R v Whyms
[2012] ACTSC 7
•January 17, 2012
R v DAVID DENNIS WHYMS
[2012] ACTSC 7 (17 January 2012)
CRIMINAL LAW – trial by judge alone – charges of burglary and theft – accused not guilty.
CRIMINAL LAW – evidence – expert evidence – evidence of DNA analysis – likelihood ratio – whether the probability of adventitious matters excludes a reasonable doubt – evidence shows a reasonable doubt.
Supreme Court Act 1933 (ACT), ss 68B, 68C
Evidence Act 1995 (Cth), s 138
Crimes Act 1900 (ACT), ss 209, 210
Buckleton JJ, Triggs CM and Walsh SJ, Forensic DNA Evidence Interpretation (CRC Press: Boca Rotan Florida, 2004)
Freckleton I, Reddy P and Selby H, Australian Judicial Perspectives on Expert Evidence: An Empirical Study (Australian Institute of Judicial Administration: Melbourne, 1999)
Macquarie Dictionary (Macquarie Dictionary Publishers Pty Ltd: Sydney, 2009) 5th ed
Fleming v The Queen (1998) 197 CLR 250
R v DM [2010] ACTSC 137
Halliday v Nevill (1984) 155 CLR 1
R v Mulcahy [2010] ACTSC 98
R v Karger (2001) 83 SASR 1
R v Jarrett (1994) 62 SASR 443
R v Mitchell (1997) 142 FLR 112
Forbes v The Queen (2009) 167 ACTR 1
R v Karger (2002) 83 SASR 135
R v Berry (2007) 17 VR 153
Riley v Western Australia (2005) 30 WAR 525
R v Doheny; R v Adams [1997] 1 Cr App Rep 369
Hillier v The Queen (2007) 228 CLR 618
Chamberlain v The Queen [No 2] (1984) 153 CLR 521
Plomp v The Queen (1963) 110 CLR 234
No. SCC 116 of 2009
Judge: Refshauge J
Supreme Court of the ACT
Date: 17 January 2012
IN THE SUPREME COURT OF THE )
) No. SCC 116 of 2009
AUSTRALIAN CAPITAL TERRITORY )
R
v
DAVID DENNIS WHYMS
ORDER
Judge: Refshauge J
Date: 17 January 2012
Place: Canberra
THE COURT FINDS THAT:
David Dennis Whyms is not guilty of count 1 on the indictment.
David Dennis Whyms is not guilty of count 2 on the indictment.
On 27 July 2008, the home security alarm at a residence (the premises) in Fadden was activated. An inspection of the premises showed that the window above the kitchen sink had been smashed and the laundry window damaged and some drawers in the master bedroom opened and disturbed. The owners and occupiers were contacted and identified certain items as missing. Blood was found on and near the damaged laundry window.
Later that day, police stopped a motor vehicle in which was found some of those stolen items. Closed circuit television from a service station appeared to show David Dennis Whyms walking into the Service Station with the persons who were in the car when it was stopped by police. Mr Whyms was later arrested.
Subsequently, Mr Whyms was subjected to a forensic procedure and the analysis of the sample then taken was compared with an analysis of the blood found on and near the damaged laundry window.
As a result, Mr Whyms was charged with certain offences and appeared in the Magistrates Court. On 19 March 2009, he was committed for trial in this Court. A draft indictment was filed on 25 May 2009.
It contained the following charges:
FIRST COUNT: THAT on the 27th day of July 2008 at Canberra in the Australian Capital Territory, DAVID DENNIS WHYMS entered a building, namely 363 Bugden Avenue Fadden, as a trespasser with intent to commit theft of property therein.
SECOND COUNT: AND FURTHER THAT on the 27th day of July 2008, at Canberra aforesaid, DAVID DENNIS WHYMS dishonestly appropriated property belonging to John Richards and Glenda Richards with the intention of permanently depriving them of the property.
Particulars of stolen property
·2 bedside table drawers
·1 set of keys with a ‘RICHARDS’ family emblem key ring
·1 set of keys, comprising of 2 master keys for the Australian National University (ANU), 1 Holden car key and 1 Mazda car key; and containing 1 ANU staff pass in the name of John Richards, and 1 garage remote control
·1 set of keys with an Ursula College key ring
·1 Australian Apprenticeship key ring
·1 silver Seiko watch
·20 pairs of cuff links
·1 seed pearl necklace
·1 silver brooch
Proceedings
As noted above (at [4]), Mr Whyms was committed to the Supreme Court for trial on 19 March 2009.
On 27 October 2009, Mr Whyms was arraigned on the indictment filed as a draft and pleaded not guilty to each of the counts. A trial date of 21 June 2010 was then set. That date was subsequently vacated and the trial came on for hearing before me on 23 August 2010.
On 18 August 2009, Mr Whyms had signed an election for trial by a judge alone under s 68B of the Supreme Court Act 1933 (ACT).Section 68C then provides in respect of such proceedings:
(1)A judge who tries criminal proceedings without a jury may make any finding that could have been made by a jury as to the guilt of the accused person and any such finding has, for all purposes, the same effect as a verdict of a jury.
(2)The judgment in criminal proceedings tried by a judge alone shall include the principles of law applied by the judge and the findings of fact on which the judge relied.
