Talking the ’CSI effect’

By Anna Sandiford 25/03/2011

Whilst filming an episode of Media 7 on Wednesday about the CSI effect, I was struck (not physically) by two things:

1. people genuinely seem interested in the CSI effect;

2. everyone has a different take on what it is and how they see it in the programmes they watch.

It was interesting to hear Marie Dyhrberg, a prominent Auckland defence barrister, talk about how less time now has to be spent by expert witnesses explaining concepts like DNA because juries already have some knowledge of them from watching TV crime fiction programmes.  That is supported by my own experience of explaining concepts to juries where the nods of the heads of understanding come quite quickly into the initial explanation. The exceptions are usually the subjects that don’t make it into CSI: Miami or Bones.   The area in which I have encountered the most problems is alcohol consumption. Explaining the difference between absorption and elimination is not glam and is doesn’t translate into a gory re-enactment of the scene beyond some swigging of lager and a bit of staggering by the actor.

A good understanding of forensic science must also help an investigative journalist like Donna Chisholm, who was involved with the David Dougherty case.  Writing a story that contains technical scientific information for a general interest publication must be easier if your readership already has an understanding and an interest in the world of forensic science.   It won’t be helped of course by the leaps of interpretation that are made on TV programmes, like those in last night’s episode of Bones.

I can’t go into detail because I shouted over it but it went something along the lines of: “Look at this previously unnoticed crack in the bone that I am suddenly and spontaneously examining with a look of revelation and excitement on my face.   Without turning over the bones or examining the interior of the damaged surfaces I know that this must have been caused six months prior to the death of the victim and she was stabbed in the arm with a knife used to cut cacao plants.”

Then there was the incredible situation where one of the scientists  managed to insert large needles into a giant bar of chocolate and draw out the gases from two bubbles that they had managed to visualise on a computer screen (on the first go – they didn’t spend five minutes stabbing around in the not-even-a-little-bit-melting chocolate: oh yes, that was because they turned the air con down in the lab….)  Not only could they say that one of the bubbles contained blood from the killer that the victim exhaled during her last breath after she had somehow managed to get the killer’s blood in her mouth but they were also able to say that the other bubble contained air from a burp caused by a specific brand of champagne.  Now, I am more than happy to be corrected if I’m wrong but that really does seem to stretch the bounds of the credible.

Many people are apologetic about watching these sorts of crime programmes, which suggests that they know they’re not realistic – and that should be of some comfort to the people who think that the CSI Effect generally has a negative impact on the geneal public’s understanding of forensic science.  I do think that a greater awareness of forensic science and, importantly, its limitations has to be good for the justice system. Without wanting to sound repetitive (but I am) forensic science needs to be transparent so that all parties can be assured that the science is being fully, fairly and accurately reported. With juries expecting science to have a place in criminal casework, I would also hope that they think about what is being presented to them.

The next question is, do juries make judgements of expert witness evidence (who to believe, basically) based on what the jury knows from the TV or based on what the expert witnesses say in court or both?

And why is there only ever one, unchallenged interpretation of the set of scientific results presented in Bones?!

0 Responses to “Talking the ’CSI effect’”

  • Anna, since DNA is well accepted in the court room today, what is your opinion about the likely acceptance in court of the use of another useful tool from probability theory known as BBN (bayesian belief network) for building causal networks? Do you think that it will ever be accepted in court at some stage in the future the same way as DNA is? Here is a document which describes how BBN can be applied in the legal domain.

    “Use of Bayesian Belief Networks in legal reasoning”
    The principles of Bayesian probability and Bayesian belief networks, and their applicability for judicial reasoning are briefly explained. As an example, a Bayesian belief network model has been constructed that compares the probabilities of two hypothesised causes of a single-car accident. It is concluded that Bayesian belief networks can be a suitable tool for lawyers to analyse the evidence in judicial cases, as well as a suitable tool to train students to perform those analyses and
    to avoid statistical pitfalls.

    Your fellow co-blogger David Winter has covered bayesian theory here at SciBlog, in one of his article with title “Ken Ring can’t predict earthquakes either “.

    BBN has been applied in various domains, such as automated tool for physicians for medical diagnosis, engineering fault detection (as in aircraft), etc,…, but its adoption in the legal domain I think has largely been unexplored. There are many open sources on the internet where BBN software is freely available and one of those is the excellent opensource software in machine learning from Waikato University called “WEKA”. BBN is available in Weka, in case your colleagues at “Forensic Group” are interested to explore it.

    I believe that had this tool being applied in the recent David Bain’s court case, the BBN model would have come up with a verdict that found David Bain guilty (ie, in terms of probabilistic reasoning).

  • Every time I read about unrealistic items on tv crime dramas I think of this scene from “Castle”
    A picture of the Victim was obtained and the protagonist said they should run it through the facial database:

    (slapping a pile of folders down in front of Castle)
    Beckett: Welcome to the department’s official facial recognition database.
    Castle: By hand? That’s like life before TIVO.

    Quite possibly my favourite scene of any crime drama in the last few years.

  • It’s always so nice to know that it’s not just me who doesn’t take these programmes seriously! It’s the ones that sail too close to reality that cause the confusion.

  • In terms of the Bayesian Belief Network, I am by no means a statistician beyond that for which we use Bayes and Likelihood Ratios. Overall though, the court system has struggled to understand LRs and it is not being done well outside DNA, particularly in NZ. I personally am very wary about any system that requires scientists to consider guilt or not as that is not our role and never should be – we are not advocates. Providing a software-based tool into which lawyers insert figures in order to establish guilty or not guilty also worries me significantly. This implies that the lawyers know the value of the “evidence” before it has been given in court. It is most often juries who make the decision of guilty or not guilty, not the judge or the lawyers or the witnesses in the trial. In my opinion, it is not currently possible to quantify the evidence of lay witnesses in a satisfactory manner in order to be able to add in numerical values for their evidence. Maybe this framework could be adapted at some point but there would need to be robust assessment of it and consensus on its use outside the court arena before it should be allowed to play a part inside the court. As with any other forensic area, the court is not the place to test things; court is the place where tested things are taken.

  • Falafulu Fisi,

    I hope I am reading your intentions re BBNs right – to replace the jury with a statistical model?

    Leaving aside technical problems, wouldn’t you end up with with a “meta trial”, a trial over the model as it were, where lawyers would argue over inclusion of evidence into the model, any weighting schemes used, etc.? If so, wouldn’t this just recreate the trial in question in another form, one further removed from the events still, one even less understandable and workable for a jury? Just loose thoughts.

    If practicable, it’s possible that related tools might in principle be used would be to try identify key questions while preparing for complex cases. (As opposed to ‘judging’ the evidence.) I don’t know how practicable this would be, though. Perhaps in some cases there would be a place for it? You’d still be stuck with showing why they are key questions in court, and having the time (and funds!) to do this modelling work. Again, just loose thoughts.

    I should add that I support better use of statistics and analytical techniques where they would assist (the example of DNA analysis that Anna gave is related to my own field, for example), and the usual disclaimer that I’m not a lawyer, etc.