Dunblane Massacre Resource Page

Version without tables (for Lynx)

Flaws in Annex G

A portion of a May 1996 submission to Lord Cullen's Dunblane Inquiry written by,
Steve Kendrick <cybershooters@compuserve.com>

[Home Office Statistical Service] Annex G purports through examination of criminological studies to find a link between the possession of guns and crime levels. Unfortunately, though many studies which refute this position are cited in the bibliography, they are glossed over in the text. (see for example Wright and Rossi's work Armed and Considered Dangerous which was sponsored by the US Dept. of Justice and Kleck's work Point Blank which has won many top awards).

On page 75, the writer dispenses with these studies because of a multitude of reasons, such as: "some studies have compared only two countries or cities". But then on page 79, the writer presents a table justifying a link between gun availability and homicide, which only compares two countries!

Much of the criminological study which the govt. glosses over does great harm to their arguments. Gary Kleck's research for example shows that gun possession actually reduces armed crime, because a person in possession of a gun is far more likely to be able to defend themselves from criminal attack than a person who is unarmed. Kleck estimates the usage of firearms in self-defence in the US at about 2.5 million times a year, including self-defence against animals. A typical self-defence situation would be a person being threatened by a criminal, the victim produces a firearm, and the criminal flees.

The govt. evidence is based mostly on the research of Martin Killias. Killias surveyed households in 18 areas, and then using homicide rates from the International Crime Survey, he attempted to show a correlation between these rates.

Killias' studies have been attacked as being flawed among the criminological establishment; most importantly, 47% of respondents refused to answer the question as to whether their household contained a gun (from Killias' own work). In addition, the methodology of asking people as to whether they own a gun is questionable. Undoubtedly many people would lie, especially if they owned a gun illegally.

On page 74, the footnote states that: "In Switzerland, for instance, where gun availability is relatively high because of the reserve militia, ammunition is kept in sealed boxes which are checked every year, and is not available for sale." This is almost verbatim from Killias' research, and is a completely untrue statement. In reality, the sale of this ammunition is subsidised by the Swiss government to encourage marksmanship practice; it is widely available at gun shops; and the ammunition kept at home is not checked every year.

The govt. then presents graphs based on Killias' research "proving" a link between gun possession and gun-related homicide. Using a Spearman Rank Order Analysis, the correlation co-efficient for all areas is a statistically significant 0.563. This is not a strong correlation, but it is a correlation. What is important however, is whether there is a correlation between total homicide and gun possession, as homicide can of course be accomplished in many different ways. There is little point in trying to reduce the number of guns possessed if in fact there is no impact on the overall homicide rate.

Presented below is a statistical analysis of some of the data presented by the government in support of their argument:

Spearman Rank Order Analysis of Killias data in Annex G p.83 of the Govt. evidence to Lord Cullen:

Country Homicide rate
per million
% households
with guns
with guns
rank d ddxd
USA 75.9 48.0 1 1 0 0
Norway 12.1 32.0 14.5 2 12.5 156.25
Canada 26.0 29.1 4 3 1 1
Switzerland 11.7 27.2 17 4 13 169
Finland 29.6 23.2 3 5 -2 4
France 12.5 22.6 13 6 7 49
New Zealand 20.2 22.3 5 7 -2 4
Australia 19.5 19.4 6 8 -2 4
Belgium 18.5 16.6 7 9 -2 4
Italy 17.4 16.0 8 10 -2 4
Sweden 13.3 15.1 12 11 1 1
Spain 13.7 13.1 10 12 -2 4
W. Germany 12.1 8.9 14.5 13 1.5 2.25
N. Ireland 43.3 8.4 2 14 -12 144
CSSR 13.5 5.2 11 15 -4 16
Scotland 16.3 4.7 9 16.5 -7.5 56.25
Eng./Wales 6.7 4.7 18 16.5 1.5 2.25
Netherlands 11.8 1.9 16 18 -2 4
Total   0 625

co-efficient r = 1 - 6(625)
= 1 - 3750
= 1 - 0.64 = 0.36

Test for statistical significance = n-2 = 18-2 = 16 degrees of freedom. Using a Rho table, the level of significance is 0.399. As 0.36<0.399 the figure is not statistically significant and there is no correlation between the percentage of households with guns and the total homicide rate.

Thus it can be seen using the Government's own statistics, that it is unlikely a reduction in the private possession of guns would lead to a drop in the overall homicide rate, as there is no correlation between the two. The logical conclusion is that when gun possession is reduced, people turn to other methods to commit homicide.

Put in the context of the Dunblane incident, this would mean that a mass murderer would accomplish his actions in another way likely to lead to mass death. Examples of such killings would include the 89 people killed by a man in New York City armed with a can of petrol and a lighter who burned down a night-club; and the numerous mass killings accomplished by people armed with home-made explosive devices and poison.

Posted: May 1996