Risk in database searching

The McKie case has some similarities to searching large fingerprint databases for suspects. In both, a single verified error in a huge number of fingerprint comparisons will lead to a false accusation. In a database search one latent print is compared with millions of records in the database and a single error will cause the wrong person to be accused of the crime. If there is a single error in the very large number of elimination comparisons connected with all crime scenes, a McKie-type case will be initiated if it is subsequently handled in the same way.

One major difference is that database searching creates a very large number of comparisons, but the number of identifications in any inquiry is limited by the number of identifyable crime scene latent prints. A McKie-type case can arise out of an unlimited population of elimination identifications, selected because of a denial. A McKie type case will be open to the risks from an unlimited number of opportunities for clerical error (e.g. mislabelling). Also, experts have different contextual information when searching databases and checking for elimination The fingerprints of people on elimination lists are expected to be found in crime scenes¹.

The above is a discussion about errors. A separate consideration is the prior probability of the identified person being guilty of the crime (before the effect of the fingerprint evidence is considered). According to Bayes theorem, the posterior probability of guilt (after the effect of the fingerprint evidence is considered) is dependent on both the error rate and the prior probability.

“g” from the CLPEX chat board offered the idea that when a database is searched, the prior probability of guilt could be a Normally distributed random variable with the assumption that those with closest proximity to the crime have more likelihood of committing it. So if a database search hits on someone who lives far away from the crime and appears to have no connection with it, then without the fingerprint evidence it would be very unlikely that this person deposited a print in the crime scene (a very low prior probability). It could be argued that the Mayfield misidentification is an example of this.

By this analysis and my own, the two most discussed and notorious misidentifications of all time, Mayfield and McKie, could have been predicted as being especially risky because both could have arisen from a large population of comparisons - a large area of opportunity for error - and they both have abnormally low prior probabilities, Mayfield because of the very weak and distant connection to the crime, and McKie because there is nothing other than the fingerprint to suggest that the proposed act of wrongdoing happened (at all, by anybody).

If a database search (and the verification) is done without knowing details of the people, and a match occurs which turns out to be someone who has a strong connection with the crime then it would be very unlikely that this could have happened by chance due to error. So perhaps a database search that finds someone who has a connection with the crime is very safe, despite the large number of records being searched and the fact that many of them will be very similar to latent mark.

1 Scottish Parliament Justice 1 Committee Report paragraph 224. “The Committee also notes the findings of Evett and Williams that ‘A fingerprint expert will generally reach an inner conviction about the correctness of an identification long before he has found 16 points.’ (A review of the 16 point standard in England and Wales, I Evett and R Williams, October 1995)

Back to previous document