University of New South Wales Law Journal
SENTENCING: FROM VAGUENESS TO ARBITRARINESS: THE NEED TO ABOLISH THE STAIN THAT IS THE INSTINCTIVE SYNTHESIS
Criminal sanctions involve the deliberate infliction of hardship on offenders. In sentencing, the state acts in its most coercive and decisive manner: ‘the state may use its most awesome power: the power to use force against its citizens and others’. Despite the importance of the interests at stake in the sentencing realm, sentencing is arguably the least coherent, predictable and principled area of law.
The High Court of Australia has not facilitated attempts to inject clarity and precision into sentencing determinations. It has repeatedly endorsed the ‘instinctive synthesis’ approach to sentencing, emphasising the need for ‘individual justice’ over the need for transparency and a step-wise systematic approach to sentencing.
The importance of the debate regarding appropriate sentencing methodology is underscored by the passionate and emotive language that has been used by senior and influential present and former Australian judges. Justice Hulme has described the instinctive synthesis approach as little more than ‘pluck[ing] [figures] out of the air’. Justice Kirby, in Markarian v The Queen, states that the instinctive synthesis encourages ‘the thought that there descends upon a judicial officer, following appointment, a mystical “instinct” or “intuition” that ensures that he or she will get the sentence right “instinctively”’. Justice McHugh, in defence of the instinctive synthesis, labels the alternative approach as an appeal to ‘junk science’, and reminiscent of a belief that ‘belongs in Lord Reid's fairytales’, which are devoid of an ‘Aladdin’s Cave’, which to his mind is necessary to provide a more accurate sentencing methodology. Justice McHugh notes that ‘chanting the magic words, “two-tier sentencing”’ cannot unlock the answers.
Tensions between the need for legal certainty and flexibility manifest in most legal areas and it is not uncommon for decision-making to involve a degree of discretion. Profoundly strong arguments can be made for the retention of discretion in most areas of the law. As noted by H L A Hart, discretion is a sound mechanism for dealing with legal indeterminacy which accords with the rule of law and is an appropriate mid-course between arbitrariness and fixed rule application. I argue that the breadth of the sentencing discretion is, however, so boundless that it violates key rule of law virtues in the form of consistency, predictability and transparency.
The High Court has recognised the importance of consistency in sentencing, and has attempted to dilute the significance of consistency in the outcome of sanctions that are imposed by judges by stating that in this realm consistency in the application of principle (not outcome of sentence) is the desired goal. I argue that this approach is flawed for several reasons. First, from the perspective of the community and the offender, what matters most in sentencing is the sanction that is imposed, not an abstract quest for legal uniformity. Secondly, the notion of consistency in the form of an application of principle is an unattainable objective in an area of law such as sentencing where there are hundreds of often competing considerations, and where the weight that is attributable to the variables is for the sentencing judge (or magistrates) to determine.
Thirdly, there has been a universal move to expose the decision-making of public officials to greater scrutiny. This is especially evident in the administrative law domain. Sentencing currently stands apart from this. It is an island of vagueness, in a world demanding, increasingly, particularity and precision.
In the end, retention of the instinctive synthesis allows (if not encourages) sloppy and outcome-oriented reasoning, and facilitates the expression of subconscious judicial biases and preferences. As noted below, research establishes that judges are influenced by their personal sentiments, and extra-legal considerations often have a considerable impact on the outcome of cases. Considerations which have been shown to influence the outcome of penalties include the attractiveness, socio-economic status and race of the offender. The political persuasion, mindset and comfort level of the decision-maker are also relevant. In a society which subscribes to even basic rule of law virtues, it is unacceptable that these considerations can influence the extent to which an individual is punished.
Many of the deficiencies of the instinctive synthesis approach to sentencing were highlighted in the United States nearly half a century ago by Marvin Frankel, who described the system as a wasteland in the law. The persuasiveness of this work was partly responsible for the (regrettable) legislative over-reaction in the United States which has resulted in a massive increase in presumptive or mandatory penalties over the last two decades, resulting in the United States imprisoning more of its citizens than any other nation – nearly 750 per 100 000 adults. Irreducible aspects of the rule of law are also likely to result in the inevitable abolition of the instinctive synthesis approach in Australia. This article attempts to accelerate this process.
The need for a rethink of the manner in which sentences are determined is made more pressing by the result of the recent High Court decision of Barbaro v The Queen, where the Court stated that it is impermissible for the prosecution to inform the sentencing judge of its view regarding the appropriate sentencing range. This is likely to result in a higher number of accused persons pleading guilty. Core to the High Court’s reasoning in Barbaro is the open-ended and unpredictable nature of sentencing outcomes. The consequence of this decision compounds the reasons in favour of abolishing the instinctive synthesis. Thus, while debate about the appropriateness of the instinctive synthesis approach to sentencing is not new, the need to reform the sentencing process is now more important.
In the next Part of the article, I provide an overview of the nature of the sentencing decision-making process. In Part III, I analyse the negative practical consequences of the instinctive approach to sentencing. In Part IV, I consider the attempt to rebut the arguments in favour of the instinctive synthesis. Part V argues that the impact of implicit judicial bias on sentencing provides an insurmountable reason for the rejection of the instinctive synthesis. Conceptually, the arguments in Part V could form part of Part III, but they deserve separate consideration because they are core to the central issue in this paper and have not been discussed in depth previously in the Australian setting, neither by the judiciary nor in academia. My concluding remarks are in Part VI.
The overarching methodology and conceptual approach that sentencing judges undertake in making sentencing decisions is ‘instinctive synthesis’. The term originates from the Full Court of the Supreme Court of Victoria decision of R v Williscroft, where Adam and Crockett JJ stated: ‘Now, ultimately every sentence imposed represents the sentencing judge’s instinctive synthesis of all the various aspects involved in the punitive process’.
The process of instinctive synthesis is a mechanism whereby sentencers make a decision regarding all of the considerations that are relevant to sentencing, and then give due weight to each of them (and, in the process, incorporate considerations that incline to a heavier penalty and offset against them factors that favour a lesser penalty), and then set a precise penalty. The hallmark of this process is that it does not require (nor permit) judges to set out with any particularity the weight (in mathematical terms) accorded to any particular consideration. A global judgment is made without recourse to a step-wise process that demarcates the precise considerations that influence the judgment.
Accordingly, a degree of subjectivity is incorporated into the sentencing calculus. Current orthodoxy maintains that there is no single correct sentence, and that the ‘instinctive synthesis will, by definition, produce outcomes upon which reasonable minds will differ’. Under this model, courts can impose a sentence within an ‘available range’ of penalties. The spectrum of this range is not clearly designated; however, if the tariff is not observed, the sentence can be overturned on appellate review as being either ‘manifestly excessive’ or ‘manifestly inadequate’.
The alternative approach to the instinctive synthesis is termed the two-tier or two-step approach. It involves a court setting an appropriate sentence commensurate with the severity of the offence and then making allowances up and down, in light of relevant aggravating and mitigating circumstances.
The two-step approach was firmly rejected by the High Court in Markarian v The Queen, where it was noted: ‘Following the decision of this Court in Wong it cannot now be doubted that sentencing courts may not add and subtract item by item from some apparently subliminally derived figure, passages of time in order to fix the time which an offender must serve in prison’.
The competing approaches were most recently considered by the High Court of Australia in Barbaro, where the plurality decisively confirmed the instinctive synthesis approach. The plurality stating:
Fixing the bounds of a range within which a sentence should fall or within which a sentence that has been imposed should have fallen wrongly suggests that sentencing is a mathematical exercise. Sentencing an offender is not, and cannot be undertaken as, some exercise in addition or subtraction. A sentencing judge must reach a single sentence for each offence and must do so by balancing many different and conflicting features. The sentence cannot, and should not, be broken down into some set of component parts. As the plurality said in Wong, ‘[s]o long as a sentencing judge must, or may, take account of all of the circumstances of the offence and the offender, to single out some of those considerations and attribute specific numerical or proportionate value to some features, distorts the already difficult balancing exercise which the judge must perform’.
Sentencing is, therefore, regarded as an attempt above all else to attain ‘individualised justice’. The High Court in Elias v The Queen, in diminishing the importance of the maximum penalty to sentence, stated:
As this Court has explained on more than one occasion, the factors bearing on the determination of sentence will frequently pull in different directions. It is the duty of the judge to balance often incommensurable factors and to arrive at a sentence that is just in all of the circumstances. The administration of the criminal law involves individualised justice, the attainment of which is acknowledged to involve the exercise of a wide sentencing discretion.
The broad rationale for instinctive synthesis was considered in Wong v The Queen, where most members of the High Court saw the process of sentencing as an exceptionally difficult task with a high degree of ‘complexity’. Exactness is supposedly not possible because of the inherently multi-faceted nature of that activity. Thus, different outcomes are tenable so long as they come within what is an acceptable range.
In addition to this, a number of specific arguments have been advanced in favour of the instinctive synthesis. The most extensive justification of the instinctive synthesis approach is by McHugh J in Markarian v The Queen. In order to understand fully the persuasiveness of his reasons and the responses to them, it is necessary to provide more background to the nature of the sentencing and the sentencing inquiry. This is canvassed in Part III of this article. Once this background has been established, the specific responses to the supposed advantages of instinctive synthesis are discussed in Part VI.
A key problem with the instinctive synthesis is that it leads to inconsistent sentences. This is an obvious shortcoming of this approach and the criticism has not been missed by the High Court. In Hili v The Queen, French CJ, Gummow, Hayne, Crennan, Kiefel and Bell JJ stated that consistency in sentencing is important, but the
consistency that is sought is consistency in the application of the relevant legal principles, not some numerical or mathematical equivalence. Consistency in sentencing federal offenders is achieved by the proper application of the relevant statutory provisions, having proper regard not just to what has been done in other cases but why it was done, and by the work of the intermediate courts of appeal.
Consistency is not demonstrated by, and does not require, numerical equivalence. Presentation of the sentences that have been passed on federal offenders in numerical tables, bar charts or graphs is not useful to a sentencing judge. It is not useful because referring only to the lengths of sentences passed says nothing about why sentences were fixed as they were. ... The consistency that is sought is consistency in the application of the relevant legal principles.
Thus, to the extent that consistency is sought in the sentencing realm, it is the uniform application of the proper sentencing considerations as opposed to numerical equivalence in the quantum and type of sentence that is required.
This response to potential inconsistency stemming from the instinctive synthesis is unsatisfactory for two reasons. The first is that what matters in sentencing, so far as offenders, victims and the community are concerned, is the ultimate penalty that is imposed, and evidence suggests patent inconsistency exists in this area. Secondly, even if consistency is defined as ‘consistency in the application of the relevant legal principles’, it is unattainable against the backdrop of the instinctive synthesis. I now expand on these matters.
(a) Inconsistency (in Outcome) is the Badge of Unfairness
Sentencing is the sharp end of the criminal law and comprises the mechanism through which criminal offenders are punished for their crimes. The punishment consists of a hardship imposed on the offender. From the perspective of the offender, the victim and community, the most important aspect of the sentence is the nature and amount of punishment that is imposed. It is, principally, on this criterion that the effectiveness or adequacy of the sentencing process is evaluated. While there is no universally endorsed meaning of consistency in sentencing, the essence of what is being appealed to is clear: ‘similar offenders who commit similar offences in similar circumstances would be expected to receive similar sentencing outcomes’.
The reasoning process by which this occurs is a secondary consideration – and a remote one at that. In order to maintain the integrity of the process, consistency in outcome is important. This was recognised by Mason J (dissenting) in Lowe v The Queen:
Just as consistency in punishment – a reflection of the notion of equal justice – is a fundamental element in any rational and fair system of criminal justice, so inconsistency in punishment, because it is regarded as a badge of unfairness and unequal treatment under the law, is calculated to lead to an erosion of public confidence in the integrity of the administration of justice.
Even in more recent times, the High Court has endorsed the importance of numerical consistency. In Green v The Queen, French CJ, Crennan and Kiefel JJ stated:
‘Equal justice requires identity of outcome in cases that are relevantly identical. It requires different outcomes in cases that are different in some relevant respect.’ ... Consistency in the punishment of offences against the criminal law is ‘a reflection of the notion of equal justice’ and ‘is a fundamental element in any rational and fair system of criminal justice’. It finds expression in the ‘parity principle’ which requires that like offenders should be treated in a like manner.
Thus, the notion of equal justice is not simply an intrinsically valuable ideal; it has instrumental benefits as well. Justice Garling in Rees v The Queen also observed that inconsistent sentences are ‘likely to lead to an erosion of public confidence in the integrity of the administration of justice’. In a similar vein, former Chief Justice Spigelman of the New South Wales Supreme Court states the absence of consistency threatens the ‘maintenance of the rule of law’, and undermines the integrity of sentencing as a judicial activity.
In theory, the instinctive synthesis approach, with its global approach and resistance to matters of weight being prescribed in the decision, is inimical to consistency in outcome. The theory is consistent with the reality. A number of reports or studies document patent inconsistency in sentencing, including some studies that go back nearly thirty years.
