Wednesday, 10 October 2012

Where is the value in SACStat?

The Sentencing Advisory Council's new SACStat program (tweeted about by my learned colleague a few days ago) is interesting, colourful and fairly intuitive. Searches of dispositions in the Magistrates' Court are recent (2009 - 2011) and statistically significant (only offences for which at least 50 examples are available will return a result). They are an extension of the Sentencing Snapshots previously compiled by the SAC.

Here's the videos, the text instructions, and the link to the program itself.

But ... who'll use it? And to what end?

For academics, policy makers and journalists (possibly for bloggers; we'll see) it will probably have lots of applications. And it's reassuring that the Powers That Be are at least trying to keep track of this sort of information, and make it publicly available.

For legal practitioners preparing their sentencing submissions, the usefulness of this new resource will be limited. While s 5(2)(b) of the Sentencing Act 1991 does require courts to take into account current sentencing practices, sentencing courts are rarely swayed by raw numbers. In the past I've talked up the Judicial College's Sentencing Manual, which links to short summaries of various types of offending. The SACStats don't go into that level of detail. It's big picture stuff.

Sentencing statistics have been collected and made available in NSW under the JIRS system for years. That system was discussed here back in 2010. But even in a jurisdiction that does not rely on instinctive synthesis, judges are frequently critical of over-reliance on raw data.

In Dodds v R [2007] NSWCCA v R 191, Fullerton J discussed the Court's concerns:

[The sentencing judge] imposed a sentence of imprisonment comprising a non-parole period of 2 years and 1 month dating from 7 December 2007 with a balance of term of 2 years and 2 months.

The challenge to the sentence is that it is manifestly excessive in light of his Honour’s finding that the offence was below the mid-range of objective seriousness. Support for that submission is said to derive from the Judicial Information Research System (JIRS) statistics which reveal that in only nine per cent of sentences for offending against s 112(1) which attracted full-time custody was a non-parole period imposed greater than 2 years and 1 month and only three per cent of sentences attracted a total term of imprisonment greater than or equal to 4 years and 3 months.

The applicant’s counsel frankly conceded that the statistics are a “blunt tool” consistent with this Court having repeatedly cautioned against the utility of referring to the JIRS statistics in assessing the appropriateness of a sentence under review. As Rothman J emphasised in Robertson v R [2007] NSWCCA 270; 177 A Crim R 121 the principal signposts for sentence are the maximum sentence provided for by the Parliament and the general principles of sentencing that apply in the sentencing exercise. His Honour went on to say:

“...The use of statistics as a measure of the appropriateness of the sentence has the effect of creating a self-fulfilling range of sentences, which may bear little relationship to the maximum sentence imposed by the legislature.”

The Crown submitted, correctly in my view, that the inherent limitation on the utility of the statistics is most pronounced in relation to offences which involve a potentially diverse range of criminal conduct as is the case with an offence [of aggravated burglary]. In addition, it is self-evident that the greater number of factors specific to an offender, the smaller the statistical base from which might be identified a range of penalties. That is exemplified in the applicant’s case. There were only four offenders within the applicant’s age range who were also sentenced for a single count following a plea of guilty accompanied by a record of previous conviction of the same type.

In Victoria, which has refused to adopt the mechanical approach to sentencing used elsewhere, the criticism is even more strident.

In Russell v The Queen [2011] VSCA 147, Buchanan JA [beginning at 2] said,

A sentencing judge is required to have regard to, inter alia, current sentencing practices and accordingly will take into account in the instinctive synthesis of relevant sentencing considerations statistics relating to sentences imposed for the offence and the results in comparable cases.

Counsel for the appellant in this case placed statistics at the forefront of his case and subjected a number of other sentencing decisions to a detailed analysis. Counsel pointed to aspects of other cases said to disclose more serious offending than the present case and relied upon lesser sentences in those cases to argue that the judge in the present case erred.

In my opinion, such an approach is misconceived. Cases said to be comparable can do no more than provide a general guide, impression or background. They are not a benchmark that acts as a straightjacket. Further, such an approach runs the risk of adding or subtracting periods of time to reflect aggravating and mitigating factors present or absent in the cases being compared. As Kaye AJA has explained, statistics are an even rougher guide to an appropriate range.

The High Court's observations in Hili v The Queen; Jones v The Queen [2010] HCA 45 (discussed here) were adopted as appropriate to State offences in Hudson v The Queen; DPP v Hudson [2010] VSCA 332 (and, subsequently, in many other cases).

So if the new resource is not particularly useful for preparing submissions (either by counsel for the accused or the prosecution), what is it good for?

I can see SACStats as most useful for advising clients on the merits of a plea or an appeal. Giving clients advice about likely penalties has often been (like the sentencing process itself) more the product of intuition than solid fact. As resources of this kind become increasingly accessible, a lawyer can advise, when asked by a nervous client about the prospect of a particular outcome, that a disposition of that kind has not been given out in years - or is given out 20% of the time, or 50% of the time, or whatever. (Always given the proviso that individual mileage may vary).

It's biggest drawback is that it fails to really differentiate between offenders sentenced for a single offence, and those sentenced for many offences simultaneously. (Unless I'm not using it correctly, which is a strong possibility). If it's to be believed, just less than 5% of people sentenced for not displaying 'L' plates received imprisonment. When I sort the statistics by charge, I see that just over 80% were fined. But what those other charges were, and what sort of history those people had, remains a mystery. Any kind of qualitative analysis is impossible.

Still, coupled with sentencing indications, this kind of information can improve the advice given to clients, and consequently their decisions are more informed - even if the numbers don't cut much ice themselves with the Court of Appeal.

No comments: