Number 31 - December 2004

The Use and Limitations of Sentencing Statistics

Ivan Potas
Director, Research and Sentencing

Introduction

 

 

The use of JIRS sentencing statistics is now commonplace in the criminal courts of New South Wales. However, the source of these statistics, what they measure and their limitations, are not always properly understood by those who may wish to use them.

 

This article aims to provide an appreciation of the value and place of sentencing statistics in the judicial process. Accordingly, it discusses the nature and purpose of sentencing statistics, as well as their strengths and weaknesses. Attention is also given to matters relevant to the proper exercise of the sentencing discretion, as it is through the prism of the judicial discretion that a proper appreciation of sentencing statistics emerges.

 

What is JIRS?

JIRS is an acronym for Judicial Information Research System. It refers to a mine of legal and statistical information stored in electronic form. The Judicial Commission developed and maintains JIRS, and it is hosted on the Commission’s intranet site. Its primary purpose is to provide judicial officers with quick and easy access to relevant information that serves as "an aid to efficient and accurate decision-making".¹ The aim of providing this information is to promote consistency of approach in sentencing.

A significant part of JIRS is devoted to the provision of sentencing information² stored electronically under a number of discrete components, such as Sentencing Principles and Practice, Court of Criminal Appeal Judgments, Court of Criminal Appeal Summaries, and Guideline Judgments. Some of these components are secondary sources prepared by Judicial Commission staff, while others contain the full text of judgments handed down by the courts. JIRS also hosts a legislative component containing both Commonwealth and New South Wales State legislation.

These primary and secondary sources of information are linked within the JIRS system and this, together with the availability of a powerful search engine, enables users to quickly locate relevant information that might otherwise be lost under a mountain of paper.

The nature of JIRS sentencing statistics

The sentencing statistics component of JIRS is divided into three subsets: Children’s Court Statistics, Local Court Statistics and Higher Courts Statistics. Each of these consists of a statistical compilation of decisions handed down by the relevant court and sorted by reference to offence (as defined by legislation) and sentence outcome. In this way the statistics provide a description, both in terms of quantum and kind, of the patterns of sentences imposed by each level of court for particular offences over a particular time frame.

The statistics are sometimes said to represent the "collective wisdom" of the courts in that they provide a statistical summary by offence type of the sentences handed down by the judges and magistrates of New South Wales.

Where do the statistics come from?

The sentencing statistics are derived from court records. The Commission obtains sentencing data for the Local and higher courts from the New South Wales Bureau of Crime Statistics and Research (BOCSAR). In the case of the Children’s Court, sentencing data are obtained directly from the Department of Juvenile Justice. Once received by the Commission, the data are further processed, audited and finally loaded onto JIRS. At that point the sentencing data become available for use by the subscribers to the system.

The principal offence

Only outcomes that satisfy the definition of a "principal offence" are presented in the statistics. If there is only one offence, then that serves as the principal offence. However, an offender who receives a number of sentences in the one case will only have one sentence included in the statistics; the rest are excluded. This means that the number of sentences in the statistics reflects the number of offenders dealt with by the courts. In other words, the statistics are offender-based rather than sentence-based.

With some variation the Commission adopts BOCSAR’s definition of a principal offence. The most important concept to grasp is that when an offender is sentenced for more than one offence, the principal offence is the one that receives the highest penalty. If two different offences receive the same sentence, the offence with the highest non-parole period qualifies as the principal offence. If these are also the same, the offence that has the highest maximum penalty is selected. The Commission retains a hierarchical listing of penalties ranging from full-time imprisonment down to no conviction recorded in order to assist it with selecting the principal offence.

In R v AEM Snr; R v KEM; R v MM,³ the Court of Criminal Appeal (joint judgment of Beazley JA, Wood CJ at CL and Sully J) spoke of an inbuilt limitation in the statistics when it set out the qualifications attendant with the notion of the principal offence. Despite this undoubted limitation, measuring sentencing patterns by reference to the principal offence appears to be the best way of presenting sentencing statistics.

Charges not proven and all non-sentencing outcomes, including forfeiture and compensation orders, are not included in the JIRS statistics. Outcomes under the Mental Health Act 1990, outcomes identified as appeals from the Local Courts and penalties imposed by a judge when exercising the sentencing powers of the Children’s Court, are also not included in JIRS statistics.

When a number of different cases (or files) are brought before a single judge for sentencing, they are treated as a single rather than a multiple case and again only one offence is selected for JIRS.⁴

Are consecutive sentences counted?

Prior to 1999 consecutive sentences qualifying as the principal offence were excluded from the statistics. It was thought that these sentences would distort the otherwise appropriate penalties because they were affected by the application of the totality principle. However, since Pearce v The Queen⁵ it has been a requirement for a sentencing judge to fix an appropriate sentence for each offence before considering the question of cumulation. Thus, since 1999 consecutively imposed principal offences have been included in the general statistics.

Consecutive sentences may also be viewed on a separate graph entitled "Term of sentence — consecutive terms only (since Pearce)". Unlike other graphs, non-parole terms are not provided for this category of principal offence because to do so would not assist a sentencing court.

When there is some kind of accumulation, the statistics do not reveal the total or aggregate sentence imposed on an individual offender. To provide such statistics would be difficult and confusing as they often involve a mix of offences and would distort any attempt to present a pattern or trend in sentencing for a particular offence.

Corrections after a successful appeal

The Judicial Commission maintains a database of Court of Criminal Appeal sentencing decisions. This database is used to correct the higher court statistics whenever sentences are quashed or varied by the Court of Criminal Appeal. However, users should be aware that corrections take time to implement.

It should be recognised that the pattern of sentences reflected in the statistics are dynamic; they do and will change over time. Further, "incorrect" sentences (erroneous sentences removed or varied on appeal) take time to correct. However the need to make corrections is small and where they relate to high volume offences their impact is insignificant and does not affect the overall sentencing patterns.

Local Court and Children’s Court statistics are not corrected when they are changed on appeal. Having regard to the volume of cases this omission is unlikely to affect the general pattern of sentences imposed in these jurisdictions.

