Do Swazis have households? Why the unit of analysis matters

Abstract

This paper starts from a very practical, empirical research issue, a very elementary issue: if one is wanting to collect basic data about the distribution of incomes, family sizes, unemployment, cattle, infant mortalities, malnutrition, wealth, illness, poverty, crop surpluses, self-sufficiency, educational attainment (or anything else) in contemporary Swaziland, what is the appropriate unit to sample? In order to answer this question sensibly we need to know something about the way a society distributes its goods and its ills. We know, for instance, that in Europe consumption is organized in nuclear family households. We assume that a woman and her husband and minor children share equitably if not equally in their pooled income. So household is the appropriate unit of analysis. The validity of this assumption has only recently been questioned by people working within a feminist perspective who have pointed to the possibility of inequities of distribution within households, linked to the earning power of family members, and how the increasing proportion of women in paid work is transforming household structure.