(3)In criminal proceedings tried by a judge alone, if a Territory law would otherwise require a warning to be given to a jury in such proceedings, the judge shall take the warning into account in considering his or her verdict.
Trial by judge alone
Under s 68C of the Supreme Court Act 1933 (ACT) a judge who tries proceedings for the prosecution of a person on indictment without a jury may make any finding that could have been made by a jury as to the guilt of the accused person and such a finding has, for all purposes, the same effect as a verdict of a jury.
The judgment of the court in such a case must include the principles of law that I as the judge apply and the findings of fact on which I rely. In Fleming v The Queen (1998) 197 CLR 250 (at 262 – 3; [28]) the High Court stated that it is necessary for the judge to expose the reasoning process linking the principles of law with the findings of fact and justify the process and, ultimately, the verdict that is reached.
Section 68C also requires me, as the trial judge, to take into account any warning, direction or comment in considering my verdict that any Territory law requires to be given or made to a jury in such proceedings.
There are certain general directions that I must take into account. These are fundamental rules designed to ensure that an accused person receives a fair trial according to law. See R v DM [2010] ACTSC 137; R v Mulcahy [2010] ACTSC 98.
As the judge of the facts in a trial by judge alone, as well as the judge of the law, I must find the facts and draw the inferences from them as well as to apply the law to the facts that I find. I must bring an open and unbiased mind to the evidence and view it clinically and dispassionately and not let emotion enter into the decision-making process. Both the prosecution and the accused are entitled to my verdict free of partiality or prejudice, favour or ill-will. I must then deliver my verdict according to the evidence.
The prosecution bears the onus of proving the guilt of the accused at all times. The accused does not have to prove that he did not commit the offences charged.
If the accused does adduce any evidence which is consistent with his innocence, he does not have to prove it; it is for the prosecution to disprove it or show that it is irrelevant, otherwise the prosecution will not have proved its case.
The standard of proof of the prosecution case is proof beyond reasonable doubt and the accused cannot be found guilty of the offence, unless the evidence which I accept satisfies me beyond reasonable doubt of his or her guilt.
The accused is presumed by law to be innocent of each of the offences unless and until the evidence I accept satisfies me that each and every element of the relevant charge has been proved beyond reasonable doubt. The accused then loses the presumption of innocence and I must find him guilty.
If, however, the evidence which I accept fails to satisfy me beyond reasonable doubt of any or all of the elements of any offence charged then he remains presumed innocent and I must find a verdict of not guilty.
If I am satisfied that there may be an explanation consistent with the innocence of the accused of any charge, or I am unsure of where the truth lies, then I must find the charge has not been proved to the standard of proof required by law and I must find the accused not guilty.
I must determine whether each of the witnesses is a reliable witness, that is whether I can rely on the evidence that the witness gives and so find the facts about which the witness has given evidence. I can accept part of a witness’s evidence and reject part of that evidence or accept or reject it all.
I must determine the facts in accordance with the evidence, considered logically and rationally, without acting capriciously or irrationally but I may use my common sense, experiences and wisdom in assessing the evidence.
As this was a case relying on circumstantial evidence, I give myself a warning about such evidence. Such a case may be just as convincing and reliable as a case based on direct evidence, depending on the nature of the circumstances relied upon when considered as a whole (not individually or in isolation), and the degree of clarity and certainty of that evidence, which may lead inevitably to the conclusion that the prosecution has established its case beyond reasonable doubt.
It is essential that such evidence is examined with care and assessed as reliable before drawing conclusions from the facts established by it. In order to draw any such conclusions, I must be satisfied that the facts are established from which I can draw the conclusions and that I can draw the conclusions from them and not merely speculate or engage in conjecture in arriving at them. They must, also, be the only conclusion in each case which, in my view, can be drawn beyond reasonable doubt. If there is more than one conclusion than that for which the prosecution contends, and that other conclusion or those other conclusions are inconsistent with the accused’s guilt or consistent with his innocence, then the prosecution has obviously not proved its case beyond reasonable doubt.
The trial commenced on 23 August 2010 and continued on 24 and 25 August and 14 October 2010. At the end of the trial, I reserved my decision.
The facts
Apart from the scientific evidence, not much of the evidence was challenged by Mr Whyms through his counsel, Mr S Gill. I found the witnesses truthful and reliable. Accordingly, I can find the following facts.
The owners of the premises left to visit England for two months from early July 2008. While away, they left the premises in the care of their daughter and son-in-law, who had arranged to visit the house on various occasions to check on it and to water the plants. They visited it on 23 July 2008. The house was then locked up and with no broken windows.
The owner’s son-in-law received a call from NRMA Security at about 4.00 am on 27 July 2008 and, as a result, he drove over to the premises, arriving at about 4.20 am. As he was driving there, he noticed a car on the side of the road, just along from the premises. The lights were off but there was a person at the steering wheel. He thought it was a four-door hatchback light in colour, possibly white.
When he arrived at the premises, he noticed that the security lights were flashing. He waited in the street, opposite the premises, for the security personnel to arrive, which they did about 15 minutes later. He then drove into the driveway of the premises and looked around the house.
The front door was closed and locked. He unlocked it and entered with the security officer and they discovered that the alarm panel was open and the panel itself smashed, which prevented the alarm from being immediately disabled. They looked further around the premises and found the kitchen window smashed with glass on the sink and the floor, having obviously been smashed from the outside.
