I Don’t Care Who You Are: Adult Respondent Selection Does Not Alter Child Deprivation Estimates

Abstract

The last decade has seen strong developments in child poverty measurement. Children are largely recognised to have age-specific needs, which has led to the development of child-specific poverty measures. One of the current ongoing debates is how best to collect that information. Questions regarding child deprivation are most often answered by a reference adult and assume that adult respondents provide accurate reports about their children’s needs. These assumptions have largely gone untested. Making use of a unique feature of the Fijian Household Income and Expenditure Survey (2019/20), this paper explores the effect of respondent selection on child deprivation estimates. We compare three approaches to the selection of adult respondents commonly used in survey design: the children’s mother, the household head, and a random adult respondent in the household. We find that adults largely agree on whether children are deprived of specific items. However, in between 5 to 24% of households (depending on the indicator), adults provide different responses regarding whether children have specific items. Despite these differences, respondent selection does not substantially alter child multiple deprivation estimates or the socio-demographic characteristics of the deprived child population. The article will be of interest to those designing surveys or questionnaires to measure child poverty and children’s unmet needs.