JOHANNESBURG - When an issue becomes tied up with nationalist sentiment it becomes almost impossible to debate it rationally. In this case a person's opinions are determined not by the facts of the matter, but by which side they identify with. Such seems to be the case in any debate on race and crime in South Africa.

To paraphrase George Orwell, the question of whether black or white South Africans are disproportionally affected by crime (and why this is so) is one that, in theory, it should be possible to give a reasoned and possibly definitive answer to.

In reality, it is very difficult to have a constructive discussion about the matter. The question is seen by certain participants as one of what Orwell called "competitive prestige" or, in this case, victimhood.

On the one side there are those, who regard it as a matter of faith that poor black South Africans must be the primary victims of crime and proceed from there. On the other, any article suggesting as much seems to attract a particularly moronic and (white) racist kind of response in the comments section.

In a recent paper Gavin Silber and Nathan Geffen take issue with my article pointing out that, contrary to the claims of the 142 academics and others, crime victimization surveys have, over the past decade, found that more white and Indian respondents report having fallen victim to crime than other groups.

The two authors state that "the view that whites are being disproportionately victimised needs to be debunked. Whites are not the primary victims of South Africa's social ills, and propagating the view that they are encourages ungenerous politics that refuses to acknowledge the responsibility whites have to address past wrongs."

This statement reflects the basic problem with the article, for it seems to reflect the starting, not the end point, of their researches. In an effort to achieve a predetermined outcome the author's tend - no doubt unconsciously and in completely good faith - to subtly distort both my argument and the evidence they think they have found to contradict it.

At the outset Silber and Geffen try to buttress their position by insinuating that, contrary to the actual phrasing of their letter, the 142 were actually referring to "serious and violent crime" (not crime in general). Having inserted this distinction, long after the fact, Silber and Geffen admonish me for failing to sufficiently distinguish (in my response) between different types and degrees of crime.

To begin with then, the authors read into the academics letter a claim they did not make, and read out of my response distinctions I did in fact touch upon (albeit mainly through quoting another author).

Silber and Geffen then use a shot gun approach to question the general validity of victims of crime surveys. These simply amount to a list of speculative suggestions, hurriedly put together, as to why these surveys should be treated with suspicion. One quotation from Michael O'Donovan that they employ in support of their case is taken egregiously out of context.

Further, the authors provide no response to O'Donovan's point that the finding that victimisation survey findings that wealthier communities and individuals are more likely to be the victims of crime are corroborated by "SAPS statistics on serious offences."

The point I made in my original article was that, for the moment, these victims of crime surveys are the most scientific evidence we have on the topic. If Geffen and Silber want to debunk their findings it is incumbent on them to come up with better data. But do they?

In their efforts to disprove the survey findings the authors focus almost wholly on a narrow category of violent crime - namely homicides.

The authors begin with StatSA's data on unnatural deaths for the period from 1997 to 2001. As Silber and Geffen note this shows no disproportion in the likelihood of a member each racial group succumbing to death through unspecified unnatural causes, although (young) black and coloured men were, it appears, more likely to die from assault than the other groups.

Silber and Geffen concede that in just under half of the cases the race of those who died during that period are unknown. Given such holes in the data it is difficult to see what weight can really be put on its findings one way or another.

There is much better data available from 2007 and this confirms that whites are more-or-less as likely as black South Africans to die of external causes of one kind or another. In the 78% of cases where racial data is available the breakdown is the following:

The authors claim that further "compelling data" is provided by a Medical Research Council (MRC) "investigation into female homicide rates in South Africa in 2004" which "used national mortuary data to determine that 2,8 of every 100 000 white women die as a result of murder, whereas 8,9 Africans and 18,3 coloureds meet the same fate. This shows, that at least for women, Myburgh is very likely wrong and the academics are right. Black women are disproportionately murdered."

How could the 142 academics be "right" about a claim they never made, and I "wrong" about something I never disputed? In any event Silber and Geffen misrepresent these findings. The report they reference makes clear that the figures they cite refer solely to rates of "intimate femicide" - i.e. cases where a woman is killed by her intimate partner - not to all female homicides. The full figures are as follows:

For murders of females by others the racial differences are much narrower, as can be seen in the table below. It is important to note though that in the 905 cases analysed the perpetrator was unknown in 182 of them.

