CLPEX Chat Board

Last week the Detail presented a paper on Contextual Information which represented a small experiment conducted under the eye of Dave Charlton. As a discussion topic I found the conclusions drawn from this research was written in strong terms that would usually be attributed to a more in-depth study and a more controlled experiment. I found it interesting that the cited case study was a US case and not a UK case. Rather than beat around the bush the details I thought were lacking in the study included; no apparent control of the degree of difficulty in the latent or inked cases, no disclosure as to the type of standard used by the participants and poor selection of participants who had no knowledge of a current topical case. For me the experiment should be given status as a scoping study to determine if proper research is needed. The language of the conclusions drawn is far to strong for the limited results as per example, "our data would still demonstrate that approximately ten percent of the experts were susceptible to misleading extraneous contextual information". If I had to respond to a defense question about this paper these would be my views. As for being a non-participant in the study I guess its because I'm interested in what happens to our science and how it is represented. Cheers from oz. Les

I have to agree with you Les and I personally feel the study is and has been flawed from the start. It is not a matter whether the case is in the UK or the US. Knowing the researchers also presents extreme egos involved and predetermined results. The findings appear to be subjective based on position, title or degree.

Les and Mr Frog, greetings. I read your comments on the research with great interest and as a scientist am never afraid to accept and learn from constructive crittcism. You have rightly observed some potential shortcomings in the experiement, it is right that you come forward with suggestions and new protocols that will further this work.

The experiment was very much a first run at trying to understand the concept of bias within our profession. It had a small scope agreed. All the more remarkable then given the minute error rate that we allow ourselves, that such a large effect was noticed in this study. I personally expected there to be no effect at all....how wrong was I??

We have noted your concerns and can inform you that we have subsequently collated another 48 data points from experts globally, and we have graded the difficulty of prints this time as well as make the suggestions more subtle. We also arranged to have 3rd party assistants administer the study to guarentee the 'ego' effect, as you put it, was negated. I have to tell you that while we are still writing up this study for publication, the results are no less dramatic. More will come from this study once the final submission has been made to academic press, hopefully JFI or another forensic journal will be interested in publication?

Please understand that as a fingerprint expert I am strictly in this research for the hell of good science. My PhD will be in the field of psychology, not how to beat up on my profession. But if the data cap fits....I must wear it!!

I welcome any comments on our research as science is all about discussion, learning and debate. Without such debate, you have no science.

Thank you Les and anyone else who wishes to contribute to this debate, and if anyone wishes to know more, then drop me a line and I will be happy to oblige.

I brought up very similar concerns to Dr. Dror when he presented his research at the NEDIAI conference earlier this month. I did not believe the strength of his conclusions were supported by the data he presented and I expressed my concerns about his sample size and sample methods.

I am pleased to say the Dr. Dror was extremely open to the concerns I expressed, he spoke about the limitations of this initial research and he did shed light on what he'd like to do next. I would suggest that you send an email to Dr. Dror (again, he was very open) and let him know you will participate in his research. I believe the only way we can increase the sample size is if we give him our names and get as involved with this research as we can.

I think this is very important research but it is in its infancy stage. More than anything I hope this research will lead us in the direction of how do we deal with bias. I agree with you Dave, we have to wear the hat if the hat fits...I'm still waiting on the fit test. Cindy

I find that we are essentially dealing with two distinct types of data correlation. The first is individualization and the second is the lack thereof. With individualization the formalized holistic process of ACEV results in a hypothesis that the impressions in question were made by one in the same source. However, with “the lack thereof” including, no match found or inconclusive as the result, the process has not culminated in a solution, but rather in the lack of correlating data. And providing sufficient information was present for evaluation we may adduce elimination.
However, the failure to find a match, or correlating data, is fundamentally different than a false positive (false individualization) hypothesis. I also am not surprised to see that ones failure to reach an information correlation threshold for a hypothesis of individualization is variable. This should be expected since the exact information set cannot be used twice. Holistic processes and its inherent complexity are the very definition of uniqueness. Each time we evaluate something our evaluation will use a different information set, however slight or great. We call this experience and environmental variability.
Image that we can recognize our spouse almost every time we see them, yet, think of all the condition that have change and all the variability (distortion) that we must evaluate. The person will have changed to some degree, their hair, clothing, the lighting will be different, and our perspective will be different. How can we do this? We are natural experts and comprehending variability.
Ultimately, the processes involved in the evaluation of individualization, elimination, and inconclusive are not reproducible due to the complexity and non-reducibility of the holistic information correlation regarding the inference reasoning process. It is the results that must be measured. Can we recognize our spouse? Can a trained fingerprint examiner accurate correlate fingerprint information? Can we drive our car 70mph down the freeway while correlating large quantities of new data with its accompanied distortion each second… and live to tell about it?
Complexity and non-reducibility does not mean that the process lacks credibility. Many, if not most, processes cannot reduce to a precise mathematical model, nor do they need to be. It is a fallacy to think that precision mathematical models (precision mathematical proof) is necessary to prove practicality. Yet, this seems to be what every lay person is asking for. Where's your proof, how can you compute uniqueness?
Ultimately, accuracy is in training, good protocol, and the evaluation of our work product by our peers... the V in ACEV. The process needs to be measure as product of ACEV. To measure only ACE, is not a measurement of the process.