View
0
Download
0
Category
Preview:
Citation preview
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 1
© Graham Jackson, Adam Baines 2016
Graham JacksonAdvance Forensic Science
and Abertay University, Dundee
The use of databases in footwear mark cases
Royal Statistical SocietyErrol Street, London15th March 2016
g.jackson@abertay.ac.ukadvanceforensicscience@gmail.com
1Footwear databases v1.8
Adam BainesForensic Specialist
Lancashire Constabulary
Adam.Baines@lancashire.pnn.police.uk
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 2
Outline
• What is the purpose of databases?
• Formal classification of databases
• What databases are available for issues involving footwear evidence?
• Two case-‐studies -‐ interactive
• Summary Not an authoritative, exhaustive review of footwear databases
Share ideas and explore issues
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 2
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 3
What is the purpose of databases?
• Provide information to assist the expert• Provide investigative leads• Classify material• Answer questions of interest
• Help the expert to arrive at an opinion in relation to an issue
• Help the expert to assign probabilities
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 4
What are the types of issue that face experts?
European Network of Forensic Science Institutes (ENFSI)
ENFSI Guideline for Evaluative Reporting in Forensic Science (2015)
http://enfsi.eu/sites/default/files/documents/external_publications/m1_guideline.pdf
Association of Forensic Science Providers (AFSP)
Standard for the formulation of evaluative forensic science expert opinionAFSP. Standard for the formulation of evaluative forensic science expert opinion, Science and Justice2009; 49: 161-‐164
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 3
© Graham Jackson, Adam Baines 2016
Both documents recognise that forensic scientistsalso contribute to, and offer opinions relating to,
investigative issues…
The AFSP Standard and the ENFSI Guideline focus on the provision of ‘evaluative’ opinion…
5Footwear databases v1.8
… through the evaluation of a likelihood ratio,preferably for ‘activity’ level issues
… but what type of opinions could, and should, scientists offer in investigative situations?
© Graham Jackson, Adam Baines 2016
Classification of opinion:‘Investigative/Evaluative’
Based on a probabilistic, Bayesian paradigm
‘Hierarchy of issues’
6Footwear databases v1.8
The AFSP Standard and ENFSI Guideline embrace two key concepts that formed the basis of an earlier model –
‘Case Assessment and Interpretation’ (C.A.I.)
Jackson, G, Aitken, CA and Robert P. 2015. Case Assessment and Interpretation of Expert Evidence (Practitioner Guide No.4). Royal Statistical Society, London http://www.rss.org.uk/RSS/Influencing_Change/Statistics_and_the_law/Practitioner_guides/RSS/Influencing_Change/Current_projects_sub/Statistics_and_the_law_sub/Practitioner_guides.aspx?hkey=2cfdf562-‐361e-‐432e-‐851b-‐ef6ff5254145
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 4
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 7
Types of issue
Categorical
Likelihood ratiosExplanations
Posterior probabilities
‘Evaluative’‘Investigative’
There are benefits and limitations to each of these types of opinion
Types of opinion
Types of opinion
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 8
Categorical
Likelihood ratiosExplanations
Posterior probabilities
‘Evaluative’‘Investigative’
Formal notation for thetypes provides guidance on what data and knowledge may help
the expert
Types of issue
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 5
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 9
Categorical
Explanations
Posterior probabilities
‘Investigative’
[H1|E,I],[H2|E,I]…[Hn|E,I]
Pr[H1|E,I],Pr[H2|E,I]…Pr [Hn|E,I]
Pr[H1|E,I] = 1Pr[H1|E,I] = 0
Types of opinion
Types of issue
© Graham Jackson, Adam Baines 2016 10Footwear databases v1.8
Evaluative
Pr 𝐸 𝐻%, 𝐼Pr 𝐸 𝐻(,𝐼
Likelihood ratios
Types of opinion
Types of issue
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 6
What level of questionis being asked?
Evaluative
© Graham Jackson, Adam Baines 2016
Did he commit the offence?
Did he do the activity?
Is he/his item the sourceof the trace material?
Is he/she the (sub)-‐sourceof the test result?
11Footwear databases v1.8
... in a judicial context
‘Hierarchy of issues’
What level of questioncan be addressed?
Evaluative
© Graham Jackson, Adam Baines 2016
Did he do the activity?
Is he/his item the sourceof the trace material?
12Footwear databases v1.8
... in a judicial context
Most footwear issues relate to whether the suspect’s shoe left the mark.
Occasionally, the issue relates to activity, e.g. kicking vs treading
‘Hierarchy of issues’
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 7
Investigative
© Graham Jackson, Adam Baines 2016
What is the sourceof the trace material?
What activity took place?
What offence was committed?
What is the (sub)-‐sourceof the test result?
13Footwear databases v1.8
‘Hierarchy of issues’
Who committed the offence?Which other offences has offender committed?
Investigative
© Graham Jackson, Adam Baines 2016
What is the sourceof the trace material?
