The Impact of Non-Coverage in Web Surveys in a Country with High Internet Penetration: Is It (Still) Useful to Provide Equipment to Non-Internet Households in the Netherlands?

Abstract

The current study focuses on the effect of the exclusion of non-Internet users in Web surveys. We raised the question whether it is worthwhile to invest in equipment to enable people without prior access to the Internet to participate in online surveys. To do so, we used data from the LISS panel in the Netherlands, which provides 'offliners' with equipment, called a SimPC. We investigate the differences between offliners (the SimPC group) and onliners (people with prior Internet access) in socio-demographics and answer scores for several questionnaires varying in theme. In addition, we look at associations between variables to investigate whether they are influenced by inclusion of the offliners. The results show that SimPC users differ significantly from the regular Internet users on most socio-demographic variables and on a variety of outcome variables in the studies in the panel. However, since the offline group is relatively small due to the high Internet penetration rate in the Netherlands, the inclusion of offliners does not cause any major differences compared to the Internet group alone. This suggests that it is not worthwhile to provide offline households with Internet access, unless one is particularly interested in doing subset analyses on the offline group. The size of the offline group in the panel is crucial. Not including offliners is ethically and scientifically problematic, however.