12.11.2012 Views

Examination of Firearms Review: 2007 to 2010 - Interpol

Examination of Firearms Review: 2007 to 2010 - Interpol

Examination of Firearms Review: 2007 to 2010 - Interpol

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

tested (whether pre-trial beliefs regarding DNA, the reliability <strong>of</strong> the labora<strong>to</strong>ry, or the<br />

style <strong>of</strong> cross-examination influenced the verdict (157)).<br />

Starting from dictionary definitions <strong>of</strong> the words ‘subjective’ and ‘objective’, Leo<br />

develops arguments in relation <strong>to</strong> different court cases where the use <strong>of</strong> the word<br />

‘subjective’ was important, according <strong>to</strong> this analysis (191).<br />

An important change occurs in the argument about fingerprint identification: it is<br />

being more and more considered an opinion, based on the specialists judgement,<br />

rather than fact (158-160).<br />

Bias<br />

<strong>Review</strong> papers in the subject <strong>of</strong> cognitive bias are now available (26, 192, 193). A<br />

short summary <strong>of</strong> the research contributions made during the reviewing period is<br />

given hereinafter.<br />

Whitman and Koppl (161) discusses biasing effects <strong>of</strong> the justice system and the<br />

integration <strong>of</strong> forensic services with the police, as well as reaction possibilities. This<br />

article also discusses ‘verification shopping’, which consists in searching for<br />

someone who will verify the original conclusions. Dror and Rosenthal describe two<br />

previously published studies and carry out a meta-analysis, reuniting the effects <strong>of</strong><br />

bias observed in those two studies; the effect size thus obtained is quite large and<br />

with a p-value <strong>of</strong> 0.015 would generally be considered <strong>to</strong> be significant (162). Bias in<br />

verification has been tested for both experts (n=43) and novices (n=86). Three<br />

groups were created for both the experts and the novices. The first group carried out<br />

comparisons without any information (control), the second group received the<br />

comparisons and had answers provided that were stated <strong>to</strong> have been given by a<br />

fingerprint examiner trained <strong>to</strong> competency (low bias group); in the third group, the<br />

answers provided were said <strong>to</strong> come from prominent and internationally recognized<br />

expert (high bias group). For experts, there was a significant increase in inconclusive<br />

opinions in the low and high bias groups with respect <strong>to</strong> the control group. This<br />

increase is larger for the three different source trials than for the three same source<br />

trials (only one <strong>of</strong> the same source trials produced inconclusive opinions). One trial<br />

lead <strong>to</strong> an increase in the relative frequency <strong>of</strong> experts’ conclusion <strong>to</strong>wards the bias<br />

prompt; in this trial the ground truth was different source, and the bias prompt was<br />

inconclusive. For novices, there were more inconclusive conclusions for the different<br />

source trials than for the same source trials, but there were not more inconclusive<br />

opinions in the bias groups with respect <strong>to</strong> the control group. There was, for novices,<br />

an increase in the relative percentage <strong>of</strong> responses in agreement with the bias<br />

prompt in the bias groups with respect <strong>to</strong> the control group (163).<br />

Dror analyses the sources <strong>of</strong> biases that are represented by the four idols <strong>of</strong> Francis<br />

Bacon, in order <strong>to</strong> improve forensic science by identifying such vulnerabilities (164).<br />

A study has been carried out with 70 fingerprint examiners, in a way resembling<br />

verification, and declared as an experiment. Examiners were in two groups, high and<br />

low emotional context (an allegation <strong>of</strong> murder and one <strong>of</strong> forgery). This study did not<br />

find any bias effects (165). The study has been criticised concerning both<br />

experimental flaws and problems with the interpretation <strong>of</strong> the data (166, 167). The<br />

authors react <strong>to</strong> both <strong>of</strong> these criticisms (168, 169).<br />

240

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!