Have you seen a change in the forensic reports you are receiving in criminal cases, lately? Have you noticed that there are some “weasel words” creeping in? Have you been left baffled by some of the wording in the conclusions? You’re not imagining it…
Mixed DNA Analysis is being relied upon more and more by the police and CPS – but it can be used to mislead, bluff, and obfuscate. When the algorithms behind this analysis are secret and the reports are full of weasel words, lawyers must be on their guard.
The statements will tell you that the mixed sample was too complicated to be analysed by a human and, instead, software has been used to pick apart the troublesome samples and to produce reliable results.
The problems with this “new” approach are many and complicated but include two really pressing concerns for any serious lawyer.
What’s under the hood of the software?
Firstly, the software that is performing these analyses is proprietary (for the most part) and, as such, the software is using an algorithm that you, the jury, and your expert do not get to see. Where this has been challenged, the companies involved have been obstructive and have asked for NDAs to be signed. But these challenges have been successful in this jurisdiction and several others. In some cases, the evidence in question has been withdrawn and no evidence offered as a result of challenging the software at the core of these reports.
If you’re still not convinced, ask yourself how many other types of evidence you would let anywhere near a jury without knowing how the evidence came about. Imagine an identification case where there was no information about the VIPER procedure – just a note saying that the defendant was the one picked out. Would you be satisfied with just being told that? Might you want to see the I,II, and III documents? Maybe even the video? Would you be happy with a statement from a police officer simply claiming that he was told that the defendant was your client? Not by whom. Not in what circumstances.
Or, in the world of forensics, imagine always relying on the SFR without question. You simply would not.
Secondly, the software is only as good as the question it is asked. This is the “lies, damned lies, and statistics” element, you might say. Let us consider an example:
A report is produced, based upon a low-level, mixed sample from a moveable object. The best possible sample is taken from the suspect and is compared. The initial report (by a human) states that the nature of the mixed sample is such that it is impossible for her to provide any meaningful conclusions about whether the samples suggest any particular scenario. Maybe she classifies it as “weak support” that the suspect is a contributor.
Step up, genotyping software. A new report appears and the conclusion (everybody reads it first!) says, “I have considered whether the DNA on the exhibit is more likely to have come from the suspect and an unknown person or from two unknown people and it is billions of times more likely to be the former than the latter.”
There it is! Billions of times! Did you hear that, members of the jury? That’s a big number, isn’t it!
Wait a minute, though…
We don’t know why that’s the case. For example, if the sample from the exhibit contained components 1 and 2 and only 20 percent of the population possess those components, that’s not really telling us anything about whether the suspect was a contributor. At best, the analysis is loading the dice against the suspect because it is effectively ignoring the existence of the hundreds of thousands (or maybe billions) of others who may have contributed.
But how do you break that down for a jury? How do you show how meaningless that statistic really is, in the context of the proceedings? You need expert examination of the maker of the report and you need the right expert knowledge.
Experts dealing with experts
There are experts in the field of genotyping and there are experts in the field of statistical analysis. There is some overlap between the two. Pro-active defence is the key to success in this type of case and that begins with the right legal team. A barrister who understands the flaws in the way evidence is being passed-off as demonstrating something that it cannot truly be said to show.
If you think that this is limited to DNA analysis, think again. Only this week, I received an expert report relating to footwear in which the author abandoned the usual grading of “levels of support” and preferred to use the weasel words approach. His approach was to say that it was billions of times more likely that the footwear prints found were “billions of times more likely to have come from the shoes seized from D than not”. But, re-reading the report, he at prefaced this (a paragraph earlier) by saying that his conclusions would be based on the proposition that the footwear print must have been made by one of the items of footwear seized.
So, that ignores the distinct possibility that the shoe that caused the print might have been caused by some other item of footwear that was not seized. That is to say, his conclusion may be read as, “if it absolutely had to be one of the shoes sent to me by the police, it is much more likely to be shoe A than it is to be shoe C – which is a different size and make”. That report, if I may remind you, ends with the words “billions of times more likely”. And it is relied upon by the prosecution, despite it showing something so statistically irrelevant as to be potentially misleading to a jury.
It is interesting that, in these heady days of “disclosure” being a word on the lips of the Man On The Clapham Omnibus – sometimes followed by the word, “scandal” – the police and prosecution are exploring such a hazardous method of evidence presentation ; one where the defence is obstructed from looking at the algorithms manufacturing the evidence.
About the author and about Central Chambers
Benjamin Knight is a senior member of the Central Chambers Criminal Team. He has a proven track record in trials and proceedings involving expert witnesses and is well-regarded for his expertise in areas such as IT evidence, medical evidence, DNA, and ballistics.
Wherever possible, Central Chambers is pleased and proud to offer a “one case, one counsel” policy in the Crown Court. We believe that professional and lay clients ought to have continuity of representation because we respect the care taken in selecting counsel for any individual case.