Who should be interviewed in surveys of household income?

Monica Fisher, Jeffrey J. Reimer, Edward R. Carr
ifpri discussion paper

This study tests the null hypothesis that it is sufficient to interview only the household head to obtain accurate information on household income. The results show that using a husband's estimate of his wife's income does not produce statistically reliable results for poverty analysis. Estimates of the wife's income separately provided by the husband and wife agree in only 6 percent of the studied households. This indicates that although limiting interviews to one person can reduce the time and expense of household surveys, this appears to be detrimental to accuracy, and may lead to incorrect conclusions on the determinants of poverty.