Where am I?
Matched Pairs  Guidelines

Merged Pairs. Name suffix 0 indicates merged pairs with probabilities based on your prior estimates of model parameters. The Merged Pairs tables for all imputations include exactly the same candidate pairs with exactly the same comparison outcomes (agree, disagree, or missing)  only the match weights and match probabilities are different.

Imputed Pairs. Name suffix indicates the imputation number. Pairs are drawn from the corresponding table of Merged Pairs. Match probability for each pair is compared to a random number between 0 and 1. Pairs with probability greater than random number are selected (imputed). Consequently, most high probability pairs are selected, but not all. Most low probability pairs are skipped but not all. Each imputed will be different. Selected pairs are grouped into sets by collecting all pairs that share a common record: (A,B) and (A,C), (A,B) and (C,B), etc. This can take many steps in an iterative process and you might receive a warning if grouping is not complete after 100 iterations. Usually, this means that one or more sets have a large number of pairs and you should investigate why.

Linked Pairs. Name suffix indicates the imputation number. For dual matches, Imputed Pairs are reduced to one to one matches in Linked Pairs. First, the best pair (highest probability) is selected from each set. Second, residual pairs that share a common record with the selected pair are removed from each set. Third, the process is repeated with the second best pair, etc.
