GRIPES OR
GRIEVANCES?
What the Independent
Complaints Directorate
statistics tell us (or not)
David Bruce,
Centre for the Study of Violence and Reconciliation
dbruce@csvr.org.za
When it was created, the Independent Complaints Directorate (ICD) was seen primarily as a mechanism for
investigating and deterring human rights abuses by members of the SAPS. It was motivated by evidence of high
levels of police brutality, including unjustified killings and the use of torture. However, according to ICD
statistics, most complaints received by the ICD are not from victims of police brutality, but from members of
the public dissatisfied with the quality of policing service provided to them. ICD statistics cannot therefore be
used as indicators of overall levels of police brutality.
O
ne of the fruits of the transition to
democracy in South Africa has been the
Independent Complaints Directorate (ICD)
which opened its doors in April 1997.1
The ICD operates independently of the South
African Police Service (SAPS) and has a separate
budgetary allocation (R36.8m in the 2003-4 year),
although the key legal provisions dealing with the
ICD are provisions of the SAPS Act (68 of 1995).2
Understanding the ICD mandate requires looking
both at the legal provisions, and how these have
been interpreted by the ICD in practice. While the
legal mandate allows some flexibility on this issue,
the ICD has tended to identify itself as an
investigative body and places less emphasis on the
question of ensuring that police carry out internal
investigations effectively. Despite this, the ICD does
not in fact investigate most of the cases that it
receives, but tends to focus its investigative
capacity on more serious cases, arguing that it
‘monitors’ police internal investigations in relation
to the rest.
SA CRIME QUARTERLY No 4 JUNE 2003
The ICD is also required by law to itself investigate
all deaths in police custody and as a result of
police action. The high number of these cases
(although they have declined significantly since
1997) has imposed a substantial burden on its
investigative capacity. In addition to its investigative
and monitoring roles, the ICD also makes policy
recommendations to the police. The ICD mandate
incorporates both the SAPS and municipal police
services.
The ICD director is nominated by the Minister of
Safety and Security, although the nomination has to
be confirmed by the Parliamentary Standing
Committee on Safety and Security. The ICD is
required to report to the minister and parliamentary
committee. The ICD has a staff of roughly
130 personnel located in nine provincial offices
and a national office in Pretoria. This includes over
60 investigative staff.
Analysis of all cases reported to the ICD
Since its establishment the ICD has been relatively
well used, having received a large number of cases
25
Figure 1: All cases and complaints received by ICD,
April 1997-March 2002
3,000
2,500
2,000
Number
in its first five years of operation. The
overall number of cases has increased
over the years, which may be interpreted
as a positive sign. Column A of Table 1
reflects the total number of cases and
complaints received by the ICD over the
period April 1997 to March 2002.
Another positive sign is that the number
of deaths in custody or as a result of
police action have, according to ICD
reports, decreased by 21% between
1997/98 and 2001/02 (Figure 1).
1,500
1,000
500
In analysing ICD statistics it is however
0
1997/98 1998/99 1999/00 2000/01 2001/02
important to distinguish between the
following three categories (see Table 1):
Deaths from
Complaints
Cases rejected:
• Police action and custody deaths.
police action
against police
outside ICD’s
The SAPS and municipal police are
and deaths
(excluding
mandate
in custody
deaths)
required by law to notify the ICD of
Source: ICD
all deaths in police custody and as a
result of police action. It is therefore
inappropriate to refer to these cases as
the police, therefore made up 69% of cases which it
‘complaints’, as the bulk of them are reported to
received in this period (column E in Table 1).
the ICD by the police as required by law. Only
a very small number of these cases are received
ICD’s categorisation of cases according to
by the ICD in the form of complaints from the
classes
public. During the 1997-2002 period police
Class I: A report or complaint of an alleged
action and custody deaths constituted 17% of
death as a result of police action or in police
cases received by the ICD.
custody.
• Cases outside the ICD’s mandate. A high
number of complaints or cases lodged by
Class II: A complaint referred to the ICD by the
members of the public are rejected by the ICD
provincial MEC.
as not falling within its mandate. Altogether
31% of cases received by the ICD during the
Class III: A complaint alleging that a member of
1997-2002 period were classified in this way.
the SAPS committed a serious criminal offence.