What activity took place?
14Footwear databases v1.8
‘Hierarchy of issues’
Most issues in footwear relate to what type of footwear could have left this mark?
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 8
© Graham Jackson, Adam Baines 2016 15Footwear databases v1.8
What databases and information are available?
Who stores data/information about footwear?
• Manufacturers• Shops/Distributors• SATRA• Police Forces• National Footwear Reference Collection
(NFRC)• National Footwear Database (NFD)• Forensic Providers
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 16
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 9
What data/information do they collect?Manufacturers
• Pattern designs• Sizes• Moulds used• Numbers produced
Source of the most comprehensive and definitive information about any particular shoe design. It could be used for frequency statements such as:“The suspects shoe is a Nike Air Max ‘95, UK size 8.5 and has been produced in mould ‘C’. 500,000 soles have been produced in this mould. This particular style is available in 61 different colour variations”© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 17
Shops/Distributors• Distribution figures• Sales figures
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 18
What data/information do they collect?
Source of frequency data such as:200,000 of this particular type of trainer were distributed around the UK.
Taking into account the total number of shoes distributed around the UK, the relative frequency of the trainer could be assessed as 1/10 -‐ a gross simplification.
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 10
SATRA• Size information
• Male• Female
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 19
What data/information do they collect?
The data are presented in percentages, e.g. 30% of women are size 6. The statistics are derived from a sample of approx. 900 women
The data are based on foot measurements and not shoe size. This is because people often wear an incorrect shoe size for their feet. SATRA have said that women tend to wear shoes 1 size too small for them and men tend to wear shoes 1 size too big.
Police Forces/Forensic Providers
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 20
What data/information do they collect?
Suspect database• Names/DOB• Date of arrest• Crime• Shoe sole pattern • Shoe size• Gender Specific data• Frequency
• Pattern• Size
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 11
Police Forces/Forensic Providers
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 21
What data/information do they collect?
Suspect database
The data are derived from the shoes of people suspected of crime.In the case of databases held by Police Forces, the data would be specific to the Force’s geographic area and, for some Forces, could even be broken down into specific Divisional areas.Databases compiled by Forensic Science Providers would be populated by the footwear from all of the Forces submitting work to the FSP. Therefore it would cover a larger area, possibly national.
Crime Scene database• Address• Date of Offence• Crime• Sole patterns identified• Frequency of sole patterns
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 22
Police Forces/Forensic Providers
What data/information do they collect?
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 12
The data are derived from footwear marks identified and collected at crime scenes. The data do not include footwear marks that could not be identified or those that weren’t recovered. The database does not take into consideration whether the identified marks are marks made by offenders’ footwear.
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 23
Police Forces/Forensic Providers
What data/information do they collect?
Crime Scene database
Reference Collection• Sole pattern• Size• Wear• Acquired features
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 24
Police Forces/Forensic Providers
What data/information do they collect?
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 13
The reference collection is often used to inform the significance of a comparison. For example, if a suspect’s shoe appears to have an unusual pattern of wear, the Reference Collection could be used to assess how many other shoes of the same sole pattern and size also show this particular wear pattern.
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 25
Police Forces/Forensic Providers
What data/information do they collect?
Reference Collection
National Footwear Reference Collection (NFRC)
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 26
Database• Shoe sole patterns• Frequencies
• Crime data• Custody data
What data/information do they collect?
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 14
National Footwear Database (NFD)• Suspect data
• Names/DOB• Date of arrest• Crime• Sole pattern• Shoe size
• Crime Scene data• Address• Date of Offence• Crime• Sole patterns identified
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 27
What data/information do they collect?
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 28
• Police called to the home of an elderly woman• A neighbour had found her body in the front living room• The woman had been physically, sexually assaulted• Point of entry appeared to be via a forced, ground-‐floor,
transom window to the rear of the property• Crime scene examiners observed a fresh-‐looking footwear
mark in the flower bed immediately below the forced window• Photographs and a cast of the mark were taken
Investigative Case Study
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 15
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 29
What is the issue(s)?
What type of opinion would be appropriate?
• What type of shoe left the mark?• What type of person left the mark?
Explanations?
Posterior probabilities?
Categorical?
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 30
• Shoe had a plain, or very finely-‐patterned, broad sole
• Most likely size was UK size 7
• Most probable style would be a suede boot
• Likely worn by an ‘older’ person
Opinions after lab examination
Posterior probabilities
What prior hypotheses? What prior probabilities?
What probabilities for the observations?
What data and knowledge?
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 16
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 31
Evaluative Case Study
• A stolen car was used to transport three offenders away from the scene of a robbery
• The car is abandoned about 1 mile away from the scene• Crime Scene Examiners observe a footwear mark in soil on the
brake pedal of the car. They believe it to have been left by the last person to have driven the car
• The owner of the car will say that he did not leave the mark• A gel lift of the mark is taken
• A suspect was arrested about three hours after the incident• The suspect’s footwear was taken• The suspect denied all involvement with the crime and denied
being the driver of the car
What is the issue(s)?What type of opinion would be appropriate?