Variations up and down

With regard to successful Crown appeals, corrected sentences tend to understate or be at the lower end of the appropriate range because of the application of the principle of double jeopardy in sentencing,⁶ and sometimes also because of considerations relating to delay in bringing forward the appeal.⁷ Thus even when these sentences are corrected they will tend to reflect a conservative sentence rather than one which ought to have been imposed at first instance.⁸ For example, in R v Price⁹ the court held that the overall sentence imposed was "infected with error" and was "manifestly inadequate". It substituted a conservative sentence which it said was not to be regarded as appropriate for the punishment of the offences committed by the offender.¹⁰

Similarly, some sentences imposed at first instance involve cases where the offender has served some period or intermittent periods of pre-sentence custody. Again, in some cases this consideration may result in the imposition of a shorter sentence than otherwise appropriate in order to take into account time already served.¹¹ In such circumstances the sentence reflected in the statistics will understate the length of an appropriate sentence for the offence.

Users of the statistics therefore should be aware of how these considerations may tend to understate the otherwise appropriate level of sentences imposed. Against this, they should be aware that the statistics presented in the graphs are rounded upwards and may tend to overstate the level of sentences imposed. Thus, for example, a sentence of imprisonment of six years and one month for manslaughter will appear as seven years in the graph thereby appearing to inflate the actual sentence imposed. Strictly speaking, the statistical category represented by seven years in the previous example encapsulates all sentences of more than six years and up to and including seven years imprisonment. In many cases, of course, sentences will fall precisely on the number represented in the graph because courts often sentence in whole numbers (for example, full years of imprisonment or in three monthly segments when sentencing in months).

In this respect it should be noted that statistics are intended as a general guide to past sentencing practice only and should not be understood, or approached, as if they exhibit the precision of a Swiss watch.

Fine-tuning the information

A unique feature of JIRS sentencing statistics is that users can specify with some degree of particularity the kind of statistical information they wish to obtain. The range of options provided in the pull down menus is quite extensive. These variables include:

•  circumstances relating to the offender, such as age group, prior record and liberty status at time of offence

•  circumstances relating to the penalty, such as penalty type, term of sentence, term of non-parole period, fine amounts, periods of licence disqualification, and so on.

For example, a user may wish to know what the overall sentencing pattern is for the offence of malicious wounding pursuant to s 35(1)(a) of the Crimes Act 1900 (NSW). A JIRS graph showing the range of penalties imposed for this offence in the higher courts is reproduced in Figure 1.

Figure 1: Sentencing pattern for malicious wounding — all offenders

It shows that 366 offenders were sentenced for malicious wounding as their principal offence in the higher courts over the period January 1997 to December 2003. Of these, 63 per cent were sentenced to a term of full-time imprisonment. Further analysis of these data is possible, including ascertaining the terms of imprisonment and non-parole periods imposed.

For the same offence, one may wish to know, for example, the pattern of sentences of imprisonment imposed in the case of a first offender who pleaded guilty, had no prior convictions and was under 21 years of age at the time of the offence. These variables, amongst others, can be selected from the pull down menus in JIRS to provide a refined set of statistics that filters out all those cases that do not match the chosen criteria. When the preceding variables are selected it can be seen from Figure 2 that the number of cases has reduced considerably, that is from 366 cases to just 19 cases.

 

Figure 2: Sentencing pattern for malicious wounding — offenders who have no priors, plead guilty and are under 21 years of age

Thus the trade-off in attempting to fine-tune the inquiry by specifying more variables is that the number of cases retrieved may be too few to suggest an established pattern of sentencing that can be relied upon with any degree of confidence.

Yet when the fine-tuned pattern in Figure 2 is viewed alongside the overall pattern of sentences displayed in Figure 1 one might infer, for example, that a penalty of imprisonment is generally less likely to be imposed in Figure 2 type cases than in Figure 1 type cases. Intuitively, this conclusion is fortified by the fact that the trend appears to be in the right direction, namely young first offenders who plead guilty are less likely to go to goal than most other offenders. If imprisonment were contemplated, the range of imprisonment terms, as well as the range of non-parole period terms, could be displayed and considered along with all the other material that the court ordinarily takes into account when sentencing.

There are a large number of variables that may be selected from the JIRS pull down menus. In fact, there are many thousands of possible permutations of data that can be obtained for any one offence. In practice an investigation of a pattern of sentencing for a particular offence may involve the examination and analysis of only a small number of graphs. Generally, and as illustrated above, an examination of a number of graphs will give the user a better appreciation (or intuitive feel) for the data and for the general trend (if any) of sentences imposed for offences of the same kind.

The purpose of sentencing statistics

The sentencing discretion

Sentencing statistics inform the exercise of sentencing discretion by assisting the court confirm the appropriateness of a sentence about to be handed down in a particular case. However, to understand how they may help, it is first necessary to appreciate the discretionary nature of the sentencing process itself.

It is trite law that the determination of sentence is a discretionary exercise undertaken in accordance with judicial principles. Sentencing is "not an uncharted sea" — the sentencing judge is aided by the application of well-established common law principles that guide the discretion.¹² The discretion is wide and must be so, as the Court of Criminal Appeal noted recently in R v Way:¹³

"Very clear direction would be expected if the courts were to be required to depart from the breadth of discretion which has been accepted as available in passing sentence and which was confirmed in R v Whyte (2002) 55 NSWLR 252 where Spigelman CJ said at 276 [147]:

‘The maintenance of a broad sentencing discretion is essential to ensure that all of the wide variations of circumstances of the offence and the offender are taken into account. Sentences must be individualised.’"

The need to individualise sentences is important and may help to explain why sentencing statistics cannot be other than a very crude tool in assessing the appropriateness of a particular sentence.

