read Dalhman's paper on weight

[marcellodibello]

link to dahlman paper on weight

https://lucris.lub.lu.se/ws/portalfiles/portal/140710068/Dahlman_Nordgaard_Information_Economics_in_the_Criminal_Standard_of_Proof_LPR_elektroniskt_s_rtryck_.pdf

below quotations from the article that summarize its main claims

[marcellodibello]

information gaps / information deficiency

Decision makers are sometimes troubled by information deficiency, sensing that their probability assessment is based on insufficient information, and wishing to gather more information before committing to a decision. ... Information gaps come in different forms. It can be a question that was not asked to a witness, or a witness that was not called to testify at all. It can be a forensic trace that was not investigated at the crime scene, or an alternative perpetrator that was never investigated by the police. (p.3)

[marcellodibello]

case specific evidence missing versus sampling statistics

In the Case of the Tin Box the information gap concerns case specific evidence (fingerprints or DNA on the tin box). In the Case of the Missing Fingers there is an information deficiency with regard to the sampling of reference data (people affiliated with ISIS).

nothing can be done about missing evidence v. something can be done

The two cases also illustrate another difference. In the Case of the Tin Box, it is too late to remedy the information gap. In the Case of the Missing Fingers, it is practically possible to gather more information. (p. 6)

[marcellodibello]

Two purposes of sufficient informativeness

Should the requirement of sufficient informativeness be restricted to investigations that are still possible at the trial, when the requirement is applied as a part of the criminal standard of proof? Or should it include all investigations that were available to the police and prosecution at one time or other? If the purpose of the requirement of sufficient informativeness is only to minimize the risk that investigations after a conviction produces new evidence that re-opens the case and exonerates the defendant

If, on the other hand, the purpose is also to incentivize the police and prosecution to optimize informativeness by acquitting the defendant as a ‘punishment’ for sloppy police-work in cases where a cost-efficient investigation was overlooked, sufficient informativeness should include investigations that are no longer possible. The acquittal in The Case of the Tin Box is an example of the latter view. (p. 10)

[marcellodibello]

probability of switching the decision from conviction to acquittal

The probability that an investigation will lead to a switch from conviction to acquittal, Pð Þ Sc!aji , is the probability that the probability of the factum probandum given the evidence produced by the further investigation does not meet the required threshold. This is a probability on a probability, and it is therefore referred to in probability theory as a second order probability (p. 11)

[marcellodibello]

Single value forensic reporting or intervals?

We partially agree and partially disagree with both sides. We agree with the advocates of single value reporting that the first-order probability of a random match and the resulting likelihood ratio should be reported as a single number. A degree of belief is susceptible to a Dutch Book if it is represented as an interval.

But we agree with critics of single-value reporting that the forensic expert should report more than this. The report should also help the legal fact-finder to assess the informativeness (p. 18)

report also sufficient informativeness, whether probability could change in light of further inquiry

We therefore propose that a forensic report should help the fact-finder assess sufficient probability of the factum probandum as well as sufficient informativeness. With regard to sufficient probability of the factum probandum, it should report the random match probability and likelihood ratio for the forensic evidence in question in the form of a single number. With regard to sufficient informativeness, the report should survey further investigations and report (1) their cost, and (2) what results they have the potential to produce. (p. 19)

[marcellodibello]

switching and sufficient informativeness

We have demonstrated that sufficient informativeness is fundamentally a question of information economics and switch-ability. In our model, sufficient informativeness is a cost-benefit-analysis of further investigations that involves a prediction of the possibility that such investigations will produce evidence that switches the decision from conviction to acquittal.

Critics of the Bayesian approach to legal evidence have claimed that ‘weight’ and sufficient informativeness cannot be captured in a Bayesian model. Our model shows that this is wrong. As we have demonstrated, sufficient informativeness can be modelled as a second order probability. (p. 22)