We study the relationship between individuals' participation in household panels, their health and employment states and the design of survey fieldwork procedures, using a comparative approach based on data from the UK BHPS and Australian HILDA Survey. We simulate the impact of alternative procedures for controlling fieldwork effort on the bias and precision of empirical findings, ranging from simple summary statistics to inferences from complex dynamic modelling. We allow for adaptive fieldwork designs linked to household size and the possible long-term consequences for sample integrity.