SA CRIME QUARTERLY No 18 DECEMBER 2006 9

O
fficial police statistics attempt to provide
an objective picture of the extent of crime.
But public opinion surveys can provide an

important complement to official statistics by
assessing people’s reported experiences with both
crime and the police, as well as their subjective
(in)security and evaluations of the police and
government attempts to reduce crime. And if
regularly repeated, these ‘snapshot’ surveys can
create a ‘moving picture’ of critical trends.  

A review of Idasa and Afrobarometer surveys
conducted in South Africa since 1994 reveals
several such trends. In contrast to what might be
inferred from the noise of partisan debate and
expert commentary, these trends show that levels of
reported experiences with crime are unchanged
over the past six years, though at exponentially
higher levels than what is reported to the police
and thus included in the official crime statistics. 

Furthermore, public perceptions of overall safety
and the performance of the police are actually

improving, albeit from a fairly dismal base. At the
same time, these surveys also demonstrate that
almost half of all citizens think that most police
officials are involved in corruption, most find it
difficult to get help from the police and some even
have to pay bribes to get this help. 

Do personal experiences match the official view?
Given the heat regularly generated by news media
coverage of crime, and opposition parties’ critiques
of government performance in reducing crime, the
first question to ask is whether ordinary South
Africans share these concerns. Since 1994 Idasa
and Afrobarometer surveys have asked South
Africans: “What are the most important problems
facing this country that the government ought to
address?”, to which respondents can provide up to
three answers.  

The results confirm that in the aftermath of South
Africa’s successful though turbulent transition to
democracy in 1994, crime quickly became one of
people’s most prominent concerns, replacing their

Robert Mattes,1 Centre for Social Science Research
University of Cape Town
Rmattes@commerce.uct.ac.za

GOOD NEWS AND
BAD

Public perceptions of
crime, corruption and
government

Idasa and Afrobarometer public opinion surveys conducted since 1994 reveal that levels of reported

experiences with crime are unchanged over the past four years, but that public perceptions of overall safety

and the performance of the police are actually improving. Of greatest concern is that the January-February

2006 survey found that almost half of all South Africans think that “all” or “most” police officials are involved

in corruption. These are the highest rates recorded across ten different types of public servants. In contrast to

the improving trends in public perceptions of crime, citizen views of corruption in the police (and other

government institutions) are becoming worse.



fears of political violence. Between 1997 and 2000,
six in every ten South Africans rated crime or some
related aspect as one of the most important problems
that government should address, consistently ranking
behind only job creation as the most prominent
issue (Figure 1).

Since reaching their highest point ever in the run-up
to the 1999 election, however, levels of public
emphasis on crime – relative to other concerns –
have actually plummeted, falling by one half from
65% to 23% in 2006. Moreover, crime has lost its
long held position as the nation’s ‘number two’
problem and now ranks behind housing (28%) in a
statistical third place along with poverty at 27%, and
HIV/AIDS (25%).  

Over the years, government has fended off questions
about these public concerns, often chalking them up
as the whingeing of previously protected white
communities now exposed to the same realities long
endured by black South Africans. The data, however,
paint a far more complex reality. It is true that
blacks’ rating of crime as a priority has consistently
lagged considerably behind those of white (but also

of Coloured and Indian) respondents. Yet, at the
height of public anxiety about crime, as many as
55% of black respondents listed crime as a priority
problem (Figure 2).  

Indeed, South Africans of all race groups have
consistently placed the same four or five issues at
the top of what might be called the ‘public agenda,’
though they have often ordered them quite
differently. This is still the case. Crime is one of the
five most cited problems by all groups in 2006, but
although it is rated as the second most prominent
problem by whites (48%), it is only the fifth most
frequently mentioned priority by black Africans
(17%).

Part of this decline is simply a function of the
emergence of other problems competing for public
attention, like HIV/AIDS, which was first mentioned
by respondents in 1997, and cited by almost one in
three (30%) by 2004. Yet part of it is also
undoubtedly a reflection of an actual decline in
crime over this period.    

