|CHAPTER 8 TESTING THE CORRELATION|
The most basic tenet of the argument for strict gun control is that licensed civilian gun ownership contributes to the level of violent crime in society.1 As Adele Kirsten of Gun Free South Africa put it "it is clear... that the availability of guns in South Africa contributes significantly to the incidence of violent crime." 2 Given that this alleged positive correlation is a fundamental building block upon which the argument for strict gun control is based, it is unsurprising that this issue has become the subject of much controversy abroad. However, in South Africa it is remarkable how this relationship is seen as common knowledge and often taken for granted. Dr. Fanaroff states that "there is no doubt that the easy availability of firearms contributes to the high level of violence and violent crime." 3
If this is the case then there should be a demonstrable correlation between levels of licensed civilian gun ownership and levels of violent crime. Most often, comparisons have been utilised to demonstrate a positive correlation between the prevalence of civilian gun ownership and violent crimes such as homicide, aggravated assault and robbery and single case studies, have been used to illustrate the point.4
International comparisons have been a popular way of demonstrating a relationship between these variables.5 Outside of South Africa, one of the typical comparisons has been between the U.S. and Japan and Britain. The U.S. has higher levels of civilian gun ownership and higher levels of violent crime than both Britain and Japan. This fact has been used to imply a positive relationship between the two variables. In addition to international comparisons, particularly within the U.S. there have been international and inter-regional comparisons and inter-city comparisons that have claimed to demonstrate this relationship.6
International comparison is of limited value when trying to establish the relationship between privately owned licensed firearms and violent crime. It is almost impossible to control for all the other factors that may impinge on the rate of violent crime such as a country's history of violence, cultural factors, religious influences, level of economic development, political circumstances etc. Wesson suggests that only once a "National Violence Index" has been developed that weights relevant sociological, historical, cultural and other factors, will international comparison be viable in this respect.7
As many authors have pointed out,8 the vast majority of these studies are marred by severe methodological flaws that cast doubt on the validity of the data generated. One common shortcoming is the failure to control for extraneous variables. The data produced by the studies has also been used to infer erroneous conclusions. One such dubious conclusion is that the alleged correlation between civilian gun ownership and violent crime indicates causality.9
However, if there is indeed a positive correlation between these variables, it could be that rising crime rates caused increased civilian firearm ownership and not the other way around.10 Thus, even if the alleged correlation exists, it certainly does not prove causality. Nonetheless, international comparison is a tool often used to "prove" the value of strict gun control policies and, therefore, the available evidence needs to be examined despite its shortcomings.
Guns can be used in a multiplicity of ways and circumstances. However, when analysing the relationship between private licensed firearms and violent crime by means of international comparison, murder rates tend to be the preferred indicator of levels of violent crime. The reasons for this include the very high reportage of murder and the availability of murder statistics across countries.
Evidence from the SADC region with regard to the alleged correlation between civilian gun ownership and violent crime is scant, with the local debate being informed largely by Katherine McKenzie's case studies of gun control legislation in the SADC region.
McKenzie examined the legislation regulating firearm ownership in Botswana, Lesotho, Malawi, Mozambique, Namibia, South Africa, Swaziland, Tanzania, Zambia and Zimbabwe, and also conducted interviews with police, academics and NGO representatives in the respective countries. For each country examined in her report, McKenzie supplies general information pertinent to the discussion such as economic prosperity, recent history of conflict and population size. She also describes the domestic gun control policy, the crime situation and available firearm statistics.
From a comparison of these ten case studies she concludes that "the convergence of poverty, unemployment, a gun culture and the availability of firearms is a lethal combination which results in high levels of gun crime. ...Countries in the region with effective gun control policies and fewer firearms in circulation have less gun crime and are safer than countries with permissive gun control policies and more firearms in circulation. As Wesson puts it "the argument is ... that the removal of private licensed firearms will reduce violence and especially death".12
In her submission on the Firearms Control Bill. McKenzie makes special reference to Botswana, which she claims has implemented a total ban on the issuing of handgun licences.13
Provisional police statistics for 1998 reveal that only 11 armed robberies were reported that year. Comparatively, in the same year in South Africa, 69 501 robberies were committed with firearms.14 In Botswana the murder rate is a quarter that of South Africa.15
McKenzie does not take into account those extraneous variables that may explain the apparent relationship between two factors. For instance Botswana has no history of a liberation struggle, while South Africa does. This could explain the variance in murder rates and violent crime without any reference to gun control laws. Secondly one cannot take only two cases and make an argument for the correlation between two factors, which could well, be spurious. For that one would have to demonstrate a trend over a number of different cases. In general the statistical ability to control for extraneous variables increases with sample size.16
