Understanding Complex Social Survey Data
The course will introduce the elements of sample design. It will discuss the use of weights to correct for the survey design as well as other common weight adjustments, e.g. for nonresponse and to “calibrate” the sample to external population totals. We will also consider the importance of taking the sample design into consideration when estimating standard errors, in particular in the case of multi-stage samples. Besides learning a set of tools necessary to get reliable estimates from social surveys we will also develop an appreciation of some of the trade-offs faced by survey organisations in collecting the data. Each lecture is accompanied by a Stata practical designed to apply the concepts learned in the lectures.
- Outline the theory of weighting, clustering and stratification,
- Discuss how social surveys are practically implemented in South Africa,
- Introduce the survey analysis tools available in Stata,
- Explain how to get estimates from sample surveys that are representative of the population,
- Show how to calculate the correct standard errors for such estimates.
Prerequisites: Course participants should be familiar with Stata and with basic statistical tools (such as regression analysis). Don't know Stata? Have a look at our Introduction to Stata course.
Date: The next course will be 2022.
Mode of Delivery: This course will be provided online.
Course Instructors: Muna Shifa and Emma Whitelaw
Course Fees: R10 500.00. Partial scholarships are available to bona fide students and academics.
This is a joint SALDRU and DataFirst course.
"This is a very relevant course for students, researchers as well as policy makers. This course would not have come at a better time for me. My only problem is that one week is not enough." (anonymous)
"It is an informative course that's gives insight into complex surveys. It enables us to understand the importance of sample design when performing an analysis and even though you may not deal with the methodology part you get to understand all that goes in to data cleaning." (anonymous)
"Very valuable material transferred in an interactive setting that promoted critical engagement and learning." (anonymous)