Indeed, South Africans appear to agree that the
country is at least somewhat safer than it was a few

SA CRIME QUARTERLY No 18 DECEMBER 200610 MATTES

Figure 1: South Africa’s public agenda over time - the top five problems

0

10

20

30

1994 1995 1997 1998 1999 2000 2002 2004 2006

50

70

40

60

80

90

100

67

74

68
73

79
76

84

77

63

46

54

44

22

32
25

22
27

28

6
9

32

58 64 65
60

35
31

23

2 3

13

%

Job 
creation

Crime and
security

AIDS Housing Poverty

26
30

25

6
3

9 9
11

28

24

27

1



SA CRIME QUARTERLY No 18 DECEMBER 2006 11

years ago. They are less likely today to report that
they or their family felt insecure in their home in the
last 12 months than they were in 2000 (51% as
opposed to 61%) (Figure 3). They are also more
likely, by a factor of two, to say that safety from crime
is better now than “a few years ago” than they were
in 2000 (Figure 4).  

At the same time, the proportion of respondents who
said that they, or someone in their family, had
something stolen from their house or were the victim
of a physical attack in the previous 12 months has
remained constant (at least since this question was
first asked in 2002) (Figure 3).  

These partial successes appear to be yielding more
positive views of government efforts to fight crime. As
of January-February 2006, just over one third (36%) of
all South Africans felt the government was handling
crime “fairly” or “very well” (Figure 4). This was the
third lowest rating across 14 different performance
areas tapped by the survey (beating only reducing
inequality and creating jobs), and also represented a
five percentage point reduction since 2004. 

Still, public evaluations are significantly more positive
now than six years ago when just 18% approved of

MATTES

Figure 2: Prioritisation of crime by race

0

10

20

30

1994 1995 1997 1998 1999 2000 2002 2004 2006

50

70

40

60

80

90

100

5

26

42

49

55

48

25
22

31

0

33

58

71

77
75

56

71

17

7

45

39

59
62

54

39

50

34

10

45

58

73
79

67 66

67

46

%

Black White Coloured Indian

Figure 3: Insecurity and personal experience of
crime

0

10

20

30

2000 2002 2004 2006

50

70

40

60

%

Feared crime in home (at least once)

Something stolen from home 
(at least once)

Physically attacked (at least once)

61

51 51

56

30
3233

16 17

19

government performance on crime. This upward
trend mirrors the one described above, namely that
people have the sense that safety from crime is
better than a few years ago.



The results also strongly suggest that actual levels of
crime are far higher than what is reflected in the
official police record. To be sure, the survey
questions were not designed to measure the crime
rate in any precise fashion, but rather to identify
those respondents who had experienced crime
directly or indirectly, and then to gauge the impact
of these experiences on other attitudes. But even
after taking into account the obviously blunt nature
of the question, as well as the complications added
by asking about “you, or anyone in your family,”
the data indicate that, at best, a minority of crimes
like burglary and assault are actually reported to the
police.  

Take for instance the 19% of respondents in the
2004 survey who said that they or someone in their
family had been physically attacked. In terms of
how crime statistics are calculated, this would yield
a rate of approximately 13,000 assaults per 100,000
population.2 Even after adding together the data on
all reported assaults, rape, and attempted murders
included in the 2003/04 official crime statistics, the
figure was only 739 per 100,000. 

Also consistent with the trends in improving
perceptions of safety is the steadily improving
overall image of the police. In early 2006, just
fewer than one half of the public (48%) said they

SA CRIME QUARTERLY No 18 DECEMBER 200612 MATTES

trusted the SAPS, up substantially from the 35%
who did so in 2000. But at least two important
caveats are in order here:
• Trust in all of the country’s political and state 

institutions has been following an upward trend
since 2000, a trend that closely follows the
steadily upward trajectory in South Africans’
confidence in the national economy over this
same time period.