As a general rule, in exercising the sentencing discretion the sentencing judge or magistrate will pay due regard to the maximum penalty and understand that it is generally reserved for the worst category of cases for which it is prescribed.¹⁴ In R v Way¹⁵ the court summarised this principle in the following manner:

"The statutory maximum penalty has been regarded as an expression of the policy of the legislature in providing for the offence (R v Oliver (1980) 7 A Crim R 174 at 177; Gilson v The Queen (1991) 172 CLR 353 at 364), or as a reflection of the seriousness of that offence as perceived by the public (R v H (1980) 3 A Crim R 53 at 65). It has been reserved for the ‘worst type of case falling within the relevant prohibition’: Regina v Tait and Bartley (1979) 46 FLR 386, R v Fernando [1999] NSWCCA 66 at para 227 and Ibbs v The Queen (1987) 163 CLR 447, although the adoption of that phrase is not an occasion for the imposition of a lesser sentence if it is possible to envisage a worse case: Veen v The Queen [No 2] (1988) 164 CLR 465 at 478."

The court proceeds in the knowledge that a lesser penalty than the maximum or a different sentence altogether may be imposed for most offences.¹⁶ Thus the task of the sentencing judge is to consider "where in the range of the conduct covered by the statutory offence, the particular criminal conduct committed by the offender falls".¹⁷

This involves an awareness of the sentencing options that may be relevant to the case at hand and, if the offence carries with it a potential prison sentence, a recognition of the principle that imprisonment is a sentence of last resort.¹⁸

The sentencing judge must consider specified aggravating and mitigating factors if relevant to the case,¹⁹ including common law principles of sentencing, many now reduced to a statutory formulation.²⁰

In the exercise of the sentencing discretion there must always be consideration of the purpose (or purposes) for which the sentence is to be imposed.²¹ Sentencing statistics are silent on this very important consideration and can only record the outcome, not the rationale for the sentence. Further, it has been held that it is wrong to approach sentencing as if automatic consequences flow from the presence or absence of particular factual circumstances:²²

"In every case, what is called for is the making of a discretionary decision in the light of the circumstances of the individual case, and in the light of the purposes to be served by the sentencing exercise."

The sentencing court will generally have regard to both the type and quantum of sentence imposed in prior cases of a similar type. If a pattern of sentencing exists "this will assist the court in interpreting and carrying into effect the policy of the legislature".²³

These considerations provide the road to consistency of approach or even-handedness in sentencing. Put simply, they place emphasis on the principle of proportionality (the idea that, all other things being equal, the more serious the offence the higher the penalty) and the principle of equity or parity (the idea that like cases should be decided alike and unlike cases differently).

The sentence to be imposed should accord also with the moral sense of the community²⁴ — a feat easier to state than apply in times of multiculturalism and changing community standards.²⁵

There is an untested assumption that the pattern of sentences revealed by the statistics reflects the application of these principles. It stands to reason that they should, given that the individual cases of which the statistics are comprised are themselves the product of principled decision-making. It is in this sense that sentencing statistics can be regarded as reliable and helpful. They open up a range of possible sentences on the basis that similar penalties have been imposed in the past because they were considered appropriate. If the pattern of sentences for an offence are sufficiently established and there is nothing exceptional about the instant case, the court may feel confident that a sentence placed appropriately within the range will be an appropriate sentence. Having regard to the discretionary nature of the sentencing decision there are two critical and closely related propositions that should be understood as applying to JIRS statistics:

•  First, such statistics are not intended to fetter, and do not fetter, the exercise of sentencing discretion.

•  Second, sentencing statistics are not intended to be, nor in any sense are they, prescriptive.²⁶ The pattern of sentences is the product of the individual cases. It is the courts that define the pattern, not the pattern that defines what the courts should or should not do.

 

Whether the statistics establish a range and, if so, whether the sentence under consideration should fall within or outside the range are all matters for the proper exercise of the sentencing discretion. The sentencing statistics are provided simply to assist the court in deciding whether the sentence about to be handed down (or indeed, in the case of an appeal, after it has been handed down) is in any way inside or outside the normal range of penalties imposed for similar offences in past cases. They inform but do not govern the sentencing decision.

Principles of accountability

Closely aligned to the proper exercise of sentencing discretion are the accountability principles that apply to the sentencing process. As Gleeson CJ has explained, "in the administration of any law there comes a point beyond which discretion cannot travel" and "no judge has a choice between implementing the law and disobeying it".²⁷ Further, judges do not have a right "to cast off the constraints of legal methodology" simply because they do not approve of the policy of the law.²⁸ Thus although judicial officers are free to decide cases "without fear or favour", they are accountable to the law and lose legitimacy if they do not apply it.

The exercise of the sentencing discretion is subject to many restrictions that are designed to ensure that the policy of the law is applied and the interests of justice are advanced. In this sense the sentencing discretion is very much structured — by the legislation, by the applicable common law principles of sentencing, and by a process designed to afford procedural fairness. Amongst the most important accountability principles are the requirements that a court give reasons for sentence, that the sentencing hearing is conducted in open court and that parties may appeal (or seek leave to appeal) against sentence on the ground that the judge or magistrate "got it wrong".

Sentencing statistics contribute to all these aspects of the process: they are increasingly found in the remarks on sentence and in submissions on appeal. They contribute to the philosophy of "truth in sentencing" by making available statistical information about sentences handed down for particular offences in the past, including information relating to non-parole periods. The cards are on the table; there are no hidden agendas. What is done in a particular case can be understood by reference to the reasons given. The sentence imposed can be compared openly with other cases, with the general pattern of sentences reflected in the statistics and of course with the penalties expressed in the legislation. A reference to the statistics in the remarks on sentence adds to the justification and transparency of the decision and this in turn facilitates the appeal process; the latter itself is governed by the same considerations.

These forms of accountability, together with the well-understood common law principles that epitomise our sentencing system, ensure that courts hand down carefully crafted sentences that in most instances fall within generally accepted boundaries. In this way the policy of individualisation can be pursued without sacrificing the goals of consistency of approach in sentencing, fairness to the accused, and accountability to the law and to the community.

How do statistics help the decision-maker?