(b) Evidence of Inconsistent Outcomes
More recently, the Australian Law Reform Commission report, Same Crime, Same Time: The Sentencing of Federal Offenders, looked at sentences across Australia involving the same offences (focusing on drug and fraud offences where the courts were all applying the federal sentencing regime), and noted considerable differences in penalties across the jurisdictions.
For example, the report looked at 63 instances of trafficking a commercial quantity of MDMA (3,4-Methylenedioxy-N- methamphetamine, or Ecstasy) during the five-year period 2000–2004. The jurisdictions where most cases occurred were New South Wales, Western Australia and Victoria. Overall, the mean terms (maximum and minimum) combined for all three states were 136 and 66 months, respectively, while for each individual state they were as follows: in New South Wales: 154: 72; Western Australia: 132: 69; Victoria: 66: 39.
For a commercial quantity of heroin there were 155 cases, of which 86 per cent involved this charge only. The mean term for these three states combined was 87: 48, but, once again, there were considerable regional differences for each state; ie, in New South Wales: 81: 48; Western Australia: 169: 70; and Victoria: 65: 43.
The level of inconsistency is also demonstrated by research reports which compare similarly placed offenders who are subjected to vastly different penalties. The most recent example of this is a 2013 report by the Victorian Advisory Council entitled Reoffending Following Sentencing in the Magistrates’ Court of Victoria. One of the purposes of this report was to ascertain whether offenders who were sentenced to imprisonment reoffended at different rates from those sentenced to other sanctions. To this end, the empirical data show that harsh punishment does not discourage offenders from further offending and, in fact, may even result in a higher incidence of future offending. The theory of specific deterrence is false. The report supported this finding – ie, offenders sentenced to imprisonment generally reoffended at a higher rate than those subjected to more lenient dispositions. This information is not new.
The most illuminating aspect of the report for the purposes of this article is the manner in which the conclusion was derived. The methodology involved comparing the recidivism rates of offenders who had been sentenced to imprisonment with those who were subjected to more lenient dispositions, including wholly suspended sentences. This involved controlling the respective samples for factors that could influence the result (eg, prior criminal record, age, offence type and sex). Thus, the methodology involved using matched sub-samples.
It is striking that identically situated offenders could be subjected to such vastly different outcomes. This can only occur against the backdrop of a largely unfettered judicial sentencing discretion, without adequate regard to the need for consistency in the outcome of sentences.
(c) Parity: Even Principles Designed to Achieve Consistency are Admitting of Failure
Further evidence of the limitations of the current sentencing process to achieve outcome consistency is the loose manner in which the parity principle is applied. Parity in sentencing is the principle that offenders who are party to the same offence should, all things being equal, receive the sanction. Formally, the principle can be clearly and simply stated. One of its clearest expressions is by Gibbs CJ (Wilson J agreeing) in Lowe v The Queen, where his Honour states:
The true position in my opinion may be briefly stated as follows. It is obviously desirable that persons who have been parties to the commission of the same offence should, if other things are equal, receive the same sentence, but other things are not always equal, and such matters as the age, background, previous criminal history and general character of the offender, and the part which he or she played in the commission of the offence, have to be taken into account.
However, the relevance of this principle is considerably limited by the fact that it only applies where the differences in the respective penalties are considerable. The importance of this limitation is underlined by the number of different adjectives that have been used to designate the required differences between sentences in order for the parity principle to be used as a basis for altering a sentence. In Green v The Queen, the High Court stated that the difference must be ‘marked’.
In England v The Queen, the Court noted that other adjectives that have been used include: ‘“gross”, ... “glaring” or “manifest” disparity’.
More fully, in Arenilla-Cepeda v The Queen, Johnson J (Macfarlan JA and Davies J agreeing) adopted the comments of Garling J in Rees v The Queen, where it was noted that:
Hence, the discrepancy required to be identified between sentences is one which is not merely an arguable one, but one which is ‘marked’, or ‘clearly unjustifiable’, or ‘manifest ... such as to engender a justifiable sense of grievance’ or else it ‘[appears] that justice has not been done’ ...
A stream cannot rise higher than its source. In the obscurity that is the instinctive synthesis, the application of a principle that is based on consistency and equality is necessarily limited; and, in fact, so limited that it loses much of its utility.
A system which requires gross or glaring error to justify appellate intervention is grossly and glaringly wrong. A minor or fine error should justify sentence reduction. Until the sentencing methodology is sufficiently fine-tuned and clear for that to occur, it will remain jurisprudentially and doctrinally compromised.
Accordingly, the instinctive synthesis leads to inconsistent sentencing in outcomes, and the differences are so stark that the courts unreasonably stretch the bounds of principles that aim to curtail inconsistency.
(d) Moves to Greater Consistency are Negligible
It is relevant to note that there are some aspects of sentencing law which incline to greater consistency in outcome. These are principally in the form guideline judgments; mandatory minimum and standard penalties; and the use of sentencing statistics by courts. However, the overall contribution of these matters to forging greater consistency is negligible. Despite legislative endorsement in four jurisdictions of guideline judgments, the courts have been very reluctant to develop them, and, in a relative sense only, a very small number of offences are subject to a mandatory or presumptive penalty. Further, while the courts are prepared to receive sentencing statistics, they have been steadfast in maintaining the capacity to impose sentences outside the range of the trends represented in the data. It is also illuminating to note that the main moves towards injecting a degree of predictability in sentencing outcomes have emanated from the legislature, not the judiciary. This gives some insight into the measures that need to be taken to abolish the instinctive synthesis. The one exception to this is a degree of convergence that has emerged with regard to the most relevant sentencing objectives in relation to some offence classifications. Thus, we see that for large scale drug and dishonesty offences, the courts have repeatedly stated that general deterrence is a cardinal consideration. However, the prospect of such measures leading to a degree of mathematical consistency is greatly undermined by the fact that the weight accorded to sentencing objectives and considerations is a matter for the sentencing judge.
Inconsistency in outcome, while undesirable, is unlikely to sway the courts or legislature to abolish the instinctive synthesis because, as we have seen, this shortcoming is recognised by the judiciary who instead aim for a different form of consistency: the uniform application of principle. However, even this form of consistency is unattainable in the context of the instinctive synthesis. To understand why requires a brief overview of the sentencing landscape and inquiry.
1 Overview of Sentencing Considerations and Process
The broad considerations that guide sentencing decisions are similar throughout Australia. Each of the respective principal sentencing statutes sets out the purposes and aims of sentencing. Typically, they include deterrence, community protection, rehabilitation and denunciation. While these objectives are often clearly set out, they often point to sentences of a different nature (especially rehabilitation in contrast to general deterrence and specific deterrence) and there is no attempt to prioritise which aim is the most important.
In addition to the purposes of sentencing, the sentencing discretion is guided by the principle of proportionality. A clear statement of the principle of proportionality is found in the High Court case of Hoare v The Queen:
[A] basic principle of sentencing law is that a sentence of imprisonment imposed by a court should never exceed that which can be justified as appropriate or proportionate to the gravity of the crime considered in the light of its objective circumstances.
In Veen v The Queen and Veen v The Queen [No 2], the High Court stated that proportionality is the primary aim of sentencing. It is considered so important that it cannot be trumped even by the goal of community protection, which at various times has also been declared as the most important aim of sentencing.
Proportionality has also been given statutory recognition in all Australian jurisdictions. However a number of statutory incursions into the proportionality principle have also occurred, mainly stemming from the trend towards tougher sentences. In Victoria, for example, serious sexual, drug, arson or violent offenders may receive sentences in excess of that which is proportionate to the offence. Indefinite jail terms may also be imposed for offenders convicted of ‘serious offences’, where the court is satisfied ‘to a high degree of probability’ that the offender is a serious danger to the community. Similar provisions to those operating in Victoria regarding serious violent and sexual offenders, and indefinite sentences also exist in other jurisdictions. The Dangerous Prisoners (Sexual Offenders) Act 2003 (Qld) goes one step further and allows for preventive detention of offenders who have completed their sentence if there is a high degree of probability that they are a serious danger to the community.
In addition to these considerations, the courts in sentencing are required to give effect to aggravating and mitigating factors. There is a considerable degree of variation in the extent to which these factors are set out in the respective legislative schemes. These considerations are set out most expansively in Crimes (Sentencing Procedure) Act 1999 (NSW), which lists 30 relevant factors. Most sentencing statutes only sparsely deal with these considerations. This is not, however, indicative of a legal divergence between the respective jurisdictions; it is because aggravating and mitigating factors are mainly defined by the common law, which continues to apply in all jurisdictions.
There are well over one hundred mitigating and aggravating factors. Mitigating factors can be divided into four categories. The first are those relating to the offender’s response to a charge and include pleading guilty, cooperating with law enforcement authorities, and remorse. The second are factors that relate to the circumstances of the offence and which contribute to, and to some extent explain, the offending. These include mental impairment, duress, and provocation. The third category includes matters personal to the offender, such as youth, previous good character, old age, and good prospects of rehabilitation. The impact of the sanction is the fourth broad type of mitigating factor and includes considerations such as onerous prison conditions, poor health, and public opprobrium.
Important aggravating factors are: prior criminal record; significant level of injury, or damage caused by the offence; high vulnerability of victim; high level of planning; offences committed while on bail or parole; offences committed with others – gangs; breach of trust, and monetary motive for the crime.
Thus, there are broadly three different phases of the sentencing calculus. First, the court must ascertain a penalty that is commensurate with the objective seriousness of the offence. The judge must also identify the key objectives of sentencing, so far as they apply to the particular offence and offender. Thirdly, aggravating and mitigating factors need to be identified and the penalty needs to be adjusted up or down to accommodate those considerations. The complexity of the process is overlaid by the fact that, as discussed below, the content of the limbs that constitute the principle of proportionality is poorly defined.
2 The Sentencing Judge Determines Matters of Weight
Predicting or anticipating the likely outcome of this process is made much harder by the fact that it is for the court to determine the weight to be accorded to any particular aggravating or mitigating factor. There is no effective fetter to prevent courts from giving, say, 40 per cent or 2 per cent weight to a particular consideration, such as remorse, in order to mitigate a penalty, or an aggravating factor such as prior criminality in order to increase the penalty. As noted in DPP (Vic) v Terrick: ‘The proposition that too much – or too little – weight was given to a particular sentencing factor is almost always untestable. This is so because quantitative significance is not to be assigned to individual considerations.’
In Pesa v The Queen, the Court acknowledged that the absence of the attribution of weight to considerations in sentencing decisions made them ‘opaque’:
[So far as weight is concerned] the ultimate sentencing decision is entirely opaque. While the sentencing reasons record the judge’s consideration of the various matters relevant to sentence, the sentencing decision itself is a conclusion arrived at by the process of intuitive synthesis, without the attribution of weight to any individual factor.
3 Consistency in Principle is Untenable: Courts Ignore Decisions of Other Courts or Make No Attempt to Engage with Alternative Jurisprudence
However, the degree of convergence and the extent to which there is a development of principle is overwhelmed by the manifest flexibility reposed in judges, and is highlighted by the glaring inconsistencies regarding matters of supposed principle.
A good example of the inconsistent application of principle relates to the identification and application of mitigating factors. The nature and type of these factors is not closed. However, the growth and expansion of these considerations is often devoid of a clear justification and frequently there is no attempt to reconcile conflicting decisions.
A few examples illustrate the point. Offenders who are not Australian citizens risk deportation if they fail a ‘character test’, which occurs, among other circumstances, if a person is sentenced to imprisonment for a year or more. Deportation is an additional burden that would be faced by the offender. Hence, arguably it should be mitigatory. This was the position taken in Valayamkandathil v The Queen, Guden v The Queen, and DPP (Vic) v Yildirim. However, a different position was taken in Ponniah v The Queen.
In Victoria, it has been held that the consent of a child sex victim does not mitigate a penalty in relation to child sex cases. A different approach is taken in Western Australia. It is not simply that the differences in ‘principle’ emerge across state boundaries.
In Avdic v The Queen, the Victorian Court of Appeal held that an offender who was pregnant at the time of sentence with her first child was not entitled to a sentencing discount on account of the fact that she would be required to raise the child in the prison setting. To this end, the Court simply stated:
The evidence on the plea was that she would be accommodated in a special unit for mothers of young children. I am not satisfied that the evidence before the sentencing judge demonstrated that her pregnancy would render imprisonment more burdensome than for other prisoners. This is not a case where the appellant will be separated from her child by reason of imprisonment.
The reality is that pregnancy is difficult. Raising children is difficult. Undergoing and undertaking these activities is necessarily more difficult in a custodial setting. In Hancock v The Queen, the same Court recognised that pregnancy could be a mitigating factor. The decisions are barely a year apart. In Hancock there is no reference to Avdic.
It is not uncommon for judges to differ in their views about matters of law and principle. But what is irregular and unacceptable is that the same court can take a wholly different position without even acknowledging the existence of an alternative position previously taken by the same court. It is inconsistent with the rule of law for there not to be even an attempt to engage with the alternative jurisprudence. The instinctive synthesis approach undermines the need for jurisprudential rigour in the development and application of the law. The capacity for legal coherence is diminished in a decision-making construct which can at its highest be described as approximate in nature. Acceptance of approximation in the outcome of any particular decision naturally leads to approximation in that field. The concept of consistency in principle is meaningless in this construct.
