THE SIEVE EFFECT South Africa ’ s conviction rates in perspective

Figures that suggest that only six out of every 100 violent crimes recorded by the South African Police Service result in a conviction are cause for concern. But every country experiences a ‘sieve effect’ – most reported crimes never even make it to court. Comparing South Africa’s success rates to those of England and the United States suggests that the country’s criminal justice system is not as dysfunctional as one might think.

much less a conviction.There are no countries where the majority of cases reported to the police wind up in court.
During the course of investigation, many cases will be determined to be something other than what they appeared at the time of reporting, and some will turn out not to be criminal matters at all.Even in those cases where a suspect is identified and charged, it sometimes turns out that the accused is innocent of wrongdoing.For this reason, a 100% conviction rate would never be desirable.
But it is important to ensure that the criminal justice system diligently applies itself to the cases that come before it.This requires the use of some form of performance measurement, usually the calculation of 'detection rates' to measure the success of investigations, and 'conviction rates' to measure the success of trials.

Definitions
International comparisons between 'success rates' in criminal investigations and prosecutions are difficult because of variations in definitions.Most jurisdictions tally some sort of 'detection rate' or 'arrest rate' as a performance measurement for the police detectives, but this indicator is generated in many different ways, and there are grey areas where it is unclear into which category a case should fit.
T he release of a Law Commission paper on conviction rates dominated South African news headlines recently and has caused further consternation about the functioning of the country's criminal justice system. 1 The study tracked over 15,000 cases initiated in 1997 and 1998 for a period of two years. 2It found that only 6% of serious violent crimes tracked resulted in a conviction within the study period.
Three quarters of the violent crimes reported did not make it to court, and of those that did, nearly as many were withdrawn by the prosecution as went to trial.Of those that did go to trial, nearly as many defendants were acquitted as convicted.
Internationally, however, only a minority of cases reported ever make it to court, much less result in a conviction.The extent of this 'sieve effect', where weak cases are filtered out at various stages of the investigation and prosecution, varies between countries.But everywhere, substantially more cases come into the system than go out the other end.
There are various points from the reporting of an incident until the perpetrator is convicted at which a case can be lost.Many reported cases are never even recorded as crimes by the police.
Internationally, the majority of reported cases never result in the identification or arrest of a suspect, Ted Leggett, Institute for Security Studies ted@iss.co.za

THE SIEVE EFFECT South Africa's conviction rates in perspective
Figures that suggest that only six out of every 100 violent crimes recorded by the South African Police Service result in a conviction are cause for concern.But every country experiences a 'sieve effect' -most reported crimes never even make it to court.Comparing South Africa's success rates to those of England and the United States suggests that the country's criminal justice system is not as dysfunctional as one might think.
For example, where a case is withdrawn because the victim refuses to cooperate, should this case be counted as a police success or failure?Where a uniformed police member makes an arrest on the street for possession of drugs, should this be counted as a 'detected' case?What if the case is cleared up by the confession of an offender incarcerated for something else? Different jurisdictions answer these questions in different ways.Take, for example, one definition from the developed world.In order for a crime to be deemed 'detected' in England and Wales: • a suspect must be identified and notified of the investigation; • there must be sufficient evidence to charge the suspect with a crime; and • the suspect must in fact have been charged, or there must be one of a number of clearly specified reasons why a charge should not be brought. 3 contrast, the South African definition of crimes detected allows the inclusion of three kinds of cases: • cases referred to court after identification of a chargeable suspect; • cases affirmatively determined by the police to be without merit; and • cases withdrawn before charging a suspect, at the request of the victim and with the approval of the prosecutor. 4 In the latter instance, the relevant police standing order makes no reference to the identification of a suspect at the time of withdrawal.

Detection rates
In England and Wales in 2001/2, 5,527,082 crimes were reported to the police, of which 1,291,396 were 'detected' under the definition above (23%). 5This is comparable to the rate at which cases in South Africa were referred to court (a more strict definition of detection) as set out in the Law Commission research, which suggests that South Africa is performing well in this area.
A more direct comparison can be made using annual figures for both countries.Unfortunately, the latest publicly available annual figures on case disposition from the SAPS are from 2000.In that year, the police recorded 2,575,617 serious cases, of which 609,928 were referred to court -also a 23% success rate. 6us, while it is frightening to think that three quarters of crimes recorded by the police never make it to court, this phenomenon is not unique to South Africa.Indeed, insofar as the comparison is valid, it appears that South African detectives are doing well by international standards.
According to the 2001/2 Annual Report of the South African Police Service (SAPS), a total of 3,110,787 serious crimes were under investigation, with a remarkably high overall detection rate of 35%.But this rate is difficult to compare to the British figures, for definitional reasons.
The British total includes a wide range of serious and minor offences, while the South African count only includes 20 major crime types.In addition, the British government calculates detection rates by comparing incoming reports for the financial year to the number of detections in that year, while the SAPS bases its calculations on total cases under investigation.
Keeping these caveats in mind, comparisons of individual crime types yield a mixed picture.South Africa appears to have performed poorly in murder investigations in 2001/2, with a reported detection rate of 25%, compared with the British claim of 87%.
Using the year 2000 annual data, the South African figure rises to 49%, but due to the decline in incoming murder cases in recent years, this figure is artificially high, and still short of the British standard.But the British police only had 886 incoming cases in the last financial year, compared to the more than 21,000 that the SAPS had to deal with.
With regard to the rape of women, on the other hand, England and Wales recorded 9,008 incidents and detected 3,723 (41%), a rate that is lower than that claimed by the SAPS (49%), despite the fact that the SAPS had over five times as many incoming incidents to handle.
The British recorded 121,375 robberies of personal property and cleared 20,199 (17%), while the SAPS detection rate for aggravated robbery is 15% and for common robbery 27%.Thus, as in England and Wales, crime detection rates in South Africa vary considerably by crime type.While definitional issues make direct comparisons difficult, it does not appear that South African detection rates are egregiously poor when compared with the British, with the possible exception of murder.This is remarkable given the incredible differences in development and resources between the two countries.South African murder detectives deal with caseloads that would be unthinkable in the United Kingdom.
Moreover, there are many factors that impede the participation of the South African public in seeing their cases through to the end, such as the inability of poorer South Africans to absorb the costs involved in multiple court appearances.