Sentencing statistics have been referred to as one of a number of "extrinsic aids", such as relevant case law, specified maximum penalties, standard non-parole periods and guideline judgments. They operate as reference points, benchmarks or sounding boards to enable comparisons to be made with the case at hand.²⁹

As noted in Way, "the concept of a sentencing spectrum is well known to sentencing judges and criminal law practitioners".³⁰ Thus sentencing statistics may assist sentencers by:

• providing information on the range of penalties imposed for past offences of a similar kind — or put a little more cautiously "The use of sentencing statistics is one tool which a court can use to assist it in its task of ascertaining the pattern of sentences"³¹

• identifying penalties that fall outside or at the very extremities of what may appear to be the normal range of penalties imposed for the offence in question.

 

By themselves the spread of penalties (deviations from the mean) does not provide information about unjustified disparities in sentencing. This is because some offences encapsulate a very broad range of behaviours calling for a wide range of penalties (for example, manslaughter) while other offences do not (for example, mid range prescribed concentration of alcohol offences).

In the former circumstance the courts often comment that the statistics are unhelpful. But are they? The statistics confirm what may well be known, namely, that the range of sentences for an offence such as manslaughter is notoriously wide.³² Conversely, the statistics themselves put the court on notice that particular care is needed when sentencing for certain offences precisely because of the wide range of penalties customarily imposed for them and revealed in the statistics.

The availability of sentencing statistics per se promotes informed debate and aids development of legal policy. For example:

• the decisions to promulgate guidelines in R v Jurisic³³ and R v Henry³⁴ were much influenced by a finding that there existed a pattern of inconsistency and leniency in sentencing for dangerous driving and armed robbery offences³⁵

• statistics have been used to indicate that an offence is prevalent and that a guideline judgment is likely to be of greater utility in such circumstances³⁶

• statistics have also been used to identify a pattern of sentences that have been described as "inadequate to a point verging on irresponsibility" and the Court of Criminal Appeal has held that in such circumstances the principle of consistency in sentencing cannot be applied ³⁷
  

• sentencing statistics have also been consulted by the legislative drafting officer in order to set standard non-parole terms for specified offences under Division 1A of Part 4 of the Crimes (Sentencing Procedure) Act 1999 (NSW).³⁸

 

In Way, the Court of Criminal Appeal considered the approach to be taken when sentencing for standard non-parole offences. It is respectfully submitted that the approach there advocated of standing back after a provisional sentence is reached and then reviewing it, has much to commend it.³⁹ It may be that it is at this point that the sentencing statistics are at their most useful, particularly if there is an identifiable sentencing pattern and the proposed sentence can be assessed (and possibly adjusted) against the background of where it sits in the spectrum of past decisions.

The changing patterns of sentencing

When the law changes, the pattern of sentencing generally also changes and consequently the older sentencing patterns are of limited use.⁴⁰ The repealed s 16G of the Crimes Act 1914 (Cth) is but one example of this. Originally construed in R v El Kaharni,⁴¹ s 16G generally required the court to adjust the "starting point" of the sentence by about one-third in order to take into account the absence of any remission rules that previously applied to sentences imposed in New South Wales. When s 16G was repealed, however, cases where there was a plea or conviction recorded after 16 January 2003 were no longer to be discounted. In R v Quoc Phong Dang⁴² Hidden J, with whom Spigelman CJ and Buddin J agreed, said:

"This Court has held that reliance can no longer be placed upon patterns of sentence established before the repeal of s 16G, at least not without an appropriate adjustment: see R v Studenikin [2004] NSWCCA 164 and R v Kevenaar & Ors [2004] NSWCCA 210."

More recently in R v Mas Rivadavia; R v El Akkaoui; R v Aksu,⁴³ Wood CJ at CL (Adams J and Smart AJ concurring) said that the proper approach to sentencing for a federal offence is to determine the correct sentence without taking into account that s 16G existed and was now repealed. However, if regard is had to pre-existing sentencing patterns, it should be understood that they were sentences passed after an s 16G discount. Further, the court held that preconceptions as to how new sentences should compare with the old (for example, by application of a mathematical formula such as increasing the old sentences by a factor of 50 per cent), cannot be allowed to dominate the approach taken under current legislation.⁴⁴

Even without legislative change, community attitudes towards particular offences change over time, as do the penalties. For example, this is well illustrated by the increasing level of punitiveness manifested towards those who commit child sexual assault offences.⁴⁵ It has been held that when sentencing patterns have moved adversely against an accused, a situation often faced by sex offenders who are prosecuted for "stale" offences committed many years ago, such offenders are entitled to be sentenced in accordance with the law and sentencing patterns that existed at the time their offences were committed.⁴⁶ In these circumstances old sentencing statistics, in so far as they exist, may assist the sentencing court.

In many cases JIRS statistics distinguish between sentences imposed before and after a significant change in the law. For example, offences affected by the introduction of sentencing guidelines have been separated so that a user can see the pattern of sentences imposed before and/or after the commencement of the guideline.

In JIRS statistics, a similar policy of separating the sentences has been adopted with regard to standard non-parole period offences. New patterns, however, take some time to evolve, and in the meantime sentencers must make do with whatever information is available, taking care to make appropriate adjustments or adopt a fresh approach if required.⁴⁷ This leads to the incremental creation of new sentencing patterns that become increasingly useful over time.

The normal range

It may sometimes be of assistance to know where within the range the proposed sentence falls, having regard to the fact that an exceptional case will generally be found at the extremities of a range.

Of course there may be cases where it is inappropriate to speak of an established statistical range. For example, a newly created offence will have no pattern of sentencing of its own, the only range being a limit that is provided by the statutory maximum penalty.

Similarly, an offence infrequently prosecuted will have no discernible range, just isolated instances of cases that may or may not be of assistance to the court. Some offences, although relatively common, will have no useful range to speak of because they encompass such a diverse range of behaviour that statistics provide little guidance. The classic example of this is the crime of manslaughter, although even with this offence it may be of some assistance to know that about one in ten manslaughter cases results in a penalty other than full-time imprisonment.

