In biometric verification, a goat is a system end-user who is refused access to the system because their biometric data pattern is outside the range recognized by the system. The term comes from a research paper on speech recognition published in 1998 by George L. Doddington. The paper, "Sheep, Goats, Lambs and Wolves - An Analysis of Individual Differences in Speaker Recognition Performance" used a menagerie analogy to explain the differences in speech recognition. Sheep were speakers whose voice patterns were easily accepted by the system, goats were speakers who were exceptionally unsuccessful at being accepted, lambs were speakers who were exceptionally vulnerable to impersonation, and wolves were speakers who were exceptionally successful at impersonation. Because false rejection rates are often high when testing a biometric verification system, goats are probably better known than the other animals in Doddington's menagerie.