Institute for Advanced Development Studies
Development Research Working Paper Series
No. 17/2009
Gun control and suicide:
The impact of state firearm regulations, 1995–2004
by:
Katherine Hempstead
Antonio Rodríguez Andrés
December 2009
The views expressed in the Development Research Working Paper Series are those of the authors
and do not necessarily reflect those of the Institute for Advanced Development Studies. Copyrights
belong to the authors. Papers may be downloaded for personal use only.
Gun control and suicide:
The impact of state firearm regulations, 1995–2004♦
Katherine Hempstead*
Antonio Rodríguez Andrés$
December 2009
Abstract:
Suicide is a major cause of preventable death. Restricting access to lethal means has
been identified as an effective approach to suicide prevention, and firearms
regulations are one way to reduce gun availability. This study examines the
relationship between state firearms regulations and suicide among males, using
negative binomial regression and state panel data for the years 1995–2004. Results
suggest that firearms regulations which function to reduce overall gun availability
have a significant deterrent effect on suicide, while prohibited persons categories
have less of an effect. Overall, the results suggest that gun control measures such as
permit and licensing requirements might have public health benefits.
Keywords: suicide; guns; state regulations; panel data.
JEL classification: I18.
♦
The authors thank Sara Markowitz, Camelia Minoiu, and Gary Kleck for helpful comments and
suggestions.
*
The Center for State Health Policy Rutgers, The State University of New Jersey 55 Commercial
Avenue, 3rd Floor New Brunswick, NJ 08901-1340 (E-mail:khempstead@ifh.rutgers.edu). (Tel: 732932-3105; Fax:732-932-0069)
$
School of Public Health, Department of Health Services Research, University of Aarhus, Bartholins
Allé 1, DK- 8000 C, Aarhus, Denmark, (E-mail: ara@folkesundhed.au.dk). Tel: (+55) 89423122.
1. INTRODUCTION
1.1 Firearms and suicide
Suicide is a major cause of preventable death. In 2006, more than 32,000 suicides
occurred in the United States, as compared with approximately 18,000 homicides.
Suicide is the 8th leading cause of death for males, and the 19th leading cause for
females in the United States. For every suicide, there were more than ten
hospitalizations from non-fatal attempts. In 2006, an average of 46 Americans
committed suicide daily with a firearm, accounting for approximately 50 percent of
all suicides (CDC, 2009).
There is a considerable body of empirical work that has documented a positive
relationship between access to firearms and suicide (Miller and Hemenway, 2008;
Miller et al., 2005). In fact, much of the decline in suicide in the United States over
the past decades has been linked to the reduced prevalence of firearms (Miller et al.,
2005; Miller et al., 2007; Cook and Ludwig, 2006). Although the respective roles of
self-selection and availability in explaining the relationship between guns and suicide
have not been completely resolved, the implication in either case is that reducing
access to firearms should reduce suicide (Duggan, 2003) Restricting access to
firearms has been recommended as a suicide prevention strategy by national and
international organizations such as the CDC and the WHO. Gun control policies can
serve to reduce overall gun availability by creating barriers to firearm ownership.
Additionally, firearms policies can also prevent individuals who are at a relatively
higher risk of suicide from purchasing firearms.
Gun Control
Gun control is a highly contentious issue in the American political debate. Guns are
common in the United States – 40 percent of Americans reported having a gun in
their home in 2009 (Gallup, 2009). Calls for increased regulation are based on the
belief that restrictions will reduce gun violence. Regulation is opposed by those who
claim infringement on the constitutional right to bear arms, and/or argue that firearm
2
ownership deters crime. In the academic literature, the efficacy of gun control in
reducing violence has received considerable attention, although little consensus has
emerged from the empirical work (Cook and Ludwig, 2006; Lott and Whitley, 2000;
Kleck and Kates, 2001; Kleck, 2004).
The current era of gun control in the United States originated with the Brady
Handgun Violence Prevention Act (1993)1, more commonly known, as the Brady
Bill. The Brady Bill established a federal requirement for a waiting period of up to
five days before the transfer of a handgun to a purchaser. During this period, a
background check is performed, which is intended to prohibit individuals with
criminal backgrounds from purchasing firearms. The transfer of the handgun is
completed whether or not the background check is finalized within the five-day
period. The federal waiting period was phased out in 1998 with the development of
the National Instant Criminal Background Check System (NICS), administered by the
Federal Bureau of Investigation (FBI). Over time, many states have passed laws
which matched or surpassed the federal minimums.
There are many different types of state firearm regulations. Some seek to establish
general oversight over individuals owning firearms, and mainly consist of permit,
registration, and/or license requirements, and bans on the purchase of firearms by
minors. These laws also facilitate the tracing of firearms used in crimes to original
purchasers. Other state laws seek to prevent gun trafficking and the use of firearms in
crimes. These consist of bans on the sale of certain types of firearms, and restrictions
on the number of firearms that can be sold to individuals. Restrictions on carrying
concealed weapons serve a similar purpose. A number of laws are designed to prevent
firearm ownership by individuals considered disproportionately likely to commit gun
crimes. These laws include prohibitions on gun ownership by those with criminal
histories, such as conviction for a felony, misdemeanor, or domestic violence offence,
as well as those with a history of mental illness, and alcohol or drug problem, and
minors. The requirement of a “cooling off” period of some specified period before the
purchase can be completed is a measure designed to reduce the consequences of
impulsive firearm purchases.
