michigan State University. He specializes in latent variable modeling, longitudinal data analysis, multilevel modeling, survival analysis, multivariate statistics, behavioral and social measurement. He is a Consulting Editor for the British Journal of Mathematical and Statistical Psychology, Multivariate Behavioral Research, and Psychological Methods, and a member of the Editorial Board of Structural Equation Modeling. Recently published (with George A. Marcoulides) books of his are: A first course in structural equation modeling (2006, Erlbaum) and An introduction to applied multivariate analysis (2008, Taylor & Francis). Prof. Raykov has more than 20 years of experience in co-working with behavioral and social scientists.


Abstract:

This workshop provides an introduction to the increasingly popular methodology of survival/event history analysis in the behavioral, social, medical, and educational sciences. A main concern of the methodology is with modeling the time to occurrence of an event of major substantive interest - such as developing dementia, stroke, onset of a disease, relapse, discharge from clinic, first drug, alcohol or tobacco use, death, dropping out of school etc. - in relation to studied covariates (that in this workshop may be observed or alternatively latent variables). A major challenge when analyzing this type of data is the presence also of censored observations, i.e., subjects who do not experience the event by the end of the study period, are lost to follow up, or have withdrawn from it.

In the past, all too frequently empirical researchers in these disciplines have discarded censored observations, or handled them as if they were regular (uncensored) observations, or alternatively treated them as missing values. Either of these traditional procedures is wasteful of information, inefficient, and in general yielding misleading statistical results and substantive interpretations. The present workshop is concerned with optimal and principled methods for analyzing and modeling this type of data, including in particular the censored observations. These methods lead to efficient parameter estimates and results utilizing all available information in the collected data from all subjects in the study.

To this end, the fundamental concepts of censoring, survival function and hazard function are initially discussed, and a nonparametric procedure is outlined of survival function estimation as well as its comparison across groups via the Kaplan-Meier product limit estimation approach. The widely used Cox' proportional hazards model is then focused on. Methods for examining possible violations of its assumptions are also indicated, as are extensions of the model when they are not fulfilled. A generalization of conventional survival analysis models to accommodate fallible latent covariates (dimensions) measured by multiple indicators is next outlined, followed by a discussion of discrete time survival analysis. Throughout the workshop, a number of illustrative examples are utilized for demonstration of main concepts in survival/event history analysis and their interrelationships. The popular statistical packages STATA, SPSS, R (Splus), and Mplus are employed for analytic and modeling purposes. Input instructions for their applications are discussed in detail, as are resulting output sections.


Intended Audience:

Researchers in psychology, sociology, epidemiology, medicine, criminology, education and economics, as well as graduate and advanced undergraduate students in these and related sciences. No prior knowledge of survival analysis is necessary or expected from the participants.



Um die herunterladbaren Videos anzuschauen, benötigen Sie den VLC Player VLC media player, den Sie hier herunterladen können. Die Videos werden bereitgestellt über die Digitale Bibliothek Thüringen 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.

Die Videos dieses Kurses sind passwortgeschützt über die dbt zugänglich.
Wenn Sie Angehörige(r) der FSU Jena, TU Ilmenau oder Uni Erfurt sind können Sie sich in der dbt direkt mit Ihrem Login der Hochschule anmelden.
Wenn Sie kein(e) Angehörige(r) der genannten Hochschule sind können Sie sich hier eine Benutzerkennung in der dbt erstellen, sofern Sie dies noch nicht gemacht haben.
Nach erfolgreichem Login in der dbt müssen Sie einmalig für jedes Video dieses Kurses folgenden Leseschlüssel eingeben: (sie müssen sich erst in diesem Kurs einloggen, um den Leseschlüssel hier angezeigt zu bekommen).

 

Videos


Session Video-presentation with slides
Session 01 Video with slides 01
Session 02 Video with slides 02
Session 03 Video with slides 03
Session 04 Video with slides 04


Material