Analysis of Survey Data: 101 and Beyond | Statistical Consulting and Survey Research Unit

Though the workshop will not go into mathematical details and proofs, it will provide some reasoning and intuition behind the key formulas used in survey sampling. It will also look at various examples using the statistical software SAS (free to students, faculty and staff at the University of Waterloo). In addition to illustrating basic data analysis techniques, those examples aim to provide guidance on some of the pitfalls of analyzing survey data. Part 3 is a practical hands-on opportunity for participants to conduct basic data analysis using SAS on their own laptop. Participants are not expected to know about survey sampling, but should have some knowledge of basic statistical concepts (ideally have taken at least one course in statistics).

This two-days workshop will be held at M3 3127. Registration is free for all members of the university.

Workshop outline

Part 1: Dec 6th, 9:30 AM – 12:00 PM

Introduction (vocabulary, notation, and basic concepts)
Sampling designs and selection probabilities
Introduction to sampling weights

Part 2: Dec 6th, 1:00 PM – 4:00 PM

Estimating a population total (i.e., Horvitz-Thompson estimator)
Estimating a population mean (i.e., Hajek estimator)
Estimating a population proportion
Selection bias and other non-sampling errors
Rao-Scott Chi-Square test

Part 3*: Dec 7th, 9:30 AM – 12:00 PM

Introduction to SAS OnDemand for Academics, basic SAS commands and procedures
Using the SURVEYFREQ and SURVEYMEANS procedures
Basic data manipulations, better looking output (e.g., formats, titles, footnotes) and ODS

Part 4: Dec 7th, 1:00 PM – 3:30 PM

Linear and logistic regression using survey data
Using the SURVEYREG and SURVEYLOGISTIC procedures

* In Part 3, participants will conduct basic data analysis using SAS OnDemand for Academics (free to students, faculty and staff at the University of Waterloo). Hence, participants must bring their own laptop and register. Installation instructions will be provided, but in person help will also be available from 4:00 to 4:30 pm on day 1.

Our speaker: Christian Boudreau

Dr. Christian Boudreau is a Research Associate Professor at the Department of Statistics and Actuarial Science, University of Waterloo. He is also the Senior Methodologist of the Statistical Consulting and Survey Research Unit.

Our speaker: Gradon Nicholls

Gradon Nicholls is a second year Ph.D student at the Department of Statistics and Actuarial Science, University of Waterloo. He is currently a research assistant at the Statistical Consulting and Survey Research Unit.