• Regardless of the trend, the police are still seen as 
one of the least trustworthy institutions in the
country, ahead only of local councils and
opposition parties (Figure 5). And the far higher
levels of trust in the National Directorate of
Public Prosecutions (62%) are a real eye-opener,
given the recent controversies over the structural
relationships of the NDPP and the SAPS (see P
Mashele in SA Crime Quarterly No 17, September
2006). 

Why is the SAPS viewed in such a bad light?
Besides the continued, though decreasing, public
dissatisfaction with the level of crime in the country,
one important reason behind the SAPS’s poor image
may be its poor community relations record. Across
five different types of government line agencies, the
Afrobarometer asks people whether, based on their
experience, they found it easy or difficult to obtain
public services from that agency.  

In 2006, just 40% of all respondents said they found
it “easy” or “very easy” to “get help from the police
when they needed it” and 49% said it was difficult
(with 19% of these saying it was “very difficult”).
These are easily the worst results of all the services
asked about in the survey (Figure 6). 

Eighty-three percent of those who say it is “very
easy” to get help also say they trust the police,
compared to just 35% of those who find it “very
difficult.”3 Even more disturbing, given the
substantial resources poured into the transformation,
training and extension of the SAPS over the past few
years, these low levels of user satisfaction are
unchanged from those measured four years ago.

In contrast to the racial patterns in ranking crime as
a priority issue, white (41%) and Indian (50%)
respondents are slightly more likely to report

Figure 4: Evaluations of police performance and
improvements in safety

0

10

20

30

1998

50

40

60

%

Government doing  fairly well/
very well at reducing crime

People safer from crime and
violence than a few years ago

27
26

23
18

1999 2000 2002 2004 2006

40
36

21
25

43
40



SA CRIME QUARTERLY No 18 DECEMBER 2006 13

positive experiences when getting help from the
police than black (39%) or Coloured (37%) South
Africans. But the most important differences are
along provincial lines. The SAPS in the Free State
(74%) and the Western Cape (62%) are rated as the
most user-friendly, while police in KwaZulu-Natal
(37%), Northern Cape (37%) and Gauteng (33%)
get the lowest ratings.

Another reason for the low levels of trust in the
police – beyond the current level of crime – may be
the fact that almost half of all South Africans (48%)
think that “all” or “most” police officials are
involved in corruption (Figure 7). These figures are
much higher than for other parts of the criminal
justice system: just 22% believe that there are high
rates of corruption among judges and magistrates.
Moreover, they are the highest recorded rates across
ten different offices or types of public servants.  

In contrast to the improving trends in public
perceptions of crime, citizen views of corruption in
the police (and other government institutions) are
becoming worse. While the Afrobarometer has a
shorter time series of data about the police on this
question, the high levels of public cynicism in
elected officials evidenced in the late 1990s fell by
approximately one half in 2002 and 2004. Yet even
at that point, levels of perceived police corruption
were about one third higher than for elected

MATTES

Figure 5: South Africans’ trust in public institutions, 2006

0

20

40

60

80

100

%
 s

ay
in

g 
“a

 l
o
t/
so

m
ew

h
at

”

SA
B

C

P
re

si
d
en

t

C
o
u
rt

s 
o
f 

la
w

N
D

P
P

R
u
li
n
g 

p
ar

ty

P
ro

vi
n
ci

al
 

p
re

m
ie

r

IE
C

P
ar

li
am

en
t

P
ro

vi
n
ci

al
 

go
ve

rn
m

en
t

SA
N

D
F

SA
P
S

Lo
ca

l 
co

u
n
ci

l

O
p
p
o
si

ti
o
n
 

p
ar

ti
es

71 70
69

62 62
58 57

54 52 50 48
42

29

Figure 6: Ease of access to public services,
2006

0

10

20

30

P
la

ce
 i

n
 p

ri
m

ar
y 

sc
h
o
o
l

fo
r 

ch
il
d

Id
en

ti
ty

 d
o
cu

m
en

t

M
ed

ic
al

 t
re

at
m

en
t 

at
go

ve
rn

m
en

t 
cl

in
ic

H
o
u
se

h
o
ld

 s
er

vi
ce

s 
(w

at
er

,
p
h
o
n
e,

 e
le

ct
ri

ci
ty

)

