How do different strategies of including the offline population affect the representativeness of probability-based online panels over time? Mixing modes versus providing Internet equipment
Jul 26, 13:45
Over the past decade there has been a vast increase in the number of online panels. While they offer a number of advantages, such as quick and flexible data collection, online panels have often been criticized with regard to the accuracy with which they represent the intended target population. As a response to this criticism, probability-based online panels have developed strategies to cover the offline population as well. There are two dominant approaches on how to include the offline population in a probability-based online panel. One approach is to apply a mixed-mode panel design where the offline population is surveyed using an offline mode of data collection, while the online population is interviewed online. The other approach to including the offline population in a probability-based online panel is to provide offline households with the necessary equipment to participate online. In our study, we investigate the consequences that these two approaches have on the representativeness of the gathered survey data over time. We use data from two probability-based online panels in Germany: the GESIS Panel, which applies a mixed-mode panel design where the offline population receives postal mail surveys, and the German Internet Panel (GIP), where offline households are provided with an Internet connection and Internet-enabled devices to participate in the online panel waves. The aim of the paper is twofold: 1) we provide insights on how the representativeness of the GESIS Panel and the GIP develop over the course of the panel waves, and 2) we demonstrate how the representativeness of these probability-based online panels is affected by the way in which the offline population is included.