As can be seen from these two tables these estimates suggest that black (African) women are not "disproportionately murdered" though Coloured women may be.

Silber and Geffen then cite a Centre for The Study of Violence and Reconciliation (CSVR) study of a number of high risk areas in South Africa. This apparently found that "85 per cent of homicide victims were black, nine per cent were coloured, five per cent Asian and one per cent of victims were white."

Again, this is undigested and misleading. It fails to make clear that the sample was representative not of the national population (as a careless reader may assume) but of the population of the particular areas analysed, and explain what that was. Moreover, the figures it refers to were not the percentage of homicide victims, but rather the racial profile of the population of that area. The complete and correct figures follow in the table below:

Finally, the authors cite murder and rape statistics from Cape Town as a "case study." Given the political and demographic exceptionalism of that city is difficult to see why they do so other than that this very uniqueness supports the pre-ordained conclusion they are straining to reach.

But here again the authors fail to process the data properly. They make much of the fact that, according to police statistics, "five police districts account for over 44 per cent of murders - Nyanga (13,18 per cent), Harare Khayelitsha (8,67 per cent), Khayelitsha (8,47 per cent), Gugulethu (7,58 per cent), and Delft/Belhar (6,1 per cent)." But they fail to take us into their confidence by telling us what proportion of the population of Cape Town actually lives in these particular areas.

Finally, the authors do cite a source which indicates a considerably higher homicide rates in poor black and coloured areas than in the wealthier and whiter suburbs of the city.

After all this, the authors claim that this all somehow nullifies the more general findings of the victims of crime surveys and "indicates that the burden of serious crime is disproportionately absorbed by black and poor South Africans." It is difficult to see a rational connection between the evidence they have produced and this claim.

As noted earlier there are problems both with the data they cite and the way they have presented it. The selection of data is problematic as well - given that they have cherry picked a crime category and regions which tend to support their claims. Khayelitsha and the Joburg CBD, for example, are undoubtedly some of the most dangerous areas of the country. But the deep rural areas are often some of the safest. Limpopo is the province with by far the lowest murder rate.

Assuming that the evidence they have cited does in fact show, as is very possibly the case, that the murder rate nationally is in reality higher among black and Coloured South Africans all they have shown is ... that the murder rate is higher among these two racial groups. They have neither proved their own claim (about "serious crime") nor that of the 142 academics (about crime in general.)

There is a danger too in simply assuming that general murder rates reflect a more general burden of victimhood at the hands of our bandit class. Firstly, as the MRC findings indicate, many killings (and assaults) are carried out by intimates or acquaintances often in moments of drunken rage. Such so-called "social fabric crimes" are not really relevant in determining whether individuals from vulnerable minorities are seen as legitimate or soft targets by predatory criminals.

Secondly, those who live by the sword are likely to die by the sword. Violent offenders are very likely to themselves to be affected by violence and to die violent deaths. A 2004 American study noted that "the extant literature has established that delinquents and victims are often one and the same...Gang members who engage in violent offending experience higher levels of violent victimization than all other youths, even nongang youths who engage in violence." High homicide rates in gang ridden areas may thus partly be a product of inter-gang violence.

Thirdly, as O'Donovan cautions, death rates from violent crime are partly determined by the quality of emergency services and hospital care the victim has access to. He states, "The derived murder rates for an area may thus rest less on the inclination of a community to violence than on the quality of services in that area. The murder stats should thus be used with greater caution than is usual."

Finally, the authors suggest - while twisting what I actually wrote - that it is "far-fetched" to raise the possibility that the anti-white sentiments of the top ANC leadership in the 1990s could have given the green light to cross-racial crime. This is obviously a hypothesis that needs to be tested. But I fail to see why it is inherently implausible. Would Silber and Geffen also claim that Thabo Mbeki's equally "obscure" statements on AIDS and anti-retrovirals had no effect on people's behaviour on the ground or on the conduct of ANC apparatchiks in state and society?

ENDNOTE:

[i] Naeemah Abrahams (et al), "Mortality of Women From Intimate Partner Violence in South Africa: A National Epidemiological Study", Violence and Victims, Volume 24, Number 4, 2009

Click here to sign up to receive our free daily headline email newsletter