Panel response and attrition in a disaster-recovery survey: trajectories and determinants following the Great East Japan Earthquake

Abstract Panel surveys of disaster victims are essential for providing disaster management with insights into the status, processes, and causal factors of recovery at the individual level. However, previous research using such survey data has largely overlooked detailed trajectories of survey response and attrition, which affect data quality. This study systematically identified these trajectories and their determinants using panel data from the Great East Japan Earthquake. It specifically examined disaster-related impacts, addressed missing values through multiple imputation, and analyzed three survey processes using a nested logit model (for processes 1 and 2) and a group-based trajectory model (for process 3): (1) the willingness to cooperate in the second wave survey (Wave 2); (2) the actual response in Wave 2; and (3) response and attrition trajectories from Wave 2 onward. The results show that (a) three trajectories—continuous response, sustained attrition, and gradual attrition—were observed in the disaster panel data; (b) resilience-vulnerability and life-recovery attitudes and participation were associated with the first process, while environmental shift was associated with the second process; and (c) in the third process, compared to the continuous response trajectory, sustained attrition was associated with resilience-vulnerability and life-recovery attitudes and participation, while gradual attrition was associated with environmental shift and life-recovery attitudes and participation. These findings suggest that response and attrition trajectories and their determinants are closely related to victims’ life recovery processes.