Consistency and quality check of survey data in India.

Abstract

Reliability of survey responses on topics such as utilization of health facilities by mother and child has long been a subject of concern. This paper explores consistency of responses from the same individuals over time on utilization of health services involving child delivery and child care. A sub-sample survey was carried out by an independent monitoring agency in 13 states as a part of a larger Coverage Evaluation Survey of all states of India in 2009 by UNICEF, to recheck the responses to improve data quality. Our randomly chosen sub-sample consisted of 510 questionnaires regarding mothers and 497 regarding children. Differences of responses were noted and conveyed to field agencies to rectify recurring errors. Statistical analysis was conducted to find consistency of responses. Matching between the original and rechecked responses varied. Generally, however, the overall match was greater than 90%. Findings suggest that response inconsistencies and the manner in which they are resolved are shown to have important implications for the overall estimate of indicators of utilization of health facilities. The monitoring exercise has, therefore, addressed the quality and consistency issue, which needs further consideration in large-scale surveys. Otherwise, it poses a validity threat to data quality and results on which national policy is framed.