Transcript PositiveTest-DNAevidence

m
a
D t
N c g
A h
i
n
A sample of DNA
from a crime
scene ‘matches’
a sample from
the suspect.
How strong is this
evidence?
The data:
• An evidence sample from the crime
scene is compared with a reference
sample taken from the suspect.
• DNA profiles are unique to each
individual person (other than
identical twins, triplets, …)
• The whole DNA profile is not
matched, but only around 13 points.
• But matching 13 particular points of
comparison, say, means that the
probability of a match by chance is
very small – less than the probability
that you choose all 6 lottery numbers
this week!
A sample of DNA
from a crime
scene ‘matches’
a sample from
the suspect.
How strong is this
evidence?
• In fact, the probability that two
samples of DNA match by chance
may be as small as 1 in a billion.
• So surely, this must be absolutely
certain evidence that the person
is guilty?
• Well, with other evidence, it may
be, but there are several
arguments which should be
considered before a decision is
made.
If the chance of
a match is 1 in a
million, that
means the
probability that
someone else is
guilty is 1 in a
million.
The correct argument:
• The chance is 1 in a million that
some other particular person left
the evidence at the crime scene.
• This is NOT the same as the
chance that someone (anyone)
else left that evidence.
• If the population size with access
to the crime scene is 1 million,
then a chance match is perfectly
possible.
Since there
thousands of
people who might
be a match for
this DNA, the
chance that this
particular person
is guilty is only 1
in several
thousands.
This would only be correct if all
these other people had the
same access to the crime
scene.
Dependence and
independence
• The calculation of the probability
1 in a billion uses the product rule
for independent events.
• It does not apply when events are
not independent.
• So if the local population is mixed
or inbred, so that the distribution
of DNA is not typical of the
population generally, then the
strength of the DNA evidence
may be over-stated.
Close relatives
have very similar
DNA.
• If the suspect is not the
guilty party, then the
similarity between the DNA
samples needs to be
explained.
• A relative may be the guilty
person.
• There is always a small
probability of a false positive
test.
There is also a small
• This does not mean that two
chance of error in
people might have the same
the tests because of
the way that the
DNA, but it does mean that
evidence was
there can be error in the test
collected, or how the
process.
test was conducted,
or how the test
results were
reported.