Exploring the limits of household surveys in Africa

In this post, LSE’s Ernestina Coast and UCL’s Sara Randall outline the importance of accuracy of data taken in international surveys to ensure poverty-related data are high quality.

Poverty statistics often depend on household-level measurements from survey data, making the definition of household of critical importance. Many policy-makers, government agencies and researchers see poverty as fundamentally a household level problem. Thus, data for poverty analyses need to be collected on households.


We know that many African households are flexible, permeable, evolving, and differentiated along lines of gender and generation. Much of life in Africa remains inherently unpredictable and valuable counteractions to poverty depend on strategies focusing on building networks of obligations and support; these are difficult to capture using household surveys. The more anthropological and qualitative the research, the more likely it is that the researchers explain what they understand by household and use locally grounded definitions of household. Such research however rarely feeds into national indicators and quantitative analyses using nationally representative survey data rarely explicitly consider the nature of the “households” they are studying, especially when doing secondary analyses of survey data sets.

Our research, conducted in Tanzania, Burkina Faso, Uganda and South Africa, reflects on what sorts of poverty-related issues may be missed or inaccurately represented because of the ways “household” and household membership is defined in surveys. A key constraint is that most surveys require individuals to be enumerated as members of one household and one household only, whereas the reality in much of urban and rural Africa is that individuals have affiliations, links, relationships and draw or contribute resources with two or more households. We develop an analytic framework of “open” and “closed” households. “Open” households cope with poverty using flexibility, movement and extra-household networks, but both household membership and affiliations and their strategies to counter or insure against poverty are poorly represented by survey data. Closed households where household members do not have flexible links with other households, and exchange few people, resources or support are likely to be better described by survey data. Closed households often encompass the very poor because they are not part of larger networks of exchanges, movement and support. Closed households though are not necessarily poor; they may also include some of the emerging middle classes, although these were not the subject of this research.

Survey definitions of a household often refer to those who eat together assuming that this indicates an economic unit of consumption. However, eating together is often more a consequence of practicality or cooking traditions rather than an indication of a basic economic unit. Furthermore it can be practically difficult to identify who does eat together and urban poverty often entails eating street food rather than cooking.

The idea of “open households” should not be seen as a new definition and a burdensome tool to add to the complexities of data collection in African surveys. The challenge is to develop simple additional questions which could be added to surveys which would allow openness (or some dimensions of it) to be identified and explored in flexible ways, while retaining comparability. Our data collection experience suggests this should not be too challenging. Openness does not constitute “anyone” who contributes or benefits but is about individuals who are generally perceived to have some claim to membership.

There are two ways forward from our research. It is probably unrealistic to expect large international surveys to change their definitions and ways of recording households. However we can expect analysts of such data to articulate much more clearly in their analyses and publications the limitations of the definitions used in data collection and to reflect on the implications of these limitations – and in particular the validity of their analyses. More triangulation with work produced by other disciplines on study populations would be a step forward as would a frank reflection on the biases generated by specific approaches and data. Surveys that are not constrained by being part of international comparative series could experiment with different approaches to recording household membership, allowing multiple and diverse types of membership and wider definitions of household. This would allow a differentiation of closed and open households and an exploration of the implications of different degrees of openness via the economic measures being used. Quantitative data that truly reflect diverse strategies for confronting and managing poverty are probably unachievable – and we may have to accept that some things are real and important but ultimately unmeasurable.

Visit the Household survey project website.

This post is based on research papers, Poverty in African households: the limits of survey representations by Sara Randall and Ernestina Coast and From design to practice: how can large-scale household surveys better represent the complexities of the social units under investigation by Antoinette Kriel, Sara Randall, Ernestina Coast and Bernadene de Clerq.

This post was originally posted on the Africa at LSE blog here.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: