European forensic labs develop statistical software together
The Netherlands Forensic Institute (NFI) is joining forces with other forensic institutes and universities to develop software that will help calculate the probative value of traces.
The first version should be available in two years, and will help forensic experts determine the likelihood of their results in different scenarios.
Forensic experts present their findings based on the likelihood ratio (LR) method: they say something about the probative value, or, put differently, how much their findings support one hypothesis over another. Experts have long expressed this LR in specific verbal definitions such as ‘much more likely’.
People from an increasing number of fields of forensic science are working on numerical substantiation based on datasets and statistical models. They state, for instance, that the results are 10 times more likely if the victim was hit with a club on the back of his head (Hypothesis 1) than if the victim slipped and fell on his head (Hypothesis 2).
The software will help forensic experts calculate numerical LRs. “To achieve that, we will have to develop scripts based on statistical models and validate existing scripts in the next few years, so that they produce reliable results. After that, we will need to build a firm and user-friendly shell around it”, project manager Annabel Bolck (NFI) says.
Many of the statistical basic models have already been developed, and mayneed further tweaking. “We want a flexible range of models that can be used for the various fields of forensic science and to which we can subsequently add new models. We expect that experts can begin to use the software in about two years.”
The various participants in this project have been given a European grant by the European Network of Forensic Science Institutes (ENFSI) to develop the software in the next two years. Statisticians and software engineers from Great Britain, France, Norway, Poland and Sweden will visit the NFI to attend the kick-off-meeting on Monday 23 and Tuesday 24 March 2015.
“At that meeting, we want to take decisions on the kind of models we want to initially include in the software, on how we are going to write and validate scripts, and on what the user interface will look like”, Bolck (NFI) explains. The kick-off meeting will be the first of a total of four meetings in the development stage.
It would be great if this software became the standard software in European forensic science, Bolck says. “If the software enables forensic researchers and statisticians to calculate LRs in a simple way, without having to develop software themselves, it will allow us to work faster, thereby saving valuable time.”