Expectancy (3) Flashcards
Dror, I. E., Charlton, D., & Peron, A. E. (2006). Contextual information renders experts vulnerable to making erroneous identifications.
Aim
Methods
Results
Limitations
- examine the influence of contextual information on the result of forensic testing
- -> hint about the outcome, fingerprints
- 5 fingerprint experts were asked to examine a fingerprint and judge whether it was a match
- they were told that it was the fingerprint from the Madrid train bomber case (which was famously mismatched)
- however they were actually given a print and a reference from an own old case which they previously matched
(with-in subject design) - only one of them found that it was a match again
- 1 said that the print was insufficient = inconclusive
- 3 said it was not a matched
- -> 4 experts changed their conclusion (“because” of the contextual info which suggested beforehand that it would not be a match)
- -> contextual information influence the judgment of forensic experts
- very small sample so cannot make inference (however the trend of findings were replicated in other studies)
- other factors then the contextual info could be confounding (for ex. time effect)
- NO CONTROL GROUP!!!
Hamnett, H. J., & Dror, I. E. (2020). The effect of contextual information on decision-making in forensic toxicology.
stages where bias influences decisions making
3 stages in toxicology where biases can occur/influence:
1) test selection (expected frequency bias)
2) identification during drug screening (affected by the target reference)
3) decision making/ interpretation of results (tunnel vision/ confirmation bias)
(siehe Pyramide)
Kukucka, J., & Kassin, S. M. (2014). Do confessions taint perceptions of handwriting evidence? An empirical test of the forensic confirmation bias.
Aim
Methods
Results
Limitaion
- show that confession that the expert is aware of will influence their judgment about the handwriting
- lay participants - with university education - (n=117) read a case summary in which the defendant did not confess or confessed
- then they evaluated two handwriting samples— one from the known perpetrator, and one from the defendant
- participants who read a case summary in which the defendant had previously confessed were more likely to erroneously conclude that handwriting samples from the defendant and perpetrator were the same
- they were more likely to judge the defendant guilty (compared with the no-confession control group)
- -> findings repeated in a second study - not done with experts
Nakhaeizadeh, S., Hanson, I., & Dozzi, N. (2014). The power of contextual effects in forensic anthropology: a study of biasability in the visual interpretations of trauma analysis on skeletal remains.
Aim
Methods
Results
(Limitations)
- examine the influence of contextual information on the result of forensic testing
- -> context of source of test item, anthropology
- 99 participants (experts and students)
- they were given 14 pics of skeletal remains on a website: 5 with no trauma, 3 with trauma, 6 ambiguous ones
- they needed to indicate for each pic if no trauma/possible trauma/high likelihood/definitive trauma
- they were randomly assigned to 3 different conditions
1) “Mass grave” (high association with trauma)
2) Archaeological findings in Dorset (low ass. with trauma)
3) no contextual info (control) - no difference in evaluation for 2) and 3)
- significantly more trauma reported for 1)
• Changing expectations –ignoring evidence against preconceived ideas
• Cognitive bias –very indicative in ambiguous images →pronounced effect of bias
12 Sources of bias in forensic neuropsychology with ways to mitigate
(Richards et al., 2015)
1) Role bias = Conflating Clinical and Forensic Roles
- Clinical neuropsychologist asked to provide testimony about their patient = fact witness / treating expert
- -> role conflict, which does not arise with an expert who has clear boundaries and understanding about confidentiality with the examinee
2) Financial/payment bias
- contingency payment –> depends on the outcome of the case => could lead to a conflict of interest
- Lien/letter of protection –> payment if settlement is reached no matter the outcome
- Free for service –> pro bono
3) Referral Source Bias (& Retaining Attorney Pressure)
- Rule 26 of the Federal Rules of Civil Procedure requires an expert to provide references
- leads to attorneys selecting certain expert who maybe is very one-sided and the experts again to be concurrent with their previous work because this is what is expected of them
- -> mitigate the bias through not have early attorney meetings
4) Self-Report Bias
- can be biases (retrospective inflation etc.) or even not true
- look at collateral for corroboration –> don’t cherry pick! everything is important!
5) Underutilisation of base rates / base rate neglect
6) Ignoring normal variance in test scores
- siehe ?
7) Confirmation bias
- think of alternative hypothesis!
8) Personal or political bias
9) Group attribution error
- overgeneralisation
- to assume own traits are reflective of the group
10) Diagnostic momentum
- tendency of diagnosis to become certain as it passes from person to person
11) Good-old-days
- tendency to view oneself as healthier higher functioning in the past
- tendency to misrepresent preinjury functioning as better than average
- negatively impacts perception of current behaviour, recovery, ability to return to work
12) Overconfidence bias
Cooper, G. S., & Meterko, V. (2019). Cognitive bias research in forensic science: a systematic review.
