We illustrate how a national household survey can be applied to analyse a subject that requires data from a small sub-sample. Even for small sub-samples meaningful statistical tests can be conducted. The benefits of applying sub-samples from larger household surveys are that primary data collection is not needed and there is a large comparison group. The illustrative case is livelihood strategies among fishermen in Malawi. We find that boat-owners do not diversify more than other households, while crew-members appear to diversify more as they combine fishery with agriculture. Copyright © 2010 John Wiley & Sons, Ltd.