The prospect of there being a convergence in principle in circumstances where there are hundreds of potential competing considerations and different layers of guidance, in the form of objectives (which are often inconsistent) and no requirement for decision-makers to articulate the emphasis that should be accorded to objectives and considerations, is remote. Certainly, there is no evidence that such consistency does exist.
Another flaw with the instinctive synthesis approach is that it is conceptually inconsistent. Current practice, which was initiated by the judiciary and endorsed by the High Court, enables, and in some situations requires, judges to provide mathematically specific adjustments to the sentence. There are two factors which are already subjected to a precise numerical application.
The High Court in Cameron v The Queen approved of a discount for offenders who plead guilty (and, in the process, the majority of the Court rejected a number of arguments against the discount, including that it constitutes a form of discrimination against offenders who elect to pursue their right to a trial). The normal range of the discount is between 10 per cent and about 30 per cent, depending on the circumstances of the case. In several jurisdictions it is now either conventional or a statutory requirement to indicate the size of the discount.
For example, in R v Thomson, the New South Wales Court of Criminal Appeal issued a guideline judgment stating that a guilty plea will generally be reflected in a 10 to 25 per cent discount on sentence, depending on how early the plea is entered and the complexity of the case. This suggested range relates only to the utilitarian value of a guilty plea to the criminal justice system and does not include additional discounts that may be available – for example, where the guilty plea may be said to evidence remorse. In Lee v The Queen, it was held, where the plea was taken on the second day set for trial, that a 12.5 per cent discount was appropriate. The co-offender received a 20 per cent discount for pleading on arraignment and it was held that the difference was appropriate.
In Western Australia, section 9AA of the Sentencing Act 1995 (WA) permits a court to reduce a sentence by up to 25 per cent for a plea entered into at the first reasonable opportunity. In South Australia, recent legislative changes allow for a guilty plea reduction of up to 40 per cent for an early guilty plea.
Providing assistance to authorities is treated in a similar way to guilty pleas, particularly where it results in the detection and prosecution of other offenders. This benefit is given independent of any reasons or remorse that might be demonstrated by assisting the authorities. Arguably, criminals, in principle, should not be dealt with less severely because they opportunistically decide to give evidence against co-offenders. However, as a matter of public policy, the law encourages those involved in criminal behaviour to betray the confidence reposed in each other by providing a significant discount at the sentencing stage of the criminal justice system. This is especially apposite given that it often places the individual in personal danger.
Assistance to law enforcement officials now enjoys recognition in a number of statutory regimes. In terms of the size of the discount available, it has been held that the discount for a plea of guilty and assistance to authorities should be up to 50 per cent.
The key aspect of these discounts is that they are precise and normally require the court to stipulate the concession that is accorded. In order to make a mathematical subtraction, it is logically necessarily to have an exact figure as the starting point. This is a two-step approach, as is expressly noted by Kirby J in Markarian v The Queen.
Thus, the two-step approach is workable. Moreover, it is arbitrary to accord some sentencing considerations concrete weight, while glossing over the impact of others. If this approach can be implemented for pleading guilty and assisting authorities, there is no conceptual limitation to extending it to all sentencing considerations.
A central problem with the instinctive synthesis is that it leads to unpredictable outcomes. From the rule of law perspective this is inherently undesirable. From the pragmatic perspective, as discussed below, it is even more undesirable because it discourages accused persons from pleading guilty. The level of uncertainty in sentencing determinations has been elevated to new levels recently following the High Court’s decision in Barbaro. In Barbaro, the High Court held that it was impermissible for the prosecution at plea to make a submission regarding the appropriate sentence or even the range of sentences that was appropriate.
In this case, the applicants, Barbaro and Zirilli, pleaded guilty to serious drug offences after the prosecution during plea negotiations expressed its views about appropriate sentencing ranges for each offender, and that it would convey these views to the sentencing judge at the pleas. In the case of Barbaro, the prosecution agreed that the appropriate range was a head sentence of 32 to 37 years with a minimum term of 24 to 28 years, while with Zirilli it was a head sentence of 21 to 25 years with a minimum term of 18 years. The sentencing judge imposed sentences which exceeded these ranges. Barbaro was sentenced to life imprisonment with a non-parole period of 30 years. Zirilli was sentenced to 26 years’ imprisonment with a non-parole period of 18 years.
The sentencing judge at the plea did not seek and refused to receive submissions from the prosecution regarding appropriate sentences. The applicants submitted that this resulted in the sentencing hearing being procedurally unfair and that the sentencing judge precluded herself from taking into account a relevant consideration to sentencing. The High Court rejected the appeal.
In doing so, it overturned earlier authority that stated that the prosecution can, and in some cases is required to, make submissions regarding the appropriate sentencing range. In R v MacNeil-Brown, a majority of the Victorian Court of Appeal (Maxwell P, Vincent and Redlich JJA, Buchanan and Kellam JJA dissenting on this issue) held that ‘the making of submissions on sentencing range is an aspect of the duty of the prosecutor to assist the court’. The Court stated that such a duty applied in two circumstances: (i) when requested by the court, and (ii) when the prosecutor believes there is a significant risk that the court would otherwise make an error regarding the appropriate range.
Thus, the High Court turned the practice of the prosecution indicating a sentencing range on its head. Prior to Barbaro, existing orthodoxy was that in some cases the prosecution had a duty to inform the sentencing judge of the bounds of the appropriate sentence; now the prosecution can never make such a statement.
The Court gave several reasons for this profound reassessment of the prosecution function at sentence. The first is that the prosecution’s view of the appropriate bounds of the appropriate sentence is supposedly a ‘statement of opinion’; not a submission – it ‘advances no proposition of law or fact which a sentencing judge may properly take into account in finding the relevant facts, deciding the applicable principles of law or applying those principles to the facts to yield the sentence to be imposed.’
Further, the Court stated:
A statement of bounds, on its face, purports to identify the points at which conclusions of manifest excess and manifest inadequacy of sentence become open. Leaving aside the evident difficulties which attend such pretended accuracy, it is important to recognise that manifest excess or manifest inadequacy of sentence founds an inference of error in the exercise of the sentencing discretion. But the nature of the error that has been made is not, and cannot be, identified. All that is known is that, because the result ‘upon the facts ... is unreasonable or plainly unjust, the appellate court may infer that in some way there has been a failure properly to exercise the discretion which the law reposes in the court of first instance’. Hence, stating the bounds of the available range of sentences states no proposition of law.
The observation by the Court that the sentencing range is a statement of opinion and hence is irrelevant is flawed to the point of being unintelligible. A statement regarding the appropriate sentencing range is a submission of law which relies on assumptions and conclusions regarding relevant facts. The submission might be wrong. The prosecution might not be aware of all the relevant facts or might misconceive the law. However, the same applies in relation to most submissions made by lawyers in Australian courts in the adversarial process. More importantly, the terminology used to describe the prosecution’s sentencing submission, whether it is an ‘opinion’, ‘view’ or ‘statement of law’, is irrelevant and reflects an obsessiveness to formalism which has the capacity to place the law into disrepute. The irreducible position is that in the adversarial context each party should be permitted to make submissions regarding the outcome of the case. Additionally, Gageler J, in his separate judgment, noted that it is not sensible that on appeal the prosecution can submit that a sentence is outside the available range, but cannot make submission at the plea (before the event) regarding the appropriate range.
The plurality in Barbaro also stated that a prosecution statement about the appropriate range is inappropriate because it could be erroneous. The prosecution view is supposedly not ‘dispassionate’, and is often based on assumed, not actual, facts.
If a party makes a submission to a sentencing judge about the bounds of an available range of sentences, the conclusions or assumptions which underpin that range can be based only upon predictions about what facts will be found by the sentencing judge. ... This serves to demonstrate that bare statement of a range tells a sentencing judge nothing of the conclusions or assumptions upon which the range depends.
This reasoning applies no less to all submissions in an adversarial process. It is rare that each party is aware of all the relevant facts. Yet parties are permitted (and often required) to make submissions on the basis of the material that is available to them. The essence of the adversarial system is that parties advance their interpretation of the facts and law in a manner that they believe will best further their position. In the sentencing realm, all pleas commence with a prosecution account of the manner in which the offence occurred. Any misunderstanding of the facts or misconceptions of the law can be responded to by defence counsel, and any contested facts or legal disputes are decided by the judge. This does not make legal submissions based on contested facts worthless. It simply means that if certain facts are not established the conclusion contained in the submission is compromised and adjustments may need to be made to accommodate the actual state of affairs. But this hardly negates totally the utility of the submission.
Finally, the Court stated that if a judge imposes a sentence within the prosecution range it may suggest that ‘the sentencing judge has been swayed by the prosecution’s view of what punishment should be imposed’.
This observation is verging on the incredulous. The purpose of legal submissions is precisely to sway the court. If the community or offender believes that a court has been persuaded by the views of the prosecution, it cannot in any manner derogate from the integrity of the process – in fact, it is evidence of the workings of the system.
It is, of course, impossible to anticipate the future development of any legal area with total certainty. But legal and logical howlers made in Barbaro represent the low water mark as far as High Court jurisprudence on sentencing is concerned. The exact reasons for the confused reasoning in the case are unclear, but it is evident that the obscurity of sentencing outcomes and opaqueness of the instinctive synthesis is a key premise in the Court’s reasoning. This is an exemplar of bad principle damaging the legal system. Not only are sentencing outcomes even more obscure as a result of the inability of prosecution and defence to discuss their views on a sentence (such discussions are now pointless given that they cannot be expressed in court), but, from a pragmatic perspective, it is likely to result in even fewer accused pleading guilty.
Governments and the community have a strong interest in encouraging more guilty pleas. At the time of writing this article the issue of encouraging more guilty pleas was the topic of a paper by the New South Wales Law Reform Commission. The data that is available, though not conclusive, suggest that plea negotiations are connected to an increase in the rate of guilty pleas. The Office of the Director of Public Prosecutions (NSW) has maintained that prosecutors should be able to negotiate the quantum of the sentence (which is not binding on the court) because it would be ‘a more effective incentive than advice given concerning a discount on sentence, because of the inherent illusoriness of a discount on sentence’. In Canada, the prosecution regularly makes sentence agreements with the defence whereby a defined sentencing range is recommended. Courts are not bound by the joint submission regarding the sentence, but will rarely depart from it because it is assumed that, without a high degree of certainty that the submission will be accepted, defendants will not be induced to plead guilty.
Australian sentencing law now has moved to even higher levels of unpredictability and uncertainty which can only discourage the entering of guilty pleas. Remarkably, this is for reasons that are devoid of any positive countervailing benefits – whether in principle or practice.
In light of the above, it is appropriate to consider in more detail the supposed benefits of the instinctive synthesis. These have been most comprehensively set out by McHugh J in Markarian v The Queen.
First, McHugh J claims that it is artificial and wrong to start with a benchmark sentence because ‘[i]nstead of sentencing this accused for his or her criminality, the judge sentences the person for another crime and adjusts the notional sentence by reference to factors that are additional to the objective circumstances’. Related to this assertion is his claim that, ‘the judge who commences with a notional sentence downplays the importance of mitigation, reformation and rehabilitation in the sentencing process’.
This reasoning is flawed. The first part of the two-tier process is to define a proportionate penalty. This involves setting a penalty that is commensurate with the objective considerations of the offence – not some other offence. The second is to then adjust up and down for aggravating and mitigating considerations, taking into account the objectives of sentencing. Once this has occurred the offender has been sentenced for his or her crime, not some other offence. It is irrelevant whether all of the circumstances of the offence and the offender have not been calibrated into the first step. The assertion that mitigating factors are given a lesser role in a two-tier approach is without foundation. There is no evidence that rehabilitation and mitigation are given a considerable role in the instinctive synthesis approach. The evidence that does exist is to the contrary. As we saw above, the only sentencing variables that carry designated weight are the discounts accorded for pleading guilty and cooperating with authorities. And the reason we can be confident that this is the case is precisely because their implementation is via a two-step process, which stands outside the instinctive synthesis orthodoxy.
Secondly, McHugh J believes that the
first tier of the two-tier approach – unless it is the maximum sentence – is itself derived by an instinctive synthesis of the “objective circumstances” of the case. Or on another view of the two-tier approach, the first-tier sentence is the product of a value judgment that is proportionate to the offence.
There is some validity to this criticism.
As we have seen, in terms of fixing the amount of punishment, the cardinal determinant is the principle of proportionality. The key aspect of the principle is that it has two limbs. The first is the seriousness of the crime and the second is the harshness of the sanction. Further, the principle has a quantitative component – the two limbs must be matched. In order for the principle to be satisfied, the seriousness of the crime must be equal to the harshness of the penalty.
