In modern social science research, surveys are mainly conducted online. The high cost of recruiting units into probability-based surveys means that nonprobability online panels are used ever more often. These are large databases of potential respondents who agree to participate in various web surveys for a certain incentive. Most market research and some public sector research has already moved to nonprobability online panels that are significantly cheaper than probability-based surveys. The latter, in turn, ensure the better accuracy of survey estimates. The article reviews evaluations of nonprobability online panels, illustrates typical problems encountered, and presents guidelines for deciding whether to use them.
|