The Concept of the Household: From Survey Design to Policy Planning Ernestina Coast, Tiziana Leone (LSE) Sara Randall (UCL) Funded by ESRC survey methods initiative

Do household definitions matter? More variables being added in ‘household section’ Way of measuring wealth / poverty / access to facilities which influence health New level of analysis / explanation More use (researchers & policy makers) made of publicly available data Recognition of importance of society’s basic unit as influence upon members’ well-being Increasing use of ‘indicators’ based on household data (e.g. MDGs, asset indicators) Increasing importance of poverty mapping which uses household level data

The Issue Why does the definition matter? What are consequences of household definition? –Data commissioners –Data collectors –Data analysts –Data users Policy makers Planning / implementing targeted interventions What are the implications for “household” members?

Data designers & collectors have: clear ideas about why need something called ‘household’ clear aims clear understanding of household definition BUT what about analysts / users / consumers far removed from collection? MIGHT: look at definition and assume this is the unit of production, consumption, socialisation central to the development process MIGHT: not even look at definition because they assume they know what a household is

We are not….. Redefining the definition of the household

Methods 1.Document review (1950-present) Sub-Saharan Africa 1.Review census reports, enumerators manuals, questionnaires > Review major household surveys since Key informant interviews (International) interviews in Tanzania, UK, USA 3.Ground truthing fieldwork (Tanzania case study) 1.Maasai area Northern Tanzania 2.Dar es Salaam 2 low income neighbourhoods 3.Rural areas: Planned for next year 4.Modelling differences, to include: 1.Female headed households 2.Household dependency ratios 3.Household size

Census Data Collection: issues in household definition AIM : complete enumeration of population along with individual level characteristics for planning purposes Recurrent themes in definitions of African households Eating together Common housekeeping Living together Answerable to head

Census Data Collection: issues in household definition AIM : complete enumeration of population along with individual level characteristics for planning purposes DIFFICULTIES EVOKED Servants – are they part of household or separate? Boarders / lodgers Absent household head Polygamy Complicated patterns of male female residence (Ghana) Children in boarding school Seasonal migration

Census Data Collection: issues in household definition Summary: household definition is practical solution to census aims of total enumeration recognition (usually) that is a reduced social unit recognition that compromises are made set of rules for enumerators to follow continuity over time – comparability Creation of what van de Walle (2006) calls ‘a statistical household’

Sample surveys: issues in household definition (eg: WFS, DHS, WHS) Household definition practical: to enable the identification of individuals for individual questionnaires “The household is a device used to get at the individual. The household is the sampling unit while the individual is the observational unit.” World Health Survey 2002 ‘ main purpose of household questionnaire was to identify women who were eligible for the individual interview’ Zambia DHS 1992, 1996

Sample surveys: issues in household definition (eg: WFS, DHS) Much more standardised (still some local variations) WFS left more space for interpretation Little variation between core questionnaires and those used by countries Little development over time Emphasis on comparability across time and space

Interviews with data producers, collectors, users, analysts Clear distinction between the ‘Operational’ household of the data collectors and the unit of analysis of the users Data collectors have very clear idea of household definition –Loss of information rather minimal –Not a major issue for comparative purposes –Contrasting preferences when it comes to decide the focus of the definition (eg social, eating, economic) Users not aware of definition issues –The main concern is to have a survey at all –Need for updated information is the strongest drive

Emerging themes from interviews Single person households Urban affluent Household headship? Migrants and mobility Low-income rental neighbourhoods Occupations Mining Agribusiness Construction

Defining a household=‘headache’? ‘An example: A woman, whose husband lives and works in Kumasi, lives with her son, his wife and two children in one house that has two rooms. Both have their own farms, own income and make independent decisions but share the same kitchen and most days it is the mother who is cooking for the rest of the family, often using her own food crops and the son contributes meat. These two are classified as independent households and contributions to meals are recorded as respective gifts. Difficult enough but now the woman's daughter has moved in with her husband and two children for two months. No clue where they managed to squeeze in but for now most food that the mother and her son's family consume comes from the son-in-law's farm with various contributions from the two households. As the daughter's family is joining the meals, not all the farm production can be recorded as gifts…..’ Bjorn Schulte-Herbruggen

In depth interviewing on the concept of household Mix of cognitive interviewing and in depth Household grid sheet-flexible data collection –Longido in prevalently Maasai area (9 ‘households’) –Urban Dar es Saalam (23 ‘households’)

An example from 2007 fieldwork in Tanzania = StevenVictoria

An example from 2007 fieldwork in Tanzania = StevenVictoria Ernest Judy (13)Joy Mary Anna

An example from 2007 fieldwork in Tanzania = StevenVictoria Ernest Judy (13)Joy Mary Anna 1 Male headed household 6 adults and 9 children Dependency ratio =1.5 Maria (13) Steven’s household

An example from 2007 fieldwork in Tanzania = StevenVictoria Ernest Judy (13)Joy Mary Anna 1 Male headed household 6 adults and 9 children Dependency ratio =1.5 Maria (13) Steven’s household

An example from 2007 fieldwork in Longido = StevenVictoria Ernest Judy (13)Joy Mary Anna Maria (13) Maasai 3 households: 1 male & 2 female headed 3 adults + 6 children (DR= 2) 1 woman+2 children (DR=2) Sleeping last night (Census)

An example from 2007 fieldwork in Longido = StevenVictoria Ernest Judy (13)Joy Mary Anna Maria (13) Maasai 1 household: male headed 5 adults + 10 children (DR= 2) Eating last night (DHS)

Modelling definition differences ‘Translated’ the household grid interviews into SPSS dataset We allowed for extra columns to include variables such as: –Would this person make it into DHS –Would this person make it into Census Created simple demographic indicators such as –Dependency ratio –% female headed household –Household size

Scenarios: from low income Dar es Salaam Whole household Slept there last night Ate there last night Mean household size Dependency ratio Sex ratio Female HH30%32%35%

Summary of fieldwork experience Dar urban: very high mobility between households of children and young people –Children: Instability / death of parents, get resources from somewhere else Often get resources from several different households where they may live for short periods –Young men: work mobility, sharing costs, poverty or sharing responsibility among relatives while they find a job Share some costs but not others Kin solidarity when someone has no money

Summary of fieldwork experience Maasai have interdependent groups that are split up in surveys but considered by themselves to be one economic unit of production and consumption The Tanzanian definition of household Reduces the average household size Increases the proportion of female headed hhs Distorts the characteristics of household heads Disassociates people from resources to which they have access

Discussion and few thoughts Data collectors clear about what a household is, users less so Definition does matter for analysis and policy intervention –Not just a matter of leaving ‘unusual’ groups out Household members relationships need to be spelt out. No need to change definitions but possibly more flexible data collection? –Need to think about data collection, back to the 70s? How can we add warnings about household data for users? ‘The household is central to the development process. Not only is the household a production unit but it is also a consumption, social and demographic unit’ Kenya: Ministry of Planning and National Development 2003, p59