1
The Gun Control Act (1968) was the first firearm act in the USA.
3
There is considerable variation in the comprehensiveness of firearm regulation across
U.S. states. Some states have almost no firearm regulation of their own. Forty-four
states have a provision in their state constitutions similar to the Second Amendment
of the Bill of Rights (the exceptions are California, Iowa, Maryland, Minnesota, New
Jersey, and New York). Firearm license holders are subject to the firearm laws of the
state in which they are carrying and not the laws of the state in which the permit was
issued. Reciprocity between states may exist for certain licenses such as concealed
carry permits. These are recognized on a state-by-state basis.
Some firearms regulations are more relevant to suicide prevention than others.
Restrictions banning the purchase of guns by convicted felons, or laws banning the
sale of “Saturday Night Specials”, for example, have little obvious applicability to
suicide. Yet other categories of restriction are potentially more salient, particularly
those that reduce overall firearm availability. Permit requirements create barriers to
gun ownership and may also serve to prevent impulsive purchases. The prohibition of
purchases by minors serves a similar function. Some of the “prohibited persons”
categories, such as those related to mental illness, a drug or alcohol problem, or
history of domestic violence problems may theoretically be relevant to suicide
prevention, but in all likelihood are not.2 Mental illness is the single most important
risk factor for suicide, and substance abuse and domestic violence are also risk
factors. However, while the criteria for “prohibited persons” categories varies by
state, they are generally based on fairly serious incidents, such as hospitalization
against one’s will or conviction records. Such bans are likely to identify only a
fraction of the population with mental health, substance abuse, or domestic violence
problems.
At the state level, the comprehensiveness of gun control laws tends to be correlated
with firearm prevalence. The causality most likely runs in both directions, since
restrictive gun control regimes reduce gun ownership, yet these laws are more likely
to be passed in states where overall gun ownership rates are relatively low and the
2
In some states, the alcohol regulation means that sale of firearms are prohibited to people who are
intoxicated at the time they are trying to buy them, while in other states it refers to people with a
documented alcohol problem. Indeed, in some states it covers both situations.
4
population of gun rights advocates is relatively small. In general, Western and more
rural states have fewer gun control restrictions and higher rates of gun ownership as
compared with more urbanized states in the Northeast. These states also have
significantly higher rates of suicide, particularly firearm suicide.
1.2 Gun Control and Suicide
The literature on gun control and suicide has in general found a negative relationship
between firearm restrictions and suicide. However, most of these studies lack a strong
design and are essentially pre- and post- comparisons (Killias et al., 2001). Lambert
and Silva (1998) perform a literature review of studies in the United States and
Canada and conclude that available information generally supports the notion that
gun control can reduce suicide rates, particularly among males. Another review of
gun control in the United States, framed within the context of historical and rational
choice theory, covers attempts to curb firearm violence in that country and the
success of such measures (Cook et al., 2001). A recently published analysis suggests
that states where background checks are conducted locally have lower rates of firearm
suicide and homicide (Sumner et al., 2008)
Several other studies find no empirical evidence in favor of a relationship between
firearms regulations and suicide. However, one study has a weak design, while the
other does not capture the most relevant types of firearms regulations. Price et al.
(2004) use cross-sectional state data for 1999 to perform a simple partial correlation
analysis between several types of gun control laws and suicide rates. Their results
suggest that gun control laws were not significantly related to suicide in 1999, even
after controlling for firearm prevalence. Rosengart et al. (2005) conduct a study of the
relation between firearm regulations and homicides and suicides using state panel
data over 1979–1998. They fail to uncover a statistically robust link between suicide
rates and firearm regulations. However, most of the regulations they examined- such
as bans on carrying concealed weapons, “junk gun” bans, and quantity sales
restrictions are not particularly relevant to suicide.
5
Several studies in other countries where regulations restricted general access to
firearms have found evidence of an effect on suicide. Cheung and Dewa (2005)
examine the relationship between suicide and the implementation of new restrictions
on firearms (Bill C-17), using time series data from Canada. They concluded that
there was a relationship between means used by young people and the imposition of
the restrictions. In the case of New Zealand, Beautrais et al. (2006) find that after the
introduction of legislation restricting ownership and access to firearms, firearm
suicides significantly decreased, particularly among the young. Ozanne-Smith et al.
(2004) conclude that the implementation of a strong reform in New Zealand lowered
firearms deaths, particularly suicides. An evaluation of the 1996 National Firearms
Agreement (NFA) in Australia documents a decline in firearm suicides after the
implementation of the agreement (Klieve et al., 2009). However, these findings may
be confounded with an overall decline in gun ownership that preceded the NFA.
Additionally, there was some evidence of increased suicides by hanging.