There has been no systematic, doctrinally sound approach to defining the factors that are relevant to proportionality. Rather than positively defining the factors that are relevant to offence severity, it has proved easier to dismiss some considerations as being irrelevant. Factors, such as ‘good character, ... repentance, restitution, possible rehabilitation and intransigence’, have been excluded. However, some factors have been positively identified as relevant to offence seriousness. These include: the consequences of the offence, as well as the level of harm; the victim’s vulnerability and the method of the offence; the offender’s culpability, which turns on such factors as the offender’s mental state, and his or her level of intelligence; the level of sophistication involved; the protection of society; and even the offender’s previous criminal history.
The problem with such a list is that despite its non-exhaustive character it is too particular, and is no more than a non-exhaustive list of common aggravating factors. There is not even an attempt to explain how these considerations relate to and inform the proportionality thesis. Once considerations such as the method of the offence and the victim’s vulnerability are included, there appears to be no logical basis for not including other considerations that are typically thought to increase the severity of an offence, such as breach of trust, the prevalence of the offence, profits derived from the offence, and an offender’s degree of participation. Moreover, once these considerations are accepted as being part of the proportionality principle, there is a possibility that they will be counted twice when it comes to sentencing: first as part of determining the proportionate penalty and again when it comes to offsetting and weighing aggravating and mitigating considerations. The opaqueness of the instinctive synthesis provides no safeguard against this double counting.
It is for this reason that despite the widespread recognition of the principle, there is no convergence in sentences either within or across jurisdictions – not even in those that ostensibly place cardinal emphasis on proportionality in sentencing determinations. The vagaries are so pronounced that it is verging on doctrinal and intellectual fiction to suggest that an objective answer can be given to common sentencing dilemmas, such as how many years of imprisonment is equivalent to the pain felt by an assault victim, or whether a burglar should be dealt with by way of imprisonment or fine, or the appropriate sanction for a drug trafficker.
Some commentators have argued that proportionality is so obscure as to be meaningless. Jesper Ryberg notes that one of the key and damaging criticisms of proportionality is that it ‘presupposes something which is not there, namely, some objective measure of appropriateness between crime and punishment’. The most obscure and unsatisfactory aspect of proportionality is that there is no stable and clear manner in which the punishment can be matched to the crime. Ryberg further notes that to give content to the theory it is necessary to rank crimes, rank punishments and ‘anchor the scales’.
Thus, when it comes to matching the harshness of the punishment to the gravity of an offence, there is considerable speculation about whether it can be done with any degree of objectivity or precision. The main difficulty here is that the two currencies are different. The interests typically violated by criminal offences are physical integrity and property rights. At the upper end of criminal sanctions, the currency is (deprivation of) freedom. The only conceivable way to give content to the proportionality principle is to ascertain the extent to which offenders and victims are set back by various offence and penalty types. This requires research and modelling into the life trajectories of both victims and offenders and for the courts to make clear assumptions regarding the impact of both crimes and punishments. To this end, considerable progress has been made in setting out a framework for injecting coherency and content into the proportionality principle. It has been suggested that the common variable that should be used to calculate both parts of the proportionality thesis is the extent to which key interests of individuals are set back by the crime. This requires collation of the results of existing data and further strategic research into the effects of crime on victims and the impact of the main forms of punishments on offenders.
However, for current purposes the most important point is that the obscurity of one aspect of the sentencing process (proportionality) cannot be used as a basis for further mystifying the process. As we have seen, against the backdrop of the instinctive synthesis the courts have declared proportionality as being the main sentencing determinant. If it is accurate that proportionality is itself devoid of any core content, then any level of rigour or objectivity associated with the instinctive process is illusory and the whole process is exposed as being a judicial whim. The absence of any content to proportionality and the prospect of better refining the concept cannot be used to shore up the use of the instinctive synthesis; to the contrary, it is a forceful reason to abolish it. Given the current shortcomings of proportionality, the only tenable approach is to more fully define its content and this can best occur in the context of a detailed and rigorous assessment and evaluation of the doctrinal underpinnings of the doctrine and empirical research into its constituent features. The most effective catalyst for this approach is a transparent approach to sentencing – not one that is accepting of approximation and subjectivity. Of course, many judgments involve a degree of subjectivity and generalisations. A two-tier approach will not eliminate them totally. It will, however, make them clearer and thereby provide an enhanced opportunity for testing their validity and revisiting or refining their relevance or scope.
The third reason McHugh J supports the instinctive synthesis approach is that he believes it is not possible to make objective additions and subtractions from the starting point:
[E]ven if the judge can correctly assess the first-tier sentence, the judge must still correctly assess the quantum of the increment or decrement for each factor in the process. ... [I]t would require a judge to have the statistical genius and mental agility of a Carl Friedrich Gauss to arrive at the correct sentence using these methods. ... [M]athematical increments and decrements to some pre-determined notional sentence are ‘apt to give rise to error’.
This is patently flawed. As noted in Part IV above, in the context of some mitigating factors, judges already make precise mathematical reductions in penalties. This process can be expanded to include all relevant factors. There is no basis for assuming that it would lead to greater error than is the case with instinctive synthesis. Rather, because of the extra transparency it is likely to expose more error. This is desirable from the perspective of all relevant stakeholders.
Fourthly, McHugh J believes two-tier sentencing is inconsistent with proportionate sentencing:
At the end of the process, the two-tier sentencer must ask whether the result of the additions and subtractions from the objectively determined sentence is proportionate to the accused’s offence. What happens if the judge concludes that the result is not proportionate to the offence? It would be almost a miracle if it was. If the judge tinkers with the quantum of each component in the sentence to achieve a result compatible with the concept of proportionality, the two-tier structure is meaningless, if not a charade.
Once again, this criticism is too strident. If at the end of the process it emerges that a disproportionate sentence evolves from the calibration, there is cause for the sentencer to readjust the calculations – there is no charade associated with this approach. A disproportionate outcome (unless the statute expressly provides for the acceptability of a disproportionate outcome) evinces an error in the calculation, in the same way that an examiner is put on notice of error if the marks add up to more than 100 per cent.
Finally, McHugh J defends the rigour of the instinctive synthesis by stating the measure selected by the judge is not plucked out of the air but is an informed judgment:
One reason why the idea of instinctive synthesis is apparently abhorrent to lawyers who value predictability and transparency in sentencing is that they see the instinct of a sentencing judge as entirely subjective, personal, arbitrary and unconfined. In fact, although a sentencing judge does ultimately select a number, it is not from thin air that the judge selects it. The judicial air is thick with trends, statistics, appellate guidance and, often enough these days, statutory guidance. First, the sentencing judge almost never imposes a sentence for an offence that has been committed for the first time. A sentencing judge may have seen dozens or scores of such cases and develops, through experience, a sense of the relative gravity of offences and the relative circumstances of offenders that dictate the weighting of different factors in the sentencing process. The need to give greater weight to general or specific deterrence in response to crime trends is one factor to which a sentencing judge has special sensitivity. A sentencing judge also has the benefit of collegiate knowledge, both formally through reading the judgments of other judges and informally through interaction with other judges. No one suggests that the judicial robe carries in its seams the wisdom of Solomon, but judicial experience in sentencing is a skill to be respected by the community and other judges.
This statement provides one of the strongest reasons for discarding the instinctive synthesis. Sentencing decisions are too important to leave to chance. The accuracy or probity of the decision should not be dependent on how long the particular judge has been at the bench – the law should be sufficiently clear so that it can be applied properly by even novice judges. Moreover, the considerations that inform a sentencing judge should not be as fickle as the disposition of a judge to consult a colleague and, in the event, should always be transparent and clear. They must be known to the prosecution and defence. Informal dealings with other judges which have the capacity to shape the outcome of a case have no role in an open system of justice; nor, indeed, in any legal system that ascribes to concepts of justice.
Central to the dispute about the instinctive synthesis is the degree of trust and confidence that the community should have in the wisdom and intellect of judges. Judges are intelligent, have been successful lawyers and do not have a vested interest in cases. The community can take this for granted. Yet, despite that, they are first and foremost people, whose judgments and standards are shaped by their experiences and values. And it is for this reason that it is better to have rules (even bad ones) governing human activity than intelligent, well-intentioned, men and women prescribing situational edicts. The most basic rule of law virtue is that the community must be governed by rules, not men. The breadth of the instinctive synthesis stands apart from this. It is a curious artefact, in a modern legal world in which particularity has rightly prevailed over hunches.
As noted by Kirby J in Markarian v The Queen:
With all respect to those of the different opinion, the phrase ‘instinctive synthesis’ sends quite the wrong signals for the law of sentencing in Australia. Who are those who have the ‘instincts’ in question? Only the judges. This is therefore a formula that risks endorsement of the deployment of purely personal legal power. It runs contrary to the tendency in other areas of the law, notably administrative law, to expose to subsequent scrutiny the use of public power by public officials. It is contrary to the insistence of Australian courts, including this Court, that judicial officers must give reasons for their decisions. At this stage in the development of the Australian law of sentencing, this Court should be encouraging, not impeding, transparency and accountability of judicial decision-making. I remain of the view that ‘[i]t is too late (and undesirable) to return to unexplained judicial intuition’. Talk of ‘instinctive synthesis’ is like the breath of a bygone legal age. It resonates with a claim, effectively, to unexplainable and unreviewable power.
While the maxims relating to the undesirability of largely unfettered discretion are ancient, the empirical evidence is recent and it is definitive. Judges are influenced by their political views and values, and life experiences shape their attitudes. It has been established that key drivers of judicial determinations include the judge’s preference for the outcome of the case and their desire for it not to be overturned on appeal.
Judges are understandably outcome driven, but what often makes the outcomes unacceptable are the hidden influences which underpin them. All humans have preferences and biases. The most difficult to negate are those of which the holder is unaware. Judges, like all people, view themselves as being objective and fair while having a bias blind spot when it comes to their own decision-making. Judge Richard Posner in his seminal work, How Judges Think, states that: ‘We use introspection to acquit ourselves of accusations of bias, while using realistic notions of human behavior to identify bias in others’. The default position of people ‘is to assume that their judgments are uncontaminated’ with implicit bias and that ‘judges are inclined to make the same sorts of favourable assumptions about their own abilities that non-judges do’. The truth is otherwise. All people are influenced by their life journey and are ‘more favourably disposed to the familiar, and fear or become frustrated with the unfamiliar’.
The evidence regarding the impact of implicit judicial bias is considerable. The range of traits which influence the outcome of decisions is wide-ranging. Thus, we see that attractive offenders receive more lenient penalties than other accused – except when the attractive appearance is used to facilitate the crime. In one study, 77 per cent of unattractive defendants received a prison term, while only 46 per cent of attractive defendants were subjected to the same penalty. Thus, unattractive people are approximately 50 per cent more likely to be imprisoned than attractive people.
Gender also influences sentences, with a United States study examining over 20 000 records showing that females are treated more leniently than males.
There is firm evidence of judicial bias on the basis of race. Jeffrey Rachlinski et al show that white judges display a strong white preference in their decisions, while black judges display no overall preference. They note that a key way to deal with this is to bring the biases to the surface: ‘[W]hen judges are aware of a need to monitor their own responses for the influence of implicit racial biases, and are motivated to suppress that bias, they appear able to do so’.
Racial discrimination in sentencing has been long documented. In one of the most wide-ranging surveys, using data from over 77 000 offenders that were sentenced, the data revealed that a black defendant who is sentenced in the same court and who commits the same offence and has the same criminal history as a white accused, will receive a 12 per cent longer prison term than a white offender.
Judicial bias extends well beyond race to matters such as socioeconomic background. A recent analysis of child custody cases showed that judges favour wealthy litigants to those who are impoverished, leading to worse case outcomes for people on low incomes.
Victim traits also impact sentencing outcomes. Black offenders who harm white victims were found to receive heavier penalties than when the victim was black, presumably because ‘the judges were also White, and their in-group or worldview was more threatened by criminal conduct against persons from their in-group’.
The political preferences of judges are also important. It has been demonstrated that in the United States, Democrat-appointed judges punish more leniently than Republican-appointed judges. This occurs even in a system where there are mandatory sentences. The judges achieve this by distorting fact finding. As noted by Nicola Gennaioli and Andrei Schleifer, ‘there is now enormous literature indicating that race, gender and the party of the nominating president affect the decisions of appellate judges, especially in politically sensitive cases’.
The mindset of a judge also influences the outcome of criminal cases. In one study, a mock file (where the offender was charged with prostitution) was assigned to judges who were requested to set bail. Half of the judges were instructed to think about their own death before setting bail. It transpired that they set bail at a much higher amount (US$450) compared to the control group (US$50).
The comfort level of a judge affects case outcome. In a recent study, offenders were better treated after rather than before a judicial meal break. A study examining the decisions of a parole court in Israel over a 10 month period, and taking into account over 1000 rulings, ascertained that the single biggest influence on whether a prisoner was granted parole was the length of time that had passed since the judge had a meal break. After the meal breaks, judges would grant parole at the rate of 65 per cent and it would drop to between 0 and 10 per cent as time wore on. The researchers speculated that the reason for this was because
all repetitive decision-making tasks drain our mental resources. We start suffering from ‘choice overload’ and we start opting for the easiest choice. ... And when it comes to parole hearings, the default choice is to deny the prisoner’s request. The more decisions a judge has made, the more drained they are, and the more likely they are to make the default choice. Taking a break replenishes them.