However, even when a statistical range is available caution is required in using this information. Firstly, the range itself may be inappropriate and should not be followed.⁴⁸ Secondly, in a particular case a sentence falling outside the range may be quite appropriate while another falling in the middle of the range may be quite inappropriate. Hence statistics should never be employed uncritically and certainly not without proper regard to all the relevant circumstances of the particular case.

Beware of simply aiming indiscriminately for a sentence in the middle of the range. In R v AEM Snr; R v KEM; R v MM,⁴⁹ the Court of Criminal Appeal said:

"There is another fallacy which sometimes creeps into discussion surrounding the use of statistics in that emphasis is often placed on the median sentence imposed. It is not the Court’s function to sentence at the median range of sentences handed down over a period of time. Its function is to sentence the particular offender by the application of correct sentencing principles commencing with the gravity of the offence."

In Way the Court of Criminal Appeal considered the approach to be taken when sentencing offenders who have committed offences with standard non-parole period terms. In such cases the statutory reference point for offence seriousness shifts from the maximum penalty specified for the offence to the standard non-parole period, the latter marking a reference point for the minimum term of actual imprisonment.⁵⁰

A complication arises because standard non-parole periods are said to apply to trial cases "falling in the middle range of objective seriousness".⁵¹ On the other hand, sentencing statistics reflect outcomes after all relevant considerations, including objective and subjective factors, have been taken into account. Hence it is important to appreciate that there is a distinction between the currently established pattern of non-parole period sentencing statistics as an indicator of offence seriousness and the statutory standard non-parole period as an indicator of offence seriousness: they do not measure, and are not indicators of, the same thing.

Although the principle that the non-parole period must reflect the minimum term of custody an offender must serve as punishment for an offence remains, it is not known how the new statutory marker of the standard non-parole period will affect sentencing practice. For this reason, the Judicial Commission has separated the statistics for offences subject to the standard non-parole period legislation from the same offences before the new legislation applied to them. In time, new patterns of sentencing for these offences may emerge and become a more reliable indicator of the trends in sentencing for these offences.

The usefulness of statistics

Bald statistics cannot provide details relating to the particular circumstances of each case: as such their application must be approached with care and circumspection. While the court can consider the statistics in determining whether a sentence is out of line with existing patterns,⁵² caution is needed when reviewing such statistics.⁵³ Statistics alone cannot be used to determine the appropriateness or inappropriateness of a particular sentence. They can, however, sound warning bells or confirm that the sentence is within range.

How then can sentencing statistics assist the courts achieve sentencing consistency? In many cases sentencing statistics function as one test or indicator that the exercise of sentencing discretion may have miscarried or appears appropriate. Equally, sentencing statistics may be used together with other criteria for determining whether the Crown or the offender should appeal against sentence. Similarly, they may assist an appeal court deliberating on the adequacy of a sentence to determine whether some error has occurred in terms of the principles enunciated in House v The King⁵⁴ and Dinsdale v The Queen.⁵⁵

In R v Bishop,⁵⁶ Hunt CJ at CL held that the sentence being challenged on appeal was not outside the appropriate range of the Judicial Commission’s statistics, noting further that the two decisions cited as authorities to the contrary were decided "without the benefit of the statistics provided by the Judicial Commission of New South Wales". This suggests that a failure to consider statistics at the sentencing hearing may, in some circumstances at least, reduce confidence in the appropriateness of such decisions.

R v Bloomfield

In R v Bloomfield,⁵⁷ Spigelman CJ reviewed a number of decisions from New South Wales, South Australia and Victoria and proceeded to set out the following eight points on the use of sentencing statistics:

"(i) The sentence to be imposed depends on the facts of each case and for that reason bald statistics are of limited use.

(ii) Statistics may be less useful than surveys of decided cases, which enable some detail of the specific circumstances to be set out for purposes of comparison.

(iii) Caution needs to be exercised in using sentencing statistics, but they may be of assistance in ensuring consistency in sentencing.

(iv) Statistics may provide an indication of general sentencing trends and standards.

(v) Statistics may indicate an appropriate range, particularly where a significant majority or a small minority fall within a particular range. Also when a particular form of sentence such as imprisonment is more or less likely to have been imposed.

(vi) Statistics may be useful in determining whether a sentence is manifestly excessive or manifestly inadequate.

(vii) Statistics are least likely to be useful where the circumstances of the individual instances of the offence vary greatly, such as manslaughter.

(viii) The larger the sample the more likely the statistics are likely to be useful."

At the time of writing, a search of the Court of Criminal Appeal judgments component of JIRS revealed Bloomfield had been cited with approval at least 35 times in connection with the use of statistics. It is clearly the leading case on this topic.

In R v Ryan,⁵⁸ Grove J (with whom Ipp JA and Shaw J agreed) commented generally on the proposition from Bloomfield that bald statistics are of limited use and added:

"...however they may provide indications of general sentencing trends and standards, assist in assuring consistency and be useful in determining whether a sentence is manifestly excessive or manifestly inadequate."

In R v BGS,⁵⁹ Virginia Bell J noted that the Chief Justice in Bloomfield had drawn together many of the court’s decisions and had given detailed consideration to the use to which statistical material may be put. However her Honour observed that statistics may be less useful than surveys of decided cases for the purposes of comparing sentences.

Cases versus statistics

While Bloomfield and BGS recognise that analogous cases may be more useful when seeking to treat like cases alike and unlike cases differently, it must be kept in mind that factual assessments in one case must not be used as legal precedents in another. Indeed, in R v George⁶⁰ the Court of Criminal Appeal was critical of what it described as a recent practice of approaching sentence appeals by a search for, and comparison with, sentences passed in other cases. It was there considering what was described as a wholly exceptional case of manslaughter and warned against attempting to correlate degrees of culpability between cases (but not between co-offenders) and of ignoring the proper exercise of judicial discretion in the individual case. Similarly in R v Zhang⁶¹ Wood CJ at CL warned that it is not always helpful to trawl for comparisons with other decided cases and illustrated the futility of attempting to gauge the element of manifest seriousness from a single decision that forms part of a range of cases with widely differing objective and subjective circumstances.

