Why we do it
Survey professionals tend to have clear understandings of what a survey household is. This "statistical household" often poorly represents residential or consumption units important in people's lives. Household survey analysts rarely consider the implications of household definitions for their analysis, perpetuating an uncritical approach to household survey data; harmonisation of major international datasets means that such uncritical use of household survey data is increasing. Many users of published results of household surveys are unaware of the limitations and implications of "households" generated by survey definitions, tending to use reports "off the shelf", accepting concepts as unproblematic. A more flexible and reflexive approach to households in surveys is needed. Aiming to compare "like" with "like" implies that all populations are structured in similar ways: but they are not. The assumption that the majority of individuals live in / contribute to / obtain resources from just one domestic unit is often false, especially in contexts of rapid social change or where wider kin support remains important. Analytical implications of different conceptualisations of the household generate substantial variation in standard indicators such as household size and characteristics of household head.