In addition to dating from 1992, as RTFirefly correctly pointed out, by counting only people released from prison this method of measurement systematically fails to count every single person who gets a life sentence without parole and is therefore never released from prison. It also doesn’t count death penalty cases.
Systematically excluding all the longest sentences and then claiming the average length of sentence is equal to the average of the unexcluded ones is absurd. Therefore this claim wasn’t even close to correct in 1992.