Matching experimentation in a dual record system for births and deaths in Morocco

Summary In an effort to improve the quality and completeness of birth and death reporting in Morocco, a sample of 84,000 persons was established, for whom birth and death data were collected by two methods. Each household was contacted once a month by interviewers in a registration method; independently, each household was interviewed once every six months by a survey team. Interviewing continued from April 1972 to July 1973. Results of these two methods were then compared through matching the vital events recorded by the two sources. First a 'true match status' was established by teams of 'experts', supplemented by field verification of uncertain cases. Of the 15 variables on the vital event reporting forms, eight were studied to establish the optimal tolerance limits for deciding whether any two completed recording forms (one from each method) were recording a single birth (or death) or two different ones. Next, those characteristics were selected which introduced the fewest matching errors; for births, this was the dwelling unit number, the name of the mother, and the name of the baby; for deaths, the dwelling unit number, the name of the person, and the name of the head of household. Aside from its manifest function, the matching operation also improved the fieldwork and helped to clean the data.