(OR01-05-23) How Participation in Assessments During Data Collection for an Agricultural Livelihood and Food Security Study in Kenya Affected Caregivers and Young Children

Objectives: Evaluations commonly assign participants to an intervention or control arm and assess them before and after intervention implementation. Assessments can influence participants’ later responses to assessments because of reactivity or influence on participants’ behavior. This study aimed to understand how participation in assessments during data collection influenced caregivers in the control arm of an agricultural livelihood study, their actions and appraisal of their child’s development, and the mechanisms through which influence occurred.

Methods: The Shamba Maisha study was created to address drought, food insecurity, and HIV in Western Kenya. Intervention participants received an irrigation pump, farming lessons, and a microloan. Control participants could receive the intervention after completing the two-year study, receiving no other formal benefits. Caregivers with a child under 5 years old were assessed every six months over two years using questionnaires and observations regarding foods eaten, home environment, and child growth, psychological development, and illness. After assessments at the two-year follow-up visit, 100 intervention and 100 control caregivers were interviewed on changes in their child’s play, eating habits, interactions with others, and caregiver mental health. The data were coded in Dedoose, and a thematic analysis was conducted.

Results: Despite receiving no formal benefits, control caregivers experienced enhanced mental health and knowledge of their child’s health and reported changes in the child’s play and diet that they attributed to participation in study assessments. Furthermore, caregivers attributed their changed behavior to repetitive questioning, instrumental support, emotional support, and increased health knowledge.

Conclusions: In the control arm, repeated assessments altered participant behavior, which may have made inference of intervention effects more difficult. In designing future studies with intervention and control arms, a trade-off between the gains in statistical efficiency provided by repeated visits and avoiding alterations in participants’ behavior that affect responses to assessments must be considered when deciding on the number of visits for assessment.

Funding Sources: National Institutes of Health (1R01MH107330-01, 1R01MH109506-01)