Making statistical inferences about linkage errors

This long-standing problem is addressed when modeling the linkage errors through the number of links of a record, and the proposed solution bases the inferences on a subset of observations that are approximately independent.