This study examines the consistency of reporting of terminated pregnancies in DHS reproductive calendars. We examine two measures of consistency: within-survey consistency and between-survey consistency. We find that 70-80% of surveys are inconsistent by either or both of these measures. Reporting of terminated pregnancies deteriorates in earlier periods of the calendar. Africa has a higher proportion of inconsistent surveys (78%), whereas in the other regions about two-thirds of surveys are classified as inconsistent. Consistency appears to vary with type of history (pregnancy or birth), direction of the history, and the inclusion of questions that distinguish between miscarriage and induced abortion. Survey length and mode of data collection appear unrelated to either consistency or inconsistency of reporting terminated pregnancies. Our findings suggest the need for more experimentation with and rigorous assessment of modifications to the calendar and modes of data collection.
Presented in Session 4. Marriage, Family, Households, & Unions