Content uploaded by Matthew Miller
Author content
All content in this area was uploaded by Matthew Miller
Content may be subject to copyright.
100 JAMWA Vol. 57, No. 2
DAVID HEMENWAY, PHD
TOMOKO SHINODA-TAGAWA,
MD, MPH
MATTHEW MILLER, MD,
MPH, SCD
Objective: to determine the association
between firearm availability and female
homicide victimization among high-
income countries.
Methods: Data were assembled for the
most recent available year (1994-1999)
from the official reports of the min-
istries of health for those countries that
had more than 2 million inhabitants
and were classified as high income by
the World Bank. Twenty-five nations
provided sufficient information for the
analysis. Rates of female victimization
from homicide, firearm homicide, and
nonfirearm homicide were compared
with a validated proxy for household
firearm ownership (the percentage
of total national suicides that are
committed with firearms). Possible
confounding variables included in
the analysis were the percentage of the
population living in urban areas and
income inequality.
Results: The United States is an
outlier. It had the highest level of
household firearm ownership and the
highest female homicide rate. The
United States accounted for 32% of
the female population in these high-
income countries, but for 70% of all
female homicides and 84% of all
female firearm homicides. Female
homicide victimization rates were
significantly associated with firearm
availability largely because of the
United States.
Conclusion: Among high-income
countries, where firearms are more
available, more women are homicide
victims. Women in the United States
are at higher risk of homicide victim-
ization than are women in any other
high-income country. (JAMWA.
2002;57:100-104)
Crossnational studies usually
1-3
(but not
always
4
) find that countries with higher
levels of household firearm ownership
have significantly higher homicide rates.
Women have much lower rates of homi-
cide victimization than men do, so the
data in these analyses are dominated
by male deaths. To our knowledge, no
study has examined female homicide
rates crossnationally.
A few studies have investigated the
association between household firearm
ownership and homicide among US
women. One study found that house-
hold firearm ownership was significantly
correlated with female firearm homicide
rates, but not with nonfirearm homicide
rates.
5
Another study found that higher
levels of firearm ownership were signifi-
cantly associated with both firearm and
overall homicide rates of women.
6
In this article we examine whether
women in countries with higher levels of
household firearm ownership are also at
higher risk of homicide victimization.
Our study analyzed late 1990s data from
countries that had more than 2 million
inhabitants and were defined by the
World Bank as high-income nations.
We analyzed only high-income nations
because, compared with lower-income
nations, their surveillance data are more
reliable and their socioeconomic condi-
tions are more comparable. We were par-
ticularly interested in the victimization
rates of women in the United States.
Methods
The World Bank classifies countries by
income level based on their gross national
product per capita.
7
The World Health
Organization assembled data on homicide
victimization and suicide, broken down
into firearm and nonfirearm categories,
from official reports by the ministries
of health of those countries classified as
high income by the World Bank. We
used data from only those high-income
countries that had populations over 2
million. Many of the less populous high-
income countries did not provide com-
plete data, and results of rare events such
as homicides among women might be
unstable if we used data from countries
with relatively few people (eg, Iceland,
Kuwait, Luxembourg, Macau, Slovenia,
Northern Ireland). We counted the
United Kingdom as 2 populous countries
(England/Wales and Scotland), for a
total of 25 nations with more than 2
million inhabitants that met the World
Bank definition of high income.
We used data from the most recent
single year (1994-1999) for which they
were available. The names of the coun-
tries, size of female population, absolute
numbers and rates of overall female
homicide, firearm homicide, and non-
firearm homicide are given in Table 1.
Data were not available on handgun
fatalities.
Perhaps the most preferred proxy for
firearm availability is survey information
on the percentage of households with
firearms, but many high-income coun-
tries do not record such data. Various
other proxies have been used, and a recent
study tried to determine which of these
was the most valid and reliable. The
authors concluded that the percentage
of suicides committed with guns was
consistently better than the other proxies
and that it had a high degree of validity
when tested against survey-based esti-
mates.