H
el

p
 f

ro
m

 p
o
li
ce

50

70

40

60

80

90

100

%

Very easy Easy

18

51

18

49

13

46

16

41

9

31



SA CRIME QUARTERLY No 18 DECEMBER 200614 MATTES

In fact, asking whether perceptions are based on
reality is probably the wrong question. In the realm
of politics, perceptions of corruption, even if
completely distorted, become a reality in and of
themselves that, among other things, erode popular
trust in the police. For example, 62% of people
who see no police corruption say they trust the
police – compared to only 23% of those who see
high levels of corruption.4

But perhaps of most importance is that comparing
perceptions and actual experience begs the far
more fundamental question of why such a
shockingly high number – one in ten adult South
Africans, or 10% – say they were asked for a bribe
by the police in the past 12 months (Figure 8).
Consistent with the police’s dismal rating in user
satisfaction, those who encounter police officials
are more likely to face demands for extortion than
any of the other public services we asked about.

As with the user-friendliness of the police, there
were enormous provincial differences in reported
experiences of being asked to pay a bribe by the
police, far larger than any racial or rural-urban
variations. While no respondents in the Free State,
and only 2% of people from the Northern Cape
reported being asked for a bribe, the figures were as
high as 13% in KwaZulu-Natal, 17% in Gauteng,
and 20% in Limpopo (Figure 9).

officials. From that point, they increased sharply in
2006, as did perceived corruption in local
government. 

Perceptions of corruption are far more prevalent
than citizens’ actual experiences with corrupt
police. While 48% of respondents felt most police
were involved in corruption, only 10% said they
had to pay some form of a bribe in the past year in
order to “avoid a problem with the police” (like a
fine or arrest). However, this should not be seen as
a reason to dismiss perceptions as totally subjective
and unreliable, or to chalk them up to biased news
media reporting, for two reasons:
• The survey question only asked people about 

their experience in the past year. If a different 9%
were forced to pay a bribe each year, the
perceptions measured this year might easily
consist of people who have been victimised by
the police at some point over the past five years.  

• Perceptions of police corruption do not have to 
stem from direct personal experience to have
some basis in reality. One act of extortion can
easily have far broader consequences once the
bribe payer returns home and tells their family
and friends about it; the same tale is frequently
passed on to someone else. Moreover, it takes just
one or two cases of high-level official fraud
reported in the media to establish widespread
perceptions of corruption.  

Figure 7: Perceptions of government corruption, 2006

0

20

40

60

80

100

%
 s

ay
in

g 
“a

ll
/m

o
st

”

Teachers/
school
officials

Office 
of the

President

Judges/
magistrates

Tax
officials

Health
workers

Members
of

parliament

National
govt

officials

Local 
govt

councillors

Local 
govt

officials

Police

19
22 22 23

25 26

36
45 45 48



1.9

1.9

2

SA CRIME QUARTERLY No 18 DECEMBER 2006 15MATTES

variables reviewed in this paper, it emerges that
perceptions of corruption are the most important,
and negative, determinant of trust, followed by
public approval of government’s performance in
handling crime, the user-friendliness of the police,
and the perception that people are safer today than
a few years ago. 

