Optimal Split Questionnaire Designs based on Multivariate Normal Model

Tuesday, March 3, 2020 - 12:30pm to 1:30pm
Event Type: 

Speaker: Dae-Gyu Jang

Abstract: A questionnaire is a broadly used tool in surveys for obtaining data for statistical inference. Split questionnaire design (SQD) is an alternative tool utilized in full questionnaires, which can debase the response quality by increasing response burden. Theories in experimental design have been applied to find an optimum SQD. But, it has not been done to find the optimal SQD among probabilistic designs. For multivariate normal data, we investigate the relationship between A-optimality criteria and the sum of mean squared errors of point estimation using multiple imputation. Then, we contrive a simulation study to evaluate the gain of A-optimal design compared to simple random sampling. The results of this simulation study establish the relationship between the gain of A-optimal design and other factors such as correlation structure and the numbers of questions in SQD and full questionnaire.