They then inspected the bedrooms and saw the top drawers of the dresser pulled out and on the floor and two bedside drawers removed from the house, but the contents of the drawers had been tipped on to the floor. There seemed no disturbance in some of the other rooms but the laundry window had been smashed. The son-in-law noticed what seemed to be blood on the smashed window.
The police were then called and the owners’ son-in-law noticed the car he had seen earlier now being driven down the street with its lights still off. He could clearly see two people in the front seats of the car but he could not see into the back and did not see whether there were other passengers in it. He then looked outside the house and noticed that the handles to the French wooden doors had been pulled off and there was a flashlight on the ground outside the doors. It did not, he said, belong to the owners; nor did a packet of tobacco which was later found by police.
He then contacted the owners who identified a number of items that were missing from the master bedroom. These were the items particularised to count 2 on the indictment (see [5] above). They included some items from the disturbed drawers and also which had been taken from the top of the dresser.
Later, crime scene investigation personnel arrived from the police and took a number of photographs of the premises and, in particular, took swabs of the apparent blood that was found on and near the broken window glass in the laundry. No fingerprints were found at the premises nor on the tobacco product found by police. A cast was made of a tool mark that had been made on the side of the security screen door. No tool was located then or later in the investigation against which any comparison with the cast could be made.
Later on 27 July 2008, at about 6 to 7 pm, Mr Whyms was photographed on closed circuit television at the service station at Kaleen, associated with a white coloured car, apparently a Pulsar.
A friend of Mr Whyms gave evidence that at about 8.30 to 9 am (incorrectly appearing in the transcript as “8.39 am”) Mr Whyms knocked at her door, having driven to her place in a white Pulsar. A few hours later, they left in that car to visit a friend in Chifley, driving there in the Pulsar, arriving at between 11 am and noon. They picked up the friend and his two brothers and drove to Tuggeranong where one of the brothers was dropped off. They then drove to another friend’s place in Wanniassa, being Mr Whyms' ex partner. They stayed there for a few hours and then travelled to Gungahlin and drove elsewhere, including Fadden, Tuggeranong, Civic and Mitchell. At about dusk, they went to the Gungahlin Labor Club and then proceeded to Kaleen. They stopped at the service station and then drove to another friend’s place in Kaleen. Before they arrived there, a police vehicle intercepted the car. The car turned into a private driveway and Mr Whyms, who had been driving, left the car and his friend did not see him again that day.
The police came over to the vehicle and spoke to the occupants. As they approached, they saw a movement in the front part of the car. They found two occupants, the friend of Mr Whyms, who had been picked up at about 8.30 to 9 am that morning, sitting in the back seat, and the friend who had been collected at Chifley, sitting in the front passenger’s seat.
The police then searched the car and found between the handbrake and the seat a leather wallet containing a driver licence in the name of David Whyms. It also contained a Medicare card and Handycard in the name of Mr Whyms. They then located a number of items, being a number of keys found within a plastic bag inside a green environmentally friendly shopping bag. These were on the floor of the front passenger seat. Further inquiries showed that these keys were part of the property that had been identified as stolen from the premises in the early hours of the morning of that day.
Initially, a challenge was mounted to the search as it was submitted by Mr Gill that it had been conducted illegally, which would have brought into play s 138 of the Evidence Act 1995 (Cth). Mr Gill had submitted that as the car was in the driveway of a private residence, the police could not rely on the powers they had under s 209 of the Crimes Act 1900 (ACT) to conduct the search, given the limits expressed in s 210, especially par (b) which provides:
In exercising power under section 209 in relation to a conveyance, a police officer –
...
(b)shall search the conveyance in a public place or in some other place where members of the public have ready access; ...
I referred, however, to Halliday v Nevill (1984) 155 CLR 1, where Gibbs CJ and Mason, Wilson and Deane JJ said (at 7):
The most common instance of such an implied licence relates to the means of access, whether path, driveway or both, leading to the entrance of the ordinary suburban dwelling-house. If the path or driveway leading to the entrance of such a dwelling is left unobstructed and with entrance gate unlocked and there is no notice or other indication that entry by visitors generally or particularly designated visitors is forbidden or unauthorized, the law will imply a licence in favour of any member of the public to go upon the path or driveway to the entrance of the dwelling for the purpose of lawful communication with, or delivery to, any person in the house.
The objection was not pressed further.
A search was then conducted of the area and Mr Whyms was located nearby and arrested. A forensic procedure was conducted in respect of Mr Whyms when a buccal swab was introduced into his mouth.
Detailed evidence was given of the procedure used and the subsequent transfer of the swab to forensic personnel. It is not necessary to detail the procedures as, in the event, no challenge was ultimately made to it.
When police interviewed Mr Whyms at the police station, they did not observe whether he had been injured in any way. He did inform the Watch House police that he had a head injury but that this had been inflicted some time in the past. The police did not look at Mr Whyms' hands.
Forensic procedures were also conducted on the two occupants who were found in the Pulsar car when police stopped it in Kaleen. The results of those procedures were also analysed by forensic scientists.
The forensic evidence
While there are circumstances pointing to Mr Whyms’ possible involvement in the burglary and theft at Fadden, the critical evidence was the forensic evidence and so it is appropriate that it be dealt with separately.
Although detailed evidence was given of the procedures followed by the forensic officers involved, ultimately nothing turns on that as no challenge was made to the procedures and so it is not necessary to set out the details here. A detailed examination of the process has been undertaken by Mullighan J in R v Karger (2001) 83 SASR 1 (at 12 – 16), and I do not need to repeat it.
Suffice to say, a visual examination is carried out of the item submitted for examination and, if blood is thought to be involved, a presumptive test may be done. The DNA analysis then commences with extraction where the cells are broken open to release the DNA into a solution. Next, the DNA extracted is loaded into a machine to determine the amount of DNA in the sample, to ensure that the correct amount is available for subsequent analysis. This is called quantitation.
The DNA is then subject to replication using the polymerase chain reaction (PCR) technique. This was described by Mullighan J in R v Jarrett (1994) 62 SASR 443 (at 446). It is called amplification. Finally, a process called capillary electrophoresis is undertaken where the sites that have been copied in the PCR process are run through a gel and passed in front of a laser which detects the DNA that is in the sample. The end product is a DNA profile which can be represented as a series of numbers or, more commonly, as a graph with peaks. These form the DNA profile. This was the process followed in this case.
In the Territory, in common with all other Australian forensic laboratories, the forensic scientists look at ten locations (or loci) along the DNA molecule. These ten locations are the same one’s investigated on each occasion.
As a result of this process, the forensic biologist was in this case presented with three profiles, two from the blood samples on and near the broken window at the premises and one from the buccal swab taken from Mr Whyms’ mouth.
One of the ten sites identifies the sex of the person from whom the DNA has been taken. In this case, the blood came from a male in both cases.
It was also clear, in this case, that the DNA from the blood came from one individual and the problems sometimes associated with mixtures was not relevant here.
The forensic biologist then compared the profile produced by analysis of the DNA of the blood taken from and near the window at the premises and the profile produced by analysis of the DNA from the buccal swab taken of Mr Whyms and found them identical, in that, in all sites, the same alleles were found in both profiles.
She then compared the profiles produced from the analysis of the material taken in the forensic procedures undertaken with the two occupants of the Pulsar motor vehicle stopped by police in Kaleen. She excluded them as contributors to the DNA found from analysis of the blood found at and near the window of the premises.
The forensic biologist then conducted a statistical calculation validated in the AFP forensic science laboratory based on a set-population database of 620 individuals from the ACT. The database assigns a frequency for the 9-loci (not including the sex marker) and the calculation is based on a formula which uses the material from the profile from the DNA analysed from material at the crime scene for how common it is in the ACT population, producing a likelihood ratio.
The likelihood ratio is an approach that has been accepted in this court: R v Mitchell (1997) 142 FLR 112 (at 124). This acceptance was confirmed by the Court of Appeal in Forbes v The Queen (2009) 167 ACTR 1 (at 9). The likelihood ratio was described by Doyle CJ, with whom Prior J agreed, in R v Karger (2002) 83 SASR 135 (at 140) as follows:
For the purposes of expressing the likelihood ratio, two factual hypotheses are compared. One is the prosecution hypothesis that the appellant was the source of the bloodstain and of the incriminating DNA. The other hypothesis is the defence hypothesis that an unrelated person is the source of the bloodstain and of the incriminating DNA. The statistical evidence calculates the probability of the occurrence, in either case, of the match found between the incriminating DNA and that of the appellant, and expresses the competing probabilities as a ratio.
The statistical evidence interpreting the significance of the DNA match is not evidence of the probability that the appellant was the source of the incriminating DNA. To so regard it would be to make an error. However, the statistical evidence interpreting the DNA match is expert evidence that the jury could use in deciding whether it was satisfied beyond reasonable doubt that the appellant was the source of the incriminating DNA. The statistical evidence is undeniably strong evidence pointing to a conclusion that the accused was the source of the incriminating DNA, but is not direct evidence of that fact. And, as is obvious, the statistical evidence must be considered in the light of other evidence in the case.
The likelihood ratio is now accepted as a proper calculation to assist the jury in the interpretation and evaluation of the evidence from DNA analysis. See R v Berry (2007) 17 VR 153 (at 161) per Redlich JA with whom Buchanan and Kellam JJA agreed. It is not, of course, a substitute for proof of guilt beyond reasonable doubt and must be considered along with all the other evidence: Riley v Western Australia (2005) 30 WAR 525 (at 534) per Steytler P.
The formulas used to calculate the ratio can vary according to the nature of the profile and the loci in it. For example, if there are two peaks at one loci, then the individual is said to be heterozygote, having inherited the characteristics from each of the individual’s parents, a different formula would be used where only one peak is found (homozygote). Different formulae would be used for mixed profiles. The formulae are standard formulae used throughout Australia, the United Kingdom and in parts of America. Part of the training of a forensic biologist is to know which formula to use.
In this case, the formulae used were those specified in Buckleton JJ, Triggs CM and Walsh SJ, Forensic DNA Evidence Interpretation (CRC Press: Boca Raton Florida, 2004).
Applying this calculation, the forensic biologist came to the opinion that it was at least 23 billion times more likely that the DNA analysed from the material found at the premises came from Mr Whyms than from another unrelated individual selected at random from the ACT population.