Importantly, once these variables are taken into
consideration, people who were personally asked
to pay a bribe by the police, or who were
personally touched by crime, are not any more or
less likely to trust the police.5

Impact of perceptions on the new democracy
Do the attitudes reviewed in this article have any
larger significance for the country’s new democratic
system? While a final answer awaits a more
complex analysis of these opinions in the broader
context of other values, knowledge and
evaluations, initial tests, using only the data
reviewed here, confirms the general thrust of the
first complete analysis of the dynamics of public
opinion in Africa.6 That is, views of crime and
policing do not make much of an impact on South

Figure 8: Services for which people were asked 
to pay bribes by public officials, 2006

0

2

4

6

10

8

12

14

%

Often A few times Once or twice

1

2

2

1

3

3

1

2

4

3

3

3

5

6

P
la

ce
 i

n
 

sc
h
o
o
l 

G
et

 m
ed

ic
in

e 
o
r

m
ed

ic
al

 a
tt
en

ti
o
n

H
o
u
se

h
o
ld

 
se

rv
ic

e

Id
en

ti
ty

 
d
o
cu

m
en

t

A
vo

id
 p

ro
b
le

m
w

it
h
 p

o
li
ce

5.5

1.8

0

2

4

6

10

8

12

14

16

18

20

Often A few times Once or twice

2.1

Free State Northern
Cape

Eastern
Cape

Mpumalanga Western
Cape

North West KwaZulu
Natal

Gauteng Limpopo

%

1.2

0.4

2.0
1.6

1.5

2.5

2.0

4.9

2.3

5.8

4.2

6.9

5.6

10.7

Figure 9: People who were asked by police to pay a bribe, by province, 2006

3.6

Ultimately, with this data, it is possible to build an
effective explanation of why people do or don’t
trust the South African police. Testing all the

0.3



standardised regression coefficients are: increased
safety from crime and violence (.211), trust in the
police (.116), user friendliness of the police (.115),
government performance handling crime (.110) and
victim of a physical attack (-.099). The same model
also explains 9% of the variance in popular
confidence in the future of democracy. The
standardised regression coefficients are: increased
safety from crime and violence (.114), government
performance handling crime (.112), trust in the police
(.101), user friendliness of the police .100) and victim
of a physical attach (-.079).

SA CRIME QUARTERLY No 18 DECEMBER 200616 MATTES

Africans’ support for democracy, or the extent to
which they reject non-democratic alternatives.  

There are, however, important links between a
specific set of views of crime and policing, and
popular satisfaction with democracy as well as
popular perceptions of the consolidation of
democracy. In January and February of 2006, 63%
of South Africans said they were “fairly” or “very
satisfied” with “the way democracy works in South
Africa,” and 68% felt that it was “likely” or “very
likely” that South Africa would remain a democratic
country.” 

Further analysis confirms that South Africans are less
likely to be satisfied with democracy, or to say that
the country would remain a democracy, if they had
been victims of a physical assault in the previous 12
months. They are more likely to be satisfied, and
confident in the future of democracy, if they feel
people are safer now than in the past, approve of
government efforts to combat crime, find it easy to
deal with the police, and express trust in the police.7

Endnotes
1 Mattes is professor in Political Studies, and director of 

the Centre for Social Science Research at the
University of Cape Town. He is also a co-founder of,
and senior advisor to, the Afrobarometer. 

2 These comparisons can only be illustrative since the 
Afrobarometer draws a representative sample of all
adult citizens, not the entire population. Thus, to reach
this figure, I calculated the actual population figure
represented by 19% of all adult South Africans and
applied this against the total population.

3 The rank order correlation of the reported ease of 
working with the police and trust in the police is (Tau
B) .198, p=.000.

4 Perceptions of corruption and trust in the police are 
correlated at (Tau B) -.238, p=.000. In contrast,
personal victimisation at the hands of the police has
no effect on trust (Tau B = -.034, p.=.117).

5 A model consisting of just these four variables 
explains 17% of the variance in trust in the police. The
standardised regression coefficients are: perceptions of
police corruption (-.227); government performance
handling crime (.161); user friendliness of police
(.159); and increased safety from crime and violence
(.107).

6 M Bratton, R Mattes and E Gyimah-Boadi, Public 
Opinion, Democracy and Market Reform in Africa,
Cambridge, Cambridge University Press, 2005. 

7 The model consists of five variables. It explains 16% 
of the variance in satisfaction with democracy. The