Household surveys are essential for the production of data for policy design and interventions in developing countries. Little attention is paid by commissioners, producers and consumers of data from household surveys to the issue of what the household unit used in the survey is, how it is defined, and what this definition might means for analysis and interpretation. The household, as defined and used in household surveys, refers to a basic social unit. If a survey-defined concept of a household differs systematically from locally understood and lived basic socio-economic units, then research based on this minimal social unit definition becomes much less useful for subsequent analysis either at household level or for aggregates of households. If standard definitions of the household do not adequately capture local realities of the social unit, then this raises significant issues in terms of survey validity. This project will use qualitative methods to systematically identify the extent of difference between the 'household' units used in household surveys and locally meaningful terms for social units. Differences will be subjected to a series of scenario models for a range of development indicators to provide substantive evidence of the impact of household definition on survey measurement and validity.

Key informant interviews: In-depth interviews (n=39) with 54 key individuals situated at different places on the chain of demographic data collection and analysis. Recorded interviews were transcribed verbatim and coded using N6 to facilitate analysis. All interviews (suitably anonymised) will be made available via www.esds.ac.uk. Permission has been requested from all respondents to quote from their interview. Key themes for coding were developed based on research hypotheses but further codes were developed inductively after reading and re-reading all interviews. All interviews were coded independently by two researchers.