Summary
Canadian homicide data over the time interval 1961-1994 is analyzed by
means of commission. Firearms homicides are further broken down into
pistol homicides and non-pistol homicides (which are referred to as rifle
homicides for reasons of brevity - my apologies to purists for including
shotguns in the rifle category).
During the time period of these graphs, all pistols were required to be registered in Canada with stringent carry restrictions. The rate of pistol homicides may thus be considered as a measure of the efficacy of firearm registration. Registration of pistols has been in effect in Canada since the 1930's.
Until 1977, rifles and shotguns could be purchased in Canada without any controls whatever. Starting in 1977, it was first necessary to obtain a Firearms Acquisition Certificate (FAC) in order to purchase any rifle or shotgun. Also, some models of rifles were reclassified as restricted weapons, and thus required registration in the same means as pistols must be registered.
Data Sources
Assorted Juristat publications were used to obtain numbers of homicides
used in construction of the following graphs. For full details about
data, see
Appendix1
which contains references as well as raw data in numeric form for those
who wish to perform their own analyses.
Something dramatic happened in 1975 - after this year there was a dramatic decrease in numbers of firearms homicides. This trend persisted until 1991 when there was a dramatic increase in use of pistols in homicide. In contrast to the marked fall in firearms murders, the non-firearm murder count continued to rise steadily, but more slowly after 1975. There has been a 3 year steady decrease from 1992-1994, and this may represent a real trend.
Here the 1975 homicide peak is more clearly seen. Since 1975 there has been a
steady decline in firearms homicide rates with a non-firearms homicide rate
which has been static, or perhaps slowly decreasing. The massive decrease in
firearms homicide rates which occurred in 1975 is even more dramatically shown
on this graph. It is not clear what happened in 1975 which was responsible
for this drop. I would be interested in hearing various hypotheses.
First note that fraction of homicides committed without firearms has steadily
risen since 1961 - although not done in this graph, the fraction of
non-firearms homicides has been rising more rapidly since 1975 than before.
Use of rifles in homicides has fallen steadily since 1961. The correlation
coefficient is 0.829 for this relationship (linefit given with graph gives
square of correlation coefficient). If one does two separate fits, the
fraction of rifle murders has decreased much more rapidly after 1975 than
in the interval 1961-1975. Pistols were used in a surprisingly constant
fraction of murders until 1991 - then there was a large increase which has
persisted. If one ignores the last 4 years of data, the fraction of pistols
would have been almost steady at about 8-9% of homicides for 30 years.
When one looks at the previous graphs, one cannot but fail to notice that
the rate of homicides involving rifles has been steadily decreasing and
is now comparable, if not less than, the rate of homicides involving pistols.
The graph below shows a more detailed view of rates of firearms homicide.
Here:
As the "rifle" homicide rate also includes homicides using sawed off shotguns and machine guns, the decrease in homicide rates involving legally owned rifles has been even more dramatic than the graph above would imply.
One has to wonder why rifles, which are not registered and can be bought by anyone with an FAC (without having to wait several months for paperwork to clear) are not used in homicides in proportion to the difficulty of acquiring one. If registration of firearms had any effect on reducing rate of homicide then one would expect registered firearms to be associated with lower homicide rates than unregistered firearms. Thus, the rifle homicide rate should be far greater than the pistol homicide rate. Instead, we see the converse - highly regulated and registered pistols are now involved in more murders than "unregulated" and unregistered rifles. In 1994, there were 66 homicides using a rifle/shotgun in comparison to 90 using pistols - the total of 106 classified as "rifle" homicides in the graphs above included 26 murders in which sawed off shotguns/rifles were used and 14 murders involved machine guns.
Dr. B. Centerwall compared pistol homicide rates in US border states and adjoining Canadian provinces, and came up with the (to many people surprising), conclusion that, among comparable populations, pistol registration makes absolutely no difference in homicide rates. [1]
While obtaining the data on fraction of legitimately owned and registered pistols that are used in homicide is almost impossible, the numbers of such homicides in Canada yearly are less than 10 and probably even smaller. The overwhelming number of pistol homicides are carried out using pistols which were either smuggled into Canada from other countries, or have been stolen. Registration has not had the slightest effect on such criminal use of pistols as registration of pistols has been in effect over the whole 34 year span that is shown in the above graphs. Similarly, the rifle homicide rate began to precipitously fall in 1975 while there was no requirement for an FAC and anyone could walk into a gun store and walk out with a rifle 10 minutes later. The reasons why such trends have occurred in homicide rates likely have nothing to do with any legislation passed by the Canadian government.
Some of this data for early years has been derived from using a screen digitizer to convert the Stats Canada derived graphs on Ian Jefferson's web page back into numbers. The numbers of firearms homicides obtained from various sources all agreed, but total homicide numbers did not always tally. Also, there appears to have been some changes in how homicide rates were calculated during the 1970's. Homicide numbers also change from year to year as deaths thought to be for other reasons may be reclassified as a homicide years later.
The full raw data used in construction of these graphs is given in tabular form at the end of this appendix. Table 1 consists of raw homicide numbers and table 3 consists of rates.