Conviction rates
In the Law Commission report, only six out of every 100 cases tracked resulted in a conviction after two years, but it must be kept in mind that 75 of these cases never made it to court in the first place.Of cases that made it to court, about a quarter produced convictions within the period of investigation, and another 16% were still on trial.
Most completed cases that make it to court fall into one of three major categories: cases withdrawn prior to plea, convictions, and acquittals.In the Law Commission research nearly half the completed cases were withdrawn by the prosecutor before plea.This phenomenon is also reflected in the statistics kept by the National Prosecuting Authority (NPA).
Unfortunately, the reasons for these withdrawals remain unknown.Since police cooperation is essential from the inception of the case until its final disposition, it is impossible to assign responsibility for these lost cases to the prosecution alone.
But there has been recent progress in this area.The share of cases withdrawn as a percentage of total cases disposed was reduced from 2001 (54%) to 2002 (51%).Despite the fact that the number of cases to court increased dramatically (more than 30%, due to police crime sweeps), the number of cases disposed increased between the two years, and the number of withdrawals actually decreased slightly (Figure 1).
Of the cases where the defendant was asked to plea, 81% resulted in a conviction in 2001 and 82% resulted in a conviction in 2002.But these totals include district court cases, which deal with lesser offences.Looking strictly at the regional courts, which deal with more serious offences, two thirds of the cases for which pleas were entered in 2001 and 2002 resulted in a conviction.This stands in contrast to the cases tracked by the Law Commission, where nearly as many cases resulted in acquittals as convictions.The NPA figures are of course calculated on an annual basis, however, not on the case tracking system used by the Law Commission.
In addition, it might be possible that the sample used by the Law Commission, which only included eight police areas, is not representative of the nation as a whole, or that the definitions of case disposition categories may differ.The Law Commission report focused on violent crimes, and the NPA figures include all crime types.How do these rates compare internationally?Of British 1998 Crown Court cases that made it to plea, 89% resulted in convictions.But of those that actually went to trial (that is, excluding guilty pleas), only 57% resulted in convictions. 7The British success rate is  clearly dependent on a high rate of guilty pleas.
More research would be required into the share of cases resolved by guilty plea in the two jurisdictions in order to make a fair comparison.It is possible that a higher share of pleas in the British system could distort direct comparisons.
The rates captured in Table 1 are derived from comparative annual data on the ratio of convictions per recorded cases. 8 Again, with the exception of murder, South African conviction rates do not seem to be out of kilter with those of more developed countries.In addition, despite increasing caseloads, the number and share of withdrawn cases have declined, and the number of convictions has gone up.

Cold comfort?
Of course, the fact that we are not alone in this situation is little comfort for those hoping to see an offender brought to book.With only six chances in 100 of seeing a violent criminal locked behind bars, some may argue that pressing a case is unlikely to be worth the effort.
It is important to keep in mind, however, that the criminal justice process is supposed to act like a sieve.Many cases recorded by the police turn out to be baseless, or simply insoluble based on the available evidence, regardless of the competency of the investigation.
In addition, a single conviction can incapacitate an offender guilty of a string of offences.While the chances of getting caught on any given reported offence may be low, the chances of a serial offender getting caught sooner or later may be much higher.
The crime-preventive impact of a single arrest can be great.
Lack of confidence in the system creates a vicious cycle, where complainant apathy causes cases to be lost, which in turn generates more complainant apathy.For this cycle to be broken, complainants need to see the big picture, and participate out of a sense of civic duty, not a personal drive for vengeance.
Of course, it is also incumbent on the criminal justice system to make this participation as simple as possible.Compelling victims and witnesses to attend court multiple times without result is a sure formula for reducing the willingness of the public to come forward.
In the end, it will require the joint commitment of the South African public and the state to reverse the momentum of disillusion and re-claim the right to a safe society.

Table 1 :
This approach is more accurate than simply looking at the number of convictions as a share of cases recorded for a particular year, because cases started in one year may be finalised in another.3 Home Office of the United Kingdom, Counting Rules for Recorded Crime, Section H: Detections. 4 South African Police Service Standing Order (General) 325; South African Police Service Annual Report 2001/2.5 Home Office of the United Kingdom, Crime in England and Wales 2001/2.Table 8.04.6 SAPS Crime Information Analysis Centre.7 National Statistics Office of the United Kingdom, Digest 4: Information on the Criminal Justice System in England and Wales.8 Patrick A. Lanagan and David P. Farrington, Crime and Justice in the United States and in England and Wales, 1981-96, US Bureau for Justice Statistics.Comparative conviction rates