A type of bias that has potentially important implications for sentencing and which, in fact, supports the outcome (though not the reasoning) in Barbaro, is what is known as the ‘anchoring effect’. Research shows that judges (like all people) are affected by the requests and demands of others, including prosecutors and even inexperienced people, regarding their expectation of sentence.
There is a phenomenon called the anchoring effect, which is a bias people form towards evaluating numbers by focusing on a numerical reference point and making adjustments from that point. Most people place disproportionate emphasis on the initial anchor, so far as it impacts on their final figure. One study showed that experienced judges were influenced by submissions regarding sentence, even if they were not made by experts. In the study, a computer science student who was acting in the role of a prosecutor made either a demand for a high sentence (34 months) or low sentence (12 months) for the identical crime (rape). The judges who received the high demand gave a sentence which, on average, was eight months longer than those who received the demand for the lower sentence. This study confirms results in other studies focusing on damages awards in civil cases and in non-legal settings that show that even arbitrary and irrelevant numbers have an anchoring effect.
More subtle studies undertaken in the criminal law setting suggest that prosecution sentencing submissions have a considerable influence on judges, which are only partly moderated by contrary defence submissions. This may be partly because ‘defense attorneys assimilate their sentencing demand to the demand from the prosecutor’, and that is an ‘unintended process’. The suggested reason for this is the sequence in the courtroom:
By granting the defense attorney the right of the last word, the legal system simultaneously grants the prosecutor the right of the first word. This allows the prosecution to introduce a judgmental anchor that determines the final sentence, by influencing the judge not only directly, but also (and predominantly) indirectly via its influence on the defense attorney’s demand.
The above findings provide reasons for proscribing prosecution submissions on sentencing ranges. Thus, the outcome of the decision in Barbaro is not without any empirical foundation. However, while anchoring bias supports the outcome in Barbaro, it cannot justify the decision because legal acceptance of anchoring bias would require a profound alteration to core aspects of the adversarial system, given that the bias also operates in civil cases – selective application to sentencing matters would be an exercise in expediency. Moreover, as noted above, subconscious bias can at least be partly negated by an awareness of its existence.
Most of the above studies relate to implicit biases or extra-legal influences of United States judges. No similar studies have been undertaken in Australia. While Australian judges are appointed differently from their United States counterparts, the judicial role is identical in both countries and hence there is no reason to believe that the reasoning of Australian judges is any less impacted by such considerations. Certainly, there is evidence of irregular patterns of sentencing when it comes to the most disadvantaged sector in the Australian community.
In Australia, the most over-represented group in prison are Indigenous Australians. Their incarceration rate is 15 times higher than the rate for non-Indigenous Australians. A Victorian study comparing the rate of imprisonment for both Indigenous and non-Indigenous offenders confirmed that the former were more likely to have a criminal record (84 per cent of Indigenous offenders had a prior conviction; compared to 75 per cent for non-Indigenous offenders), but even when controlling for this and other known sentencing variables, Indigenous offenders were still more likely to be imprisoned. The reason for these findings is unclear. There are three possible explanations: (i) the researchers did not eliminate all variables that impacted on sentence; (ii) actual bias by judges and magistrates; or (iii) subconscious bias by magistrates and judges. There is no evidence in support of the second alternative. However, the possibility of subconscious bias cannot be ignored. Not because judicial officers have any considered dislike of indigenous offenders, but simply because of the normal tendency for humans to better understand and associate with people similar to them. The narrative, life journey and values of Indigenous accused are largely foreign to courts and hence may be less likely to influence courts than pleas made on behalf of offenders who come from backgrounds more familiar to judges. In any event, there is a need to further explore the reasons for the higher penalties for Indigenous offenders.
The key point from the above discussion is that the considerations that influence judicial decision-making are multi-faceted and many of them have no relationship to the facts of the case or the words of the statute and, often, they are even not known to the judge. In light of this, it is imperative that the actual reasoning process by which important decisions are made is clearly and precisely set out – the more detail the better. Facilitating, and – even worse – encouraging, a global, intuitive, decision-making regime is pro-hunch and anti-law.
It is a process that should be discouraged by the lawmakers. This, too, is recognised by Kirby J in Markarian v The Queen:
[A]s the present appeal demonstrates, appellate courts expounding general principles should encourage revelation at least of the important adjustments that are made by a sentencing judge. They [the courts] should not be encouraging the thought that there descends upon a judicial officer, following appointment, a mystical ‘instinct’ or ‘intuition’ that ensures that he or she will get the sentence right ‘instinctively’. That approach discourages explanation of the logical and rational process that led to the sentence, so far as it can reasonably be given and is useful.
The only way to remedy this is through legislative change. As noted by Posner, judges, like all people, are utility maximisers and gain satisfaction from different aspects of their role, including its prestige and influence. In making decisions, judges give effect to their own preferences, which are contingent upon their ‘background, temperament, training, experience, and ideology, which shape his [or her] preconceptions and thus his [or her] response to arguments and evidence’.
Judges are unlikely to make the sentencing determination process more clear and in the process reduce their capacity to craft a decision affirming what they believe is the appropriate result. This would be inconsistent with human nature. Individuals have a preference to shape the world in light of their preferences and beliefs. There is an innate desire for people to influence their surroundings. From the perspective of judges, it means retaining as much capacity as possible to impose their views on cases before them. This desire is understandable but should not be accommodated in a system governed by rules instead of by men and women.
A two-step approach would, of course, not totally negate implicit judicial bias, but by making it necessary for judges to justify more clearly and precisely their reasoning, it would necessarily curtail the opportunity for judges to shape each sentence according to what they impressionistically believe is the best outcome. There is also evidence that setting out even non-binding sentencing guidelines reduces the influence of subconscious bias in sentencing. At the minimum it will require judges to think more carefully about sentencing decisions and resist any temptation to obfuscate or ‘keep secret’ the underpinnings of their reasoning. The temptation for judges to keep secret their real thinking has been recognised by one prominent former judge. Justice Kirby in Markarian v The Queen stated:
Functional analysis also suggests that talk of judicial ‘instinct’ is ill-advised. If, in reasoning, the judicial officer does make a significant adjustment for a particular factor – measurable in the judge’s opinion in quantitative or percentage terms – the choice before the law is whether that factor should be specifically exposed in the reasons or not. There are many grounds of policy and principle, in such circumstances, why it should be. If it is not identified, the risk that arises is that identified by Hulme J in the Court of Criminal Appeal in this case. Some judges will feel that it is safer, wiser or even essential to keep the process of reasoning secret. That course is good neither for the parties, nor for the community, nor for the discharge of the functions of sentencing, nor for appellate review. With some judicial officers, talk of ‘instinct’ and pure ‘intuition’ might be understood as endorsing a process of sentencing that involves little more than plucking a figure from the air, to use Hulme J’s telling expression.
Owing to the limited nature of human foresight, future situations cannot necessarily be anticipated in advance and, accordingly, in any area of the law a strong argument can be made for maintaining some degree of discretion. Sentencing practice, however, is so nebulous and unconstrained that even the outcome of stock-in-trade cases is unpredictable. As the situation presently stands, sentencing law is so indeterminate that judges are free to switch from one rationale to another as they choose, according to the case or type of case before them, which amounts to the liberty to determine and to switch policy at a whim.
The instinctive synthesis approach leads to inconsistency, both in outcome and principle, and makes sentencing unpredictable. Given the richness and fullness of factors that can influence sentencing and the fact that judges are not required to stipulate the degree of weight or emphasis that is accorded to any particular consideration, it is misconceived to assert that it is possible to attain consistency in principle or individualised justice – certainly, it is impossible to demonstrate the attainment of such objectives. The quest for ‘individualised justice’ in this context is futile; there is no evidence that it is anything other than an individualised judicial preference. The use of the term ‘justice’ in the court’s description of the process is inappropriate – probably reflexive and possibly expedient.
In addition, the instinctive synthesis cuts across developments in other areas of law, whereby increased transparency and particularity is required. It also enables unbridled and undetectable judicial implicit bias to readily influence sentencing decisions. The anomaly that is the instinctive synthesis is jurisprudentially flawed – fundamentally so, and should be discarded as a decision-making methodology.
An incidental disadvantage of the instinctive synthesis is that it provides a disincentive to jurisprudential and doctrinal development of sentencing law. In order to achieve what is considered by the court to be the appropriate outcome, the instinctive synthesis allows the court effectively to ignore the operation of any variable by reducing its impact to negligible weight. A more regimented reasoning process would promote engagement with supposedly relevant considerations.
The instinctive synthesis makes it difficult for progress to be made in the discipline of sentencing. If we do not know why and how a particular conclusion has been reached (that is, why a particular sanction was imposed) it is not possible to subject the decision, in any serious respect, to logical evaluation. At present, we have a process where certain factual data are entered; the data are then subjected to poorly defined variables and a conclusion is formed. Given that the relevant considerations are always to some extent conflicting and are not prioritised in a process of lexical ordering, however, we are normally not adequately informed of the rationale underlying the decision. Conflicting principles without weight are vacuous, since they can be used to justify a broad range of conclusions.
A two-stage process would require more rigour and inject more complexity into an already difficult process. It would make sentencing a more exacting task,  whereby judges would be required to set out their reasoning in greater detail. It would, thus, produce more complex sentencing reasons and compel judges to think more deeply and precisely about their decisions. As a result, considerable benefit would accrue to the community and ultimately to judges, whose decisions would become more legally sound and defensible. As Kirby J noted in Markarian v The Queen, the instinctive synthesis is ‘a formula that risks endorsement of the deployment of purely personal legal power’. This outcome is intolerable in relation to such important legal decisions.
A difficult task is worth undertaking if the potential gains are high. In order to have a coherent, transparent and justifiable sentencing system, the relevant principles must not only be articulated, but also prioritised and weighted. Although this may lead to mistakes being made along the way, the gravest mistake is not to attempt the task at all. By being open and forthright about why we punish people, the courts would make their decisions subject to considered analysis and comment and, thereby, amenable to improvement and refinement, as opposed to merely promulgating existing errors.
[*] Professor and Dean of Deakin University Law School.
The author would like to thank the anonymous reviewers of this article, whose careful reading and detailed comments considerably improved the structure and content of the article.
 C L Ten, Crime, Guilt and Punishment: A Philosophical Introduction (Clarendon Press, 1987) 2; Jeremy Bentham, An Introduction to the Principles of Morals and Legislation (Athlone Press, first published 1789–1823, 1970 ed) 158; Nils Christie, Limits to Pain (Universitetsforlaget, 1981) 19, 48.
 Andrew E Taslitz, ‘The Rule of Criminal Law: Why Courts and Legislatures Ignore Richard Delgado’s Rotten Social Background' (2011) 2 Alabama Civil Rights and Civil Liberties Law Review 79, 81–2.
 See further the discussion in Part III.
 R v Markarian  NSWCCA 8; (2003) 137 A Crim R 497, 505 .
  HCA 25; (2005) 228 CLR 357, 404 .
 As discussed below, this is referred to as the two-tier or two-step approach.
 Markarian v The Queen  HCA 25; (2005) 228 CLR 357, 386 .
 Indeed, the whole area of equity is founded on the tension between legal rules and the need for wider principle to ameliorate the harshness of inflexible rules: see Samantha Hepburn, Principles of Equity and Trusts (Federation Press, 4th ed, 2009) ch 1. Rules (such as no driving over 60 km/h) apply to conclusively resolve an issue or not all, and hence according to Dworkin rules never clash. Where there is an apparent clash of rules, this is because of a failure to fully understand the scope or limits of one or more of the rules. Whereas principles (such as no person should benefit from his or her own wrongdoing) are standards observed because of a requirement for fairness or justice and more often than rules secure individual or group rights. These are of a broader compass and carry a certain amount of weight; several principles can apply to one situation, with the most relevant or important resolving the outcome: see Ronald Dworkin, Taking Rights Seriously (Duckworth, 1977) 22–8, 76–7. To this end, Dworkin uses the concepts of rules and principles in a different manner to theorists such as Neil MacCormick, who distinguishes rules from norms (as opposed to principles). According to MacCormick, norms are rules with a normative component: see Neil MacCormick, Rhetoric and the Rule of Law: A Theory of Legal Reasoning (Oxford University Press, 2005). For an application of MacCormick’s theory, see Bibi Sangha and Robert Moles, ‘MacCormick’s Theory of Law, Miscarriages of Justice and the Statutory Basis for Appeals in Australian Criminal Cases’  UNSWLawJl 11; (2014) 37 University of New South Wales Law Journal 243. For a discussion regarding the action-guiding nature of normative concepts, see J L Mackie, Ethics: Inventing Right and Wrong (Penguin Books, 1977).
 H L A Hart, ‘Discretion’ (2013) 127 Harvard Law Review 652. For a critique of Hart’s view on discretion, see Geoffrey C Shaw, ‘H L A Hart’s Lost Essay: “Discretion” and the Legal Process School’ (2013) 127 Harvard Law Review 693.
