Abstract
There are generally three stages to the development of rules for matching vital events data from two sources covering the same population: (a) establishing a set of “true” matches and nonmatches; (b) determining the best tolerance limits for each single characteristic which might be used in matching; and (c) experimenting to determine the set or sets of characteristics and the weights to be used in classifying a pair of records as matched or nonmatched. Specific examples, based on early matching experiments with data from the dual record system of the Mindanao Center for Population Studies (MCPS), are presented. Successive application of different sets of characteristics (differential valence rule) to the remaining unmatched events produced an acceptable rule for matching in this study.
