Director of Public Prosecutions v Tangaloa
[2013] VCC 621
•3 June 2013
| IN THE COUNTY COURT OF VICTORIA | Revised |
AT MELBOURNE
CRIMINAL DIVISION
Case Nos. CR-09-01829
CR-09-01830
CR-09-01831
CR-09-01832
| DIRECTOR OF PUBLIC PROSECUTIONS | |
| v | |
| TANIORA TANGALOA | Accused |
| AND | |
| JACK VAOTANGI | Accused |
| AND | |
| HENRY OWEN DAY | Accused |
| AND | |
| JASMIN DESTANOVIC | Accused |
---
JUDGE: | HIS HONOUR JUDGE STUART | |
WHERE HELD: | Melbourne | |
DATES OF VOIR DIRE: | 6-9 May 2013 | |
DATE OF RULING: | 3 June 2013 | |
CASE MAY BE CITED AS: | DPP v Tangaloa & Ors | |
MEDIUM NEUTRAL CITATION: | [2013] VCC 621 | |
RULING
---
Catchwords:
CRIMINAL LAW – DNA evidence – Admissibility of a likelihood ratio where a binary model is utilised and there exists a non-concordant allele of a person of interest – Evidence found to be inadmissible – Disclosure of matters which may affect the analysis of a DNA profile, the calculation and reliability of the likelihood ratio.
---
APPEARANCES: | Counsel | Solicitors |
| For the Crown | Mr A Albert | Office of Public Prosecutions |
| For the Accused Tangaloa | Mr M Sharpley | Valos Black & Associates |
| For the Accused Vaotangi | Mr T Bourke | Haines & Polites |
| For the Accused Day | Dr M Marich | Marich Legal Pty Ltd |
| For the Accused Destanovic | Mr W Toohey | Stephen Andrianakis & Associates |
HIS HONOUR:
The Application
1 This is an application by Mr Sharpley on behalf of his client Mr Tangaloa during his defence case to exclude DNA profile evidence and likelihood evidence that was presented to the jury during the prosecution case.
Background Facts
2 On the 15th of January 2009, the Accused Tangaloa together with three other co-Accused are alleged to have engaged in a home invasion at Stephen Jones’ premises. During the course of this home invasion, among other things, Mr Tangaloa is said to have produced a handgun which he placed into the victim’s mouth and pulled the trigger. The firearm did not discharge as there were no bullets in the breech. On the 5th of February 2009, Mr Tangaloa was arrested. He was wearing a bum bag which contained a loaded pistol.
3 All four of the accused are charged with aggravated burglary, intentionally causing serious injury and armed robbery.
DNA Analysis and the Likelihood Ratio
4 This pistol was subsequently examined and a trace or low level DNA profile was obtained from the gun muzzle. As occurs in these cases, two propositions were put forward. The first may be described as the ‘Prosecution Proposition’ that the victim Jones’ DNA was on the gun muzzle DNA sample; the second, ‘Defence Proposition’ that the victim Jones’ DNA was not on the gun muzzle DNA sample. It was statistically estimated that it was 1200 times more likely that the Prosecution Proposition was true, than the Defence Proposition which provided ‘strong’ support for the proposition that the victim’s DNA profile was on the gun muzzle. These propositions and the likelihood ratio were set out in “Exhibit 15” tendered in the trial.[1]
[1]See Annexure 1 to this Ruling.
Allele Dropout in Trace DNA Profiles
5 The DNA obtained from the gun muzzle was a very small amount and is described variously as ‘trace’, ‘low template’ or ‘low level DNA’. Forensic scientist Mr Maxwell Jones described the sample obtained from the gun muzzle as “…. this particularly low level case.”[2] On the other hand “Exhibit 16”[3] tendered in the trial provided a full profile of Stephen Jones’ DNA at all 10 sites. Of critical importance to this Ruling on the voir dire is that the full DNA profile of Stephen Jones at site D18S51, at alleles 12 and 15, there were peak heights of 329 and 232 Relative Fluorescence Units (“RFU”). RFU is in effect a measure of the peak height of these alleles. These measurements can be seen as the lowest figure in the rectangular boxes at site D18S51 alleles 12 and 15. One of those alleles would have been given by Mr Stephen Jones’ mother and the other by his father.
