Are We Done Yet?
Response Fatigue and Rural Livelihoods
with Kate Ambler and Mywish Maredia
Response Fatigue and Rural Livelihoods
with Kate Ambler and Mywish Maredia
|
Most Recent Version:
(February 2021) |
| |||
Abstract:
Effective policy requires an accurate understanding of peoples’ livelihoods activities. The data for this evidence is often generated via lengthy surveys where designated respondents provide information about their household members. This burden on respondents may lead to both losses and biases as they grow fatigued during the interview. We test these hypotheses with an experiment in rural Ghana where we randomize individual household members’ position in the labor module. We find that moving a household member back by one position reduces their reported number of productive activities by 2.2% with average aggregate losses of 8%, or approximately one out of every twelve activities. For women and youth, losses are closer to one in nine. These biases result from both differential exposure to response fatigue (being positioned later in rosters) and differential vulnerability (greater impacts conditional on position). These results have important implications for data quality across many settings and topics.
Effective policy requires an accurate understanding of peoples’ livelihoods activities. The data for this evidence is often generated via lengthy surveys where designated respondents provide information about their household members. This burden on respondents may lead to both losses and biases as they grow fatigued during the interview. We test these hypotheses with an experiment in rural Ghana where we randomize individual household members’ position in the labor module. We find that moving a household member back by one position reduces their reported number of productive activities by 2.2% with average aggregate losses of 8%, or approximately one out of every twelve activities. For women and youth, losses are closer to one in nine. These biases result from both differential exposure to response fatigue (being positioned later in rosters) and differential vulnerability (greater impacts conditional on position). These results have important implications for data quality across many settings and topics.