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

Web Survey Bibliography

Title The Use of Paradata to Predict Future Cooperation in a Panel Study
Year 2014
Access date 29.03.2014

Relevance & Research Question: In panel studies, low attrition is especially important, because only complete data sets across waves can be analyzed. Survey paradata (e.g., response times, item nonresponse, or break-off) can be collected at moderate cost. In this explorative study, we analyze if paradata lend themselves to predict non-participation in future waves.
Methods & Data: Members of a commercial online panel were invited to participate in an academic study on the stability of personal preferences and traits. The questionnaire consisted of questions regarding biometrics (e.g., size, eye color, handedness), preference for certain pictures, preference for certain foods, and items to measure personality traits. Invitations to identical questionnaires were sent in December 2011, June 2012, December 2012, and June 2013.
Results: Overall, 807 respondents participated in Wave 1 and 249 respondents in all four waves. The completion rate continuously rose from 91.8% (Wave 1) to 98.4% (Wave 4). Break-off in one wave proved to be a good indicator for refusal to participate in the next wave: The chances that a participant completed wave n were more than 3.5 times higher for those who completed wave n-1 than for those who did not complete wave n-1. Reminder emails were most effective in the first wave. Neither item nonresponse nor response times were good indicators to predict future participation.
Added Value: In longitudinal studies there is a high risk of loosing respondents. Break-off in one wave proved to be a predictor for break-off in subsequent waves. We discuss how paradata can be taken advantage of to lower panel attrition.

Access/Direct link

GOR Homepage (abstract) / (presentation)

Year of publication2014
Bibliographic typeConferences, workshops, tutorials, presentations

Web survey bibliography - General Online Research Conference (GOR) 2014 (34)