8
The percentage of suicides with
ALSO IN THIS ISSUE
Firearm Availability and Female Homicide Victimization
Rates Among 25 Populous High-Income Countries
Dr. Hemenway is director, Dr. Shinoda-Tagawa is
a researcher, and Dr. Miller is associate director, all
at the Harvard Injury Control Research Center.
Dr. Hemenway is also a professor in the Harvard
School of Public Health.
firearms has been shown to be very highly
correlated with the percentage of house-
holds reporting firearm ownership
among US Census regions (r=.93),
9
21 US states (r=.90),
9
170 US cities
(r=.86),
10
and 14 areas within a single
US state (r=.76).
8
The proxy is also highly correlated
with surveys of gun ownership among
high-income nations; the correlation was
over 90% (r=.91) in 16 developed
nations in the early 1990s.
2
For this arti-
cle, we analyzed data from 18 high-
income nations for the late 1990s
4
and
found a similar high degree of correla-
tion (r=.92). Given the high degree of
correlation for those 18 countries, the
association between firearm availability
and female homicide rates was virtually
identical whether we used firearm own-
ership or the percentage of suicides with
firearms as our proxy for firearm avail-
ability (not shown).
Thus, we used the percentage of total
(male and female) suicides with firearms
as a measure of firearm availability. The
percentage of male suicides with firearms
was virtually the same as the percentage
of total suicides with firearms (r=.99) for
these 25 countries. We did not use the
percentage of female suicides with firearms
as a proxy for firearm availability in this
analysis because most firearms are owned
and used by men, and most female
homicide victims are killed by men.
Analyses were conducted using both
the homicide victimization rate and the
natural log of the homicide victimization
rate, the latter better approximating a
normal distribution. Analyses were also
conducted using both unweighted data
and data weighted by each country’s
female population.
We also included 2 control variables,
urbanization
11
and income inequality,
12
which have been associated with homi-
cide rates in previous studies. Data on
the percentage of the country that is
urbanized come from the Population
Reference Bureau. As a proxy for income
inequality, we use the Gini coefficient;
data come from the World Bank’s World
Development Report. Unfortunately,
Gini coefficient data were not available
for New Zealand or Hong Kong, and
data for England/Wales and Scotland
were combined. As a result, multivariate
analyses, which included both urbaniza-
tion and income inequality as indepen-
dent variables along with gun availability,
had only 22 observations.
We did not use poverty as a predictor
variable because data were available for only
17 of the populous high-income countries.
We provide the Pearson correlation
coefficient between the gun availability
proxy and female homicide, and in
the regressions we used the t-test for
significance.
Spring 2002 101
Table 1. Female Homicide Rates for 25 Populous High-Income Countries, 1994-1999
Gun Suicides
Female as Percentage
Population Total Homicides, Gun Homicides, Nongun Homicides, of All Suicides
Country Year (in millions) n (rate) n (rate) n (rate) (men and women)
Australia 1997 9.3 106 (1.14) 25 (0.27) 81 (0.87) 11.3
Austria 1999 4.2 36 (0.86) 8 (0.19) 28 (0.67) 17.5
Belgium 1995 5.2 69 (1.33) 22 (0.42) 47 (0.91) 13.4
Canada 1997 15.1 146 (0.96) 44 (0.29) 102 (0.67) 22.2
Denmark 1996 2.7 23 (0.86) 4 (0.15) 19 (0.71) 9.0
