Biology Reference
In-Depth Information
Match probability statement
DNA analysis of the bloodstain from the crime scene gave a full DNA profile
that matched that of the suspect. If this blood did not come from the suspect
then the STR profile must match by chance. It is estimated that the chance
of obtaining these matching profiles if the blood came from a random person
unrelated to the suspect is in the order of 1 in 1 billion (a billion is a thousand
million).
The frequentist approach has the advantage that when dealing with small numbers
the phraseology can be understood. It should be noted that part of the duty of the
forensic scientist is to make the strength of the evidence understood by the jury or a
judge and, therefore, to that extent the frequency approach succeeds. Thus, quoting
a random match probability figure of 1 in 1 million, for instance, is relatively simple
for the jury to understand and picture. There are however a number of disadvantages
with the approach. Take the example above, where the random match probability is
1 in 4 trillion. This leads to the observation that there are not 4 trillion people in the
UK so what population is being described? Indeed there are only around 6.5 billion
people on the planet at present.
Source attribution
One approach to the extremely strong evidence is once it reaches a defined point the
prosecution can state that the bloodstain originated from a particular individual, that
is they can attribute, with reasonable scientific certainty, the source of the forensic
evidence to a given individual. Budowle
et al
. [6] provided some estimates for the
profile frequencies that would be necessary - so, for example, a profile frequency of
3.9
×
10
−
11
or less would be unique in the 260 million individuals of the USA (with
99% confidence) [6]. In England, with a population of 50 million the target profile
frequency for 99% confidence would be 2.0
10
−
10
or less [6]; all SGM
Plus
×
profiles shown in Figure 8.1 would meet this target.
However, current methods for determining source attribution have been criticized
for not being sufficiently robust [7] and despite the advantages that source attribution
may have, namely that it removes the need for a court to understand the meaning of
extremely high random match probabilities, it has not gained widespread use.
Likelihood ratios
A likelihood ratio is the ratio of two competing hypotheses. In terms of a criminal
case, it is the ratio of the prosecution hypothesis (
H
p
) and the defence hypothesis
(
H
d
). The likelihood approach is a more logical way to interpret and present the
profile frequency information as it considers an alternative scenario.
