I just got back from attending the American Academy of Forensic Sciences Convention, which was held in Atlanta this year. This was my second year to go, so I knew more about what to expect. Last year the sheer number of people and presentations to attend was overwhelming. It was no different this year - and going in I knew there was no way I could attend everything I wanted to.
In future posts I may write about specific sessions, but here I wanted to set out some general thoughts. The first impression was the quality of forensic scientists out there. The people that attend these meetings take their job seriously; they are not the people who blindly follow whatever direction the police set out. They are aware of the criticisms that have been leveled over the last several years, and want to fix the problems. If you work with them consider yourself fortunate.
One of the concerns I have is the use of forensics to determine guilt or innocence - we have seen too often what happens when the testimony turns out to be wrong. I'm happy to see that the good forensic scientists appear to share the same concerns. They appear to be frustrated about the way their testimony is manipulated and interpreted to fit the theory of one side or another. There were a number of sessions on how such testimony should be admitted and used - unfortunately, I was not able to attend them all, since almost every discipline had sessions on the topic.
Several of the presenters addressed the way such testimony should be presented. Not surprisingly error rate and uncertainty was a frequent topic. One of the scientists wondered why others were so concerned about talking about error rates; it's a fact that exists. The real issue is what it means, and how to explain it. For example, what does an error rate of +/-5% mean for a particular result? It might mean that the odds that the true result is within that range is 68%; we've all seen the "bell curve", but how many people actually understand it. If that is understood, then the question is whether that is enough to make a test admissible, or to base a conviction.The use of statistics and how they can manipulated and misused was also a topic of concern.
I'm starting to believe that the problem with the use of forensics cannot be laid entirely on the scientists; it's on the lawyers. We not only have to understand it, but we have to understand it's limitations. Clearly the scientists have an obligation to make sure we understand the limitations, but its up to us to make sure they do. And it's also on us to make sure juries and judges understand it. If properly explained, then maybe judges and juries can make the right decision.