Complex sampling designs in epidemiology, April 3, 2009


Teacher:

Thomas Lumley

Location:

All teaching takes place at the Center for Health and Society (Kommunehospitalet) in room 7.0.18, which is in building 7, ground floor. Enter via Entrance A on this Map and continue through the hallway next to the bookstore.

Laptops:

Everybody should bring a laptop. I will assume that most will be using Windows XP or Vista. If you need help with Macs or Linux machines please contact the course director separately. Make sure that the computer has not been locked by your administrator. We'll bring the necessary files along, but it will be good if as many of you as possible install the software in advance.

Basic installation:

R for Windows can be found here: http://cran.dk.r-project.org/bin/windows/base/. Just download and run

R-2.8.1-win32.exe

Also install the survey, RSQLite, quantreg, hexbin, mitools packages, either via the menu, or by typing

install.packages(c("survey", "RSQLite", "quantreg", "hexbin", "mitools")).

If you are unfamiliar with R, it is suggested that you go through the interactive session at the end of the manual “An Introduction to R”, which comes with the software.

Schedule:

9:15-12:00

Traditional designs for health surveys

  • Introduction

  • Sampling designs and basic analysis

    • exercises

  • Graphics and regression modelling

    • exercises

12:00-13:00

Lunch break

13:00-15:45

Modern methods for two-phase samples

  • Two-phase designs

  • Calibration, post-stratification, raking

    • exercises

  • How efficient are design-based methods compared to model-based methods?

Literature:

We shall primarily use handouts of the slides, but the following literature may be relevant:

Also useful are the manuals that ship with R, especially

Participants:

The list of participants can be found here (PDF file)

Further materials:

Slides and various supporting files can be found here (PDF files and data files). Documentation of the “nhanes” survey can be found here.


Contact:

You can mail the course director at p.dalgaard@biostat.ku.dk or phone me at 35 32 79 18.