lowmanio.co.uk title image

Problems faced in forensic science

Sun, 04 Oct 2009 06:37PM

Category: Forensics & science

Written by Sarah | No comments

In my introductory lecture of Fundamentals of Forensic Science, the lecturer spoke about the problems currently faced in the field. Forensic “science” is not always an exact science. Things such as drug analysis and DNA comparison is exact and very scientific, but clothing damage, blood stain patterns and scene reconstruction is not. It requires a lot of domain knowledge, common sense and just a little scientific method. They are, however, very useful no matter how much actual science goes into them. There are three other problems faced – 1) not enough research is happening in the field; 2) things are treated as unique when they are not and 3) confirmation bias.

It is well established that DNA is not unique, but there is an incredibly low probability that it will match someone else. However the idea that nothing is unique, only rare, hasn’t filtered through to other artefacts such as fingerprints, tool marks, shoe prints etc. Are they unique? No, it is just often unlikely to find another match. More research needs to be done how “unique” these things are. If a bullet matches a gun, how likely is that it came from that one? It’s not 100% conclusive as many TV shows and actual CSIs/lawyers/detectives believe. But how possible is it to test every gun and every bullet?

The most important problem faced today is confirmation bias. Confirmation bias is 'an irrational tendency to search for, interpret or remember information in a way that confirms preconceptions or working hypotheses'. In other words, if you are certain x caused y, you might only look at the evidence that supports that and conclude you are right. There are easy ways to stop this. For example, when comparing a fingerprint to five suspects (and one is a previous offender and the likely owner of those fingerprints) - get someone else to label the comparisons A to E in random order, so you don’t know what is being compared. Blind tests mean that any bias is gone and the test can be carried out scientifically.

The lecturer said that many of these problems are caused by the older generation, who went into forensic science when there were no degrees or specialised training. Many even deny confirmation bias exists or do not believe that matching evidence to a suspect is probabilistic and not unique.  More training and research needs to be done in these areas to keep forensic science as scientific as possible.

Itiel Dror wrote a short, but interesting article on confirmation bias in the forensic world which you can read here. One experiment he carried out was the fingerprint example above using leading fingerprint experts. Later he repeated the test randomised and with no obvious suspect, and the experts gave different results.

Comments

No comments.

Add a comment

captcha