If only two cases were sufficient then the argument could easily be made the other way by comparing Botswana and Namibia or Namibia and South Africa. Botswana with its total ban on private protection firearms has a murder rate (14 murders per 100 000) almost three times that of Namibia (5 murders per 100 000) with its availability of firearms for personal protection. Furthermore, Botswana does not have a history of a liberation struggle, while Namibia does.17
Namibia and South Africa have a similar history of liberation struggle. Final figures are not yet available but again the official estimates suggest that Namibia has a similar, and perhaps higher, per capita ratio of licensed firearm ownership than South Africa and yet the murder rate in Namibia is about one twelfth that of South Africa.18
The SADC country with the strictest firearm legislation is Swaziland. Firearms for protection are effectively banned and even hunting firearms are severely restricted. However, Swaziland has by far the highest violent crime and murder rate within SADC at 25 percent higher than South Africa's.l9 Swaziland has a higher murder rate than South Africa at 80 per 100 000, yet ownership of firearms is only 1.1 percent. Furthermore, traditionally very few of these murders are committed with firearms.20
Wesson points out that Botswana, Lesotho and Zambia have similar murder rates of 14, 15.5 and 13 per 100 000 respectively. Lesotho and Zambia allow ownership of licensed personal protection firearms while Botswana does not. In addition to this Namibia has the lowest murder rate of all six countries (5 per 100 000) but allows private ownership of licensed firearms.21
These observations all undermine the assertion that there is a direct correlation between availability of privately owned licensed firearms and the murder rate.22
Furthermore, Wesson considers all six countries for which McKenzie provides murder rates in an attempt to establish whether the alleged correlation does in fact exist. McKenzie supplies murder statistics for Botswana, Namibia, Lesotho, RSA, Swaziland and Zambia. Murder statistics for Mozambique, Malawi, Tanzania and Zimbabwe were unavailable. In addition to the information provided by McKenzie relating to these four countries, Wesson calculates an estimate of the murder rate for Zimbabwe (5 per 100 000) based on combined murder and attempted murder figures provided by the Zimbabwe Government Census and Statistics Office.23
|COUNTRY||POPULATION||No. MURDERS||MURDER RATE|
|Botswana||1.60m||217||14 per 100 000|
|Namibia||1.62m||82||5 per 100 000|
|Lesotho||2.10m||325||15.5 per 100 000|
|RSA||43.0m||12267||64 per 100 000|
|Swaziland||0.97m||n/a||80 per 100 000|
|Zambia||10.0m||1347||13 per 100 000|
(Source: McKenzie, "Domestic Gun Control Policy" cited in Wesson, "Murder and Private Firearms.")
When comparing cases it is vital that one compare like with like and control for extraneous variables. Wesson argue that only four of the six countries make for suitable comparison based on cultural similarities, economic activities, degree of urbanisation and population distribution within these countries. RSA and Swaziland are excluded as being atypical 24 with regard to the factors under consideration.25
Wesson attempts to analyse the significance of these figures in relation to gun control laws by devising a quantitative measure of the qualitative expression "ease of obtaining a licence for self protection purposes" can be obtained. He does this by weighting factors in relation to the difficulty of obtaining a licence in relation to both Botswana and South Africa using a scale of 0-10. 0 represents impossibility of obtaining a private licence for a personal protection firearm whereas 10 indicates no controls.26
|COUNTRY||WEIGHTED AVAILABILITY OF LEGAL PERSONAL PROTECTION FIREARMS||MURDER RATE PER 100 000 OF THE POPULATION|
(Source: Wesson, "Murder and Private Firearms")
Plotted on a graph, these figures appear to demonstrate a trend. For the five countries under consideration, murder rates tend to be higher for those countries with perceived increased difficulty in obtaining licences for firearms for personal protection purposes. In conclusion then it appears that, notwithstanding the limitations) there is no evidence from the data collected with regard to SADC countries that supports the alleged positive correlation between civilian gun ownership and homicides. If anything the data suggests the exact opposite. 27
The figures in table 13 have been used, to support the thesis that there is a positive correlation (0.43) between firearm ownership and murder rates.28
|COUNTRY||YEAR||TOTAL HOMICIDES PER 100 000 POP.||FIREARM HOMICIDES||NON-GUN HOMICIDES||% HOUSEHOLDS WITH GUNS|
|England & Wales||1992||1.41||0.11||1.30||4.7|
(Sources: Murder rates from International Journal of Epidemiology (1998:27:216), percentage households with firearms from Canadian Medical Association Journal, M. Killias (1993) cited in Wesson, "Murder and Private Firearms") (South Africa's data from UN United Nations Economic and Social Council: Commission on Crime Prevention, Measures to Regulate Firearms (United Nations: Vienna, 1997),18,23.)
Kleck 29 demonstrates that by removing two countries from the data set, there is a much stronger trend the other way. The US and Northern Ireland should be excluded because the US has a different history of violent crime to Europe, as well as other sociological factors, and Northern Ireland is subject to a peculiarly volatile situation that distorts crime figures.
The U.K. is often heralded as the supreme example of the efficacy of strict gun control. Of all the countries represented in table 12, England and Wales have the lowest rate of firearm homicides (0,11 per 100 000).