Further, in R v Campillo Vaquuere⁶² Hulme J (with whom Ipp JA and Howie J agreed) said:

"This Court has said on many occasions that it is not a proper approach to sentencing to seek to compare a sentence under challenge directly with that imposed on another offender (who is not a co-offender) simply because the 2 offenders may have similar characteristics and may have committed similar crimes — see eg Morgan (1993) 70 A Crim R 368 at 371; R v Lawson (1997) 142 FLR 323 at 324, per Hunt CJ at CL; R v Speechley [2002] NSWCCA 300 at [39]; R v Merritt [2004] NSWCCA 19 at [62]; R v Trevenna [2004] NSWCCA 43 at [50, 89 and 99]. It is equally, or more, inappropriate to seek to compare fact findings or characterisations under challenge with those made in one or a few of the thousands of past cases and, upon the basis that there are similarities between the cases, urge that the same findings of fact or characterisation are required. Invariably of course, differences between the case where the sentence is being challenged and the one or few in the past are ignored."

This criticism, it is respectfully submitted, must surely relate to the process of selectively isolating individual cases simply because they may favour one side or another and not because they represent a trend or pattern that the court might follow. It must surely not mean that the use of analogous cases generally cannot assist the court in pursuing even-handedness in sentencing, or in pursuing the goal of treating like cases alike and unlike cases differently. However this line of decisions indicates that the sentencing court should focus on the range appropriate to the subjective and objective circumstances of the offence rather than on determining whether the sentence is more or less severe than a particular statistic or a particular sentence imposed in another case.⁶³

Further, it is suggested that as an aid to the sentencing exercise, statistics and analogous cases should not be seen as in competition but as operating in tandem. What works best is to analyse and compare individual cases with other cases in conjunction with the statistics. This enables other decided cases to be located in the statistical range and helps to provide valuable reference points. Assuming the range and the cases are regarded as relevant and useful, placement of the sentence to be imposed within the statistical range or checking the sentence to be imposed against the statistical range then becomes a more concrete and simpler task. However in pursuing this task the sentencer should be satisfied that:

• any prior decisions that are relied upon appear to be good or appropriate decisions

•  the pattern of sentences exposed by the statistics also appears fair and reasonable.

These assessments should be regarded as ancillary to the primary task of determining the sentence by reference to the merits of the case before the court and by application of the relevant law and sentencing principles.

Finally, users of sentencing statistics should have regard to the cautionary observation encapsulated in the following passages from the joint judgment of Gaudron, Gummow and Hayne JJ in Wong v The Queen:

"The actual sentence which a court imposes on an offender reveals very little about the reasons which the court had for fixing that sentence. Contrary to submissions...the sentence itself gives rise to no binding precedent. What may give rise to precedent is a statement of principles which affect how the sentencing discretion should be exercised, either generally or in particular kinds of case. It is, therefore, fundamentally wrong to speak of ‘quantitative aspects’ of discretionary decisions."⁶⁴

"To focus on the result of the sentencing task, to the exclusion of the reasons which support the result, is to depart from fundamental principles of equal justice. 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."⁶⁵

In R v Bezan,⁶⁶ Wood CJ at CL referred to the latter passage from Wong, noting that sentencing is not "a crude numbers exercise" and that a range indicated by the Court of Criminal Appeal "is not a template within which a sentence must necessarily fall". He explained that it nevertheless remains appropriate:

"to look at the sentence imposed by reference to comparative sentences to see whether there is a reason, including a general sentencing discretion, to support the result of a substantial and unexplained departure from an established range in order to ensure that a reasonable degree of consistency is maintained.

…Prima facie, a departure from an established range manifests error in failing to maintain consistency: Griffiths v The Queen (1977) 137 CLR 293; see also R v Whyte (2002) 55 NSWLR 252 at 278 [168] to 280 [179], and 282 [190]."

Some examples of how statistics are used

In R v McCaffrey; R v Rowsell,⁶⁷ Bloomfield was cited and statistics used to show that the sentences imposed were amongst the highest with respect to both the full and minimum terms imposed in that case. Again the court proceeds with care:

"The sentencing statistics before the court, as well as a summary of s 27 cases heard by the Court of Criminal Appeal, are indicative that the sentences imposed on both applicants were among the highest with respect to both the full and minimum terms. While the court can consider the statistics in determining whether a sentence is out of line with existing patterns (R v Visconti (1982) 2 NSWLR 104) nonetheless caution is needed when reviewing such statistics (R v Bloomfield (1998) 44 NSWLR 734)."

In the following passage from R v Hanslow,⁶⁸ Howie J demonstrates the weakness of sentencing statistics, in that they are silent about the factors contributing to the objective seriousness of the offences that make up the range. However, his Honour also illustrates the strength of sentencing statistics, in that, despite shortcomings, they still function as a reference or sounding board of some value for assessing the adequacy of sentence:

"The applicant also relied upon statistics in both the Local and District Courts. The statistics in relation to the Local Court matters are in my view irrelevant. Those in relation to the District Court showed that the sentence imposed upon the applicant came within the top 38 per cent of sentences for offences of this nature. I do not find that result supportive of the argument that the sentence was outside the appropriate range when nothing is known about the amount of money involved, the nature of the frauds or the extent of the criminality in those cases. It does show, however, that the sentence was a heavy one and at the very top of his Honour’s discretion, but I am not satisfied that it exceeded it. I find no assistance from a comparison of non-parole period’s so as to suggest that there is a range in relation to that part of a sentence against which a particular sentence should be judged."

We have seen that an analysis of prior decisions of a similar kind may work in conjunction with sentencing statistics and need not be in opposition. We have also noted that reliance on outcomes alone, be they statistical or case based, will not provide a sufficient justification for determining the existence of an applicable range or the appropriateness of a particular sentence.

