Rejoinder to Comments on "Gun Control and Rates of Firearms Violence in Canada and the United States." (Response to C. Kwing Hung in This Issue, P. 37)
Mundt, Robert J., Canadian Journal of Criminology
It is gratifying that the author of the accompanying critique read my article so carefully; that critique identified some minor data problems, and pointed to some alternative interpretations of those data. I hope to demonstrate that the totality of the criticisms does not challenge the conclusions reached in "Gun control and rates of firearms violence in Canada and the United States." (Mundt 1990).
The critique presents the data mostly in terms of percentage changes in crime rates. This is a perfectly acceptable approach to describing the data, but obviously it will tend to show a greater change than does the gap in points shown in the line graphs in my article, especially starting from the lower base of Canadian rates. Furthermore, the critique never refers to the sense of "parallel" change that is conveyed in the line graphs, i.e., whether the direction of change is the same over time, and whether the gap between the rates narrows or widens. There is more than one way to present and describe a given set of data; the approach in my article was not "wrong."
(1) The critique identifies a "factual inaccuracy" in the comparison of victim rates depicted in Figure 1 and the accompanying narrative. Indeed, as records are revised, these rates will vary slightly over time. The critique finds that the average victim rate was 2.71 per 100,000. Using current figures from Canadian Centre for Justice Statistics (CCJS)(1), I find the mean to be 2.62. Not "exactly" 2.5 1, but (to use the critique's approach to data analysis) the 1987 rate was only 4 percent lower than the 1974-1987 mean. The "bottom line" in all this is that the ratio of U.S. - Canadian homicide victim rates, at a mean of 3.5 annually from 1980 to 1988, was 3.5 in 1988. The conclusion of the cited paragraph in the article, that U.S. and Canadian homicide rates have remained largely parallel, remains true.
(2) The critique suggests that the description of robbery rates is "inaccurate" because totals for armed robbery are discussed, rather than firearm robbery. This is disingenuous criticism, for firearms robbery rates are discussed in the next paragraph, and a comparative line graph is presented for each (Figures 3 and 4, pp 143 and 145). The critique states that the use of "armed robbery" is unjustified. However, in each category of violent crime discussed (including suicide), the comparison of national rates is given both for overall rates and firearms rates: Overall homicides and homicides by firearm; all armed robberies and robberies by firearm; and total suicides and suicides by firearm. Presumably the purpose of restricting firearms is not only to reduce the use of firearms in violent crime, but to reduce the level of such crime. If a murderer uses a knife because a gun is not available, is the purpose of the legislation served? I am not concluding that such is indeed the result. But I think the reason for looking at both rates -- total violent crime and those involving firearms -- is obviously relevant to the discussion. The presentation can hardly be faulted, in any event, when firearms crime is presented and discussed in each category.
(3) The critique finds that the 1985 rate of armed robbery in the U.S. was not significantly" lower than the 1974-1978 mean. Although "significance" is a slippery concept when not referring to exact statistical tests (which it could not here -- I should perhaps have said "meaningful"), the difference between the 1974-78 mean and 1985 was 6.6 percent.(2)
(4) The critique finds fault with a shift from 'armed robbery' to |firearm robbery' "without clear indication of the shift." If the purpose explained in 2) above in unclear, I apologize to the readers of CJC. The critique asks why Figure 3 ends in 1985; given the time that has passed, I cannot answer the question, except to assume that I used different sources of data in the two cases. But as revised Figure 3 demonstrates, the addition of 1986 to 1988 does not change the conclusion in the slightest. …