To overcome these criticisms, the National Research Council (NRC) proposed two years ago that in calculating the likelihood of a random match, litigants should use the "ceiling principle
": use only the highest rate of incidence found in any group for each segment.
The report concluded that the statistical methods used by forensic laboratories should be abandoned in favor of a more conservative approach called the ceiling principle, which produces less extreme match estimates.
Some scientists have attacked the ceiling principle as unnecessarily conservative,(17) some have said it is unclear and subject to conflicting interpretations,(18) some have defended it as a workable compromise,(19) and some have argued that it may not be conservative enough.(20)
A major theme of critics from all perspectives is that the "ceiling principle" is not a principle of science.
In one case, an FBI agent-examiner purported to follow the modified ceiling principle while computing the frequency of the defendant's DNA profile six different ways.(25) The results ranged from 1 in 1.26 million to 1 in 877.
Cohen, Violations of the Ceiling Principle: Exact Conditions and Statistical Evidence, 53 AM.