Manager of Institutional Data Analytic and Reporting, Qatar University, United States
Session Abstract: Nowadays, online surveys are an increasingly popular form of survey research that are usually suffering from the nonresponse bias. We developed a new technique named “Two-phase Post-Stratified Sequential Sampling (TPSS) design” to deal with the nonresponse bias and comparing this technique with simple random sample for a study where the link of a questionnaire is sent to all members of the population by email. This sampling method is developed to be used in surveys where every member has an institutional email. The design has features that can control the nonresponse bias and guarantees large enough observations from each subpopulation in order to run a valid statistical analysis at the subpopulations level. An unbiased estimator and an approximate variance estimator are introduced. Using a simulation study, we show that the sampling design is more efficient than conventional sampling designs. We also provide some guidelines on sampling sizes at the end.
