Relevance & Research Question: This presentation examines how including offline households (households without access to the internet) in a probability-based online panel counters the problem of undercoverage which online surveys are prone to. Taking into account coverage error is a central issue when conducting a survey which aims to be representative for the general population.
Methods & Data: We use data from the recruitment and first 8 waves of the German Internet Panel (GIP), a longitudinal internet survey run by the University of Mannheim. The GIP is based on a probability sample of individuals living in private households. The recruitment was done offline by short face-to-face interviews in 2012. All offline households were offered a simple PC and a 3G internet connection.
In a first step we use data from the recruitment interview to visualize the differences between respondents with and without internet. In a second step we compare online panel registration rates between those two groups. To evaluate the impact of the participation of offline households on sample representativeness we compare the panel composition with and without offline households to the German entire population. We use German Census data as benchmark for this comparison. Using longitudinal data of the GIP we analyze differences in participation and attrition rates between online and offline households over time.
Results: Analyses show that offline and online households differ in several aspects from each other. People without internet are older, more often female, more often low and medium educated and life more often in single households. These results illustrate the coverage error which is usually connected to online surveys. Sign up rates to the panel are considerably lower for offline households (15%) compared to online households (50%). However, we can show that adding these people to the panel improves sample representativeness with respect to the variables (age and gender) we compared.
Added Value: Researchers and practitioners will learn about the benefits and quirks of our approach to counter noncoverage of non-internet households in an online survey.