The potential (negative) influence of observational biases at the analysis stage of fingermark individualisation

Recent cases of erroneous identification have strengthened critical comments on the reliability of fingerprint identification. This goes hand in hand with recent publications regarding the lack of scientific foundation of the discipline. Combined with "legislative" needs, such as for instance the admissibility criteria under Daubert, or experimental studies revealing potential bias, the call for research on the identification process has become more urgent. That background set the basis of this research project financed by the Swiss National Science Foundation (SNSF) which includes, among other parts, experimental tests to study potential observational biases in the analysis stage of fingerprint individualisation. These tests have been submitted to several groups of forensic science students at the University of Lausanne. The aim is to study factors potentially influencing the analysis of fingermarks, more specifically the influence of training/education (test I) as well as the potential impact of case contextual information or known print availability (test II). For all tests students were given 11 or 12 fingermarks of a medium to difficult quality, with a range of 8-15 minutiae. For all tests the task was always the same for the participants but carried out in different contexts: to analyse the marks, to annotate the minutiae observed, to designate them and to decide on the status of the mark in two categories, exploitable and identifiable. The aim was to see how the fingermarks were annotated by different individuals so as to have an idea of the variation in annotation and counting in the analysis stage only. For test I, students were submitted the same 12 fingerprints before and after having followed specific training in fingermark individualisation. The aim was to see how training/education impacts the analysis of fingermarks. For test II, were participants given eleven fingermarks so as to study whether the presence of a comparison print changes the amount of minutiae found and whether low/high-profile background information influences the analysis stage. Results show that for test I the effect of training can be observed, among other, in an increase of minutiae annotated and a higher consensus between participants. For test II no effect of the stimuli used to induce observational biases has been observed by all of the factors studied.