[2]See Annexure 2 to this Ruling – Transcript of the portion of Mr Maxwell Jones’ evidence at page 1351, line 24.
[3]See Annexure 3 to this Ruling.
6 In the gun muzzle profile “Exhibit 14”[4] tendered in the trial, a 98 RFU peak height is seen at allele 12 but no peak is seen at allele 15. This is referred to as an allele dropout which is common in profiles obtained from trace DNA. Had the DNA from the gun muzzle profile been a full profile, and no peak seen at allele 15, this would have excluded Stephen Jones as a contributor to the DNA profile.
[4]See Annexure 4 to this Ruling.
7 A forensic scientist, Mr Maxwell Jones, gave evidence before the jury that given the low peak height of 98 RFU at allele 12, it was unremarkable that there was no peak height at allele 15. The logic here is because it is a low level DNA profile, there will be dropouts at different alleles. That is, the other allele is assumed to be present but simply not detected because of the low level DNA in the sample. Mr Maxwell Jones pointed out that whereas the full profile of Mr Stephen Jones at allele 12 showed a peak height of 329 RFU, the low level DNA profile from the gun muzzle at that same allele 12 has a peak height of 98 RFU. According to Mr Maxwell Jones the absence of any peak at allele 15 at site D18S51 is explicable on the basis that the allele has dropped out.[5] The statistical analysis which then followed assumed that to be the case.
[5]See Annexure 5 to this Ruling – Transcript of the portion of Mr Maxwell Jones’ evidence at pages 1409 to 1410, 1415 to 1416.
The Law
8 Redlich JA stated in R v Berry:[6]
The likelihood ratio is now an accepted calculation which has the capacity to assist a jury. Based upon the mainstream of forensic science, an appropriate way for DNA mixtures to be evaluated is to consider competing hypotheses – usually the prosecution and defence case – and to assess the likelihood of the DNA evidence occurring if each hypothesis were true.
[6](2007) 17 VR 153 at 161.
9 Such evidence has on occasion been excluded for a variety of reasons including, “… there was no proper foundation upon which the jury could assess the reliability of the expert opinions or on the basis that the prejudicial effect of the evidence outweighed its probative value.” [7]
[7]R v Berry (2007) 17 VR 153 at 165 per Redlich JA.
10 Maurice J. observed in R v Lewis:[8]
For my part I think that whenever the Crown wishes to rely upon forensic evidence the prosecutor has a clear duty, not just to his client, the Crown, but to the trial judge and the jury to acquaint them, in ordinary language, through the evidence he leads, with those aspects of the expert’s discipline and methods necessary to put them in a position to make some sort of evaluation of the opinions he expresses. Where the evidence is of a comparatively novel kind, the duty resting on the Crown is even higher: it should demonstrate its scientific reliability. It is not an answer to considerations that dictate these things be done to say the defence may draw it out in cross-examination; that is an abdication of the Crown’s primary function in a criminal prosecution (my emphasis added).
[8](1987) 29 Australian Criminal Report 267 at 271.
11 If the methodology utilised has been rejected by the scientific community then the evidence may be excluded either because it is unreliable and therefore irrelevant, or in the exercise of a discretion to exclude the evidence.
The Scientific Debate
12 In the abstract to the article written by van Oorschot, Ballantyne and Mitchell, ‘Forensic Trace DNA: A Review’ Investigative Genetics (2010) 1-14, the authors stated that:[9]
[9]Note: This article provides a helpful overview of the history of trace DNA and summarizes some of the methods and techniques “used to collect, amplify and interpret the smallest samples encountered in forensic biology.”
DNA analysis is frequently used to acquire information from biological material to aid enquiries associated with criminal offences, disaster victim identification and missing persons investigations. As the relevance and value of DNA profiling to forensic investigations has increased, so too has the desire to generate this information from smaller amounts of DNA.
Later at page 11, the authors say that:
The increased discussion on the appropriate interpretation for low level DNA profiles has helped to move the field towards a more rigorous strategy for interpreting the evidence and presenting appropriate statistical measures to the courts. While this is undoubtedly a step in the right direction, there remains much work to be done. The incorporation into the LR (likelihood ratio) framework of more criminalistic aspects of trace DNA, such as the possibility of transfer, background contamination and deposition rates, will produce more conservative estimates and help to eliminate some of the critics’ concerns over the validity of trace DNA as a prosecutorial tool (bracketed section added).