England and Wales 1997 26.5 106 (0.40) 6 (0.02) 100 (0.38) 2.9
Finland 1996 2.6 51 (1.94) 10 (0.38) 41 (1.56) 22.3
France 1997 30.0 216 (0.72) 84 (0.28) 132 (0.44) 22.9
Germany 1998 42.0 278 (0.66) 47 (0.11) 231 (0.55) 7.8
Greece 1997 5.3 32 (0.60) 12 (0.23) 20 (0.38) 20.1
Hong Kong 1996 3.2 24 (0.76) 1 (0.03) 23 (0.73) 0.5
Ireland 1996 1.8 14 (0.77) 2 (0.11) 12 (0.66) 9.1
Israel 1997 3.0 6 (0.20) 1 (0.03) 5 (0.17) 19.3
Italy 1997 29.6 159 (0.54) 52 (0.18) 107 (0.36) 13.3
Japan 1997 63.7 284 (0.45) 4 (0.01) 280 (0.44) 0.2
Netherlands 1997 7.9 63 (0.80) 11 (0.14) 52 (0.66) 3.2
New Zealand 1998 1.9 25 (1.30) 1 (0.05) 24 (1.25) 12.5
Norway 1997 2.2 13 (0.58) 5 (0.22) 8 (0.36) 23.8
Portugal 1998 5.2 33 (0.64) 17 (0.33) 16 (0.31) 9.0
Scotland 1999 2.6 20 (0.76) 1 (0.04) 19 (0.72) 2.2
Singapore 1997 1.5 16 (1.04) (0.00) 16 (1.04) 1.3
Spain 1997 20.1 99 (0.49) 22 (0.11) 77 (0.38) 6.6
Sweden 1996 4.5 36 (0.80) 2 (0.04) 34 (0.76) 13.0
Switzerland 1994 3.6 48 (1.34) 22 (0.61) 26 (0.72) 26.4
United States 1997 136.6 4384 (3.21) 2106 (1.54) 2278 (1.67) 57.6
Total 430.5 6287 (1.46) 2509 (0.58) 3778 (0.88)
Total excluding United States 293.9 1903 (0.65) 403 (0.14) 1500 (0.51)
102 JAMWA Vol. 57, No. 2
Results
The female homicide rate per 100 000 for
the 25 populous high-income countries
varied from 0.20 (Israel) to 3.21 (United
States). The weighted mean was 1.46
(6287 homicides divided by the total
female population); the unweighted
mean was 0.92, with a standard deviation
of 0.60.
The United States was an extreme
case in terms of both gun prevalence and
female homicide (Figures 1 and 2). The
United States accounted for 32% of the
total female population among these
high-income nations and for 70% of
all female homicide victims (Table 1).
The US female homicide rate was 5
times that of all the other high-income
countries combined (3.21 per 100 000
v 0.65 per 100 000), and the female
firearm homicide rate in the United
States was 11 times higher than that of
the other countries in our sample (1.54
per 100 000 v 0.14 per 100 000); the
US nongun homicide rate was 3 times
higher (1.67 per 100 000 v 0.51 per
100 000) (Table 1).
When the United States was excluded,
Finland and Israel were the main outliers
(Figures 1 and 2). Both have fairly many
guns, but Finland had a high female
homicide rate and Israel a very low rate.
In the bivariate analysis, for unweighted
data, the total female homicide rate
was significantly associated with firearm
availability (r=.71) among these 25
countries. When the United States was
excluded, however, the association was
not significant (Table 2). Results using
the natural log of the dependent variable
were always very similar (but with corre-
lation coefficients a bit lower) and are
not shown.
For unweighted data, the female
firearm homicide rate was very highly
correlated with firearm availability
(r=.87). US women accounted for 84%
(2106/2509) of women who were killed
with firearms (Table 1). Even when the
United States was excluded, there was
still a strong correlation between firearm
availability and firearm homicide rates
(r=.66) (Table 2).
For unweighted data, the nongun
female homicide rate was significantly
associated with gun availability (r=.42).
When the United States was excluded,
Homicide Rate
% Suicides With Guns
0
20
40
60
0
1
2
3
Figure 1. Female homicide victimization rates versus a proxy for firearm
availability for 25 populous high-income countries, 1994-1999. See Figure 2
for names of countries.
