| In the early days of the use of genetic | | | | if using a full profile) It should be |
| fingerprinting as criminal evidence, | | | | noted that figures of this magnitude are |
| juries were often swayed by spurious | | | | not considered to be statistically |
| statistical arguments by defense lawyers | | | | suportable by scientists in the UK, for |
| along these lines: given a match that | | | | unrelated individuals with full matching |
| had a 1 in 5 million probability of | | | | DNA profiles a match probability of 1 in |
| occurring by chance, the lawyer would | | | | a billion (one thousand million) is |
| argue that this meant that in a country | | | | considered statistically supportable |
| of say 60 million people there were 12 | | | | (Since 1998 the DNA profiling system |
| people who would also match the profile. | | | | supported by The National DNA Database |
| This was then translated to a 1 in 12 | | | | in the UK is the SGM+ DNA profiling |
| chance of the suspect being the guilty | | | | system which includes 10 STR regions and |
| one. This argument is not sound unless | | | | a sex indicating test, this test updated |
| the suspect was drawn at random from the | | | | the SGM DNA profiling system on which |
| population of the country. In fact, a | | | | the National DNA Database was founded in |
| jury should consider how likely it is | | | | 1995. The SGM system included 6 out of |
| that an individual matching the genetic | | | | the 10 STR regions used in the SGM+ |
| profile would also have been a suspect | | | | system and the same sex indicating test, |
| in the case for other reasons. Another | | | | however the discriminating power of the |
| spurious statistical argument is based | | | | SGM system was only considered to be |
| on the false assumption that a 1 in 5 | | | | supportable at 1 in a million) . |
| million probability of a match | | | | However, with any DNA technique, the |
| automatically translates into a 1 in 5 | | | | cautious juror should not convict on |
| million probability of innocence and is | | | | genetic fingerprint evidence alone if |
| known as the prosecutor's fallacy. | | | | other factors raise doubt. Contamination |
| When using RFLP, the theoretical risk of | | | | with other evidence (secondary transfer) |
| a coincidental match is 1 in 100 billion | | | | is a key source of incorrect DNA |
| (100,000,000,000). However, the rate of | | | | profiles and raising doubts as to |
| laboratory error is almost certainly | | | | whether a sample has been adulterated is |
| higher than this, and often actual | | | | a favorite defense technique. More |
| laboratory procedures do not reflect the | | | | rarely, Chimerism is one such instance |
| theory under which the coincidence | | | | where the lack of a genetic match may |
| probabilities were computed. For | | | | unfairly exclude a suspect. |
| example, the coincidence probabilities | | | | When evaluating a DNA match, the |
| may be calculated based on the | | | | following questions should be asked: |
| probabilities that markers in two | | | | Could it be an accidental random match? |
| samples have bands in precisely the same | | | | If not, could the DNA sample have been |
| location, but a laboratory worker may | | | | planted? |
| conclude that similar -- but not | | | | If not, did the accused leave the DNA |
| precisely identical -- band patterns | | | | sample at the exact time of the crime? |
| result from identical genetic samples | | | | If yes, does that mean that the accused |
| with some imperfection in the agarose | | | | is guilty of the crime? |
| gel. However, in this case, the | | | | Fake DNA evidence |
| laboratory worker increases the | | | | The value of DNA evidence has to be seen |
| coincidence risk by expanding the | | | | in light of recent cases where criminals |
| criteria for declaring a match. Recent | | | | planted fake DNA samples at crime |
| studies have quoted relatively high | | | | scenes. In one case, a criminal even |
| error rates which may be cause for | | | | planted fake DNA evidence in his own |
| concern [1]. Because of this, arbitrary | | | | body: Dr. John Schneeberger of Canada |
| ceilings were put on match probabilities | | | | raped one of his sedated patients in |
| used in RFLP analysis rather than the | | | | 1992 and left semen on her underwear. |
| theoretically computed ones. Today, RFLP | | | | Police drew Schneeberger's blood and |
| has become widely disused due to these | | | | compared its DNA against the crime scene |
| difficulties in interpretation. | | | | semen DNA on three occasions, never |
| STRs do not suffer from such | | | | showing a match. It turned out that he |
| subjectivity and provide much better | | | | had surgically inserted a Penrose drain |
| powers of discrimination, for unrelated | | | | into his arm and filled it with foreign |
| individuals (of the order of 1 in 10^29 | | | | blood and anticoagulants. |