I disagree with his attitude about prior probabilities, but am beginning to appreciate his system of verbal predicates, even though they could be said to rest on an inappropriate assumption of prior probability. Hummel computes a paternity index L, then converts to the probability of paternity (Vaterschaftswahrscheinlichkeit), W, with the blind formula that assumes "equal priors." Corresponding to the probability range, he then recommends a set of "verbal predicates" for the expert to employ in explaining the significance.
Of course, we can as well associate the verbal predicate with a range of paternity indices, thus somewhat sidestepping the issue of whether or not we approve of the paternity probability calculation.
Hummel's chart | Equivalent in terms of Likelihood Ratio | ||
---|---|---|---|
Vaterschafts- wahrscheinlichkeit | verbales Prädikat | verbal predicate | likelihood ratio |
(99,9) | Vaterschaft praktisch erwiesen | Paternity practically proven | |
99,8 | 399 | ||
Vaterschaft höschst wahrscheinlich | highly likely | ||
99 | 99 | ||
Vaterschaft sehr wahrscheinlich | very likely | ||
95 | 19 | ||
Vaterschaft wahrscheinlich | likely | ||
90 | 9 | ||
(50) | (Ohne Prädikat) | (no verbal predicate) | |
10 | 1/9 | ||
Vaterschaft unwahrscheinlich | unlikely | ||
5 | 1/19 | ||
Vaterschaft sehr unwahrscheinlich | very unlikely | ||
1 | 1/99 | ||
Vaterschaft höchst unwahrscheinlich | highly unlikely | ||
0,2 | 1/399 | ||
(0,1) | Vaterschaft praktisch ausgeschlossen | practically excluded |
Thanks to Dr. Peter M. Schneider, Institute for Legal Medicine, Mainz (but now in Koeln), Germany, for passing along Hummel's original table. See Hummel, Biostatistical opinion of parentage, Table Part I.