In addition to this evidence, the co-ordinator of criminalistics and identification sciences of the Australian Federal Police gave evidence about the statistical calculations that were led. He confirmed the evidence referred to above (at [57]), about formulae appropriate to the relevant tasks in the statistical evaluation of the DNA analysis, the source of them and their use.
His evidence was that the formulae were used “because they deal most comprehensively with population genetic effects that are observed in populations and need to be taken account of in those sorts of estimates.” He also described them as “the most conservative”, avoiding less conservative ones which “exhibit a mild bias towards or in favour of the prosecution.”
He gave a detailed description of elements of the formulae but, in the event, it is not necessary for me to address those issues in any detail. I note, however, his evidence that the material on which he relied was based on recommendations from the National Research Council of the US National Academy of Sciences in a 1996 report on the interpretation of statistical assessment of DNA evidence but supplemented by extensive Australian studies of populations, both indigenous and non-indigenous populations in Australia. He had personally reviewed and conducted a statistical validation of the ACT population database. His opinion was that the ACT population database of 620 individuals was a statistically valid database.
The evidence was that the population database is used as a component of assessing how frequently a particular profile might appear in the general population. As such it is a statistical estimate. It is, of course, not necessarily the actual frequency of a combination that forms a DNA strand within the population, but, through the observed frequencies it provides, as a representative database, the expected frequencies in the broader population. The database, he said, cannot justify a forensic scientist predicting the DNA profile of any particular individual.
He accepted that the database did not take into account direct relatedness where a person whose profile has been prepared may have a relative, close or otherwise. He accepted that the likelihood ratio to be calculated, whereby a relative, such as a brother or sister, may have been the source of the DNA, would require a variation in the hypothesis in the calculation of the likelihood ratio. Thus, if a suspect, whose DNA profile was being compared with a DNA profile of material found at the crime scene, had a brother, cousin or children, the likelihood ratio would change.
He also accepted that a likelihood ratio does not require the existence or presence of the number of persons in order that there be a match. That is to say, if the likelihood ratio is that it is a million times more likely that the DNA profile from material at a crime scene came from the accused than from another unrelated person, it is not necessary that one has encountered a million people before the actual match is encountered. That, it seems to me, is an obvious fact from probability theory. It does not, however, undermine the information that, as a statistical probability, the likelihood ratio provides. It cannot completely exclude the one, it does not require the presence or existence of a million, but it expresses the likelihood of the existence of such a one.
He did explain, however, that a likelihood ratio could be provided when relatedness was included. He said:
[S]o I’m saying that we could give an estimate – a theoretical estimate of the probability under the circumstance that it was not the accused who left a sample but that it was the accused’s brother or son or cousin but they are estimates of an unknown. The circumstance is where a profile is presented and we are asked the question what’s the likelihood that this profile is from the brother of a particular individual. Then obviously we can make an assessment of that. It wouldn’t be the sort of assessment that would be common in this type of case but it can be for relationship testing for identification purposes or those sorts of things.
Would the margin of error be different? --- Yes, it would, yes. Because the change in the formula is really reflective of, again, in itself another probability and that’s the probability that a sibling will share DNA types with their sibling and as you’ve observed a sibling may share all of their DNA types or they may share none. On average they share a quarter so, you know, the probability that two alleles chosen from a pair of siblings, each member of a pair of siblings, the probability that they would the [sic] same is a quarter. But in actual fact they may not be the same or they may always be the same but on average the probability is a quarter. I hope that sorts [sic] of reflects what I’m trying to say.
He was asked about database research which analysed the occurrence of matches within databases. He was referred to some research (the Buckleton Study), in which some matches were found initially at nine of nine loci though these were later excluded because the information about the databases was not such as to rule out that these matches were, in fact, made by the same individuals appearing more than once in the database. There were, however, also matches at eight out of nine loci and seven out of nine loci.
His recollection of the study was that “the concordance between the expected number of partial matches, whether it be at the eight sites or seven sites or less, concorded well with the observed number” (which I understood to be the likelihood ratio).
He was asked about other studies, including one in Arizona where a database of about 65,000 people was evaluated and found 122 matches at nine out of the ten loci, 20 matches at 10 out of 13 loci and 1 each at 11 out of 13 and 12 out of 13 loci, though the latter two were related persons.
The statistical evidence and the evidence of probability then became problematic and, I have to say, for me, quite murky. I am aware that in a recent study, about 10% of judicial respondents reported statistics as being a most difficult area of expert evidence: Freckleton I, Reddy P and Selby H, Australian Judicial Perspectives on Expert Evidence: An Empirical Study (Australian Institute of Judicial Administration; Melbourne, 1999) p 66. I suspect that I fall within that 10% category.
His evidence was that, when assessing the probability of a matched profile in a database, it is necessary to compare each profile with every other profile. Thus, in a database of 100, the first profile is compared with the other 99, the second with the remaining 98 (since it has already been compared with the first one) and so on. Thus, it is not the same as looking simply at 100 profiles; the actual number of comparisons is much larger.
He then referred to a process he had undertaken in New Zealand, when dealing with a 6 loci test for DNA analysis. He said the match probability was in the order of one in a million, but then said that “[i]t was anticipated that when the database reached a size of about 8,000, we were into the realm where there was the expectation of a chance match.” The evidence then continued:
If I could then ask you this about the Buckleton and Arizona tests? --- Yes.
