While victim surveys gather a variety of opinion data, their core purpose is to determine real rates of criminal victimisation. Only a fraction of all crime is reported to the police, and thus police crime statistics are inherently inaccurate. Some types of crimes are more under reported than others. Crimes for which a case number is required in order to enter an insurance claim, such as auto theft, are more likely to be reported, as are any property crimes in which there is some hope of the police recovering the stolen goods. For crimes where the likelihood of property recovery is viewed as quite low, such as robberies involving small amounts of cash, reporting levels are also low.

Surveys have found that interpersonal crimes, especially sex crimes and domestic violence, are also less likely to be reported, although many people are just as reluctant to discuss these matters with a pollster at the door as they are to report them to the police. The exception, of course, is murder, which is nearly always reported because disposal of the body requires official sanction.

Levels of victimisation

In the present survey, respondents were asked about experiences of crime in inner Johannesburg since 1994 and then experiences in the last year. Because the survey was done close to Easter, the previous Easter was used as a cut off point for the last year’s experiences. Survey respondents generally like to report crime and often have a faulty sense of when events occurred. Past surveys have found that victims tend to ‘telescope’ the events of past years into the most recent one if only asked about recent victimisation. It is for this reason that experiences since 1994 were queried, although this data was not used for analysis. Because a large percentage of the respondents were new to the area in any case, most of the crime reported occurred in the last year.

As is standard methodology, certain crimes were considered ‘household crimes’ and certain crimes ‘individual’ crimes. Household crimes include those that any member of the household could be expected to report on reliably, such as burglary or theft of a vehicle. Murder, of course, is a household crime – one cannot speak to the victim personally. Individual crimes include those that only the individual can be expected to report on reliably, such as robbery and assault. In the present survey, respondents were questioned about the household crimes of murder, burglary and vehicular theft, and the individual crimes of robbery, assault, and car hijacking. As was mentioned above, due to ethical issues and low levels of success in past victim surveys, sex crimes were excluded from the survey.

Figure 4.1 shows the victimisation rates found in the 1999 Johannesburg survey done as part of the International Crime Victim Survey of the United Nations; the national survey done by Stats SA and the Secretariat for Safety and Security in 1998; and the present survey. There are methodological differences between these surveys, of course, but the figures do give some basis for comparison.

Figure 4.1: Percent of respondents experiencing crime in the last year, multiple surveys



As might be expected, the survey results echo the official statistics in labelling inner Johannesburg as one of the most dangerous parts of the city and, indeed, of the nation as a whole. Most striking is the extremely high level of robbery, a phenomenon that will be explored further in Chapter 7. Murder is not covered in all victim surveys, and vehicular crimes are generally cited without regard to the percentage of the population that owns a vehicle. Figure 4.2 shows the victimisation levels found in this survey across crime categories, with the vehicular theft rates calculated only for those households owning a vehicle. Hijacking, which can occur in any vehicle and was deemed an individual crime like other robberies, is calculated for the sample as a whole. Remarkably, nearly as many hijackings (32) were recorded in this survey as vehicular thefts (48), a phenomenon that will be discussed further in Chapter 10.

Figure 4.2: Percent of respondents experiencing crime in the last year, inner Johannesburg, 2002

Reporting rates

Variation in reporting levels between surveys is to be expected, since different areas have different relationships with the police and these attitudes vary over time. People with easy access to the police are generally more likely to report. Access is determined by a variety of factors, including:

All other factors being equal, then, high rates of reporting could be expected of poor urban areas like inner Johannesburg, where 92% of the people polled know where the local police station is situated. While rates of insurance are likely to lower in these areas, the police stations servicing these areas are within walking distance of most of their jurisdiction. In the present case, these stations are supplemented by several ‘contact points’ (satellite sites where crime can be reported). Telephones and public transport are readily available for the less mobile. While the necessity of paying cash for accommodation and the lack of access to subsistence agriculture requires some cash income, many people are engaged in informal enterprises and have some flexibility to their working hours. Poverty means that loss of property is of major importance, and hopes of recovery may prompt reporting. Diversity and the lack of community cohesion mean informal methods of dispute resolution may be lacking. This may promote reliance on more impersonal means of dispute resolution – the services of the state criminal justice system.

Figure 4.3 shows reporting rates for major crime types found in several polls. The methods used by these studies varied quite a bit (the ISS surveys were street surveys, for example), but while this might affect victimisation rates, there is little reason to expect it would affect reporting rates. Respondents are asked if they have been the victims of these crimes and then asked if they reported this crime to the police. These figures show some consistency in under-reporting patterns by crime type, with between 36% and 48% of robberies and 33% and 48% of assaults being reported. Burglaries are more of a wild card, with 59% and 80% of cases being reported.

Figure 4.3: Percentage of victims who say they reported crimes to the police, various surveys



For most surveys, assaults are the least reported, robberies next, and burglaries the most reported of the three crime types. The present survey is remarkable in bucking this trend: assaults were more likely to be reported than robberies. In fact, the inner Johannesburg residents reported a lower percentage of robberies than any other area surveyed. This may be tied to the fact that robberies are far more common here than in any other area. Assault reporting was actually higher in inner Johannesburg that in any other site besides Pretoria. This may have to do with what is considered an ‘assault’ in this highly violent area; most of the assaults described to the fieldworkers in this study involved weapons and injuries requiring medical attention. If the case involved hospitalisation, there may have been little choice as to whether it was reported to the police.

The present survey also asks whether the victim received a case number. This is important for two reasons. First, as discussed in Chapter 3, this is an important police performance indicator, because improved public trust in the police should lead to a greater percentage of cases being reported. Second, it gives some indication of the difference between the number of crimes reported to the police and the number of crimes recorded by the police. Police members may elect not to start an official police docket on a case for a number of reasons, from suspicions that the claim is fraudulent to a desire to keep crime statistics low and clearance rates high by only recording cases in which a good chance of arrest exists.