ABSTRACT
The average exclusion probability is a measure of efficiency in paternity testing; it refers to the a priori ability of a battery of tests to detect paternity inconsistencies. This parameter measures the capacity of the system to detect a false accusation of paternity. Traditionally, this average exclusion probability has been estimated as the probability of excluding a man who is not the father by an inconsistency in at least one of the studied loci. We suggest that this criterion should be corrected, as currently the presumed father is excluded when at least three genetic inconsistencies are found with the child being tested, not just one. This change of criterion has occurred because of the use of microsatellite loci, whose mutation rates are much greater than those of the coding genes used previously in paternity studies. We propose the use of the average probability of exclusion for at least three loci (not only one), as an honest measure of the combined probability of exclusion of several loci, and we propose an algebraic expression to calculate it.