Homicide Rate
% Suicides With Guns
0
20 40 60
0
1
2
3
Australia
Austria
Belgium
Canada
Denmark
England
Finland
France
Germany
Greece
Hong Kong
Ireland
Israel
Italy
Japan
Netherlands
New Zealand
Norway
Portugal
Scotland
Singapore
Spain
Sweden
Switzerland
United States
Figure 2. Female homicide victimization rates versus a proxy for firearm
availability for 25 populous high-income countries, 1994-1999. See Figure 1
for placement of each country.
Spring 2002 103
there was no association between the
nongun female homicide rate and gun
availability (Table 2).
In bivariate analysis using weighted
data, results show that, when the United
States was excluded, overall female homi-
cide rates and female firearm homicide
rates were associated with gun availability,
and nongun homicide rates were not
(Table 2).
Neither urbanization nor income
inequality was significantly associated
with the overall female homicide rate,
the female firearm homicide rate, or the
female nonfirearm homicide rate in either
the unweighted or weighted analyses.
Including these 2 variables in multivariate
regressions had virtually no effect on the
association between firearm availability
and female homicide victimization (not
shown).
Discussion
Our simple regressions of 25 populous
high-income countries showed a signifi-
cant positive correlation between gun
availability and rates of female homicide
victimization. The correlation between
gun availability and the rate of female
firearm homicide victimization was even
higher.
Our results are consistent with many
ecological studies of the relationship
between firearm availability and total
(male and female) homicide rates.
Crossnational studies of high-income
countries
1-3
and cross-sectional studies
of the United States
13-17
have found that
areas with more firearms have higher
homicide rates. Our results are also
consistent with case-control studies that
have found a gun in the home to be a
risk factor for homicide victimization
and perpetration.
18-21
The US female homicide victimization
rate was 5 times higher than the rate
for the other high-income countries
combined. When the United States was
excluded from the sample, a significant
relationship existed between gun avail-
ability and female homicide in the model
when we weighted observations by each
country’s population, but not for the
unweighted analysis. Weighting the
25 observations by the population of
the country increased the association
between firearm availability and female
homicide because some of the most
populous countries (eg, the United
States, Japan, England/Wales) were at
the extremes. The United States had
many guns and many female homicides,
whereas Japan and England/Wales had
few guns and few female homicides.
Because the United States has such a
large population, weighted analyses
emphasized how it differed from other
high-income nations in homicide rates
and firearm availability. The United
States was also an outlier among high-
income countries not only in the number
of households with firearms, but even
more so in the percentage with handguns.
2
In addition, the United States has fewer
regulations governing the acquisition and
use of firearms than other high-income
countries,
22
giving residents much easier
access to firearms.
Although we know of no other cross-
national studies of firearms and female
homicide, various studies have examined
the relationship between firearms and
female victimization within the United
States. Stranger violence is not the major
threat to women, as it is for men. An
analysis of female homicides in the United
States from 1976 to 1987 found that
when the perpetrator was known, almost
half were spouses or intimate acquain-
tances; only 13% of female homicide
victims were killed by strangers. Many
more women were killed with guns used
by their husbands or intimate acquain-
tances than were murdered by strangers
using guns, knives, or any other means.
23
A recent study of the United States
showed that women in states with higher
levels of firearm availability had higher
rates of homicide victimization.
6
A case-
control study
24
of 143 women from 3
metropolitan counties who were killed
in their homes found that having a gun
in the home was a large, independent,
and significant risk factor for homicide
victimization (odds ratio 3.4). Other
factors controlled for in the analysis
included age, race, neighborhood, a
history of mental illness or depression,
and living alone.
One reason a gun in the home can be
a threat to women is that assaults with
guns are far more likely to be lethal than
are other assaults. A study of family and
intimate assaults in Atlanta found that
firearm assaults were 3 times more likely
to result in death than were assaults with
knives and 23 times more likely to result
in death than were assaults with other
weapons.