In this regard care needs to be exercised in determining whether there is a range at all, even when prior cases rather than sentencing statistics are relied upon. Thus in R v Hoerler,⁶⁹ Spigelman CJ was considering whether the sentencing judge had erroneously identified an established range of between five and 10 years imprisonment for the unlawful killing of very young children by their parents or carers. The sentencing judge had relied on prior decisions in R v Woodland⁷⁰ and R v Ditfort.⁷¹ The Chief Justice observed that the sentencing judge had adopted the starting point for the sentence in Ditfort as the starting point for the sentence before him. Meanwhile counsel for the respondent relied on Ditfort as being at the top of the range. Spigelman CJ said:⁷²

"However, that is to misunderstand the references in that case to ‘a range’. Those references concerned the appropriate range for sentencing in that specific case, including both its objective and subjective circumstances. They were not references to an appropriate ‘range’ for child killing on a charge of manslaughter."

The Chief Justice then said that often neither prior cases nor JIRS statistics would determine the existence of an appropriate range for a particular offence — rather they merely reflect what was regarded as an appropriate sentence in the wide range of circumstances in those particular cases. Whether or not a sentencing pattern emerges therefore:⁷³

"...requires consideration of the whole body of sentences. It is unlikely that any such pattern can be said to have been established unless there have been a significant number of cases covering a wide variety of objective circumstances. Unless that is so, the cases would not encompass the relevant range of objective criminality."

Hoerler reminds sentencers not only of the difficulty of sentencing offenders where there exists a wide range of circumstances calling for a wide range of penalties, but of the care that is required when adopting other sentencing decisions as an authoritative indication or benchmark for the range or appropriate starting point for sentences. Similarly, it reminds sentencers seeking to use sentencing statistics of the need to ensure that there is an established pattern of sentences covering the relevant range of objective criminality for the offence in question. It may, for example, invite consideration of whether there are ranges within the range; that is, whether there can be identified specified categories of an offence within the general range of sentencing for that offence.

Judicial notice

There are no legislative provisions authorising or governing the use of sentencing statistics in the courts yet they have emerged as an important consideration in identifying patterns of sentencing. Although, from time to time, the courts have criticised the use of statistics or have simply commented that they are unhelpful, nobody appears to have challenged the right of a court to seek to ascertain sentencing patterns by referring to them. This right seems to have been accepted under the doctrine of judicial notice which enables a fact in issue or a fact that is relevant to a fact in issue to be taken into account without the need for formal proof. In Woods v Multi-sport Holdings Pty Limited,⁷⁴ McHugh J considered at some length the various categories of judicial notice including the use of statistics in judgments and said:

"Courts have also used published statistics to resolve issues vital to the resolution of litigation and to inform themselves on policy issues. In Aqua Max Pty Ltd v M T Associates Pty Ltd, Gillard J found that one party was entitled to recover a bonus by taking judicial notice of movements in the Consumer Price Index for Melbourne. In R v Henry, the New South Wales Court of Criminal Appeal used sentencing statistics to give a guideline sentencing judgment. In Wong v The Queen, this Court has recently queried the utility of statistics in sentencing. But Wong has nothing to say concerning the right of a court to take judicial notice of statistics in an appropriate case."

Since the emergence of JIRS in the 1990s New South Wales has led other jurisdictions in the use of sentencing statistics in the criminal courts.

 

Conclusion

The High Court has said that the process of sentencing is not a syllogistic exercise but rather "a synthesis of competing features which attempts to translate the complexity of the human condition and human behaviour to the mathematics of units of punishment usually expressed in time or money".⁷⁵ Sentencing statistics alone cannot hope to unravel this complexity but may count as one of the many tools that the court may enlist in seeking to impose a just sentence.

Clearly, sentencing statistics need to be properly understood and applied if they are to contribute to informed decision-making. If they are not properly understood and applied they can lead the court into error.

Statistical information on sentencing in JIRS was never intended to fetter or in any way usurp the judge or magistrate’s discretion to impose an appropriate sentence in a particular case. As the Court of Criminal Appeal observed, there is:⁷⁶

"an enormous difference between recognising and giving weight to the general pattern as a manifestation of the collective wisdom of sentencing judges on the one hand, and on the other hand, forcing sentencing into a strait-jacket of computerisation."

Thus, subject to the qualifications set out in Bloomfield, sentencing statistics can provide useful additional information which assists the sentencing court formulate a sentence that resonates favourably with sentences imposed in the past. It promotes consistency in sentencing.

In a particular case the usefulness of sentencing statistics may vary on a scale of unhelpful to very helpful, and will often depend on the type of case under consideration and the number of cases contained in the statistics.

If the use of sentencing statistics has an Achilles heel, it is this: each case depends upon its own particular circumstances and neither a record of statistical outcomes of past decisions nor a schedule of prior decisions can reliably dictate to a sentencing court the appropriate sentence that should be imposed in a particular case. Once this limitation is appreciated the value of sentencing statistics assumes its proper place.

In a particular case statistics may play a greater or lesser role; in many cases they may make no useful contribution at all. However, when they do apply, they apply in conjunction with a diverse range of considerations that exemplify the complexity of the sentencing process.

Endnotes
1 See the Honourable Chief Justice Gleeson, quoted in "Launch of the SIS" (1990) 2 (13) Judicial Officers’ Bulletin 1.

 

2 JIRS was originally called the SIS (Sentencing Information System), but the name was changed after it was expanded to incorporate information other than sentencing.

3 [2002] NSWCCA 58 at [114], quoting from the Commission’s Explanatory Note, "Explaining the Statistics".

4 In this respect the method of defining principal offence differs from that used by BOCSAR as it would record a principal offence for each file.

5 (1998) 194 CLR 610.

6 See, for example, Dinsdale v The Queen (2000) 202 CLR 321 at 341.

7 R v Hernando (2002) 136 A Crim R 451; R v Walter and Thompson [2004] NSWCCA 304; R v Price [2004] NSWCCA 186.

8 R v Holder & Johnston (1983) 3 NSWLR 245 at 256.

9 [2004] NSWCCA 186.