 See, eg, Joseph Raz, The Authority of Law: Essays on Law and Morality (Oxford, 1979) ch 11, especially 211, 214–16; Joel Finnis, Natural Law and Natural Rights (Clarendon Press, 1980) 270–6.
 See discussion in Part III of this article.
 Markarian v The Queen  HCA 25; (2005) 228 CLR 357, 403–4  (Kirby J).
 Marvin Frankel, Criminal Sentences: Law without Order (Farrar, Straus and Giroux, 1973). For a critique of his impact, see Lynn Adelman and Jon Deitrich, ‘Marvin Frankel’s Mistakes and the Need to Rethink Federal Sentencing’ (2008) 13 Berkeley Journal of Criminal Law 239.
 Douglas A Berman and Stephanos Bibas, ‘Making Sentencing Sensible’ (2006) 4 Ohio State Journal of Criminal Law 37; Centre for Law and Global Justice, University of San Francisco School of Law, Cruel and Unusual: US Sentencing Practices in a Global Context (2012).
 This reflects a near doubling in the past 20 years: see Lauren E Glaze and Erinn J Herberman, ‘Correctional Populations in the United States, 2012’  Bureau of Justice Statistics Bulletin <http://www.bjs.gov/content/pub/pdf/cpus12.pdf> . This compares to the rate in Australia of 168 per 100 000: see Australian Bureau of Statistics, 4517.0 – Prisoners in Australia, 2012 (2 April 2013) <http://www.abs.gov.au/ausstats/abs@.nsf/Products/5087123B0CCE48C1CA257B3C000DC7CE?opendocument> .
  HCA 2; (2014) 305 ALR 323 (‘Barbaro’).
 See, eg, Terry Hewton, ‘Instinctive Synthesis, Structured Reasoning, and Punishment Guidelines: Judicial Discretion in the Modern Sentencing Process’  AdelLawRw 3; (2010) 31 Adelaide Law Review 79. Hewton argues that there is not a sharp distinction between the two-tier approach and the instinctive synthesis. However, as discussed below, developments post-Markarian v The Queen  HCA 25; (2005) 228 CLR 357 (which is the foundation of the analysis by Hewton) firmly indicate a preference by the High Court for an unfettered approach to sentencing.
 R v Williscroft  VicRp 27;  VR 292, 300 (Adams and Crockett JJ).
 With minor exceptions discussed in Part IV below.
 Markarian v The Queen  HCA 25; (2005) 228 CLR 357.
 Hudson v The Queen (2010) 30 VR 610, 616 (The Court).
 In Melham v The Queen  NSWCCA 121, , the Court stated: ‘The relevant test for the applicant to succeed on this ground [manifest excess] requires the applicant to demonstrate that the sentence was unreasonable or plainly unjust’: Dinsdale v The Queen  HCA 54; (2000) 202 CLR 321,  (Gleeson CJ and Hayne J).
 For discussion of this concept, see R v Creighton  ACTCA 13; R v Hill  SASCFC 79; R v Holland  NSWCCA 65; (2011) 205 A Crim R 429, 441–2  (Schmidt J); R v Sukkar  NSWCAA 140; R v McHarg  NSWCCA 115, – (Johnson J).
 The contrasts are also set out by McHugh J in Markarian v The Queen  HCA 25; (2005) 228 CLR 357, 377–8 , as follows:
By two-tier sentencing, I mean the method of sentencing by which a judge first determines a sentence by reference to the ‘objective circumstances’ of the case. This is the first tier of the process. The judge then increases or reduces this hypothetical sentence incrementally or decrementally by reference to other factors, usually, but not always, personal to the accused. This is the second tier. By instinctive synthesis, I mean the method of sentencing by which the judge identifies all the factors that are relevant to the sentence, discusses their significance and then makes a value judgment as to what is the appropriate sentence given all the factors of the case. Only at the end of the process does the judge determine the sentence.
  HCA 25; (2006) 228 CLR 357.
 Ibid 375  (Gleeson CJ, Gummow, Hayne and Callinan JJ).
  HCA 2; (2014) 305 ALR 323, 330  (French CJ, Hayne, Kiefel and Bell JJ) (emphasis in original) (citations omitted).
  HCA 31; (2013) 248 CLR 483, 494–5  (citations omitted). The importance of individualised justice is also emphasised in Bugmy v The Queen (2013) 249 CLR 571, 592 , 594  (French CJ, Hayne, Crennan, Kiefel, Bell and Keane JJ).
  HCA 64; (2001) 207 CLR 584.
 Ibid 612  (Gaudron, Gummow and Hayne JJ).
 See the dicta of McHugh J, who notes the difficulties of any ‘attempts to give the process of sentencing a degree of exactness which the subject matter can rarely bear’: AB v The Queen  HCA 46; (1999) 165 ALR 298, 303 .
 These reasons expand on those he set out in AB v The Queen  HCA 46; (1999) 165 ALR 298, 303–4 –.
  HCA 45; (2010) 242 CLR 520, 527  (emphasis in original). See also Barbaro  HCA 2; (2014) 305 ALR 323, 328  (French CJ, Hayne, Kiefel and Bell JJ).
 Hili v The Queen  HCA 45; (2010) 242 CLR 520, 535 –.
 Ibid 526 (French CJ, Gummow, Hayne, Crennan, Kiefel and Bell JJ).
 Ibid 527 .
 Ten, above n 1; Nigel Walker, Why Punish? (Oxford University Press, 1991) 1.
 This is reflected by the fact that it is only when courts impose what are perceived to be inappropriate penalties that there are community calls for changes to the sentencing system. A recent example is the community disquiet following the sentencing of Kieran Loveridge for killing Thomas Kelly, which resulted in legislative increases to ‘king hit’ deaths in NSW: see Rick Morton, ‘Family Fury as Thomas Kelly Killer Kieran Loveridge Gets Four Years’, The Australian (online), 9 November 2013 <http://cached.newslookup.com/cached.php?ref_id=111 & siteid=2104 & id=3645037 & t=1383917349> .
 UK Sentencing Council, Analytical Note: The Resource Effects of Increased Consistency in Sentencing (2011) [3.1]. See also Sangha and Moles, MacCormick’s Theory of Law, above n 11, 244.
  HCA 46; (1984) 154 CLR 606, 610–11.
 Green v The Queen  HCA 49; (2011) 244 CLR 462, 473  (emphasis altered) (citations omitted).
 Rees v The Queen  NSWCCA 47, .
 J J Spigelman, ‘Sentencing Guideline Judgments’ (1999) 11 Current Issues in Criminal Justice 5, 8.
 See, eg, Austin Lovegrove, ‘An Empirical Study of Sentencing Disparity among Judges in an Australian Criminal Court’ (1984) 33 Applied Psychology 161; T Vinson et al, Accountability and the Legal System: Drug Cases Termintating in the District Court 1980–1982 (University of New South Wales, 1986); Samantha Jeffries and Christine Bond, ‘Does Indigeneity Matter? Sentencing Indigenous Offenders in South Australia’s Higher Courts’ (2009) 42 Australian and New Zealand Journal of Criminology 47. For observations regarding the difficulty in firmly proving inconsistency in sentencing, see Sarah Krasnostein and Arie Freiberg, ‘Pursuing Consistency in an Individualist Sentencing Framework: If You Don’t Know Where You’re Going, How Do You Know When You’ve Got There?’ (2013) 76 Law and Contemporary Problems 265, 272–3.
 Australian Law Reform Commission, Same Crime, Same Time: Sentencing of Federal Offenders, Report No 103 (2006).
 Australian Law Reform Commission, above n 48, 878–80. In 2008, the High Court in Adams v The Queen  HCA 15; (2008) 234 CLR 143 ruled that there is no difference in drug seriousness for sentencing purposes. Thus, the disparity between sentences for MDMA and heroin is no longer justified.
 Sentencing Advisory Council (Vic), Reoffending Following Sentencing in the Magistrates’ Court of Victoria (2013) 58–9.
 See, eg, Don Weatherburn, Sumitra Vignaendra and Andrew McGrath, ‘The Specific Deterrent Effect of Custodial Penalties on Juvenile Re-offending’ (2009) 132 Crime and Justice Bulletin 1; Daniel S Nagin, Francis T Cullen and Cheryl Lero Jonson, ‘Imprisonment and Reoffending’ (2009) 38 Crime and Justice 115, 145.
 It has been argued that suspended sentences constitute no punishment at all: Mirko Bagaric, ‘Suspended Sentences and Preventive Sentences: Illusory Evils and Disproportionate Punishments’ (1999) 22 University of New South Wales Law Journal 565.
 Sentencing Advisory Council (Vic), above n 51, 25.
  HCA 46; (1984) 154 CLR 606.
 Ibid 609. For a general overview of the principle, see Geraldine Mackenzie and Nigel Stobbs, Principles of Sentencing (Federation Press, 2010), 53–8. See also Postiglione v The Queen  HCA 26; (1997) 189 CLR 295; Nguyen v Western Australia  WASCA 81.
  HCA 49; (2011) 244 CLR 462.
 Ibid 474  (French CJ, Crennan and Kiefel JJ). The majority noted that on appeal a court can reduce a sentence even when it is not manifestly excess to ‘avoid a marked disparity with a sentence imposed on a co-offender’ (citations omitted).
  NSWCCA 274.
 Ibid  (Howie J).
  NSWCCA 267.
  NSWCCA 47.
 Ibid , citing Lowe v The Queen  HCA 46; (1984) 154 CLR 606, 610 (Gibbs CJ) (Wilson J agreeing), 613 (Mason J), 623–4 (Dawson J); Postiglione v The Queen  HCA 26; (1997) 189 CLR 295, 301 (Dawson and Gaudron JJ), 323 (Gummow J), 338 (Kirby J); R v Taudevin  VICSC 7;  2 VR 402, 403 (Hampel AJA), 404 (Callaway JA); DGM v The Queen  NSWCCA 296,  (Latham J) (McColl JA agreeing); Green v The Queen  HCA 49; (2011) 244 CLR 462, 474–5  (French CJ, Crennan and Kiefel JJ), 496–7  (Bell J).
 These are discussed in some length; see Krasnostein and Freiberg, above n 47; Hewton, above n 20.
 Legislative provisions in NSW, Qld, SA, Vic and WA expressly provide for the respective Appeal Courts to provide guideline sentencing judgments: Crimes (Sentencing Procedure) Act 1999 (NSW) pt 3 div 4 ss 36–42A; Penalties and Sentences Act 1992 (Qld) ss 15AA–15AL; Criminal Law (Sentencing) Act 1988 (SA) ss 29A–29B; Sentencing Act 1991 (Vic) ss 6AB(1)(a)–(b); Sentencing Act 1995 (WA) s 143. They are used sparingly in these jurisdictions (and in Vic and WA, to date, not at all), with NSW being the jurisdiction where most guideline judgments have been issued. They are not issued in other jurisdictions.
 The main extensive use of standard penalties in NSW: see Crimes (Sentencing Procedure) Act 1999 (NSW) pt 4 div 1A s 54A(2). Section 54A(2) sets designated penalties for a range of offences, which set out standard non-parole periods where the offender is found guilty after a trial. For application of these provisions, see Muldrock v The Queen  HCA 39; (2011) 244 CLR 120. See also Criminal Law (Sentencing) Act 1988 (SA) s 32A. Most recently, see Sentencing Amendment (Baseline Sentences) Act 2014 (Vic). While presumptive or mandatory penalties can inject greater consistency, they should not be implemented if they will result in disproportionately harsh penalties: see George Zdenkowski, ‘Mandatory Imprisonment of Property Offenders in the Northern Territory’ (1999) 22 University of New South Wales Journal 302; Russell Hogg, ‘Mandatory Sentencing Legislation and the Symbolic Politics of Law and Order'  UNSWLawJl 51; (1999) 22 University of New South Wales Law Journal 262; Neil Morgan, ‘Capturing Crims or Capturing Votes? The Aims and Effects of Mandatories’  UNSWLawJl 52; (1999) 22 University of New South Wales Law Journal 267; Mirko Bagaric and Athula Pathinayake, ‘Mandatory Harsh Penalties for People Smugglers in Australia: Time for Reform’ (2012) 76 Journal of Criminal Law 493.
 See, eg, R v Nguyen ( NSWCCA 238; 2010) 205 A Crim R 106; R v Chandler  QCA 21, where statistics seemed to play an influential role in determining the penalty.