25
Guns are used against women
to intimidate as well as to wound or kill.
More than 6% of US women reported
having ever been threatened with guns,
and most assaults against women were
perpetrated by their partners.
26
National
random-digit-dial surveys conducted in
1996 and 1999 found that gun threats
in the home against women by intimates
were far more common than home self-
defense gun uses by women.
27,28
The analyses in this article have various
limitations. First, the regressions contain
only 3 independent variables: urbaniza-
tion, income inequality, and a proxy for
firearm availability. It is possible that the
associations could be explained by other
variables. Although we looked exclusively
at high-income nations and thus con-
trolled, in part, for some social and eco-
nomic variables, these countries differ
in many ways. Even among regions of
the United States, for example, cultures
may vary sufficiently to affect homicide
rates.
29
The cultural differences among
Table 2: Correlation Coefficients and Significance Levels from
Regressions of Gun Availability and Female Homicide Rates
for 25 Populous High-Income Countries, 1994-1999
Gun Ownership Level
For Female Victims Gun Ownership Level, Weighted by Population,
Dependent Variable Correlation (p) Correlation (p)
Total homicide rate 0.71 (<.001) 0.97 (<.001)
Excluding United States 0.30 (.15) 0.54 (.01)
Firearm homicide rate 0.87 (<.001) 0.98 (<.001)
Excluding United States 0.66 (<.001) 0.84 (<.001)
Nonfirearm homicide rate 0.42 (.04) 0.93 (<.001)
Excluding United States 0.02 (.91) 0.19 (.38)
nations are even greater.
Another limitation of our study is that
measures of homicide may vary from
country to country because of differences
in the specificity and sensitivity of the
surveillance systems. Such problems
with data comparability and accuracy are
particularly acute for nonindustrialized
nations. Including only high-income
countries in our analysis reduced this
problem.
Cross-sectional studies like ours do
not provide information about causality.
It is possible that high rates of lethal
violence cause some US households to
acquire firearms. For women, this does
not generally appear to be a beneficial
strategy, because many women are
murdered by intimates. Furthermore,
because men typically own household
firearms,
30
reverse causation is not likely
to be a problem in this study.
Our analyses showed that in high-
income countries, where there were more
firearms, there were more female homi-
cide victims, particularly female firearm
homicide victims. These results were
driven largely by the United States. US
women are at far higher risk of homicide
victimization than are women in any
other high-income country.
Thanks to Gyanendra Sharma of the World Health
Organization for providing the data. This research
was supported in part by grants from the Joyce
and Robert Wood Johnson Foundations, the Open
Society Institute, and the Centers for Disease
Control and Prevention.
References
1. Killias M. International correlations between
gun ownership and rates of homicide and
suicide. Can Med Assoc J. 1193;148:1721-1725.
2. Killias M. Gun ownership, suicide and homi-
cide: An international perspective. In: del Frate
AA, Zvekic U, van Dijk JJM, eds. Understand-
ing Crime: Experiences of Crime and Crime Con-
trol. Rome: United Nations Interregional Crime
and Justice Research Institute; 1993:289-303.
3. Hemenway D, Miller M. Firearm availability
and homicide rates across 26 high-income
countries. J Trauma. 2000;49:985-988.
4. Killias M, van Kesteren J, Rindlisbacher M.
Guns, violent crime, and suicide in 21 countries.
Can J Criminol. 2001;43:429-448.
5. Kaplan MS, Geling O. Firearm suicides and
homicides in the United States: Regional varia-
tions and patterns of gun ownership. Soc Sci
Med. 1998;46:1227-1233.
6. Miller M, Azrael D, Hemenway D. Firearm
availability and unintentional firearm deaths,
suicides and homicides among women. J Urban
Health. In press.
7. World Bank. World Development Report. New
York, NY: Oxford University Press; 2000.
