DNA database reports shows costs up, 'detections' down
Worst tool we've got apart from all the others...
Analysis Researchers looking to assess the effectiveness of DNA profiling in solving crime are unlikely to take much comfort in the recently released Annual Report of the National Policing Improvements Agency (NPIA).
Meanwhile, civil liberties campaigners must wait to discover what the Government plans to do about removing the profiles of innocent people from the database.
The Annual Report, which covers two years' activity on the DNA Database (NDNAD) is part propaganda, part statistical report. Early sections reference one case where a chance profile for a minor offence (a pub brawl) led to arrest and conviction for a major one (murder) as in the case of Mark Dixie. Similar cases are documented, as well as cases where individuals have been cleared of crimes as a result of DNA profiling.
However, when it comes to a fuller explanation and analysis of the effects of DNA, the report can only document the current picture, as issues such as outcomes (criminal convictions as a result of DNA profiling) are beyond the NPIA's remit.
The NDNAD holds DNA profiles, not individual details. Over the last two years, the number of individual profiles held on the NDNAD has risen from around 4.4 million to 5.6 million in March 2009.
Fuzziness in the estimates derives from the fact that figures given for the numbers loaded on the database do not quite tally with the number of profiles added year on year. The actual number of individuals for whom profiles are now held is estimated at around 4.85 million, with around 13 per cent of the database consisting of "replicates".
This report does little to answer the vexed question of effectiveness. NPIA boss, Peter Neyroud describes DNA profiling as "the most effective tool for the prevention and detection of crime since the development of fingerprint analysis".
However, during the period in question, the total number of crime scene matches dropped from 41,717 to 36,727. Given the increased number of profiles held, this is not inconsistent with some criminals changing their behaviour to avoid leaving samples.
Total detections involving DNA samples also fell from 41,148 to 31,915. The official reason given for this fall is that over the last five years there has been a major fall in recorded crime. A spokesman said: "Burglary offences are down 29 per cent over a five-year period and offences against vehicles are down by 40 percent."
Also, the topline figures hide as much as they illuminate. A "detection" means that the crime was cleared up and a DNA match was available. It does not mean a conviction was obtained, or that guilt was established.
However, the NPIA do not report to this level, as this is a matter for the Home Office – while the Home Office sees the DNA figures as an operational matter, and does not make this link either.
Furthermore, the headline figure includes all crimes where DNA aided detection, and includes a suspect confessing to other offences. The raw figure for offences where crime detection resulted directly from a DNA match show a fall from 19,949 to 17,463 in the same period.
Over the last year, costs have doubled – from £2.1m to £4.2m, due to the transfer to the NPIA and the need to accredit and monitor additional suppliers.
Meanwhile the government continues to play its cards very close on new regulations for the DNA database: what they will be and when they will be announced are both, at this moment, unclear. Following a European Court ruling last December in the case of S and Marper, the government has spent nine months consulting on the way forward, and even included proposals in the recent Policing Bill to continue to retain the profiles of innocent individuals, albeit for set periods of time, according to the seriousness of the crime alleged.
Reports that this plan has been dropped are premature. A Home Office spokesman today indicated that new proposals would return for the next session of parliament, which begins on November 18, and that nothing could be ruled either in or out at this stage.
The legislation is highly unlikely to pass in advance of a general election, which means that the Police position of adhering to the status quo and keeping DNA profiles indefinitely could yet remain unaltered until the middle of 2011.
The Home Office continues to claim that the proposals that it put forward during consultation were "embodied in research", despite the Jill Dando Institute having now comprehensively repudiated the usefulness of that research in relation to the current debate. ®
One issue...that perhaps not all readers have grasped...is that there is a difference between the odds on a random match, which is usually what gets quoted, and a false match.
The random odds are what are usually quoted These are the odds that the profile matches a particular unique individual and not some other individual, whose details may not even be on the database. Usually, these are in the realms of millions to one against, and reflect the fact that it is unlikely, barring twins, for one individual to have an exact genetic doppelganger elsewhere at large in the world.
What has not been properly studied are the odds on false mtches, which are matches that take place whenever the wrong individual's DNA gets matched for any reason - including procedural error, contamination of samples, etc., etc.
Some academics estimate the odds on false matches as considerably higher.
The basic error
Few articles on this subject ever mention the relevance of 'the birthday theorum.' (http://en.wikipedia.org/wiki/Birthday_problem)
As others have explained, a DNA profile is a summary of the person's genome. Very like an MD5 checksum. If you have a sample at a crime site and an existing suspect it's very unlikely than you'll get a false match. You can and should take a fresh sample from the suspect at that point.
OTOH, if you have samples from very many crimes and a large database of 'suspects,' things are different. The only reason for having the db is to ask 'who in this set matches any of that set?' Then the probability of a false match increases exponentially. It will very quickly fail the 'beyond reasonable doubt' test if you could only get jurors to understand that.
Or politicians for that matter...
Burglaries down, mugging up
I can't find the link, but there was an Economist article that put the fall in burglaries down to falling prices of TVs/DVD players etc, while smart phones, iPods etc meant that mugging was more profitable.
Falls in car crime are linked to improved anti-theft devices built in by the manufacturers.