 As noted above, guideline judgments are only issued in NSW and SA. Guideline judgments that have been issued are in the following cases: R v Jurisic  NSWSC 423; (1998) 45 NSWLR 209; R v Henry  NSWCCA 111; (1999) 46 NSWLR 346; R v Ponfield  NSWCCA 435; (1999) 48 NSWLR 327; R v Thomson  NSWCCA 309; (2000) 49 NSWLR 383; R v Whyte  NSWCCA 343; (2002) 55 NSWLR 252; R v Place  SASC 101; (2002) 81 SASR 395; Eldridge v Bates  SASC 1268; (1989) 51 SASR 532; Police v Cadd  SASC 6187; (1997) 69 SASR 150. There is some evidence that guideline judgments can lead to greater consistency in outcome: Lynne A Barnes and Patrizia Poletti, ‘Sentencing Trends for Armed Robbery and Robbery in Company: The Impact of the Guideline in R v Henry’ (2003) 26 Sentencing Trends and Issues <http://www.judcom.nsw.gov.au/publications/st/st26> Patrizia Poletti, ‘Impact of the High Range PCA Guideline Judgment on Sentencing Drink Divers in NSW’ (2005) 35 Sentencing Trends and Issues <http://www.judcom.nsw.gov.au/publications/st/st35/st35.pdf> .
 See R v Ioane  VSCA 84; R v Manners  VSCA 161.
 As noted in Part V below, this will almost certainly require legislative sanction.
 See Mirko Bagaric and Theo Alexander, ‘A Rational Approach to Sentencing White-Collar Offenders in Australia’  AdelLawRw 18; (2014) 34 Adelaide Law Review 317.
 Crimes (Sentencing) Act 2005 (ACT) s 7(1); Crimes Act 1914 (Cth) s 16A(1)(2); Crimes (Sentencing Procedure) Act 1999 (NSW) s 3A; Sentencing Act 1995 (NT) s 5(1); Penalties and Sentences Act 1992 (Qld) s 9; Criminal Law (Sentencing) Act 1988 (SA) s 10(1); Sentencing Act 1997 (Tas) s 3; Sentencing Act 1991 (Vic) s 5(1); Sentencing Act 1995 (WA) s 6.
  HCA 33; (1989) 167 CLR 348, 354 (The Court) (emphasis altered).
  HCA 7; (1979) 143 CLR 458, 467 (Stephen J).
  HCA 14; (1988) 164 CLR 465, 472 (Mason CJ, Brennan, Dawson and Toohey JJ).
 See, eg, R v Channon (1978) 20 ALR 1.
 The Sentencing Act 1991 (Vic) s 5(1)(a) provides that one of the purposes of sentencing is to impose just punishment, and that in sentencing an offender the court must have regard to the gravity of the offence: s 5(2)(c), and the offender’s culpability and degree of responsibility: s 5(2)(d). The Sentencing Act 1995 (WA) states that the sentence must be ‘commensurate with the seriousness of the offence’: s 6(1), and the Crimes (Sentencing) Act 2005 (ACT) s 7(1)(a) provides that the sentence must be ‘just and appropriate’. In the NT and Queensland, the relevant sentencing statute provides that the punishment imposed on the offender must be just in all the circumstances: Sentencing Act 1995 (NT) s 5(1)(a); Penalties and Sentences Act 1992 (Qld) s 9(1)(a), while in South Australia the emphasis is upon ensuring that ‘the defendant is adequately punished for the offence’: Criminal Law (Sentencing) Act 1988 (SA) s 10(1)(j). The need for a sentencing court to ‘adequately’ punish the offender is also fundamental to the sentencing of offenders for Commonwealth matters: Crimes Act 1914 (Cth) s 16A(2)(k). The same phrase is used in the NSW Act: Crimes (Sentencing Procedure) Act 1999 (NSW) s 3A(a).
 See especially Sentencing Act 1991 (Vic) s 6D(b). Serious offenders are, essentially, those who have previously been sentenced to jail for a similar type of offence, except in the case of serious sexual offenders, where the offender must have two prior sexual matters or a sexual and violent prior arising from the same incident. See R v LD  VSCA 311; R v Dooley  VSCA 269; R v Nguyen  VSCA 141 for examples where offenders were wrongly characterised as serious offenders by the sentencing judges.
 Sentencing Act 1991 (Vic) ss 18A–18P. Serious offences include certain homicide offenders, rape, serious assault, kidnapping and armed robbery: s 3.
 Sentencing Act 1991 (Vic) s 18B(1).
 Penalties and Sentences Act 1992 (Qld) s 163; Sentencing Act 1995 (NT) s 65; Criminal Law (Sentencing) Act 1988 (SA) s 23; Sentencing Act 1997 (Tas) s 19; Sentencing Act 1995 (WA) s 98.
 See, eg, Penalties and Sentences Act 1992 (Qld) pt 10; Crimes (Serious Sex Offenders) Act 2006 (NSW); Sentencing Act 1995 (NT) ss 65–78; Sentencing Act 1995 (WA) ss 98–101; Criminal Law (Sentencing) Act 1988 (SA) pt 2 div III; Criminal Code 1924 (Tas) s 392.
 The constitutional validity of this Act was upheld by the High Court in Fardon v The Queen  HCA 46; (2004) 78 ALJR 1519. For a discussion of the Queensland provision, see Mackenzie and Stobbs, above n 56, 210–11. See also Crimes (Sentences) Act 1999 (SA) s 23.
 See Crimes (Sentencing Procedure) Act 1999 (NSW) ss 21A, 24.
 See Bui v DPP (Cth)  HCA 1; (2012) 244 CLR 638, with particular reference to the federal sentencing regime.
 Joanna Shapland identified 229 factors while the Legal Studies Department in a study of Victorian Magistrates’ Courts identified 292 relevant sentencing: Joanna Shapland, Between Conviction and Sentence: The Process of Mitigation (Routledge, 1981) 55; Legal Studies Department, La Trobe University, Guilty, Your Worship: A Study of Victoria’s Magistrates’ Courts (1980). For an overview of the operation of mitigating and aggravating factors, see Mackenzie and Stobbs, above n 56, ch 4; Richard Fox and Arie Freiberg, Sentencing: State and Federal Law in Victoria (Oxford University Press, 2nd ed, 1999).
 See Victorian Sentencing Committee, Sentencing: Report of the Victorian Sentencing Committee (1988) 359–60.
 See Cameron v The Queen (2002) 209 CLR 339.
 See R v Cartwright (1989) 17 NSWLR 243, 252 (Hunt J and Badgery-Parker J); R v El Hani  NSWCCA 162,  (Howie J); TXT v Western Australia  WASCA 28; (2012) 220 A Crim R 266, 270–1  (Buss JA).
 R v Whyte  VSCA 5; (2004) 7 VR 397; Barbaro v The Queen  VSCA 288; (2012) 226 A Crim R 354; Phillips v The Queen (2012) 37 VR 594.
 See R v Tsiaras  VicRp 26;  1 VR 398; Muldrock v The Queen  HCA 39; (2011) 244 CLR 120; R v Verdins  VSCA 102; (2007) 16 VR 269.
 Tiknius v The Queen  NSWCCA 215; (2011) 221 A Crim R 365.
 Va v The Queen (2011) 37 VR 452.
 R v Neilson  QCA 369; R v Kuzmanovski; Ex parte A-G (Qld)  QCA 19.
 Although it has limited weight in relation to white-collar offenders: R v Coukoulis  VSCA 22; (2003) 7 VR 45.
 Gulyas v Western Australia  WASCA 263; (2007) 178 A Crim R 539; R v RLP  VSCA 271; (2009) 213 A Crim R 461.
 R v Osenkowski (1982) 30 SASR 212; R v Skilbeck  SASCFC 35; Elyard v The Queen  NSWCCA 43; (2006) 45 MVR 402.
 Western Australia v O'Kane  WASCA 24; R v Puc  VSCA 159; Tognolini v The Queen [No 2]  VSCA 311.
 Dosen v The Queen  NSWCCA 283; AWP v The Queen  VSCA 41.
 Ryan v The Queen (2001) 206 CLR 267.
 Field v The Queen  NSWCCA 70; Saunders v The Queen  VSCA 93.
 DPP (Vic) v Marino  VSCA 133.
 R v Halls  NSWCCA 251; Porter v The Queen  NSWCCA 145.
 R v Tran  VSCA 52; (2002) 4 VR 457; DPP (Vic) v Grabovac (1998) 92 A Crim R 258; R v Eisenach  ACTCA 2; Royer v Western Australia  WASCA 139; (2009) 197 A Crim R 319; R v El-Chammas  NSWCCA 154.
 R v Yildiz  NSWCCA 97; (2006) 160 A Crim R 218; R v Douglas  VSC 372; (2004) 146 A Crim R 590.
 R v Gray  VicRp 27;  VR 225; R v Basso  VSCA 201; (1999) 108 A Crim R 392; R v AD  NSWCCA 289; (2008) 191 A Crim R 409.
 R v Quin  NSWCCA 16.
 DPP (Vic) v Truong  VSCA 172; Carreras v The Queen (1992) 60 A Crim R 402; A-G (Tas) v Saunders  TASSC 22; Hill v The Queen  TASSC 29; R v Ottobrino  WASCA 207; R v Black  WASCA 26.
 Pesa v The Queen  VSCA 109.
 For an example of where a considerable amount of weight was given to remorse, see CD v The Queen  VSCA 95.
 The amount of weight given to a sentencing factor is only erroneous if it results in a sentence being manifestly excessive or inadequate: DPP (Vic) v Terrick (2009) 23 VR 457.
 Ibid 459  (The Court).
  VSCA 109,  (The Court).
 Migration Act 1958 (Cth) s 501; John Vrachnas et al, Migration and Refugee Law: Principles and Practice in Australia (Cambridge University Press, 3rd ed, 2011) ch 10.
  VSCA 260.
  VSCA 196; (2010) 28 VR 288.
  VSCA 219.
  WASCA 105.
 Clarkson v The Queen  VSCA 157.
 R v SJH  WASCA 40.
  VSCA 172,  (Osborn JA) (‘Avdic’).
  VSCA 199 (‘Hancock’).
 To this end, coherency is used in the same context as Neil MacCormick, to mean norms that are part of a rational and integrated body of knowledge: MacCormick, Rhetoric and the Rule of Law, above n 11. For a good illustration of MacCormick’s Theory of Law, see Sangha and Moles, MacCormick’s Theory of Law, above n 11.
 (2002) 209 CLR 339.
 In NSW and Qld, the court must indicate if it does not award a sentencing discount in recognition of a guilty plea: Crimes (Sentencing Procedure) Act 1999 (NSW) s 22(2); Penalties and Sentences Act 1992 (Qld) s 13(3). In SA, WA and NSW, the courts often specify the size of the discount given. In Victoria, s 6AAA of the Sentencing Act 1991 (Vic) states that when courts provide a discount for a plea of guilty, they must specify the sentence that would have been given in the absence of that discount. The rationale and size of the typical discount in Victoria is discussed in Phillips v The Queen (2012) 37 VR 594. There has been some judicial comment as to the artificiality of s 6AAA given the instinctive synthesis that produces the actual sentence. See Scerri v The Queen (2010) 206 A Crim R 1, 5 – (The Court); R v Flaherty [No 2]  VSC 270; (2008) 19 VR 305. See also Mackenzie and Stobbs, above n 56, 90–1; Fox and Freiberg, above n 86.
  NSWCCA 309; (2000) 49 NSWLR 383.
 See also Charkawi v The Queen  NSWCCA 159; R v Bugeja  NSWCCA 196.
  NSWCCA 169.
 The same discount was accorded in Nakhla v The Queen  NSWCCA 143.
 See the Criminal Law (Sentencing) (Guilty Pleas) Amendment Act 2012 (SA), which introduced ss 10B and 10C into the Criminal Law (Sentencing) Act 1988 (SA).
 See Mackenzie and Stobbs, above n 56, 93–4; Fox and Freiberg, above n 86.
 Malvaso v The Queen  HCA 58; (1989) 168 CLR 227, 239 (Deane and McHugh JJ).
 R v Barber (1976) 14 SASR 388, 390 (Bray CJ). See also DPP (Cth) v AB  SASC 84; (2006) 94 SASR 316.
 See Penalties and Sentences Act 1992 (Qld) s 9(2)(i); Crimes (Sentencing) Procedure Act 1999 (NSW) s 23; Criminal Law (Sentencing) Act 1988 (SA) ss 10(1)(h), 10A; Sentencing Act 1995 (NT) s 5(1)(h); Crimes (Sentencing) Act 2005 (ACT) s 36. There are also similar provisions at the Commonwealth level: see Crimes Act 1914 (Cth) s 16(2)(h).
 For an example of where a 50 per cent discount was allowed, see R v Johnston  VSCA 133; (2008) 186 A Crim R 345, 349–50 – (Nettle JA). For an application of these principles, see Wang v The Queen  NSWCCA 319; Yue Ma v The Queen  NSWCCA 320; R v Nguyen  NSWCCA 331.
  HCA 25; (2005) 228 CLR 357, 402  (Kirby J).
 For a discussion regarding the importance of legal certainty and predictability, see MacCormick, Rhetoric and the Rule of Law, above n 11; Tom Bingham, The Rule of Law (Penguin, 2011). I thank the anonymous referee for this observation.
  HCA 2; (2014) 305 ALR 323.
 Ibid 376  (French CJ, Hayne, Kiefel and Bell JJ).
 Ibid 324  (French CJ, Hayne, Kiefel and Bell JJ).
 Ibid 376  (French CJ, Hayne, Kiefel and Bell JJ), 383  (Gageler J).