13 In her report tendered as “Exhibit VD-10” on the voir dire, Ms Taupin makes the following observations:
There has been increasing debate in the forensic science community regarding the analysis, interpretation and meaning in the context of the case of trace DNA amounts. It is a complex field and the debate has not been resolved.
14 Mr Maxwell Jones appears to agree with the contents of the first sentence of the above quote.[10] Any doubt I had about the correctness of Ms Taupin’s assessment was removed after reading the above article. This is particularly so where, as here, the mixed trace DNA profile contains at least three contributors.
[10]See Annexure 6 to this Ruling – Transcript of Mr Jones’ evidence at pages 1296 to 1297, 1309 to 1313 and 1392 to 1394.
15 Based on this material and what follows, I consider that likelihood ratios produced from trace DNA profiles “… is of a comparatively novel kind.”[11]
[11]R v Lewis (1987) 29 Australian Criminal Report 267 at 271 per Maurice J.
Validity of the Statistical Analysis
16 The question before me became a refined one, namely, where there is an assumed dropout of an allele in a person of interest i.e. the victim Stephen Jones, can there be a statistical analysis producing a likelihood ratio where a binary model is utilised.
17 The distinction between a binary method and a probabilistic method is of critical importance in this area. Crudely put, an aspect of the binary method is that it utilises thresholds treating ‘… alleles as present or absent.’[12] For example, an allele below peak height 50 RFU would not be taken into account, or, where in a partial DNA profile, an allele which has a peak height greater than 250 RFU and the corresponding allele is not seen, then because of the height of the first allele, that is conclusive evidence that the second allele should be seen and if it is not, represents an exclusion of that person as contributing to the DNA profile. Putting it differently, the higher the peak height of the allele detected, the less likelihood of there being a dropout of the second allele, and the converse is also said to be so, namely, the lower the peak height RFU of the allele detected, the greater the likelihood that there has been dropout.
[12]See Annexure 7 to this Ruling – Hannah Kelly et al, ‘The Interpretation of Low-Level DNA Mixtures’ Forensic Science International: Genetics 6 (2012) 191-197.
18 The Victorian Police Forensic Services Centre (“VPFSC”) has used the binary model, the current model being Model 33 and is in the process of adopting the probabilistic model. As I have stated earlier this Ruling is confined to the admissibility of likelihood ratios where, as here, a binary model is used and there is a non-concordant allele of a person of interest being the victim.
19 The partial DNA profile from the gun muzzle was analysed by forensic scientist Mr Maxwell Jones, utilising the binary methodology set out in Model 33. Mr Jones contended that it was appropriate for him to assume allele 15 dropout at site D18S51 given the low RFU of 98 at allele 12 and proceeded to calculate the likelihood ratio. This was a contention which he vigorously maintained throughout his evidence on voir dire.
20 I also heard from a fellow VPFSC forensic scientist Ms Pollett who conducted a blind cross-check of Mr Maxwell Jones’ analysis. Counsel for the Defence, Mr Sharpley, called Ms Jane Taupin, forensic consultant. All three were accepted as expert witnesses in the field of DNA profiling and analysis.
21 Mr Sharpley’s cross-examination of Mr Maxwell Jones challenged his opinion in many respects. In the broad, it was suggested that the likelihood ratio was unreliable for a variety of reasons and should not have been calculated. The precise complaint which is the subject of this Ruling did not emerge clearly until the evidence of forensic consultant Ms Taupin called during the defence case of the accused Tangaloa. Her evidence was where there is assumed drop out of an allele of a person of interest then the probability of that drop out having occurred must be known. Further, the literature supports this contention and a binary model cannot address this issue and other issues whereas the new probabilistic methods can.[13]
[13]See Annexure 8 to this Ruling – Transcript of portion of Ms Taupin’s evidence at pages 1504 to 1505, 1509 to 1510, 1525, 1552 to 1556.
22 Ms Pollett gave evidence that Dr John Buckleton, an internationally renowned forensic statistician, had granted the VPFSC a dispensation from any such prohibition. Her evidence at pages 1664 to 1665 of the transcript:[14]
[14]See Annexure 9 to this Ruling – Transcript of portion of Ms Taupin’s evidence at pages 1664 to 1665.