On what you’ve told us about Buckleton, what we’re observing is the fact that what you see in terms of matches is not what you expect when you hear a figure of one in a billion? --- I think that’s a fair comment. It is not what is intuitively understood when you hear the probability figure, but the probability figure actually correleates closely to the observed results in that study.
And what it also means then is if a figure of 1 in 23 billion is what’s produced in this particular instance, that does not mean that we have 23 billion episodes before a match is found? --- No.
A match could readily be found within 10,000? --- Yes.
But we simply do not know? --- Yes. The statistic if you think about it in its purist form is dealing with the observed DNA types out of a finite number of markers. Now, if you were to calculate the possible combinations that could exist across all of those markers, you know, accepting how many – and this is remote from the human context if you like, then the possible number of permutations that you could come up with is an enormous number. And that is irrespective of what we’ve actually observed in the several billion humans that have inhabited the world, you know, since – well, since modern humans evolved. So the number of humans is – is not directly related if you like, or not ---
With a number of combinations? --- With a number of possible combinations and the statistic takes that into account as well.
Can I ask you this then? --- Yes.
And I think this is finally for the second time of the Buckleton, that when as a court we come to consider statistics such as one in a billion or one in 23 billion? --- Yes.
A useful check on how we conceptualise those is to understand that in a population of 18,000 such as that studied by Dr Buckleton? --- Yes.
There are likely to be multiple matches or multiple just underneath matches occurring? --- At – yes, there have been highly similar profiles observed. As I said, my recollection of the 9 locus matches was either that there weren’t any or that they weren’t considered in the study. And I can’t recall which of the two. I do recall a study undertaken in 2004 by Professor Bruce Weir which is on a very similar issue, and it was in fact his study that Buckleton et al followed in terms of the approach. And what Professor Weir did, and he used Australian data and he sued the 9 locus test that we’ve used here, what he then did was looking at Australian populations and the population genetics observed, and the probability of a random match, he then looked at what size population would be approximately needed in order to expect to observe an adventitious match.
Yes? --- And my recollection of his findings, and again this is going back a little bit in my memory, and he was talking in reference to a national database, was that a national database of around about 600,000, once you had a database of approximately that size, you could expect to see two individuals on that database with the same profile purely by chance.
And sorry, that’s a statistical hypothesis that he was advancing? --- Correct, yes. And that was you know, notwithstanding that the random match probability or average random match probability for the profile was in the order of the one billion figure that we’ve discussed earlier.
Right, so just coming back to the Buckleton Study? --- Yes.
We are to conceptualise what one in a billion means? --- Yes.
It means that there’s a likelihood of a match within a population of about 18,000? --- I just – I’m not happy to conclude on that – on that figure, because I would need to you know, estimate the number of comparisons and those sorts of things, but those studies, such as the Buckleton Study and the Arizona Study, yes, can be used to assist you in understanding what these figures mean.
Perhaps if I could put the question around the other way? --- Yes.
And if you’re still unhappy, of course feel free to disagree? --- Yes.
The upshot I think of what you’re saying is that for a population of 65,000 on the Arizona database or 18-plus thousand on the New South Wales or whatever database it was? --- Yes.
That it’s not unexpected in terms of the statistical model you use to find a perfect match, or perfect matches on each of those? --- It’s – I would probably be happy to agree with you that it’s not unexpected, but you know, it’s not an accurate answer in the sense that ---
...
It doesn’t tell you how often to expect it but it’s not what we’d call unexpected? --- It’s not what I would call unexpected.
I then asked him:
HIS HONOUR: Dr Walsh, what do you mean when you say it’s not unexpected? --- That – I mean that the possibility is real, that without being able to calculate the expected figure, I’m not able to say whether it, you know, is a – I’m not able to comment on I guess the – the nature of that expectation, but what this – these sorts of studies do assist with is an understanding that the notion of shared DNA is I guess more complex in a sense, and the figures that are produced as a random match probability can lead to intuitively incorrect conclusions. So my answer to Mr Gill which was a difficult answer to give is that if I had a nine locus file and I had a database of tens of thousands it’s not unexpected for me to observe that there may be a matching profile on that database.
He did then make the point that relatedness was very significant in this consideration. He said “where there are higher levels of relatedness, higher levels of co-ancestry, then the likelihood of observing a similar profile is obviously much higher”.
Mr Gill then put the matter squarely to the witness.
Perhaps if I backtrack? One can’t tell in advance which profile would find its perfect match elsewhere in the population? --- No. It may be that there are – there are individuals within this population set, that you’ve described, who have the same DNA profile on the nine sites we’ve tested.
Which putting it in a blunt fashion means there may be individuals who have a perfect match, even within the population of the ACT and Queanbeyan? --- Yes, there may be.
Regrettably, that answer was not further explained, at least in a way that was meaningful to me. It was not limited or qualified in any way. The witness did point to the conceptual difference between the probability of a match occurring and the contingent possibility that it occurs. As he explained:
The common analogy to this particular problem is what they refer to as the birthday problem. So there are 365 days, or 366 if you count a leap year, on which someone could be born. So the probability of anyone’s particular birthday is one in 365. In first year statistics, they often pick a day and have a room of a hundred students and they say, ‘So who’s got the birthday on this date?’ and they might have, you know, 10 people put their hand up. Now so you think well, so that birthday has a probability of a one in 10, if 10 in a hundred have it. But it doesn’t, its probability is still one in 365, it’s just that if you actually take a group of a hundred and compare all their birthdays together, you’ve done an enormous number of comparisons that far exceeds 365, so the chance that you would find any matching pairs in that, you know, is different to what you might intuitively expect. The questions that are being answered, whilst the statistics do correlate, are conceptually different and that is that is the cause for people’s misunderstanding, I guess.