 Ibid 374 – (French CJ, Hayne, Kiefel and Bell JJ)
  VSCA 190; (2008) 188 A Crim R 403, 406  (emphasis added).
 Barbaro  HCA 2; (2014) 305 ALR 323, 375  (French CJ, Hayne, Kiefel and Bell JJ).
 Ibid 380  (French CJ, Hayne, Kiefel and Bell JJ) (citations omitted).
 Ibid 382–3 .
 Ibid 378  (French CJ, Hayne, Kiefel and Bell JJ).
 Ibid 378 – (French CJ, Hayne, Kiefel and Bell JJ).
 Ibid 378  (French CJ, Hayne, Kiefel and Bell JJ).
  HCA 2; (2014) 305 ALR 323.
 New South Wales Law Reform Commission, Encouraging Appropriate Early Guilty Pleas: Models for Discussion, Consultation Paper No 15 (2013).
 Office of the Director of Public Prosecutions, Submission No PEGP06 to New South Wales Law Reform Commission, Encouraging Appropriate Early Guilty Pleas: Models for Discussion, 8, quoted in New South Wales Law Reform Commission, above n 152, 69.
 Simon N Verdun-Jones and Adamira A Yijerino, Victim Participation in the Plea Negotiation Process in Canada: A Review of the Literature and Four Models for Reform (Policy Centre for Victim Issues: Research Statistics Division, 2002) 23, cited in New South Wales Law Reform Commission, above n 152, 61.
 The reasons develop on those set out in AB v The Queen  HCA 46; (1999) 165 ALR 298.
 Markarian v The Queen  HCA 25; (2005) 228 CLR 357, 378–9 .
 Ibid 379 .
 See further the discussion below.
 Markarian v The Queen  HCA 25; (2005) 228 CLR 357, 379 .
 Veen v The Queen [No 2]  HCA 14; (1988) 164 CLR 465, 491 (Deane J).
 See also Hoare v The Queen  HCA 33; (1989) 167 CLR 348, 363 (The Court).
 This includes the matters such as use of weapons and whether there was a breach of trust: see Richard G Fox, ‘The Meaning of Proportionality in Sentencing’  MelbULawRw 1; (1994) 19 Melbourne University Law Review 489, 499–500.
 Eg, whether it was intentional, reckless or negligent.
 See Fox, above n 162, 498–501; New South Wales Law Reform Commission, Sentencing, Discussion Paper No 33 (1996) 62–4.
 Veen v The Queen [No 2]  HCA 14; (1988) 164 CLR 465, 474 (Mason CJ, Brennan, Dawson and Toohey JJ).
 R v Mulholland (1991) 1 NTLR 1, 13 where prior convictions were treated as part of the objective circumstances of the offence on the basis that they are relevant to the mens rea of the offender in committing the offence. This view was rejected in R v McNaughton  NSWCCA 242; (2006) 66 NSWLR 566.
 I thank the anonymous referee for this observation.
 See also Jesper Ryberg, The Ethics of Proportionate Punishment: A Critical Investigation (Kluwer Academic, 2004) 2.
 Ibid 184.
 Ibid 185.
 See further Andrew von Hirsch and Andrew Ashworth, Proportionate Sentencing: Exploring the Principles (Oxford University Press, 2005).
 See Mirko Bagaric, ‘Injecting Content into the Mirage that is Proportionality in Sentencing’ (2013) 25 New Zealand Universities Law Review 411; Andrew von Hirsch and Nils Jareborg, ‘Gauging Criminal Harm: A Living-Standard Analysis’ (1991) 11 Oxford Journal of Legal Studies 1.
 In relation to the development of the proportionality principle in particular, the two-tier approach is preferable over the instinctive synthesis because it is only pursuant to the former approach that courts would be required to clearly demarcate the proportionate sentencing range (and then proceed with adjustments for aggravating and mitigating considerations). This transparency is likely to result in robust analysis, refinement and improvement of sentencing ranges – at least there would be the opportunity for concrete analysis and discussion on this issue. I thank the anonymous referee for raising this issue.
 Markarian v The Queen  HCA 25; (2005) 228 CLR 357, 380  (citations omitted).
 Ibid 386 .
 See discussion in Part III(B) for examples where this occurs.
 Markarian v The Queen  HCA 25; (2005) 228 CLR 357, 388 – (emphasis added).
 Of course experience can assist to improve many judgments, however, against a backdrop of an unfettered discretionary process there is no way to measure improvement; as opposed to simply changed or different outcomes. I thank the anonymous reviewer for this observation.
 As noted by Neil MacCormick: ‘A concern for the rule of law is one mark of a civilised society. The independence and dignity of each citizen is predicated on the existence of a “governance of laws, not men”’: Neil MacCormick, ‘Rhetoric and the Rule of Law’ in David Dyzenhaus (ed), Recrafting the Rule of Law (Hart Publishing, 1999) 163, 163. See also MacCormick, Rhetoric and the Rule of Law, above n 11; Bingham, above n 137. I thank the anonymous referee for this observation. For an excellent discussion of MacCormick’s Theory of Law, see Sangha and Moles, MacCormick’s Theory of Law, above n 11.
  HCA 25; (2005) 228 CLR 357, 403–4  (citations omitted).
 For a relatively recent illustration of where the High Court criticised an unfettered discretion as being contrary to notions of ‘fairness and decency’ in the administration of criminal justice, see Lacey v A-G (Qld)  HCA 10; (2011) 242 CLR 573, 583  (French CJ, Gummow, Hayne, Crennan, Kiefel and Bell JJ). The concept on an unfettered discretion is also discussed and criticised (in the context of the limits of the criminal appeal process) in Bibi Sangha and Robert Moles, ‘Mercy or Right? Post-Appeal Petitions in Australia’ (2012) 14 Flinders Law Journal 293. See also Bibi Sangha and Robert Moles, ‘Post-Appeal Review Rights: Australia, Britain and Canada’ (2012) 36 Criminal Law Journal 300. I thank the anonymous reviewer for this observation.
 Nicola Gennaioli and Andrei Schleifer, ‘Judicial Fact Discretion’ (2008) 37 Journal of Legal Studies 1.
 Jennifer K Robbennolt and Matthew Taksin, ‘Can Judges Determine Their Own Impartiality?’ (2010) 41(2) Monitor on Psychology 24.
 Richard A Posner, How Judges Think (Harvard University Press, 2010) 121.
 Timothy D Wilson, David B Centerbar and Nancy Brekke, ‘Mental Contamination and the Debiasing Problem’ in Thomas Gilovich, Dale Griffin and Daniel Kahneman (eds), Heuristics and Biases: The Psychology of Intuitive Judgment (Cambridge University Press, 2002) 185, 190.
 Jeffrey J Rachlinski et al, ‘Does Unconscious Racial Bias Affect Trial Judges?’ (2009) 84 Notre Dame Law Review 1195, 1228.
 Rose Matsui Ochi, ‘Racial Discrimination in Criminal Sentencing’ (1985) 24 (Winter) Judges 6, 53.
 Birte Englich, ‘Heuristic Strategies and Persistent Biases in Sentencing Decisions’ in Steffen Bieneck, Margit E Oswald and Jörg Hupfeld-Heinemann (eds), Social Psychology of Punishment of Crime (Wiley-Blackwell, 2009) 295.
 John E Stewart, ‘Defendant’s Attractiveness as a Factor in the Outcome of Criminal Trials: An Observational Study’ (1980) 10 Journal of Applied Psychology 348.
 D Steffensmier and S Demuth, ‘Does Gender Modify the Effects of Race-Ethnicity on Criminal Sanctioning? Sentences for Male and Female White, Black and Hispanic Defendants’ (2006) 22 Journal of Qualitative Criminology 241. See also David B Mustard, ‘Racial, Ethnic, and Gender Disparities in Sentencing: Evidence from the US Federal Courts’ (2001) 44 Journal of Law and Economics 1, where it is shown that females receive a 12 per cent shorter penalty than males.
 Rachlinski et al, above n 186.
 Ibid 1221.
 Ochi, above n 187.
 Ronald Everett and Roger Wojtkiewicz, ‘Difference, Disparity, and Race/Ethnic Bias in Federal Sentencing’ (2002) 18 Journal of Quantitative Criminology 189; David S Abrams, Marianne Bertrand and Sendhil Mullainathan, ‘Do Judges Vary in Their Treatment of Race?’ (2012) 41 Journal of Legal Studies 347.
 Michele Benedetto Neitz, ‘Socioeconomic Bias in the Judiciary’ (2013) 61 Cleveland State Law Review 137.
 Siegfried L Sporer and Jane Goodman-Delahunty, ‘Disparities in Sentencing Decisions’ in Steffen Bieneck, Margit E Oswald and Jörg Hupfeld-Heinemann (eds), Social Psychology of Punishment of Crime (Wiley-Blackwell, 2009) 379, 390.
 Max M Schanzenbach and Emerson H Tiller, ‘Strategic Judging under the US Sentencing Guidelines: Positive Political Theory and Evidence’ (2006) 23 Journal of Law, Economics and Organisation 24.
 Gennaioli and Schleifer, above n 182.
 Abram Rosenblatt et al, ‘Evidence For Terror Management Theory I: The Effects of Mortality Salience on Reactions to Those Who Violate or Uphold Cultural Values’ (1989) 57 Journal of Personality and Social Psychology 681, cited in Sporer and Goodman-Delahunty, above n 196, 390.
 Shai Danzinger, Jonathan Levav and Liora Avnaim-Pesso, ‘Extraneous Factors in Judicial Decisions’ (2011) 108(17) Proceedings of the National Academy of Sciences <http://www.pnas.org/content/
 Ed Yong, ‘Justice is Served, but More So after Lunch: How Food-Breaks Sway the Decisions of Judges’ on Discover (11 April 2011) <http://blogs.discovermagazine.com/notrocketscience/2011/04/11/justice-is-served-but-more-so-after-lunch-how-food-breaks-sway-the-decisions-of-judges/#.UyA57YUt18F> .
 Colin Miller, ‘Anchors Away: Why the Anchoring Effect Suggests that Judges Should be Able to Participate in Plea Discussions’ (2013) 54 Boston College Law Review 1667.
 Thomas Mussweiler, Birte Englich and Fritz Strack, ‘Anchoring Effect’ in Rüdiger F Pohl (ed) Cognitive Illusions: A Handbook on Fallacies and Biases in Thinking, Judgement and Memory (Psychology Press, 2004) 183.
 See also Stephanos Bibas, ‘Plea Bargaining Outside the Shadow of Trial’ (2004) 117 Harvard Law Review 2463.
 For a discussion of the relevant studies, see Miller, above n 202.
 Birte Englich et al, ‘The Last Word in Court – A Hidden Disadvantage for the Defense’ (2005) 29 Human Behaviour 705, 717.
 Ibid 718.
 Although without positive evidence of bias, it is necessary to be cautious about transposing United States research and experience to Australia. I thank the anonymous reviewer for this observation.
 Australian Bureau of Statistics, above n 18.
 The same overall trend exists in the United States: The Sentencing Project, Reducing Racial Disparity in the Criminal Justice System: A Manual for Practitioners and Policymakers (2nd ed, 2008) 6.
 Sentencing Advisory Council (Vic), Comparing Sentencing Outcomes for Koori and Non-Koori Adult Offenders in the Magistrates’ Court of Victoria (2013).
 However, there has been no wide-ranging study undertaken to negate this possibility. For discussion of this issue, see Chris Cunneen, ‘Racism, Discrimination and Over-Representation of Indigenous People in the Criminal Justice System: Some Conceptual and Explanatory Issues’ (2006) 17 Current Issues Criminal Justice 329. For some guarded observations, see Lucy Snowball and Don Weatherburn, ‘Does Racial Bias in Sentencing Contribute to Indigenous Overrepresentation in Prison?’ (2007) 40 Australia and New Zealand Journal of Criminology 272.
  HCA 25; (2005) 228 CLR 357, 404 .
 Posner, above n 184, 35–6.
 Ibid 249.
 Richard D Hartley, Sean Maddan and Jeffery T Walker, ‘Sentencing Practices under the Arkansas Sentencing Guideline Structure’ (2006) 34 Journal of Criminal Justice 493, where a study showed that following the introduction of sentencing guidelines extra-legal factors had less influence. Thus, while values and biases cannot be eliminated from judicial decision-making, they can be reduced. I thank the anonymous reviewer for this observation.
  HCA 25; (2005) 228 CLR 357, 404–5  (citations omitted) (emphasis added).
 See also Andrew Ashworth, Sentencing and Criminal Justice (Butterwoths, 2nd ed, 1995) 60.
 Lexical ordering is the manner by which certain objects are chosen over others: see John Rawls, A Theory of Justice (Harvard University Press, 1971).
 It has already been described as a painful process: ‘Judges ... frequently confess that the longer they perform the task of sentencing, the less confidence they have that they know what they are doing. Sentencing has been described as the most “painful” and least rewarding of judicial tasks’: Australian Law Reform Commission, Sentencing of Federal Offenders, Interim Report No 15 (1980) xxiii.
  HCA 25; (2005) 228 CLR 357, 403 .