The math task (5) Flashcards
When to decide indecision?
Dror, I. E., & Langenburg, G. (2019). “Cannot Decide”: The Fine Line Between Appropriate Inconclusive Determinations Versus Unjustifiably Deciding Not To Decide.
- How is a decision reached?
- 4 issues with concluding inconclusive
- Solutions: when to decide inconclusive?
- if you have a categorical decision like in fingerprint analysis, you can decide
-> identification (sufficiently high degree of corresponding features)
-> exclusion (discordances between the two fingerprints are sufficiently high)
OR inconclusive (can neither decide) - humans do not look at all the evidence the start (either form a biased or neutral position) accumulate the evidence and way it
- whenever on threshold is reached decision for identification or exclusion is made
- if after a while no threshold is met decision: inconclusive (even if still not all evidence was examined)
- it can be a sign of caution but there are some issues…
1) inconclusive is a decision with certainty that the quantity and quality of information are not sufficient AND NOT “well, can’t decide)
2) it could be abused for a “way out” –> never seen as wrong and seldom challenged
3) loss of data (might be near threshold)
4) if sufficient data is there, one must decide! - avoid over-use (blind verification, multiple examiners)
- need for justifying the decision
- not categorical, but probable (use stats)
- use of good assessment tools, good technology (reduces bias –> see task 6)
Do legal decision makes understand the way conclusions are presented?
De Keijser, J., & Elffers, H. (2012). Understanding of forensic reports by judges, defense lawyers, and forensic professionals.
AIM
METHOD
RESULT
(LIMITATION)
- Examine the extent and nature of (mis)understanding of technical forensic reports
- Judges and justices, defence lawyers and employees of the Dutch Forensic Institute (NFI)
- were given fake forensic reports concerning a robbery
- participants either got a version with a verbal likelihood ration or with a visual representation of the likelihood
–> both were logical sound (no prosecutors fallacy) - measure were their self-reported understanding and their actual understanding
> after reading the report, had to select the correct statements about likelihoods –> possible to answer logically sound, but also unsound = prosecutors fallacy (siehe Aufgabe 7 dieses Tasks) - for none of the groups did presenting a visually expressed likelihood ratio improve understanding of the conclusion
- the actual understanding of jurists was quite poor, due mainly, but not exclusively, to the prosecutor’s fallacy
- however, their self-reported understanding was reportedly high (weak linkage)
- experts outperformed jurist by large but made mistakes themselves
- -> jurist do not seem to like the new, logical sound way and want back to “the good old days” with expressing the likelihood of the source given the evidence (prosecutor fallacy) and categorical statements
- -> they are not such fans of the Bayesian reasoning (weighing the evidence and assigning a likelihood for both hypothesis: guilty/not guilty)
De Keijser, J., Malsch, M., Luining, E. T., Kranenbarg, M. W., & Lenssen, D. J. (2016). Differential reporting of mixed DNA profiles and its impact on jurists’ evaluation of evidence. An international analysis.
AIM
METHOD + RESULT of study 1 & 2
CONCLUSION
- examine what substantive differences exist in interpretation and subsequent reporting of DNA evidence
1st
- mock case with mock DNA evidence (mixed profiles)
=> had ground truth (suspect=perpetrator)
- gave this to 19 different experts from around the 9=(wester) world
- had to interpret and write a report
–> variety in mode of conclusion between the 19 reports was large: how they came to the conclusion (technical explanations, context info etc.) and conclusions themselves differed
–> 1/2 of the report quite similar and use of likelihood ratio :)
–> but e few were quite extreme: excluded the suspect as contributor or overstated the likelihood
2nd
- picked the most differing from the 3 reports and gave them to jurists to read and interpret
- -> Dirksen is considered moderately incriminating (1.29) overall, De Boer just slightly exculpating (–0.46) and Ten Cate is considered moderately exculpating(–1.49) overall
- Ten Cate was favoured, had an activity level summary
–> what experts conclude, but also how the formulate this conclusion influences juridical decision making
Kahan, D. M., Peters, E., Dawson, E. C., & Slovic, P. (2017). Motivated numeracy and enlightened self-government.
The two theories about conflict about scientific evidence
Two theories to explain conflict about scientific evidence:
1) “science comprehension thesis”
–> defects in the public’s knowledge and reasoning capacities (numeracy) as a source of controversies
numeracy = ability to understand quantitative info
2) “identity’-protective cognitive thesis”
–> conflicts about scientific evidence as a result of differences in beliefs, values, and political ideology
- avoiding cognitive dissonance
- group identification and in-group vs. out-group dynamics –> controversies often about highly politicizes topic like climate change or gun-control
Kahan, D. M., Peters, E., Dawson, E. C., & Slovic, P. (2017). Motivated numeracy and enlightened self-government.
The Experiment
- Aim
- Methods
- Results
- Discussion
- determine which theory is correct about controversies in science
- 1111 US participants (big sample)
- they were questioned for ideology: 1/3 liberal, 1/3 moderate, 1/3 conservative
- first, they assessed participants numeracy
- they gave them two mock studies were the results were presented in a contingency table (1st: about skin treatment; 2nd gun-control)
- two conditions: either positive effect or contingencies were swapped and it presented a negative effect
- then they asked participants about the results of the study –> dependent variable: numeracy score
- for the skin treatment: people with high numeracy performed well in both conditions and those who were low in numeracy did not
- for gun-control: liberals even if they were high in numeracy performed bad if the paper reported a crime increase after the weapon-ban, for conservatives it was the other way around
- “science comprehension thesis” cannot fully explain those findings (partly correct)
- “identity’-protective cognitive thesis” –> seems to be proven that people choose how they learn and understand fact which are in line with their identify –> there is however some rationality behind it because it would otherwise hurt their self-image and their understanding of the world = expressive rationality
- dual process model and bound rationality cannot fully explain this!