10 Ibid at [65].

11 See Crimes (Sentencing Procedure) Act 1999, s 47(3).

12 See, for example, the well-cited passage by Street CJ in R v Rushby [1977] 1 NSWLR 594 at 597 and the remarks of McHugh J in AB v The Queen (1999) 198 CLR 111 at [13]–[19].

13 [2004] NSWCCA 131 at [58].

14 Ibbs v The Queen (1987) 163 CLR 147 at 451–452.

15 [2004] NSWCCA 131 at [51].

16 Crimes (Sentencing Procedure) Act 1999, s 21.

17 Cited in R v Way [2004] NSWCCA 131 at [49], quoting Howie J in R v Moon (2000) 117 A Crim R 497 at 534 [68].

18 Crimes (Sentencing Procedure) Act 1999, s 5.

19 Crimes (Sentencing Procedure) Act 1999, s 21A(1) and (2); see also R v Way [2004] NSWCCA 131 and R v Wickham [2004] NSWCCA 193.

20 See ibid and note also the power to impose a lesser penalty for a guilty plea, Crimes (Sentencing Procedure) Act 1999, s 22 and power to reduce penalties for assistance to authorities, s 23.

21 Crimes (Sentencing Procedure) Act 1999, s 3A.

22 R v Engert (1995) 84 A Crim R 67 at 68 per Gleeson CJ.

23 R v Oliver (1980) 7 A Crim R 174 at 177.

24 See, for example, R v Geddes (1936) 36 SR (NSW) 554 at 555 per Jordan CJ; R v Cuthbert (1967) 86 WN (Pt 1) (NSW) 272 at 274.

25 As for the meaning of "community" in this context, see R v Dole [1975] VR 754 at 768.

26 R v Whyte (2002) 55 NSWLR 252.

27 AM Gleeson, "Judicial Legitimacy" (2000) 12 (6) Judicial Officers’ Bulletin 41 at 42.

28 Ibid at 44.

29 R v Way [2004] NSWCCA 131 at [122]. For an unremarkable example of the use of statistics see R v Gokas [2004] NSWCCA 296 at circa [25].

30 R v Way at [49], quoting from the Attorney General’s Second Reading Speech, New South Wales Parliamentary Debates (Hansard), Legislative Assembly, 23 October 2002; R v Cakowski (2003) 141 A Crim R 446 at [41]–[42].

31 R v AEM Snr; R v KEM; R v MM [2002] NSWCCA 58 at [110].

32 R v Hoerler [2004] NSWCCA 184 at 28, citing R v Isaacs (1997) 41 NSWLR 374 at 381; see also R v George [2004] NSWCCA 247 at [46]–[50] and the cases therein cited.

33 (1998) 45 NSWLR 209 at 221.

34 (1999) 46 NSWLR 346 at [109].

35 Ibid at [110].

36 Application by the Attorney General under section 37 of the Crimes (Sentencing Procedure) Act for a Guideline Judgment Concerning the Offence of High Range Prescribed Concentration of Alcohol under section 9(4) of the Road Transport (Safety and Traffic Management) Act 1999 (No 3 of 2002) [2004] NSWCCA 33 at [54].

37 R v Aristodemou (unrep, 30 June 1994, NSW CCA).

38 R v Way [2004] NSWCCA 131 at [49].

39 Ibid at [124].

40 See, for example, R v Maclay (1990) 19 NSWLR 112.

41 (1990) 51 A Crim R 123 at 136–137.

42 [2004] NSWCCA 265 at [13]; see also R v Dang [2004] NSWCCA 269 at [9]; R v Dujeu [2004] NSWCCA 237.

43 [2004] NSWCCA 284.

44 Ibid at [81].

45 See generally G Hazlitt, P Poletti and H Donnelly, Sentencing Offenders Convicted of Child Sexual Assault, 2004, Research Monograph 25, Judicial Commission of New South Wales, Sydney, pp 5–7.

46 See R v MJR (2002) 54 NSWLR 368.

47 See R v Maclay (1990) 19 NSWLR 112 per Gleeson CJ, Hunt and Loveday JJ.

48 R v Aristodemou (unrep, 30 June 1994, NSW CCA, Badgery-Parker J).

49 [2002] NSWCCA 58 at [116].

50 R v Way [2004] NSWCCA 131 at [53].

51 Ibid at [68].

52 R v Visconti (1982) 2 NSWLR 104.

53 R v Bloomfield (1998) 44 NSWLR 734.

54 (1936) 55 CLR 499 at 505.

55 (2000) 202 CLR 321 at 325.

56 (Unrep, 23 September 1996, NSW CCA).

57 (1998) 44 NSWLR 734 at 739.

58 (2003) 141 A Crim R 403 at 441 [46].

59 [1999] NSWCCA 89 at [23].

60 [2004] NSWCCA 247.

61 [2004] NSWCCA 358 at [26].

62 [2004] NSWCCA 271 at [77].

63 See also R v Mastronardi [2004] NSWCCA 273 at [19], citing Hunt CJ at CL in R v Morgan (1993) 70 A Crim R 368 at 371.

64 (2001) 76 ALJR 79 at [57].

65 Ibid at [65].

66 [2004] NSWCCA 342 at [30]–[31].

67 [1999] NSWCCA 363 at [24].

68 [2004] NSWCCA 163 at [32].

69 [2004] NSWCCA 184.

70 [2001] NSWSC 416 per Wood CJ at CL.

71 (Unreported, 17 March 1992, NSWCCA).

72 [2004] NSWCCA 184 at [34].

73 Ibid at [35].

74 (2002) 208 CLR 460 at [69] (footnotes omitted).

75 Weininger v The Queen [2003] HCA 14 at [26] per Gleeson CJ, McHugh, Gummow and Hayne JJ.

76 R v Oliver (1980) 7 A Crim R 174 at 177.

ISSN 1449 6607


Disclaimer
This paper was prepared by officers of the Judicial Commission for the information of the Commission and for the information of judicial officers. The views expressed in the report do not necessarily reflect the views of the Judicial Commission itself but only the views of the officers of the Commission who prepared this report for the Commission.

 

BACK TO TOP