Course Coordinator:

Prof. dr. Siswanto Agus Wilopo, SU., M.Sc., Sc.D.,

Clinical Epidemiologist, Biostatistician, and Demographer of the Department of Biostatistics, Epidemiology and Population Health (BEPH), Graduate Public Health Program, Faculty of Medicine, Public Health and Nursing Universitas Gadjah Mada

 

Address           : Gedung IKM Lantai 1, Phone: +62-274-565076 or 548156

Email               : Wilopoilopo@ugm.ac.id

Class website   : https://elok.ugm.ac.id/ or http://gamel.fk.ugm.ac.id.

Course Description

Modern multivariable statistical analysis is based on generalized linear models using maximum likelihood estimation (MLE), including linear, logistic, and Poisson regression, survival analysis, fixed-effects analysis of variance, and repeated measures analysis of variance. This course emphasizes the underlying similarity of these methods, the choice of the right method for specific problems, common aspects of model construction, the testing of model assumptions through influence and residual analyses, and the use of graphical and other methods to present results that are readily understood by health researchers. This second course in biostatistics covers multi-predictor methods, including exploratory data analysis and multiple regressions (linear and logistic). This course will cover more details on the analysis of observational data, including from survey and survival analysis (time-to-event issues). In addition, new topics will be introduced: structural equation modeling (SEM), marginal effects, and analysis data with many missing observations. Emphasis is on the practical and proper use of statistical methodology and its interpretation. The statistics package STATA will be used throughout the course. Students interested in analyzing a big data set (i.e., IDHS., household survey) are suggested to take the course.

Goals and Course Objectives

The goal of this course is to provide knowledge and skill to the students for analyzing data using a multivariable technique. At the end of the course, students will be able to:

• assess characteristics of the problem to help choose the appropriate analytic technique,
• compare techniques appropriate for handling a single outcome variable and multiple predictors,
• demonstrate analysis of binary data using simple and multiple logistic regressions,
• demonstrate analysis of simple and multiple poison regressions,
• demonstrate and select analysis of time-to-event data (survival analysis), including the time-dependent covariate,
• describe analysis parametric survival analysis,
• demonstrate analysis using structural equation modeling,
• demonstrate analysis for longitudinal data,
• demonstrate analysis of data with missing values and imputation methods, and
• demonstrate analysis of data using marginal effects.