However, if the rate of homicides (both firearm and non-firearm) is considered, then there are seven countries represented in table 12 that have lower murder rates than England and Wales, with a total murder rate of 1.41 per 100 000. Of the seven, only one, the Netherlands, has fewer firearms per household than England and Wales. Spain has approximately three times the number of firearms per household, by percentage, than England and Wales, but has about two thirds of its murder rate. Belgium has the same murder rate as England and Wales, but the percentage of household firearm ownership is three times that of England and Wales.30
Interestingly, since 1988, the number of legal firearm owners in Great Britain has dropped by almost 19 percent, yet in the same period robbery with a firearm doubled, and the overall violent crime rate increased by 29 percent. In 1996 alone, violent crime rose by 10 percent.31
From table 12 it is evident that, ironically, South Africa's rate of legal firearm ownership is in fact closest to that of the United Kingdom. However, the discrepancy in homicide rates could not be larger. South Africa's rate of firearm ownership is also about one seventh of that of the United States and yet South Africa's murder rate is about nine times that of the U.S. In fact, while South Africa' murder rate is the highest of all countries represented. South Africa's percentage of legal firearm ownership is one of the lowest.
From an international comparison of homicide rates and levels of civilian gun ownership, it is difficult to see how it can be asserted that high levels of civilian gun ownership translate into high levels of violent crime. Even a cursory glance at the figures suggests that if legitimate firearm ownership is a contributing factor in violent crime, it is extremely limited.
A watershed study conducted in the United States in 1997 by Lott and Mustard was the first of its kind to overcome the difficulties encountered by previous studies by combining cross sectional and time series data and controlling for changing law-enforcement factors like arrest or conviction rates, prison sentence lengths. It was this study that formed the basis of Lott's controversial book, More Guns, Less Crime.
Lott and Mustard analysed eighteen years of FBI data collected from three thousand U.S. counties. They examined levels of crime before and after the implementation of "shall issue" right-to-carry gun laws in 10 U.S. states between 1977 and 1992 controlling for other variables that may influence crime levels. "Shall issue" laws require that authorities issue, without discretion, concealed-weapons permits to qualified applicants.32 On a continuum of gun control legislation, "shall issue" laws are probably the most liberal, guaranteeing the right of all suitable adults to carry a firearm. Especially in states that previously discouraged civilian gun ownership by stricter gun control policies, the adoption of these laws would probably have raised the level of civilian gun ownership quite significantly.33
Regression analysis revealed that the implementation of "shall issue" laws coincided with fewer murders, rapes and aggravated assaults but with higher rates of motor car theft and fraud.34 As the incidence of violent crimes fell, so the incidence of non-confrontational property crimes rose. Thus, it would appear that criminals responded to the threat of being shot by instead engaging in less confrontational crimes.35
When state concealed handgun laws went into effect in a county, murders fell by 7.65 percent, rapes by 5 percent, aggravated assaults by 7 percent. In particular, subsequent to the passing of "shall issue" concealed handgun carry laws, incidents of multiple victim public shootings declined by 84 percent. Deaths from these shootings have plunged by 90 percent and injuries by 82 percent.36 If these calculations are accurate then the net effect of allowing concealed handguns saves lives contrary to claims by the gun control lobby.37
While the substitution effect has usually been tested in the context of the decreased availability of firearms, Lott and Mustard tested for the substitution effect if the prevalence of firearms increased rather than decreased. The implementation of "shall issue" concealed cany firearm laws effectively increased the general availability of firearms among the citizenry in the counties they were studying. Lott and Mustard attempted to measure whether this caused a substitution in the methods of committing murders and whether the number of gun murders rose after these laws are passed even though the total number of murders fell.
While concealed handgun laws raise the cost of committing murders, murderers may also find it relatively more dangerous to kill people using non-gun methods once people start carrying concealed handguns and substitute into guns to put themselves on a more even basis with their potential prey. Carrying concealed handguns appears to have been associated with approximately equal drops in both categories of murders. Carrying concealed handguns thus appears to make all types of murders relatively less attractive.38
Lott and Mustard's results consistently indicate a negative and statistically significant effect in rates of violent crimes, rape, and aggravated assault and high rates of licensed civilian firearms. On the basis of their analysis it would appear that "shall issue laws" reduce murder rates and other violent crimes but are correlated with a concurrent rise in property crime. The results also imply that the gun laws immediately altered crime rates, and that crime rates continued to decrease in subsequent years. The annual decrease in violent crimes averaged about 2 percent, while the annual increase in property crimes averaged about 5 percent.39
An examination of the evidence from Southern Africa and abroad indicates that the alleged positive correlation between the prevalence of civilian gun ownership and the incidence of violent crime has yet to be empirically demonstrated. It appears that international comparison yields no such evidence and certain studies, in fact, find a negative correlation between high levels of civilian gun ownership and violent crime.
The lack of reliable evidence that consistently demonstrates the alleged positive relationship between high levels of legal civilian gun ownership and levels of violent crime, does not rule out the possibility that such a relationship exists. Also, that certain studies from America have indicated the exact opposite, does not imply that this is necessarily the case elsewhere in the world.
As this dissertation has demonstrated, there are a myriad of factors that impinge on the relationship between civilian firearms and violent crime, assuming such a relationship exists. What is fairly certain is that, despite assertions to the contrary, the relationship between levels of civilian firearm ownership and violent crime is neither simple nor direct.40
1Hansson, "Guns and Control." 9.