Quite frankly, this did not help and I found it confusing, for it did not connect the concept of the probability (365 to 1) with the possible occurrence (1 in 10) and explain how to use either in this context.
He agreed that “the two probabilities are in fact consistent because even though there may be five – adventitious matches in a particular population, the statistical probability of that remain one in 23 million.” This made an understanding of how to interpret the probability even more confusing. The reference to “million” rather than “billion” came from a question I had asked which incorrectly referred to the smaller figure rather than the larger one which had come from the likelihood ratio.
Finally, a statement of Mr Whyms’ only brother was tendered without objection, confirming that he was Mr Whyms’ only brother, that he had two sisters and that he, Mr Whyms’ brother and that he was not in Canberra on 27 July 2008, intended, as I understood it, to exclude the problem of relatedness, which has not been taken into account in calculating the likelihood ratio.
The contentions
The prosecution, represented by Mr M Thomas, accepted that the case was a circumstantial case. He referred to the following as the appropriate strands which, when taken together, led, in his submission, to an irresistible inference of the guilt of Mr Whyms. There was:
· the profile comparison of the DNA material taken from the premises and from Mr Whyms;
· the similarities between the car seen near the premises by the owners’ son-in-law and the car which Mr Whyms was driving later that day; and
· the finding by police of items from the burglary in the car being driven by Mr Whyms.
He submitted that while the evidence about the DNA likelihood ratio had included an acknowledgement that it was possible to find matches within a small population, this did not shake the witness’s “faith in the likelihood model”. Matches, he submitted, could be explained by relatedness. That was not relevant here because the evidence of Mr Whyms’ brother excluded him as the offender. The difficulty for me, however, was that this was not a complete analysis of the evidence.
He referred to the well known passage in R v Doheny; R v Adams [1997] 1 Cr App Rep 369 when Phillips LJ, for the Court, said (at 373):
The significance of the DNA evidence will depend critically upon what else is known about the suspect. If he has a convincing alibi at the other end of England at the time of the crime, it will appear highly improbable that he can have been responsible for the crime, despite his matching DNA profile. If, however, he was near the scene of the crime when it was committed, or has been identified as a suspect because of other evidence which suggests that he may have been responsible for the crime, the DNA evidence becomes very significant.
He submitted that, contrary to the suggestion of an alibi by Phillips LJ that would undermine the DNA evidence, Mr Whyms was not “at the other end of England”; he was, in fact, very near to the premises and the stolen goods.
Mr Gill initially challenged each of these strands. He pointed out that there was no actual identification of the car near the premises as being the car which Mr Whyms was driving later in the day. He also challenged the significance of the proceeds of the burglary and theft being found in the car, pointing out that there was no direct evidence of when the burglary in fact occurred as there was damage to the alarm system. He submitted that even if it were the same car, this explains why there are proceeds of the offences in it but not necessarily connecting it to Mr Whyms. The goods were, after all, at the feet of the front seat passenger.
He went dangerously close in his submissions to falling into the error identified by the High Court in Hillier v The Queen (2007) 228 CLR 618, where Gummow, Hayne and Crennan JJ said (at [48]):
Often enough, in a circumstantial case, there will be evidence of matters which, looked at in isolation from other evidence, would yield an inference compatible with the innocence of the accused. But neither at trial, nor on appeal, is a circumstantial case to be considered piecemeal.
Their Honours referred to what had been said by Gibbs CJ and Mason J (as his Honour then was) in Chamberlain v The Queen [No 2] (1984) 153 CLR 521 (at 535):
Similarly, in a case depending on circumstantial evidence, the jury should not reject one circumstance because, considered alone, no inference of guilt can be drawn from it. It is well established that the jury must consider ‘the weight which is to be given to the united force of all the circumstances put together’ ...
They also referred to what Dixon CJ said in Plomp v The Queen (1963) 110 CLR 234 (at 242):
All the circumstances of the case must be weighed in judging whether there is evidence upon which a jury may reasonably be satisfied beyond reasonable doubt of the commission of the crime charged. There may be many cases where it is extremely dangerous to rely heavily on the existence of a motive, where an unexplained death or disappearance of a person is not otherwise proved to be attributable to the accused; but all such considerations must be dealt with on the facts of the particular case. I cannot think, however, that in a case where the prosecution is based on circumstantial evidence any part of the circumstances can be put on one side as relating to motive only and therefore not to be weighed as part of the proofs of what was done.
(Emphasis added.)
Mr Gill acknowledged the point. He submitted, however, that without the DNA evidence, the “strands” of the other circumstances were not strong enough in combination but on their own to sustain a conviction.
He then submitted that the DNA evidence lacked probative value. He identified two elements to the evidence: the consistency of the profiles and the statistical evidence.