Question: So, what do you say; as we are approaching, if it’s higher than 250 or lower than 250, what’s the situation?
Answer: The advice that we have had from John Buckleton suggests that if the peak is greater than 150 RFU then Method 33 should not be used.
23 Mr Maxwell Jones appeared not to know of the existence of that advice from Dr Buckleton.
24 As I understood it, the suggested exception or dispensation was that where the seen allele is less than 150 RFU then it is legitimate to assume allele dropout because of the relatively low peak height of the seen allele. If, however, the peak height of the seen allele is of, or greater than, 150 RFU, then allele dropout cannot be discounted and where the RFU of the seen allele is 250 or greater, it is conclusively presumed that there has been no allele dropout at the unseen allele. In that latter event, an RFU of, or greater than, 250 peak height at allele 12 of site D18S51, and no seen allele at allele 15 would conclusively exclude the victim Stephen Jones as a contributor to the DNA profile taken from the pistol muzzle.
25 During the course of the voir dire, I asked, on numerous occasions, to be provided with any written statement or authorisation to this effect by Dr Buckleton to VPFSC. None has been produced.
The Scientific Literature
26 A large quantity of materials from various international publications was provided to me during the course of the voir dire. In particular, two articles are of great moment in relation to this matter. In the first article, ‘Interpreting Low Template DNA Profiles’[15] page 5 of that learned article reads as follows:
[15]See Annexure 10 to this Ruling – David Balding and John Buckleton, ‘Interpreting Low Template DNA Profiles’ Forensic Science International: Genetics 4 (2009) 1-10.
One important conclusion, here and in what follows, is that if drop out is invoked to sustain the prosecution case then estimation of drop out probabilities cannot be avoided. DNA-based prosecutions that rely on drop out and do not explicitly estimate plausible ranges for the dropout rate parameter are, in our view, defective (my emphasis added).
The binary model cannot be utilized to calculate dropout probability.
The passage is in unequivocal terms. No where in that article is it suggested that any exemption or exception to that proposition can be countenanced. That article was published in 2009 and the learned authors are David Balding and Dr John Buckleton.
27 A second article jointly written by Kelly, Bright, Curran and Buckleton was published in 2012.[16] At page 191, the learned authors provide a definition of a non-concordant allele being:
[16]See Annexure 7 to this Ruling – Hannah Kelly et al, ‘The Interpretation of Low-Level DNA Mixtures’ Forensic Science International: Genetics 6 (2012) 191-197.
A non-concordant allele is one present in the POI (person of interest) that is not visualised in the epg (electropherogram) (my emphasis and bracketed sections added).
Later, at the same page, the learned authors continue:
The binary method treats alleles as present or absent, the semi-continuous method assigns a probability to the events of drop out or non-drop out but still treats alleles as present or absent. Fully continuous methods deal with the probabilities of stochastic events (like drop out) based on the heights of the peaks visualised at a locus. These methods improve in power, flexibility and elegance in the order binary, semi-and fully continuous. However the simplicity of binary methods retains much appeal and allows manual or semi-manual implementation. There is no modification of the binary method that can deal with a non-concordant allele in a comprehensive manner (my emphasis added). In this paper, we explore methods to extend the binary method to complex mixtures that have no non-concordant alleles when compared to the POI.[17]
[17]Note: The bracketed section i.e. “like drop out” appears in the original article.
At page 196 they continue:
In this work, we have extended the binary model using two distinct methods to deal with complex mixtures that have no non-concordant alleles. Both methods are able to take into account the possibility of drop out. We have compared the performance of these models with the misapplication of the unconstrained combinatorial method, UC (unconstrained combinatorial method). We stress that these methods will not hold if there are non-concordances. (my emphasis and bracketed sections added).
According to Mr Maxwell Jones, at least three of the four authors are internationally regarded as being forensic statisticians in this area of the highest order including Dr John Buckleton.
28 It was suggested by Mr Maxwell Jones that this article did not prohibit a statistical analysis of a DNA profile where there has been assumed allele dropout. It was suggested that the four eminent authors had made a definitional mistake or more accurately, there was a lack of precision in the definition of a non-concordant allele in the passage extracted above from that second article. Mr Maxwell Jones suggested that there is a distinction between an explicable non-concordant allele and an inexplicable non-concordant allele and further that the authors of the second article were referring to inexplicable non-concordant alleles of a person of interest.[18]
[18]See Transcript of Mr Jones’ evidence at pages 1750 to 1751.
