Most major surveys use complex sampling methods such as stratification, multistage sampling and unequal selection probabilities. The result is that the sampling variances of estimates are different from those that would be obtained from simple random sampling (SRS). A convenient way to summarise the effect of using complex sampling methods on an estimate is to calculate the design effect, which is the ratio of the sampling variance for the complex design to that which would be obtained using SRS. Knowledge of the likely design effects can be very useful when designing a survey. In this presentation some of the issues associated with analysing and interpreting design effects will be discussed. Some empirical results from the Australian National Children’s Nutrition and Physical Activity Survey will be presented.
Presented by:
David Steel (Centre for Statistical and Survey Methodology, University of Wollongong)
Date & time:
November 28, 2011 4:00 pm
External seminars home