The consistency alone, he submitted, did not take the matter very far. It was only when the statistical evidence was added that it could rise to the level of supporting the inference beyond reasonable doubt that the burglary and theft were committed by Mr Whyms
He pointed to the conceptual difficulties raised by the statistical evidence as noted in the evidence so that, as he put it, the Buckleton Study and the Arizona Study could assist. He accepted that it was difficult to conceptualise how one could have a number of expected matches in a population of 18,000 (Buckleton) or 65,000 (Arizona) and yet be consistent with a probability of 1 in 23 billion being the likelihood ratio. Yet, he submitted, that is what the evidence was. He referred to the articulation of the evidence, which included:
· “... in a population of 18,000 it would not be unexpected for there to be a matching profile”
· “... if one was to conduct the search of the ACT and Queanbeyan population, there’s a likelihood that there’d be a number of perfect matches?” “There’s a likelihood that perfect matches would be observed, yes.”
· “... there may be individuals who have a perfect match even within the population of the ACT and Queanbeyan?” “Yes, there may be.”
· “... the possibility is real ...”
When I put to him that “likelihood” had to be read in the context of a probability of 1 in 23 billion, he submitted that the evidence had not actually addressed that and it was inconsistent with the actual articulation of the evidence.
Mr Gill also challenged the statistical probability because of the issue of relatedness. He submitted that a probability relying on unrelatedness was irrelevant because there may be other relatives than Mr Whyms' brother and sisters who had not been accounted for in the analysis. He submitted that, if the other relatives of Mr Whyms were not excluded, then the factual basis for a statistical calculation based on unrelatedness had not been made out.
Consideration
I accept that this is a circumstantial case and have already (at [22] – [23] above) given myself the appropriate direction in such a case.
I accept, of course, the approach directed by the High Court in Hillier v The Queen and take into account all the evidence.
Having approached it in that way, I can say that the circumstances, without the DNA evidence, are insufficient for me to be satisfied beyond reasonable doubt that Mr Whyms is the person who entered the premises as a burglar and stole the items particularised in connection with the theft charge. It depends, however, on what the DNA evidence actually is and what can be concluded from it.
The DNA evidence is apparently strong. It shows that it was at least 23 billion times more likely that the DNA analysed from the material found at the premises came from Mr Whyms than from another unrelated individual selected at random from the ACT population.
Taken in that form, the DNA evidence, following cases such as Forbes v The Queen, would ordinarily be sufficient to convict Mr Whyms.
The problem is that the actual evidence went further. In explaining that probability, the evidence was that such a probability was consistent with the finding of an adventitious match in a population of 18,000. That could be understood as meaning no more than that the probability, though one in 23 billion, is still not a certainty. In this sense, it is like “the Man on the Moon” probability. It is logically possible that there is a man on the moon but the contingent or physical possibility is so slight that it would not raise a reasonable doubt as to the proposition that there is no man on the moon.
The evidence, however, did not stop there. It went further, namely that it was described as not unexpected “that there would be a match to Mr Whyms’ DNA profile in the population of the ACT and Queanbeyan”. The term “unexpected” was not explained further either by the witness or in re-examination. Giving it its ordinary meaning of not “unforeseen”, not “surprising” (Macquarie Dictionary (Macquarie Dictionary Publishers Pty Ltd: Sydney, 2009) 5th ed p 1794) seems to me to take it well into the area of reasonable doubt.
I find it difficult to rationalise the acceptance of the probability of 1 in 23 billion with the fact that an adventitious match in the ACT and Queanbeyan area is not unexpected. That, however, was the expert evidence.
It may be that it could have been further explained. It may have referred to relatedness. It may have referred not to “unexpected” in the ordinary sense but simply that a possibility, even a remote possibility, denies the certainty of non-existence. These were, however, not explained and there was plenty of opportunity for such explanation.
I find this result unsatisfactory for it does appear to me that there was much more explanation that I could have, perhaps should have, been offered. The fact is that, of course, I have to rely on and accept the evidence as presented to the Court.
It would, of course, be possible to use even the evidence I had as a strand to prove a circumstantial case, but the other strands would have to be must stronger that they were here to combine with what I had to lead to an inference of guilt beyond reasonable doubt.
I may have grave suspicions but, as I point out to a jury, that is not enough if there is a reasonable doubt.
The clear evidence was that the DNA, notwithstanding the likelihood ratio, was such that in the ACT and Queanbeyan a population it was not unexpected, a real possibility that there would be a person whose DNA profile would match the DNA profile from the blood at and near the window of the premises and who is not Mr Whyms. The evidence was not further explained or circumscribed in any way. As a result, I could not exclude that another person left the blood found at and near the laundry window of the premises.
Accordingly, I entertain a reasonable doubt for there is an explanation which is not consistent with the prosecution case. Mr Whyms is, therefore, not guilty of the offences as charged.
I certify that the preceding one-hundred and seven (107) numbered paragraphs are a true copy of the Reasons for Judgment herein of his Honour, Justice Refshauge.
Associate:
Date: 17 January 2012
Counsel for the Crown: Mr M Thomas
Solicitor for the Crown: ACT Director of Public Prosecutions
Counsel for the defendant: Mr S Gill
Solicitor for the defendant: Legal Aid Office (ACT)
Date of hearing: 23, 24, 25, 26 August 2010, 14 October 2010
Date of judgment: 17 January 2012
0
14
3