29 That four eminent forensic statisticians should make so basic a mistake is improbable to say the least. For that reason and the reasons which follow, I reject that contention as having no substance. However even if there is substance to Mr Jones’ contention then I would be left in a situation where statements by internationally renowned forensic statisticians of apparent clarity becoming suddenly ambiguous. The end result would be that there was no clear statement within the scientific literature and no consensus within the scientific community on this matter. In that event I would reject the likelihood ratio calculation for that reason alone.
30 In the second article, the learned authors spoke to two scenarios where the binary method could be utilised. No third scenario was put forward by the learned authors in this article of the kind suggested by the oral advice said by Ms Pollet to have been given by Dr John Buckleton to the VPFSC.
31 As is plain from the extracted section, the article proceeds on the basis that where there is an allele dropout of a person of interest or a non-concomitant allele, no likelihood ratio could be calculated. This is entirely consistent with the passage quoted above from the first article.
32 The learned authors of the second article conclude at page 196 with the following:
It is important to highlight that none of these methods make full use of the available information. They are effectively methods to extend the working life of the binary model but better models are now becoming available. More intelligent models that can accommodate an assessment of the probability of dropout and drop in offer a way forward that makes better use of the available data. These could include semi-continuous models like LoComatioN or fully continuous like TrueAllele. We are also aware that the Forensic Science Service has a fully continuous model in development (DNA Insight), but this is not currently commercially available. These models, in our opinion are better suited for the interpretation of LtDNA (low template DNA) profiles than the binary model (bracketed sections added).
33 It is to be noted that Dr Buckleton is a co-author of both these articles.
34 Furthermore, during the voir dire, I asked to be referred to any scientific literature which did or might suggest that the dispensation given to VPFSC by Dr Buckleton was recognised as appropriate when dealing with a binary model such as Model 33. I was directed to no such literature by any of the three experts. At page 1759 of the transcript during the course of the voir dire, I asked the following two questions:
Question: Mr Jones, is there anything you can point to me, and I have asked you this before, since your reading – since your evidence that contradicts the statements in these two articles, other than the evidence of Ms Pollett, that Dr Buckleton has for your laboratory specified that a RFU of 150 as being the cut off after which for greater values the model 33 should not be used?
Answer: I haven’t seen any publication about that at all.
Question: You have not?
Answer: No.
This absence reflects adversely upon the contentions put forward by Mr Jones.
35 The learned Prosecutor Mr Albert and his instructing solicitor have made every endeavour to make contact with Dr Buckleton in order to obtain his expert opinion and comment upon the suggested exemption referred to in the evidence of Ms Pollett. Whilst they have been able to make contact with Dr Buckleton, for a variety of reasons, they have not been able to obtain his assistance. In these circumstances, I must act on the evidence and materials before me.
Ms Taupin’s and Ms Pollett’s Expert Opinions
36 It was Ms Taupin’s expert opinion consistent with this analysis that in this case, no likelihood ratio could have been produced. Ms Pollett, absent the suggested dispensation from Dr John Buckleton, agreed that this was so. Ms Pollett agreed the literature supports that contention and there is nothing in the literature which suggests otherwise.[19]
[19]See Annexure 11 to this Ruling – Transcript of portion of evidence of Ms Pollett at pages 1666, 1673, 1693 to 1695, 1698 to 1699 and 1717 to 1718.
Conclusion
37 Based on the evidence and materials before me, I reject the suggestion that the scientific community accept that it is valid for a likelihood ratio to be calculated where there has been an assumed allele dropout of a person of interest or to use a different terminology, where there is a non-concordant allele of a person of interest.
38 Even if I am wrong about that, there is no evidentiary material other than from Mr Maxwell Jones (and to a limited extent from Ms Pollett) that suggests that the calculation of a likelihood ratio in these circumstances is statistically reliable. Or put differently, the Prosecution have not demonstrated the scientific reliability of the proffered likelihood ratio.