Aim
Method
Results
Limitation/Recommendation for future
- examine extent to which cognitive biases may influence decision-making in forensic science
- systematic literature review
• most studies limited to (prospective) practitioners (n = 22), some done with students or general population
• Latent fingerprint analysis was in 11 studies, 1–3 other studies found in 13 other disciplines - support for influence of contextual info and confirmation bias:
• biasing influence of case-specific information about the “suspect” or crime scenario in 9 of 11 confirmed
• biasing influence of knowledge of a previous decision in 4 of 4 studies confirmed
➔ potential value of procedures designed to reduce access to unnecessary information and control the order of providing relevant information –> blinding (siehe Task 6)
➔ the more human judgment the more vulnerable to bias, happens also in DNA cases - look at under-studied disciplines,
- assess level of subjectivity in the analytical procedures in relation to magnitude of bias
- assess difficulty of decision as modifier
Task-relevant vs. task-irrelevant information (for forensic experts)
Def. task-relevant
Example task = irrelevant
- National Commission on Forensic Science: “information necessary for drawing conclusions about
the propositions in question
the from the physical evidence that has been designated for examination
(…) through the correct application of an accepted analytic method by a competent analyst” - suspects age, race, criminal history, confession etc. (all this contextual info mentioned in the other articles)
Gardner, B. O., Kelley, S., Murrie, D. C., & Dror, I. E. (2019). What do forensic analysts consider relevant to their decision making?
AIM
METHOD
RESULTS
LIMITATIONS
- examine what experts perceive as task-relevant
- 183 forensic analysts
- survey
1) attitudes regarding the acceptability of error types
2) estimates of error rates within forensic science disciplines
3) perceptions of task-relevance regarding available case information! - nearly no absolute consensus
- general consensus (in lab disciplines) about irrelevance of ethnicity, age, confession and eyewitness account (–> CSI only age)
- general consensus about relevance of evidence description and case synopsis
- lack of consensus about offense type, method of collection of evidence (“lab” disciplines)
- lack of consensus about name of investigator and victim (CSI)
- -> over all little barely any consensus interdisciplinary
- -> also not a lot consensus with in the different fields
- study did not examine forensic psychologists (Schade!)
Murrie & Boccaccini (2015). Adversarial allegiance among expert witnesses.
- Def. adversarial allegiance?
- Problems in studying it?
- What does this paper conclude?
- 3 reasons why?
= some experts’ opinions tend to drift toward the party retaining their services
- it is an understudied issue
- has been suspected (kind of makes sense)
however, in field studies it is hard to conclude this because it could also be due selection bias - in experimental it is hard to generalize it
- field studies suggest that their is an adversarial allegiance bias, but cannot prove it –> with PCL-R which has a good interrater reliability in clinical setting, prosecution retained evaluators assigned higher risk measure scores than do defense retained evaluators an adversarial legal settings
- in experiments shown same trend even more pronounced and with causal implication
- -> adversarial allegiance seems to be a thing!
1) “social-psychological process”: unconscious in-group vs. out-group mentality
2) typical errors in decision making: confirmation bias, stating from a bias not a neutral point etc.
- -> go hand in hand
3) MONEY
Hamnett, H. J., & Dror, I. E. (2020). The effect of contextual information on decision-making in forensic toxicology.
Study 1
Aim
Methods
Results
Limitations
- In two studies, examine the effect of contextual information on for once toxicological test selection and decision commonly made during drug screening
- Immunoassay screening = detection of macro-molecules (like pregnancy tests, Corona tests or in forensic toxicology to determine groups or ‘families’ of drugs)
- problem: lots of false positives
- simple colour-change tests which change colour if a certain absorbance (Abs) is reach
- forensic toxicology students were give Abs-cut-off and Abs-values of 5 post mortem cases
- did test to determine opiate-type drugs
- were give either no context (control) or context that the deceased was a heroin user
- were more conclusive if they had context and more inclined to determine that it was in fact to confirm what they new from the context
- the context group had a higher error rate and lower accuracy
- -> context biases judgment about conclusion of a test
- student not practicing expert population
- in real life there is usually always some context given
Hamnett, H. J., & Dror, I. E. (2020). The effect of contextual information on decision-making in forensic toxicology.
Study 2
Aim
Methods
Results
Limitations
- this examined how context could already influence the choice of test used
- again toxicology students
- were given multiple post-mortem cases
- had to decide which test to use
- were given either no context info or demographics (age and ethnicity)
- demographic information affected the choice of tests
- rule of thumb:
> older people, medicinal drugs tested
> younger people, drugs of abuse tested
–> context info has a biasing effect on the choice of test experts use - student not practicing expert population
- in real life there is usually always some context given
