Comprehensive data quality studies as a component of poverty assessments
Realistic poverty assessments necessitate high-quality household survey data. Such data provide the foundation for designing sound policies to sustainably reduce poverty. Despite of this, welfare measures from household surveys are often plagued by non-sampling errors in the form of non-response and measurement error. Current research, while generating important lessons, is often limited in scope and the majority of studies on determinants of data quality deal with quantifiable interviewer and respondent characteristics. A comprehensive study on data quality of an ongoing long-term household panel survey in Thailand and Vietnam is presented in this paper. Determinants drawn from respondent and interviewer characteristics, the interview and survey environment and interview paradata are found to have a significant effect on the overall quality of income-related data. We suggest that survey managers utilizing computerized questionnaires further develop and optimize validation and plausibility guidelines in order to minimize nonsampling errors. Furthermore, referring to validation data (e.g. from administrative records) during data processing is likely to be a promising approach in improving the identification of such errors.