• Complaints against the police. These constituted
Class IV: A complaint alleging that a member of
the majority of cases (52%) received by the ICD
the SAPS committed a less serious offence or act
during the 1997-2002 period. Most of these are
of misconduct in violation of SAPS regulations.
lodged by members of the public, but they also
include a number (which is unknown) lodged
Cases of police use of force or alleged brutality
by police. The most accurate figures to use
ICD statistics relating to cases or allegations
when talking about complaints lodged against
regarding the use of force or brutality3 by the police,
police, are therefore these figures that exclude
cover deaths as a result of police action (recorded
both ICD statistics on deaths, as well as cases
along with deaths in police custody in Class I)4, as
that do not fall within the mandate of the
well as various types of non-fatal assaults. These are
organisation.
recorded under a range of sub-categories, including:
• Serious assaults recorded under Class III
Legitimate cases for investigation and monitoring by
including the sub-category of torture, as well as
the ICD, comprising deaths and complaints against
26 BRUCE
SA CRIME QUARTERLY No 4 JUNE 2003
Table 1: Cases and complaints received by the ICD, April 1997-March 2002
A
Total cases
received
B
Deaths as a
result of police
action and
deaths in
custody
C
Cases rejected
as falling
outside ICD’s
mandate
1997/98
1,999
737
91
1,171
1,908
1998/99
2,874
755
279
1,840
2,595
1999/2000
4,380
681
1,237
2,462
3,143
2000/01
5,225
687
2,293
2,245
2,932
2001/02
5,675
585
2,306
2,784
3,369
20,153
3,445
6,206
10,502
13,947
100%
17%
31%
52%
69%
Total
% of all cases
received
assault GBH and attempted murder.
• Less serious assaults including common assault
under Class III, and a separate category of
assault recorded under Class IV.5
Figures for these cases for the April 1997 to March
2002 period are reflected in Table 2.
As indicated in Table 2, deaths as a result of police
action have consistently made up roughly 50% of
ICD cases relating to the use or misuse of force by
police. The ICD therefore receives roughly as many
cases dealing with the use or misuse of force from
the police themselves (in fulfilment of their
mandatory reporting obligations), as it receives by
means of complaints from the public (reflected in
the categories for torture and assault in Table 2).
Types of complaints received by ICD
Complaints (apart from complaints dealing with
deaths as a result of police action or in police
custody) are categorised by the ICD in roughly 45
sub-categories which fall mainly under either Class
III or Class IV.6 Of the 10,502 complaints received
by the ICD between April 1997 and March 2002
(reflected in column D of Table 1), only 21% were
for incidents involving non-fatal (and non-sexual7)
assaults.
SA CRIME QUARTERLY No 4 JUNE 2003
D
Complaints
against the
police (cases
falling within
ICD mandate
excluding
deaths)
E
Cases which
ICD
investigates
and monitors
(Col B+D)
The bulk of complaints (59%) were classified into
one of two sub-categories, namely: ‘neglects duty
or performs duty in improper manner’ and ‘failure
to perform duties and responsibilities’ (Table 3). In
fact, in the 2001-2002 year these complaints
constituted almost 75% of all complaints received
by the ICD (Figure 2) and 90% of complaints
received in Gauteng and North West provinces.8
The 20% of ‘other complaints’ include a mixture of
sub-categories. In the 1997-1998 year, for instance,
the other main sub-categories were an unexplained
category titled ‘abuse of power’ (48 cases),
corruption (40 cases) and various forms of
misconduct. The main sub-categories in the 20012002 year were ‘abuse of power’ (40 cases), theft
(32 cases), and corruption (30 cases).
It should be noted that the vast majority of
complaints in the categories ‘neglects duty or
performs duty in improper manner’, and ‘failure to
perform duties and responsibilities’ (and therefore
the bulk of complaints received by the ICD), are
complaints from ordinary members of the public
who are dissatisfied with the service that they have
received from the SAPS.
The ICD is therefore used much more by members
of the public who are dissatisfied with the service
BRUCE 27
Table 2: ICD national figures for cases involving police use of force or brutality, April 1997- March 2002
Deaths as result of
police action (excluding
deaths in custody)
1997-98
518
1998-99
558
1999-2000
472
2000-01
432
2001-02
371
Total
2,351
%
51
68
61
28
27
37
221
5
157
311
500
344
298
1,610
35
Torture
Assault GBH and
attempted murder
Assault (common)
Total
67
69
143
102
81
462
10
810
999
1,143
905
787
4,644
101*
* Exceeds 100% due to rounding.
they received from the SAPS, than it is used by
suspects and others who have allegedly been
victims of assaults and other abuses at the hands of
the police.
