Accounting for Clustering for Self-Reported Outcomes in The Design and Analysis of Population-Based Surveys: A Case Study of Estimation of Prevalence of Epilepsy in Nairobi, Kenya

Abstract

This methodological study explored approaches for appropriately accounting for clustering in self-reported outcomes within population-based surveys, using epilepsy prevalence estimation in Nairobi as a case example. Several analytical strategies were compared to evaluate bias, precision, and robustness in the presence of intra-cluster correlation. Results demonstrate that failure to adjust for clustering leads to underestimated standard errors and potentially misleading prevalence estimates. The study provides practical guidance for researchers designing surveys with clustered self-report data.