An optimal assignment procedure for multiple online surveys

The problem of optimal assignment of respondents to internet surveys has been considered. The task is taken from a leading online research company in Central and Eastern Europe, which runs several dozen surveys in parallel. Each survey is assigned a target number of responses to be collected and unique selection criteria for choosing respondents based on their socio-demographic data. A mathematical programming model has been proposed that enables us to find an optimal mechanism for assigning respondents to surveys so as to minimize the required total number of invitations to surveys issued. A side effect of the assignment procedure is that the samples obtained are not representative of the population under survey. Therefore, a sample weighting scheme has been developed that takes this fact into account and allows us to obtain unbiased estimators of the characteristics of the population surveyed.