Attrition in the Swiss Household Panel : are vulnerable groups more affected than others ?

Panel attrition is one of the main concerns to longitudinal surveys, and may be especially problematic when vulnerable populations are concerned. This study looks into nonresponse and attrition in the Swiss Household Panel (SHP), focusing on the following two questions: (1) To what extent does attrition create bias in means and frequencies, and does weighting correct for this? (2) Are respondents who are at risk of vulnerability more likely to drop out from the SHP compared to others and if so, why? Our results based on data from 1999 to 2012 indicate that there are different nonresponse patterns in the SHP and that attrition is not completely at random, but is related to specific characteristics that are often associated with vulnerability. In particular respondents with a migration background, a low level of education, who are unemployed or whose health status is poor are more likely to drop out. Using weights only partially corrects for the selective dropout. Although general population surveys such as the SHP provide unique opportunities to study vulnerability in the population, researchers should be aware that overall prevalence of vulnerability is most likely underestimated