Abbreviations used in graphs are:
Here: Mt - total # of murders Gtot - All murders committed with guns Pt - Pistol homicides Oth - Other homicides (non-firearms) Rifle- Homicides using rifles/shotguns/sawed off shotguns and other, non-pistol firearms Year Mt Gt Pt Oth Rifle 1961 233 85 14 148 71 1962 265 80 18 185 62 1963 249 99 22 150 77 1964 253 105 28 148 77 1965 277 113 33 164 80 1966 250 92 19 158 73 1967 338 137 35 201 102 1968 375 135 37 240 98 1969 391 154 42 237 112 1970 467 178 38 289 140 1971 473 191 48 282 143 1972 521 200 56 321 144 1973 546 216 57 330 159 1974 600 273 75 327 198 1975 701 290 88 411 202 1976 668 250 67 418 183 1977 711 252 61 459 191 1978 661 234 60 427 174 1979 631 199 51 432 148 1980 593 185 58 408 127 1981 648 185 58 463 127 1982 668 248 88 420 160 1983 682 224 78 458 146 1984 667 228 66 439 162 1985 704 222 73 482 149 1986 569 175 38 394 137 1987 642 202 58 440 144 1988 575 169 47 406 122 1989 657 218 54 439 164 1990 660 196 69 464 127 1991 756 271 135 485 136 1992 732 246 129 486 117 1993 630 193 90 437 103 1994 596 196 90 400 100
Where: NGfrac - fractions of non-gun homicides Pfrac - fraction of homicides using pistols Rfrac - fraction of homicides with rifles/shotguns/machine guns and all other "non-pistol" firearms. Year NGfrac Pfrac Rfrac 1961 0.635 0.060 0.305 1962 0.698 0.068 0.234 1963 0.602 0.088 0.309 1964 0.585 0.111 0.304 1965 0.592 0.119 0.289 1966 0.632 0.076 0.292 1967 0.595 0.104 0.302 1968 0.640 0.099 0.261 1969 0.606 0.107 0.286 1970 0.619 0.081 0.300 1971 0.596 0.101 0.302 1972 0.616 0.107 0.276 1973 0.604 0.104 0.291 1974 0.545 0.125 0.330 1975 0.586 0.126 0.288 1976 0.626 0.100 0.274 1977 0.646 0.086 0.269 1978 0.646 0.091 0.263 1979 0.685 0.081 0.235 1980 0.688 0.098 0.214 1981 0.715 0.090 0.196 1982 0.629 0.132 0.240 1983 0.672 0.114 0.214 1984 0.658 0.099 0.243 1985 0.685 0.104 0.212 1986 0.692 0.067 0.241 1987 0.685 0.090 0.224 1988 0.706 0.082 0.212 1989 0.668 0.082 0.250 1990 0.703 0.105 0.192 1991 0.642 0.179 0.180 1992 0.664 0.176 0.160 1993 0.694 0.143 0.163 1994 0.671 0.151 0.178
Here:
RMtot - total murder rate/100,000 by all means
RGtot - murder rate for all firearms
RPist - rate of murder using pistols
ROther- non-firearm homicide rate
Rifle - rifle/shotgun and other non-pistol firearm homicide rate
Year RMtot RGtot RPist ROther Rifle
1961 1.280 0.467 0.077 0.813 0.390
1962 1.430 0.432 0.097 0.998 0.335
1963 1.320 0.525 0.117 0.795 0.408
1964 1.310 0.544 0.145 0.766 0.399
1965 1.410 0.575 0.168 0.835 0.407
1966 1.250 0.460 0.095 0.790 0.365
1967 1.660 0.673 0.172 0.987 0.501
1968 1.810 0.652 0.179 1.158 0.473
1969 1.860 0.733 0.200 1.127 0.533
1970 2.190 0.835 0.178 1.355 0.657
1971 2.150 0.868 0.218 1.282 0.650
1972 2.340 0.898 0.252 1.442 0.647
1973 2.420 0.957 0.253 1.463 0.705
1974 2.620 1.192 0.327 1.428 0.865
1975 3.020 1.249 0.379 1.771 0.870
1976 2.840 1.063 0.285 1.777 0.778
1977 2.990 1.060 0.257 1.930 0.803
1978 2.750 0.974 0.250 1.776 0.724
1979 2.600 0.820 0.210 1.780 0.610
1980 2.410 0.752 0.236 1.658 0.516
1981 2.600 0.742 0.233 1.858 0.510
1982 2.650 0.984 0.349 1.666 0.635
1983 2.680 0.880 0.307 1.800 0.574
1984 2.600 0.889 0.257 1.711 0.631
1985 2.710 0.855 0.281 1.855 0.574
1986 2.170 0.667 0.145 1.503 0.522
1987 2.520 0.793 0.228 1.727 0.565
1988 2.140 0.629 0.175 1.511 0.454
1989 2.400 0.796 0.197 1.604 0.599
1990 2.370 0.704 0.248 1.666 0.456
1991 2.690 0.964 0.480 1.726 0.484
1992 2.570 0.864 0.453 1.706 0.411
1993 2.190 0.671 0.313 1.519 0.358
1994 2.040 0.671 0.308 1.369 0.363
Centerwall, B. Homicide and Prevalence of Handguns: Canada and the United States. American Journal of Epidemiology. 134(11):1245-1260, (1991)Last updated 20-Oct-1995 by Boris Gimbarzevsky. All comments and criticisms should be directed to: borisg@unixg.ubc.ca