Notice: the WebSM website has not been updated since the beginning of 2018.

Web Survey Bibliography

Title Tracking the Representativeness of an Online Panel Over Time
Year 2016
Access date 01.06.2016
One of the greatest threats to the accuracy of estimates from panel data is the occurrence of attrition of participants over the panel lifetime. Fortunately, panel recruitment interviews often provide rich sources of data that allow identifying the type of background characteristics affected most seriously by attrition. However, in practice the degree of attrition is seldom studied on these rich data sets, possibly because it is difficult to synthesize the degree of attrition across many auxiliary covariates. In this study, we address this shortcoming while studying attrition in the Dutch Liss online panel using the rich set of auxiliary variables available from the panel recruitment interviews. We estimated panel representativeness for each month of seven years of panel lifetime (2008-2015) using so-called representativeness (R-) indicators. The favorable feature of R-indicators is that they allow quantifying the degree of selectivity in a sample at a given point in time based on the information encoded in theauxiliary data. In doing so, we could include nearly forty variables in estimating representativeness,including respondents’ health, economic status, personality, political attitudes, and survey attitudes, besides standard socio-demographics. The time-series of R-indicators that resulted from this project offers a unique view into the evolvement of panel representativeness and attrition. We found that attrition occurred in the Liss panel in two stages. First, there was an onset loss in representatives that occurred already at the first panel wave. Second, representativeness decreased across two and a half years of panel lifetime and subsequently stabilized. In our presentation, we show how the general losscan be split up into the contributions of the variable groups. We also evaluate the impact that a stratified and a random refreshment sample had on panel representativeness.
Year of publication2016
Bibliographic typeConferences, workshops, tutorials, presentations

Web survey bibliography (431)