While some South Africans may be inclined to see
this as an indictment of the poor quality of service
received by members of the public from the police,
it appears to conform to patterns in the US:
In New York City, for example, only about 10
percent of all complaints in 1993 arose from an
incident involving the arrest or citation of the
complainant..... [T]he typical complainant is not
a young criminal but a middle-aged person who
feels that a police officer’s behaviour was
improper. The vast majority of these complaints
involve discourtesy, rudeness, or a failure to
provide adequate service and are not related to
a criminal incident.10
ICD statistics – an indicator of police brutality?
While the statistics in some ways reflect patterns in
other countries, it is useful to examine the
underlying factors that contribute to these patterns
locally. An examination of ICD statistics on non-fatal
assaults for the April 2001-March 2002 year (Table 4)
points to some of the difficulties in using ICD
statistics uncritically.
As indicated in Table 4, ICD figures for serious
assaults by the police consistently outnumber figures
for less serious assaults by a factor of 4 to 1.
Although in practice common assaults by police far
outnumber more serious assaults, the ICD statistics
indicate that victims of less serious assaults are highly
unlikely to report these to the ICD.
There is in fact some evidence that the SAPS itself
receives many more complaints and allegations
involving forms of assault by SAPS members, than
does the ICD. Statistics reported by the Minister of
Safety and Security indicated that over the years
1994-1997, an annual average of over 5,300 cases of
common assault, assault GBH and attempted murder
were lodged against SAPS members. Sixty-seven
percent of these were cases of common assault.11
It is furthermore clear from Table 4 and Figure 3 that
there are significant discrepancies between the
provinces. The fact that Gauteng and the Western
Table 3: Complaints received by the ICD, April 1997-March 2002
1997-98
292
1998-99
441
1999-2000
671
2000-01
473
2001-02
416
Total
2,293
%
21
Neglect of duty or performs
duty in improper manner
506
864
1,279
1,441
2,066
6,156
59
Other complaints
373
535
512
331
302
2,053
20
1,171
1,840
2,462
2,245
2,784
10,502
100
Complaints of
non-fatal assaults9
Total
28 BRUCE
SA CRIME QUARTERLY No 4 JUNE 2003
Figure 2: Types of complaints (excluding deaths as a result of police
action or in custody) received by ICD, April 1997-March 2002
2,500
2,000
Number
Cape have the highest number of
complaints suggests factors such as
levels of urbanisation, income and
literacy have an impact on the level
of use of the ICD. In remote areas
the relative inaccessibility of ICD
offices is also likely to inhibit the
lodging of complaints (although the
ICD does allow for complaints by
telephone and the internet).
1,500
1,000
500
The fact that the KwaZulu-Natal
office recorded the lowest number
0
of complaints in this period suggests
that there are serious problems with
the functioning of this office. Over
the April 2001-March 2002 period,
Source: ICD
the ICD office in KwaZulu-Natal
recorded a lower number of assaults
than the ICD in the Northern Cape. This seems
curious, given that KwaZulu-Natal is the office
which consistently receives the highest number of
reports on deaths as a result of police action from
the SAPS, and given that KwaZulu-Natal has a
population ten times the size of the Northern Cape.
According to a press report, a six month pilot
project conducted by the KwaZulu-Natal Campaign
Against Torture at six police stations (3% of the
roughly 180 stations in KwaZulu-Natal), recorded
86 cases of torture and other abuse of detainees. In
six months the project therefore recorded 12 times
as many allegations of non-fatal assault by the
SAPS, as were recorded by the ICD in KwaZuluNatal during an entire year.12
Conclusion
As indicated by Table 2, the total number of cases
received by the ICD recording the use of force or
alleged police brutality, reached their highest levels
ever in 1999-2000. Thus, while the overall number
of complaints received by the ICD has increased
steadily over the 1997-2002 period, both reports of
deaths and complaints of alleged police brutality
have declined after an initial increase in the early
years after the ICD was established.
