Voluntary Reporting Sources

One of the assumptions for the 2-sample capture-recapture method is the homogeneity of the population, that is, for a given source, all the members in the population should have the same chance to be listed in the source. However, because of, for example, the socioeconomic status or severity of the disease, members of the population may actually differ considerably in the chance of being ascertained by a particular source. Such heterogeneity or variable catchability within the population may cause problems in the capture-recapture analyses. The source based on voluntary reporting often has such problem. For example, drug abuse treatment centers as a source for identifying intravenous drug uses (ref). This source arises primarily from the voluntary efforts of the drug users to seek health care or medical treatment for their drug problem. Several factors will affect the catchability of this sources, for example the drug users' socioeconomic status, educational level or the severity of the addiction. Usually this type of source will catch more "heavy" users than "light" users. Another example is the Pittsburgh dog-bite study (ref) which used hospital, animal control and police reports to identify the injured. All these three sources relied the victims to report the injury to different facilities. Therefore, it is also likely that more severe dog-bite injuries were identified through these sources. If sources based on voluntary efforts catch different strata of subjects, according to Neugebauer and Wittes (ref), negative dependence may occur and, in turn, this will inflate the estimated number of cases perhaps considerably. It has often been suggested to use stratification when such problem occurs. However, for "hidden" and "elusive" (ref) population, the more determined members will have very low probability of capture and will cause estimable problem in the capture-recapture design. Therefore, voluntary reporting sources should be used very cautiously in the capture-recapture use.