We compile citations and summaries of about 400 new articles every week.
Email Signup | RSS Feed

HELP: Tutorials | FAQ
CONTACT US: Contact info

Search Results

Journal Article


McDowall D. Soc. Forces 1991; 69(4): 1085-1101.


(Copyright © 1991, Social Forces Journal, Publisher University of North Carolina Press)






This article examines the relationship between firearm availability and homicide rates in Detroit, Michigan. Noting the difficulties involved in measuring gun density, the analysis uses an indicator based on the relative frequency with which firearms are employed in robberies and suicides. Models estimated from time-series data are consistent with the argument that higher levels of firearm density increased the risk of homicide in the city. A variety of supplementary analyses support this finding and suggest that the effect of gun availability on Detroit's homicide rates is relatively large.

VioLit summary:

The purpose of this study by McDowall was to examine the relationship and influence between firearm availability and homicide rates in Detroit, Michigan. The question of whether firearm availability may increase the use of guns in crimes was also a focus.

A quasi-experimental, secondary analysis was used to analyze and estimate the influence of firearm availability (per 100,000 residents) on Detroit's homicide rate. Official data that annually covers the period between 1951-1986 was used. Justifiable homicides were excluded by an index summing two indicators: 1) the proportion of Detroit robberies committed with a gun, and 2) the proportion of Detroit suicides committed with a gun. Validation of the gun density index was tested in two ways: 1) for 1974+1978, values of the index were correlated with Cook's (1979) estimates of gun availability (correlations were .86 in 1974 + .85 in 1978). The fifty cities were grouped by U.S. census regions, and the 1974 mean value of the density index was computed for cities within each region, and compared with 1973, '74, and '76 estimates of 655 gun ownership (correlation were .87 for pistol ownership and .93 for total firearm ownership.) To allow for variation in demographic composition, the proportion of Detroit's population that was nonwhite and between the ages of 15 and 24 was included. To take into account the deterrent effect of crime control activity, the model includes homicide rates lagged by one year. The number of robberies per 100,00 was included to avoid indirect effects of gun density. All indicators were transformed to their natural logarithms for the analysis, and a two stage least squares multiple regression analysis was used.

In this study gun density influenced homicides. The estimates indicated that changes in firearm availability altered the risk of homicide, but increases in gun density could not completely account for Detroit's high murder rate. Regression analysis was done using the Detroit homicide rate (excluding robbery rate) as the dependent variable. Three indicators had significant positive relationships with homicide rate: proportion of population aged 15-24, proportion of the nonwhite population, and gun density index. The lagged homicide clearance rate had a significant negative relationship with the homicide rate. When the robbery rate was added to the equation, neither the direction or significance of the relationships changed except that the proportion of nonwhites ceased to be significant. Both gun density and robbery rates had independent, significant effects. Additional analysis were conducted to examine the sensitivity of the results to changes in estimation methods and variable measurements. Substituting annual numbers of gun licenses issued (per 100,000) into the equation resulted in the proportion of the population 15-24, and gun license measures, to be significant and positive. Proportionally, nonwhite lagged homicide clearance and robbery rates were insignificant. A second equation substituted per capita income in Michigan for the riot variable. The results of this were very close to the first analysis. A third equation assumed that there would be instantaneous mutual causation between gun density and homicides within one year. The findings were identical to the original findings. Another modification looking at lagged gun density found identical results with the lagged gun density. A model included unemployment rate and ratio of Detroit infant mortality to U.S. infant mortality as indicators showed virtually the same results among the original variables except that robbery and proportion of nonwhites were insignificant. The two economic variables were also insignificant. None of these modified models were significant as a whole.

The author stated that there appears to be nothing about Detroit's criminal, social, or economic environment that would influence the association between firearm availability and homicide rates in a unique way. The analysis should be replicated, both in other areas and with other measures of gun density, since knowledge about gun use is very limited.

The author present a study with many contributions. There measure of gun density, shown to have good external and internal validity, is one that provides for further replications of the study. The rigorous analysis which both tested other measures of gun density and other possible confounding factors still, in all cases, showed that firearm availability is directly connected to homicide rates. This finding, particularly if replicated on a wider scale, has strong implications for gun policy. More research is also needed into the factors surrounding gun use and ownership. (CSPV Abstract - Copyright © 1992-2003 by the Center for the Study and Prevention of Violence, Institute of Behavioral Science, Regents of the University of Colorado)

KW - Michigan
KW - 1950s
KW - 1960s
KW - 1970s
KW - 1980s
KW - Homicide Rates
KW - Homicide Causes
KW - Firearms Violence
KW - Firearms Homicide
KW - Firearms Availability

Language: en


All SafetyLit records are available for automatic download to Zotero & Mendeley