Presentation is loading. Please wait.

Presentation is loading. Please wait.

GENERIC PRINCIPLES FOR SELECTING DATABASES TO REPRESENT THE BACKGROUND POPULATION Heidi Eldridge*, Prof. Colin Aitken and Dr. Cedric Neumann.

Similar presentations


Presentation on theme: "GENERIC PRINCIPLES FOR SELECTING DATABASES TO REPRESENT THE BACKGROUND POPULATION Heidi Eldridge*, Prof. Colin Aitken and Dr. Cedric Neumann."— Presentation transcript:

1 GENERIC PRINCIPLES FOR SELECTING DATABASES TO REPRESENT THE BACKGROUND POPULATION Heidi Eldridge*, Prof. Colin Aitken and Dr. Cedric Neumann

2 Where did this group come from? ■Juror understanding of quantitative evidence and ability to reason logically ■Reporting and presenting quantitative and qualitative information on forensic evidence in the court room (Ranadive, Neumann, Kaye, Eds). http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2690899 http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2690899 ■New goal: Understanding/Studying the criteria to be satisfied for a database to be acceptable to scientists and the courts –Example: footwear database – what is the ‘relevant population’ for the denominator? –Geographical distributions, temporal distributions ■Can a list of generic principles be conceived to provide guidance in selecting the appropriate database type for different circumstances?

3 Court precedents for database selection controversy ■R v T (UK 2010) –Footwear; Use of LR to evaluate strength of evidence was challenged –Conviction quashed –Questioned analyst’s selection of “shoes received by the FSS” in contrast to “shoes in the general UK population” as the relevant population of shoes –Illuminates the question: who is part of the relevant population? All shoes sold? All mens’ shoes? All mens’ shoes in a police database? All trainers? –Illuminates the question: who defines what is the relevant population!

4 Court precedents for database selection controversy ■Passino (Vermont 1993) –DNA; defendant of Native American/French/Italian ancestry –DNA evidence ruled inadmissible: “it is unclear which, if any of the FBI’s databases is appropriate for calculating the probability of a coincidental match” –RMP is of interest under the hypothesis that the trace was NOT left by the defendant, in which case his ethnic background is irrelevant 1 –Under the hypothesis that the defendant DID leave the trace, random members of the population are irrelevant 1 Weir and Evett (1992) Whose DNA? Am. J. Hum. Genet. S0:869.

5 Quantifying the weight of forensic evidence

6 Distribution of traces left by suspects when they are truly the offender (given X) Distribution of control material provided by suspects when they are truly the offender Distribution of traces left by offenders at crime scenes (here Y is independent from X) Distribution of control material provided by potential offenders who turn out to be innocent

7 Quantifying the weight of forensic evidence Distribution of control material provided by suspects when they are truly the offender (given Y) Distribution of traces recovered at crime scenes when the suspect is the true offender Distribution of control material provided by individuals suspected of crimes and who turn out to be innocent (here Y is independent from X) Distribution of traces recovered at crime scenes when the suspect is innocent

8 Populations needed to assign these LRs ■Multiple needs –Innocent suspects (IS) –True offenders (TO) –Crime scene data ■But this is not the end of the story. We need to consider: –Who are the individuals that can reasonably be suspected in the considered case? –Who are the individuals that can reasonably truly have committed the considered case? –What is the ”population” of crime scenes?

9 Populations needed to assign these LRs ■What can influence these elements: –The set of propositions put forward by prosecution and defense –Case circumstances –Type of crime –Type of evidence –Other considerations –Geographical –Temporal ■Can we think of examples where varying one of these affects the choice of database?

10 Case Circumstances A wife is found murdered in her own home ■A shoe print is found at the scene ■Family/friends suspected A wife is found murdered in her own home ■A shoe print is found away from scene at shooter location (i.e., outside of premises) ■Contract killing suspected

11 Case Circumstances A wife is found murdered in her own home ■A shoe print is found at the scene The case is in NYC A wife is found murdered in her own home ■A shoe print is found at the scene The case is in North Dakota A wife is found murdered in the street ■A shoe print is found at the scene The case is in NYC

12 Type of Crime A wife is found murdered ■A shoe print is found at the scene A burglary is committed ■A shoe print is found at the scene

13 Type of Evidence A wife is found murdered ■A shoe print is found at the scene ■A bullet is found at the scene A wife is found murdered ■A fingerprint is found at the scene ■DNA is found at the scene Does the choice of reference population matter?

14 Type of Evidence A wife is found murdered ■A shoe print is found at the scene ■A bullet is found at the scene A wife is found murdered ■A fingerprint is found at the scene ■DNA is found at the scene Does the choice of reference population matter?

15 Type of Evidence A wife is found murdered ■A shoe print is found at the scene ■A bullet is found at the scene A wife is found murdered ■A fingerprint is found at the scene ■DNA is found at the scene Does the choice of reference population matter? It depends on the variability in reference populations for the evidence type

16 Other Considerations ■Local v. transient –DNA: US or Caribbean island or Japan? –Shoes: Hawaii or mainland US? ■Offender related vs. Innocent suspect databases 2 –What if they come from the same population? –Are the police arresting vastly wrong people? –Are victims considered “innocent suspects” of leaving the trace? 2 Champod, C., Evett, IW., Jackson, G. (2004) Establishing the most appropriate databases for addressing source level propositions. Science and Justice. 44(3):153-164.

17 General principles ■The relevant database should be selected based upon what is known about evidence type, the true offender, or the case information, not the suspect ■What is known or believed about the true offender must be considered before we look at the forensic evidence. –These knowledge/beliefs may be shaped by case information, or the defense theory of the crime ■Whether choice of database (IS, TO, etc) is affected by evidence type is dependent upon the distribution of the evidence type across different populations the type of crime, the type of evidence and geographical/temporal information ■And the work continues…

18 CEDRIC.NEUMANN@SDSTATE.EDU HELDRIDGE@RTI.ORG CGGA@ED.AC.UK


Download ppt "GENERIC PRINCIPLES FOR SELECTING DATABASES TO REPRESENT THE BACKGROUND POPULATION Heidi Eldridge*, Prof. Colin Aitken and Dr. Cedric Neumann."

Similar presentations


Ads by Google