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 32
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 17
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 33
© Graham Jackson, Adam Baines 2016
Evaluation of the observation of a ‘match’ (E) at ‘source’ level
HP -‐ The suspect’s shoe is the source of the markHD -‐ Some other, unknown shoe is the source of the mark
Assumption of numerator approaching 1
A relative frequency of what?
Is the suspect’s shoe the sourceof the mark? Evaluative Case Study
Footwear databases v1.8 34
𝐿𝑅 =→ 1
𝑅𝑒𝑙. 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦
𝐿𝑅 =Pr [𝐸|𝐻%, 𝐼]Pr [𝐸|𝐻(, 𝐼]
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 18
© Graham Jackson, Adam Baines 2016
HP -‐ The suspect’s shoe is the source of the markHD -‐ Some other, unknown shoe is the source of the mark
Is the suspect’s shoe the sourceof the mark?
Evaluative Case Study
What sort of data would give a useful relative frequency and thereby inform the denominator probability?
What would be a relevant population in this case?
Footwear databases v1.8 35
𝐿𝑅 =Pr [𝐸|𝐻%, 𝐼]Pr [𝐸|𝐻(, 𝐼]
=1
𝑅𝑒𝑙. 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦
© Graham Jackson, Adam Baines 2016
Biedermannet al (2012) refer to the analysis in Champod et al (2004)
Footwear databases v1.8 36
𝐿𝑅 =Pr [<𝑀𝑎𝑡𝑐ℎ<|𝐻%, 𝐼]Pr [<𝑀𝑎𝑡𝑐ℎ<|𝐻(, 𝐼]
=1
𝑅𝑒𝑙. 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦
𝐿𝑅 =1
Pr [𝐸A |𝐻(, 𝐼A ]×Pr [𝐸C|𝐻%, 𝐼A , 𝐼C]Pr [𝐸C|𝐻(, 𝐼C]
Informed by data on scene marks
Informed by data on offenders like the defendant
Informed by data on innocent people like the defendant
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 19
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 37
Mark on pedalSuspect’s shoe.
NFRC Pattern code: Ugg 17
Population Total UGG 17 %* LR**Suspect 12089 23 0.2 500Scene 4552 21 0.5 200
* Approximate figures** Assuming Numerator =1
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 38
Mark on pedalSuspect’s shoe.
NFRC Pattern code: Ugg 17
Witnesses report the driver who ran away from the abandoned vehicle was a “small, young woman”.
The suspect is 18 years old, 5’4”.
HP -‐ The suspect’s shoe is the source of the markHD -‐ Some other, unknown shoe is the source of the mark
HP -‐ The suspect’s shoe is the source of the markHD -‐ Some other, unknown shoe, worn by a “small, young
woman”, is the source of the mark
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 20
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 39
Population Total UGG 17 %* LR**Suspect 12089 23 0.2 500Women 1097 23 2 50Scene 4552 21 0.5 200
Mark on pedalSuspect’s shoe.
NFRC Pattern code: Ugg 17
Population Total UGG 17 %* LR**Suspect 12089 23 0.2 500Scene 4552 21 0.5 200
* Approximate figures ** Assuming Numerator =1
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 40
Suspect’s shoe.NFRC Pattern code: Adidas 509
Population Total Adidas 509 %* LR**Suspect 12089 1343 11 9Women 1097 19 2 50Scene 4552 847 19 5
Population Total UGG 17 %* LR**Suspect 12089 23 0.2 500Women 1097 23 2 50Scene 4552 21 0.5 200
* Approximate figures ** Assuming Numerator =1
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 21
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 41
Summary
• What is the purpose of databases?• Provide information on prior hypotheses• Provide data to help assign prior probabilities for hypotheses• Provide data to help assign probabilities for observations
• Formal classification of databases• ‘Investigative’• ‘Evaluative’
Share ideas and explore issues Not an authoritative, exhaustive review of footwear databases
g.jackson@abertay.ac.ukadvanceforensicscience@gmail.com
Adam.Baines@lancashire.pnn.police.uk
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 42
Summary
• What databases are available for issues involving footwear evidence?• Commercial• Police Forces• Forensic Science Providers• National
• Two case-‐studies• Importance of specifying the issue and the propositions• Impact on LR
Share ideas and explore issues
g.jackson@abertay.ac.ukadvanceforensicscience@gmail.com
Adam.Baines@lancashire.pnn.police.uk
Not an authoritative, exhaustive review of footwear databases
Footwear databases 15-3-2016 21/03/2016
© Graham Jackson, Adam Baines 2016 22
© Graham Jackson, Adam Baines 2016 Footwear databases v1.8 43
Thanks for attention and contributions
Hope it’s been of interest to you
g.jackson@abertay.ac.ukadvanceforensicscience@gmail.com
Adam.Baines@lancashire.pnn.police.uk
Recommended