39 For these reasons, I rule that the evidence of the likelihood ratio is inadmissible.
The Need for Disclosure
40 During the course of his evidence, forensic scientist Mr Maxwell Jones indicated that it would be helpful for there to be some guidance from the Court about matters which should be disclosed in statements prepared by forensic scientists employed at the VPFSC for Court purposes. The matter arose because despite the profile obtained from the gun muzzle being a partial profile, inadvertently that fact was not disclosed in Mr Jones’ statement. Indeed, within the statement, appears the following: “Note: Complete DNA profiles were obtained throughout unless otherwise mentioned.”
41 The often quoted observations of Maurice J. in R v Lewis are apposite not only to the Crown’s duty but also to scientists reporting their findings and opinions in criminal cases.
42 The need for disclosure of any underlying assumptions and methodology is illustrated by the following examples:
a) As revealed in the above reasons it was assumed that there was allele dropout of the person of interest at site D18S51 allele 15. In those circumstances the binary model cannot be used to calculate a likelihood ratio.
b) In this case the profile evidenced there were three contributors to the DNA and it was assumed there was not a fourth contributor. All three experts agreed however that it was possible there was a fourth undetected contributor. In Mr Jones’ statement concerning the sample taken from the gun muzzle area, he stated, “…A mixed DNA profile was obtained consisting of DNA originating from at least three people.” If the assumption was not made then a binary model could not be utilised and no likelihood ratio calculated.[20] This assumption may not be warranted.[21]
c) The likelihood ratio of 1,200 calculated by Mr Jones in 2010 was different from one earlier calculated by forensic officer Skye Thorpe of 36,000 in 2009. This 30 fold decrease came about as a result of Mr Jones using Method 33. Thus changes to threshold levels and other statistical evaluation methodology can have a significant effect on the likelihood ratio.
[20]See Annexure 12 to this Ruling – Transcript of portion of evidence of Mr Maxwell Jones evidence at pages 1249 to 1250, 1422 to 1424.
[21]David R Paoletti et al, ‘Empirical Analysis of the STR Profiles Resulting from Conceptual Mixtures’ (2005) Journal of Forensic Science 50 and a second article quoting from the first article in Jaheida Perez et al, ‘Estimating the number of contributors to two-, three- and four-person mixtures containing DNA in high template and low template amounts’ (2011) Forensic Science 314.
43 It is essential for Prosecutors, Defence Lawyers and the Courts to be fully aware of any matter which may affect the analysis of a DNA profile, the calculation of the likelihood ratio and the reliability of a proffered likelihood ratio.
44 Any assumptions made must be plainly stated. Any change to the analysis of the allele profile or the calculation of the likelihood ratio if such assumptions are not made must also be stated. The existence of non-concordant alleles of a person of interest in a low level DNA profile must be stated.
Acknowledgments
45 I wish to acknowledge the assistance that I have received from both Counsel and all three expert witnesses during the course of this voir dire. In particular, Mr Maxwell Jones has gone to great lengths to endeavour to ensure that I have a sufficient understanding of the scientific processes of DNA analysis and the calculation of likelihood ratios for the purpose of this Ruling.
Discharge of the Jury
46 Evidence and submissions having been concluded on the 9th of May 2013, I announced on the following day that I would rule the likelihood ratio calculation to be inadmissible and that I would deliver my reasons at a later date. I was of the view that in the circumstances, the jury had to be discharged albeit in the closing days of a relatively long trial. All Counsel including the Learned Prosecutor submitted that the jury should be discharged. I then discharged the jury.
Annexure 1
“Exhibit 15”
Annexure 2
“Transcript of Maxwell Jones”
Annexure 3
“Exhibit 16”
Annexure 4
“Exhibit 14”
Annexure 5
“Transcript of Maxwell Jones”
Annexure 6
“Transcript of Maxwell Jones”
Annexure 7
“Hannah Kelly et al, ‘The Interpretation of Low-Level DNA Mixtures’ Forensic Science International: Genetics 6 (2012) 191-197.”
Annexure 8
“Transcript of Jane Taupin”
Annexure 9
“Transcript of Jane Taupin”
Annexure 10
“David Balding and John Buckleton, ‘Interpreting Low Template DNA Profiles’ Forensic Science International: Genetics 4 (2009) 1-10.”
Annexure 11
“Transcript of Aimee Pollett”
Annexure 12
Transcript of Maxwell Jones
0
3
0