Start your brilliant career with a degree from Australia's #1 ranked university.


Find contact details for any general enquiries.

Engage with us

Find contact details for any general enquiries.

Study with us

Our graduates gain the knowledge and skills to lead organisations, develop public policy, create new companies and undertake research.

Our research

Our research is focused on issues that are highly significant for organisations, the Australian economy, and society at large.

Current Students

Whether you're a new or continuing student, you can find everything you need here about managing your program and the opportunities available to you.


Our alumni may be found in the world’s leading companies, policy agencies and universities.

Contact us

Find contact details for any general enquiries.

RSFAS Seminar | Assoc Prof Robert Clark

RSFAS Seminar | Assoc Prof Robert Clark

Speaker: A/Prof Robert Clark
Institution: ANU

Title: Complete Sample Likelihood Analysis of Complex Surveys

Abstract: Sample surveys typically have a range of objectives including enumerative inference (population means and proportions) and analytical inference (parameters of an assumed model). Sample data is often subject to informative selection, because populations are sampled with unequal intensities in order to meet these various objectives. To remove resulting biases, regression modelling of survey data is often weighted by the inverse of the selection probability. This removes the possible bias from an informative design but it increases the variance, often by a factor of between 1.5 and 3. Sample likelihood is an alternative approach based on the distribution of the dependent variable conditional on the selection event. It can avoid biases due to the complex sampling plan with relatively little penalty to the variance of parameter estimates. However, it is more complicated to apply, as it requires a “selection model” of the selection probabilities conditional on the dependent and independent variables in the regression of interest. Existing sample likelihood methods do not incorporate the variability due to the selection model in the likelihood. We propose a complete sample likelihood which propogates uncertainty about both the process model of interest and the selection model. Methods for continuous and binary variables will be described for both one and two-stage sampling. We evaluate some competing methods in simulations motivated in part by the ABS Business Longitudinal Survey.

This is a joint work with Dr Francis Hui (ANU).

Updated:   27 August 2020 / Responsible Officer:  CBE Communications and Outreach / Page Contact:  College Web Team