A Statistical Reformulation of Demographic Methods to Assess the Quality of Age and Date Reporting, with Application to the Demographic and Health Surveys

Thomas W. Pullum, University of Texas at Austin

Demographers have developed several procedures to identify systematic patterns of misreporting of ages, dates, and intervals in survey and census data. Most of these methods concern digit preference, transfers across specified boundaries, omission, and non-response. Almost all methods to identify and describe such patterns use aggregated data, such as the age and sex distribution or the distribution of time since last birth. This paper adapts, integrates, and extends these procedures within the framework of statistical methods for individual-level data, primarily logit and multinomial logit regression. This approach leads to the calculation of interpretable parameter estimates with standard errors and confidence intervals, corrected for the sample design, and the inclusion of covariates that may be related to misreporting. This research was motivated by the author's assessment of age and date reporting in all surveys conducted by the Demographic and Health Surveys project, and the paper includes illustrations from that assessment.