8. Azrael D, Cook PJ, Miller M. State and local
prevalence of firearms ownership: Measure-
ment, structure and trends. 2001. National
Bureau of Economic Research Working Paper
#8570. Available at: www.nber.org/papers/w8570.
Viewed on 2/28/02.
9. Miller M, Azrael D, Hemenway D. Firearm
availability and unintentional firearm deaths,
suicides and homicides among 5-14 year olds.
J Trauma. 2002; 52:267-275.
10.Kleck G. Targeting Guns: Firearms and their
Control. New York, NY: Aldine de Gruyter;
1997.
11. Fingerhut LA, Ingram DD, Feldman JJ. Firearm
and non-firearm homicide among persons 15
through 19 years of age—Differences by level
of urbanization, United States, 1979 through
1989. JAMA. 1992;267:3048-3053.
12.Kennedy BP, Kawachi I, Prothrow-Stith D,
Lochner K, Gupta V. Social capital, income
inequality, and firearm violent crime. Soc Sci
Med. 1998;47:7-17.
13.Brearly HC. Homicide in the United States.
Chapel Hill, NC: University of North Carolina
Press; 1932.
14.Seitz ST. Firearms, homicide, and gun control
effectiveness. Law Soc Rev. 1972;6:595-614.
15.Brill S. Firearm Abuse: A Research and Policy
Report. Washington, DC: Police Foundation;
1977.
16.Lester D. Firearm availability and the incidence
of suicide and homicide. Acta Psychiatr Belg.
1988;88:387-393.
17.Lester D. Relationship between firearm avail-
ability and primary and secondary murder.
Psychol Rep. 1990;67:490.
18. Kellermann AL, Rivara FP, Rushforth NB, et
al. Gun ownership as a risk factor for homicide
in the home. N Engl J Med. 1993;329:1084-
1091.
19.Cummings P, Koepsell TD, Grossman DC, et
al. The association between the purchase of a
handgun and homicide or suicide. Am J Public
Health. 1997;87:974-978.
20.Rowland J, Holtzhauer F. Homicide involving
firearms between family, relatives and friends in
Ohio: An offender-based case-control study.
Am J Epidemiol. 1989;130:825.
21.Kleck G, Hogan M. A national case control
study of homicide offending and gun owner-
ship. Soc Probl. 1999;46:275-293.
22.United Nations. International Study on Firearm
Regulation. New York, NY: United Nations;
1998.
23.Kellermann AL, Mercy JA. Men, women, and
murder: Gender-specific differences in rates
of fatal violence and victimization. J Trauma.
1992;33:1-5.
24.Bailey JE, Kellermann AL, Somes GW, et al.
Risk factors for violent death of women in the
home. Arch Intern Med. 1997;157:777-782.
25.Saltzman LE, Mercy JA, O’Carroll PW, et al.
Weapon involvement and injury outcomes in
family and intimate assaults. JAMA. 1992;
267:3043-3047.
26.Tjaden P, Thoennes N. Prevalence, Incidence
and Consequences of Violence Against Women:
Findings from the National Violence Against
Women Survey. Research in Brief. Washington,
DC: National Institute of Justice, US Depart-
ment of Justice; 1998.
27.Azrael D, Hemenway D. “In the safety of your
own home”: Results from a national survey of
gun use at home. Soc Sci Med. 2000;50:258-291.
28.Hemenway D, Azrael D. Gun Use in the United
States: Results of a National Survey. Report to the
National Institute of Justice. Washington, DC:
US Department of Justice; 1997.
29.Nisbett RE, Cohen D. Culture of Honor: The
Psychology of Violence in the South. Boulder,
Colo: Westview Press; 1996.
30.Cook PJ, Ludwig J. Guns in America: Results of
a Comprehensive National Survey on Firearms
Ownership and Use. Washington, DC: Police
Foundation; 1996.
104 JAMWA Vol. 57, No. 2