STAR and ROVER: the Madigan experiments on demographic data collection.

Abstract:  A nation-wide sample household survey conducted in 1964 showed that the Philippine vital registration system underreports both births and deaths by about 30% for the nation as a whole. This is caused by: 1) the difficulty barrio residents have in getting to the registry office downtown, compounded by the fact that they have to pay a fee; 2) doctors and nurses report births and deaths but local midwives often do not; and 3) the registrars are not paid extra for their duties and often send records in late. In 1969 the Research Institute for Mindanao Culture of Xavier University, Cagayan de Oro City, joined with the University of North Carolina and USAID to set up the Mindanao Center for Population Studies. This Center developed a dual record system for a systematic approach to information collection about births, deaths, and migrations. The STAR system records vital statistics as they happen. The ROVER system conducts house-to-house surveys to measure vital events. The 2 systems are cross-checked for a true estimate of vital events. A key to the system is a careful definition of geographic areas; to do this the research teams surveyed the study areas on foot to draw up accurate maps. In an urban area the ROVER system alone gave a birthrate of 28.51 per 1000, the STAR alone 37.36 per 1000, and the match between the 2 gave 43.14 per 1000. In a rural area ROVER alone gave a birth rate of 42.24/1000; STAR, 42.36/1000; together 45.76/1000. A recall study showed that a 6-month recall was all that could be relied upon if accurate descriptions of events were wanted, but that in rural areas (and only in rural areas) 18-month recall could be quite accurate. The type approach used depends upon the area. In rural areas the informant approach is most effective for collecting data, whereas in urban areas the house visit is more appropriate. The workload study showed that a full-time worker using an informant in a small area records the largest number of vital events. The maximum that can be covered is 200-240 households in an urban area and 124-187 in rural areas.