So, forensics can be, and often is, unreliable. If something as reliable as forensics can be so unreliable, why are so many judicial officials so willing to place their faith in admittedly unreliable eye-witness accounts....
In my position as an automotive quality laboratory manager, I constantly come in contact with "Bad" forensics. The "Cure" for this problem is the enforcement of MSA practices, specifically as taught in "Six Sigma" courses. MSA's are Measurement Systems Analysis' and as performed by statisticians they measure the chance of an erroneous measurement (or conclusion) by testing the system with contrived "masters" which are samples of whatever you are testing. Therefore if you are comparing tire tracks, one would make casts of a hundred random tire tracks in varying soils and environments and blindly classify them. Then reclassify them. When this data is crunched it gives a reliable appraisal of the technique in question. Surprisingly many standard techniques fail to be as reliable as previously believed. Unfortunately I can imagine why this practice, required in my field by certification agents, is probably not adhered to in police labs - Cost! It takes a lot of time and personnel hours to enforce MSA practices and to train Personnel in statistical analysis. A typical Six Sigma Black Belt training course costs $80,000, plus! This is more than most of their Salaries. Who should pay for this?
In my position as an automotive quality laboratory manager, I constantly come in contact with "Bad" forensics. The "Cure" for this problem is the enforcement of MSA practices, specifically as taught in "Six Sigma" courses. MSA's are Measurement Systems Analysis' and as performed by statisticians they measure the chance of an erroneous measurement (or conclusion) by testing the system with contrived "masters" which are samples of whatever you are testing. Therefore if you are comparing tire tracks, one would make casts of a hundred random tire tracks in varying soils and environments and blindly classify them. Then reclassify them. When this data is crunched it gives a reliable appraisal of the technique in question. Surprisingly many standard techniques fail to be as reliable as previously believed. Unfortunately I can imagine why this practice, required in my field by certification agents, is probably not adhered to in police labs - Cost! It takes a lot of time and personnel hours to enforce MSA practices and to train Personnel in statistical analysis. A typical Six Sigma Black Belt training course costs $80,000, plus! This is more than most of their Salaries. Who should pay for this?
ReplyDelete