Are we done yet? Response fatigue and rural livelihoods

Accurate understanding of peoples’ livelihoods activities is needed to inform effective policy. Existing evidence relies heavily on studies that use designated respondents to provide information about their household members, imposing significant costs on these respondents along with possible distortions in the data. In rural Ghana, we randomize the order that household members are asked about and estimate that response fatigue leads to undercounting of labor activities by 8% on average. Women are twice as impacted as men while youth are four times as impacted as older adults, distorting both within-household and population wide comparisons. These biases result from women and youth being listed systematically later in rosters and stronger effects of fatigue for them, conditional on roster position. The implications of our results extend to other topics of enquiry as well, wherever similar repetitive survey structures are deployed, such as birth records, plot-level inputs, and household consumption and expenditures