By Elliott Williams, Frontiers
For forensic science, technology that assists in the analysis of complex and crucial evidence, such as DNA evidence, has seen rapid advancements, with ever more sensitive tests being introduced.
Statistical interpretation of such complicated results has not been as swift.
There exists now a lack of communication which spans all levels of those handling and presenting crucial evidence. Statisticians develop software with little input from the forensic scientist user, who does not have the possibility to discuss the intricacies of the results with the court. And in some cases, doesn’t know how the software “black box” generated its data.
This is the long needed discussion Sue Pope (Principal Forensic Services) and Alex Biedermann (University of Lausanne) hope their topic, will spark between the statisticians developing analytical methods, the scientists who apply them, and the lawyers presenting the life changing results to the court.
We spoke to them to find out more about this important and inter-disciplinary topic “The Dialogue Between Forensic Scientists, Statisticians and Lawyers about Complex Scientific Issues for Court.”
Could you briefly give us some background on your Research Topic’s theme?
As a working forensic scientist, I have become increasingly aware of the variety of approaches between the groups of people within the Criminal Justice System (both in the UK and elsewhere). For instance, scientists and statisticians consider DNA evidence differently, and confusingly, often use similar terminology to mean different things. Lawyers are concerned with other aspects of the evidence.
I became interested in seeing whether each group could learn from what others are trying to get across. This Research Topic might help to simplify the discussions that take place inside and outside court.
Why did you choose Frontiers? What is it about the Frontiers Research Topic offering that you liked?
I like the variety of journals and wide readership of Frontiers. and hoped that this would cover the intended participants.
Frontiers has an innovative publishing and reviewing model. Reviewers and the Handling Editor are acknowledged explicitly with their name and affiliation included on the published article. I think this is transparent and fair, and encourages a thorough commitment to all phases of the publishing process.
What are the objectives you are hoping to reach by organizing this Research Topic?
Criminal Justice Systems would benefit from:
- A scientist understanding a statistician’s need for user requirements when programs are commissioned, and a scientist understanding a lawyer needs non-technical descriptions of what has been done.
- A statistician understanding of lawyer’s need for non-technical descriptions of what has been done, and a statistician understanding that a scientist needs to understand the programs so they can both use and explain it.
- A lawyer understanding the way that a scientist interprets the DNA evidence and its limitations, and the way that statisticians do the same process.
What are the ‘real’ open questions in this topic?
– Whether a black box piece of software is suitable as a basis for debating if DNA evidence is admissible, and what impact it should have in court proceedings.
– How to validate models for statistical evaluation, how to validate software for statistical evaluation, the difference between the two.
– If a defendant should have a right to a computer program’s source code. In the US, there is case law admitting testimony based on computer programs—mainly specialized programs for evaluating forensic DNA profiling results—but denying the defense access to the source code.
Why is this an important topic now?
This Research Topic is especially topical in the light of the soon-to-be-released PCAST report in the US, which is critical of DNA software. Indeed, the report has recommended that such evidence should no longer be considered valid.
The subject of this Research Topic is most visibly raised in connection with forensic DNA but it could be relevant for all forensic techniques relying, in one way or another, on computer source code (for example, devices for the measurement of alcohol concentration in breath).
What is the historical basis for the Topic?
Earlier statistical approaches were based on concepts of probability that were familiar to forensic scientists. As DNA analysis has become more sensitive, a person’s genetic profile can now be obtained from very a small amount of DNA which may be absent from some test sites. A much higher proportion of DNA results are complex mixtures of DNA from multiple donors. Issues which were once observed rarely have become everyday occurrences. Thus, the statistical approaches have become increasingly complex and high-powered, for example using Markov Chain Monte Carlo techniques, making it difficult to explain the methods to scientists and the non-experts.
How does your Topic sit within the community today?
A special semester on probability and statistics in science is taking place at the Isaac Newton Institute in Cambridge from 18th July to 21st December 2016, which has a much wider scope but does also include DNA. Lawyers, scientists and statisticians from across the world are attending and discussing this, as well as other areas.
Read more on this Research Topic at its homepage: http://journal.frontiersin.org/researchtopic/4000