Social surveys are often used to estimate unemployment duration distributions. Survey non-response may then cause a bias. We study this using a unique dataset that combines survey information of individual workers with administrative records of the same workers. The latter provide information on unemployment durations and personal characteristics of all survey respondents and non-respondents. We develop a method to empirically distinguish between two explanations for a bias in results based on only survey data: (1) selectivity due to related unobserved determinants of unemployment durations and non-response, and (2) a causal effect of a job exit on non-response. The latter may occur even in fully homogeneous populations. The methodology exploits variation in the timing of the duration outcome relative to the survey moment. The results show evidence for both explanations. We discuss implications for standard methods to deal with non-response bias.