From design to practice: how can large-scale household surveys better represent the complexities of the social units under investigation?

‘The household’ is most commonly used as unit of analysis in household surveys and as enumeration unit during census data collection. Relationships are mostly indicated with regards to ‘household head’ or ‘acting household head’. The way in which ‘the household’ is defined in these surveys has long been criticised by anthropologists and sociologists as unable to adequately capture the complexities of the social units within which people arrange themselves. However the problem often doesn’t simply lie with the definition but the assumptions made when designing a questionnaire and taking the survey from the design phase into the implementation phase. The aim of this paper is to reflect on lessons learnt from a household financial well-being survey with regards to defining ‘the household’ and operationalising the definition in a heterogeneous, complex society such as South Africa. In a follow-up study group interviews were conducted with fieldworkers who administered the survey, and willing respondents were re-interviewed, collecting qualitative information about those with familial and/or financial links. It was found that: 1. There was already confusion at the survey design stage as to exactly what the unit of analysis is. 2. This confusion translated into lack of clarity in terms of operationalising the working definition given to fieldworkers. 3. Fieldworkers therefore reverted back to what they know, namely ‘family’ or simply ‘who is there’. 4. The design and layout of the questionnaire did not easily allow for the capturing of more complex units and understanding the relationships between the individual members. 5. This affected not only the accuracy and reliability of the demographic data collected but may also have an impact on the financial well-being data collected. Considering the type of definition to be used for the household as unit of analysis is simply the starting point. Much more consideration needs to be given by survey designers to the implementation of concepts in the field instead of trying to find ‘the perfect’ definition