A Common-Sense Approach to Collecting Household Data
Household income is an obvious determinant of poverty. Traditionally, researchers have used household surveys—in which the the head of the household (typically, the husband) is interviewed—to determine household income. But a recent IFPRI discussion paper suggests that this may not always be the best approach, and that it could lead to inaccurate representations, since husbands are rarely able to estimate their wives’ income correctly.
In Who Should Be Interviewed in Surveys of Household Income? (IFPRI Discussion Paper 949), authors Monica Fisher, Jeffrey J. Reimer, and Edward R. Carr explain that a householder’s estimate of his wife’s income is often unreliable for a number of reasons, including misinformation, incomplete pooling of money, and conflict over the distribution of resources. Based on data from surveys conducted in southern Malawi throughout 2008, the authors noted that husbands and wives who were separately asked to estimate the amount of income generated by the wife arrived at the same estimates in only 6 percent of the studied households. Thus, if the husband is the only member of the household being interviewed, survey responses might not be telling the full story.
“The study highlights how important it is for researchers to have a good understanding of the local context when they design a survey instrument to collect household income data,” explains lead author Monica Fisher, a research fellow in IFPRI’s Development Strategy and Governance Division. “In some settings, it may suffice to interview the male household head only—for example, if the head is the only income earner.”
However, the authors suggest that surveying all adult members of the household about their incomes and expenditures results in a more realistic picture of household finances. “In situations where there is clear evidence of individual incomes and these incomes are not directly observable,” says Fisher, “it is recommended that interviews be conducted with multiple income earners.”
Interviewing only one person per household does save time and reduces the expense of a survey. But while cost efficiency and timeliness are certainly important factors to consider in developmental research, the authors’ findings suggest that spending just a bit more time and money can help ensure the collection of accurate data.