While it is possible that this points to some type of
change in the patterns of use and misuse of force
SA CRIME QUARTERLY No 4 JUNE 2003
1997/98
Non-fatal
assaults
1998/99
1999/00
Other
complaints
2000/01
2001/02
Neglect of duty or
failure to perform duty
by South African police, it is clear that there are
other factors which impact on the overall number
of cases of non-fatal police violence received by
the ICD. These include factors impacting on the
public use and productivity of the various ICD
offices, including questions of accessibility. At the
same time, because victims of police violence
tend to be poorer and more marginal members of
the public, they are more likely to be vulnerable
to police intimidation, but less likely to lodge
complaints against the police.
The ICD is relatively underused by victims of
torture and other assaults by the police, and ICD
statistics on non-fatal police violence cannot be
seen as significantly related to overall levels of
violence of this kind.
Potentially the ICD could impact on the patterns
of complaints by ‘outreach’ initiatives, directed at
people held in custody and aimed at improving
their access to systems for the lodging of
complaints. Alternatively the ICD could
encourage and motivate the SAPS and
metropolitan police services to improve their
capacity for receiving complaints, and ensure that
these are accessible and user-friendly. Any such
efforts would be likely to lead to a higher number
of cases of non-fatal police violence being
recorded in South Africa.
BRUCE 29
Table 4: Allegations of assaults and other complaints by province, April 2001- March 2002
Provinces
Assault
Gauteng
Western Cape
North West
Eastern Cape
Mpumalanga
Limpopo
Northern Cape
Free State
KwaZulu-Natal
Total
Torture, assault
Assault common Total
GBH & att murder and assault
assaults
75
2
77
38
37
75
13
2
15
52
12
64
43
6
49
79
19
98
13
3
16
15
0
15
7
0
7
335
81
416
Figure 3: Total number of complaints (excluding
deaths as a result of police action or in custody)
in provinces, April 2001-March 2002
KwaZuluNatal
Free State
Northern
Cape
Limpopo
Mpumalanga
Eastern Cape
North West
Western
Cape
Gauteng
0
Source: ICD
200
400 600
800 1,000 1,200
Number
Endnotes
1 Thanks to Moses Dlamini at the ICD for assistance in
clarifying questions addressed in this paper.
2 The Domestic Violence Act, 116 of 1998 also imposes
certain obligations on the ICD.
3 Brutality is here understood as unlawful uses of force or
violence.
4 Note that deaths in custody which are related to the use
of force or brutality by police are recorded by the ICD
as deaths as a result of police action. ICD statistics on
deaths in custody exclude all deaths related to lawful or
unlawful police uses of force.
5 Prior to the 2001-2002 annual report, assault GBH and
attempted murder were presented as a single category.
As a result of these and other changes to the system of
30 BRUCE
Total
complaints
1,121
519
317
228
196
167
167
52
17
2,784
% of
national
total
% of
national
population
40
19
11
8
7
6
6
2
1
100
18
10
8
15
7
13
2
6
21
100
categorisation the number of sub-categories in Class 3
has changed between 1997 and 2002 from roughly 18
to 22. Prior to 2001-20002 there was only an ‘assault’
category under Class IV and no common assault
category in Class III.
6 Few cases are classified in Class II which deals with
cases received from the Minister or MEC.
7 Sexual assaults make up a very small proportion of
cases received by the ICD. In 1997-98 they constituted
1% of complaints (10 cases of rape, 2 of indecent
assault, and 3 of sexual harassment). In 2001-2002
they constituted 0,5% of complaints (14 cases of rape,
and 2 of indecent assault).
8 The three provinces in which these categories
constitute the smallest proportion of complaints
(KwaZulu-Natal, Free State and Limpopo), are also the
provinces which, relative to population, generate the
smallest number of complaints.
9 Includes ‘torture’, ‘assault GBH and attempted
murder’, and ‘assault’ (Figures in Table 2).
10 S Walker, Police Accountability: The Role of Citizen
Oversight, Belmont, Wadsworth, 2001.
11D Bruce and G O’Malley, In the Line of Duty?
Shooting incident reports and other indicators of the
use and abuse of force by members of the SAPS,
Report for the Independent Complaints Directorate,
Centre for the Study of Violence and Reconciliation,
2001.
12 Natal Mercury, 6 November 2002.
Sources for tables and graphs
All ICD statistics are from ICD annual reports except the
1998-1999 figures which are from other statistics provided
by the ICD. Provincial population percentages are from
the South African Institute of Race Relations 1999/2000
Annual Survey. Budget data is from the National
Treasury’s budget review.
SA CRIME QUARTERLY No 4 JUNE 2003