Terms in this set (...)
Fenton & Neil, 2012
Bayes: 'basic rule for updating the probability of a hypothesis, given evidence'
Wording of statistical evidence in court - the word 'match'
R v T, 2010
Bayes/LRs should not be used in evaluating forensic evidence except for DNA or other evidence with a 'firm statistical base'
Gonzalez-Rodriguez et al., 2006
Two problems with Bayes:
1) Info in Bayes is riddled with assumptions which could be questioned in court.
2) Very difficult to explain simply and then scale up to real life processes
Fenton et al., 2014
Show how easy it is to use pairs of hypotheses assumed to be mutually exclusive but are not.
Evett & Buckleton, 1989
Assumptions in Bayes are often made to facilitate discussion
Biedermann et al., 2014
Fenton et al., 2014 risk making LRs and stats seem like an impossible framework to use
Freckleton et al., 2012
Jury generally have v little numerical background and haven't adapted to quantitative evidence
R v Clark, 2000
Nobles & Schiff, 2003
Statistics of killing both children vs them dying of SIDS.
Prof Meadows multiplied the two probabilities together, but this doesn't account for other factors eg genetic predisposition to SIDS.
Lindsey et al., 2003
Communicating stat evidence to the jury can be challenging
Beyth-Maron & Fischhoff, 1983
'People are much better at using...[statistical] evidence than they are at seeking it out...or articulating reasons for its usage.'
Kay & Koehler, 1991
Explanation of statistics is important so the jury can understand what the evidence entails.
CSI effect - juries are influenced by the 'presence' of stats and also demand more evidence.
Defence Attorney's Fallacy
Base rate statistics
Lopez Puga et al., 2015
Coin eg for the prosecutor's fallacy
Aitken & Taroni, 2004
The Prosecutor's Fallacy
Thompson & Schumann, 1987
Exposed undergrads to statistical statements framed in different ways to see how they were influenced
Balding & Nichols, 1994
Product rule ignores uncertainty and issues with sub-populations
Gill & Haned, 2002
P-values should be used to explain the strength of an LR in relation to DNA
Krujiver et al., 2015
Reject Gill & Haned's idea - p-values don't relate directly to the strength of the evidence, they just say how rare it is to find evidence that strong/stronger
Random occurrence ratios can be misjudged by jury to be an expert's opinion on the likelihood of an event.
Lecture 18 - Reasoning, Judgement and Decision Making
Research methods in psychology