In Europe, a study of the E.U. countries as a group and the U.K. in particular has
found that changes in firearm legislation have led to fewer firearm suicides (Kapusta
et al., 2007; Hawton et al., 1998; Haw et al., 2004). An analysis of firearm legislation
reforms enacted in 1997 in Austria also found a statistically strong effect on suicide
rates (Wahlbeck and Mäkinen, 2008).
2. EMPIRICAL MODEL AND DATA
2.1 Empirical model
The basic model that motivates the empirical analysis is that firearm availability
affects suicide rates, and that gun control affects firearm availability. Our hypothesis
is that regulations such as permit requirements, which create overall barriers to gun
ownership, are the most important way in which gun control may affect suicide.
While it is possible that “prohibited persons” categories can affect the likelihood that
certain persons at above average risk of suicide will obtain firearms, we believe that
the actual measurement and enforcement of those categories will result in the
6
prohibition of a relatively small proportion of people at risk. Firearms regulations
designed to prevent gun trafficking or other criminal activity involving guns are not
expected to influence suicide rates.
There are several potential complications to this simple model. The first is that of
state variation in views towards guns is likely to affect both firearm prevalence and
the comprehensiveness of gun control regulation. Additionally, views toward gun
ownership evolve over time. Finally, there is the problem of the mismeasurement of
gun ownership.
The basic model can be expressed with two equations:
1. S=α + βG + µ
2. G=δ - ∆R + Φ
Where S is suicide, G is firearm prevalence, and R is firearms regulation. The reduced
form is:
3. S= α+ βδ – β∆R + βΦ + µ
The potential endogeneity of firearm prevalence with respect to gun control is
reflected in the identifying equation,
4. R=ω + ηG+φ
However, G is not measured annually. For our main specification we estimate the
reduced form equation (3), thereby assuming that η is zero. In an alternative
specification, we instrument for G with the number of hunting licenses per capita, a
statistic which is collected annually for all states.
For the main specification, the estimating equation is:
5.
Sijt = f ( Z jt , X it , αi , λt , ε it )
where the subscript i indexes each age group with i = 15–24, 25–44, 45–64, and 65+ ,
j indexes each state with j=1,…,50 and t indexes time period with t = 1995,…, 2004.
Equation (5) specifies that the number of male suicides for age group i in state j
ε the
during the year t is a function of laws regulating the possession of firearms (Zit), other
socioeconomic characteristics (Xit), state effects (αi), year effects (λt), and
it
classical error term. The sample contains 500 state-year observations. The sample
period (1995-2004) was chosen because gun data before and after 1995 is not readily
7
comparable, and there are in fact relatively few state firearms regulations prior to
1995.
The dependent variable— male suicides by state, year and age group— violates the
Poisson assumption that the conditional mean is equal to the conditional variance.
Instead, the suicide data are overdispersed: the variance exceeds the mean. The overdispersion can cause a downward bias in the standard errors resulting from Poisson
regression. To avoid such bias, we estimate a negative binomial model via maximum
likelihood. The variance of the distribution is modeled as a quadratic function of the
mean, E [Yi] = µ, so that σ2 = µ + αµ2, where α is the dispersion parameter (Cameron
and Trivedi, 1998). In our case, the negative binomial model has a significantly better
goodness of fit than the Poisson model. A likelihood ratio test of the negative
binomial dispersion parameter leads to the conclusion that the Poisson distribution is
inappropriate for the data. We do not use zero-inflated specifications since there is no
excess of zeros in the dependent variable.
Each model includes the relevant population as a right hand side control variable to
normalize by exposure. The specification includes fixed effects that account for
unobserved heterogeneity across states. The fixed effects model is appropriate in this
case given the almost complete population coverage by the sample (Kennedy, 1998,
pp. 227) and it is likely that the omitted variables captured by the αi are correlated
with some of the included covariates. Since nearly ninety percent of firearm suicides
are committed by males, we have excluded females from the analysis (CDC, 2009).
We estimate separate models for males of all ages and in four different age groups:
15–24, 25–44, 45–64, and 65 years and older.
We face several identification challenges. The first is that gun control regulations by
state tend to change slowly, so there tends to be relatively little within-state year–onyear change. Further, once states adopt particular gun control regulations, they never
remove them. For these reasons, it is not possible to analyze leads and lags, which
would be a desirable robustness check. To maximize variation, we have created
several indices of categories of gun control regulations, which are additive measures
of individual measures.
8
2.2 Data
a. Dependent variable
Data on the number of suicides in states over the period 1995–2004 come from the
Centers for Disease Control (CDC). Deaths included in the study are those
categorized as suicides according to the International Classification of Diseases
(ICD). In 1999, there was a change in the classification system from ICD–9 to ICD–
10. In the case of suicide, this change in ICD version is thought to have a negligible
effect on the classification of suicides. For 1995–98, suicide deaths were coded as
E950–E959. Starting in 1999 and later, suicide deaths were coded as X60–X84,
Y87.0, and U03.
Table 1 displays the age adjusted male suicide rates across US states for the year
2004.3 Reported suicide rates in the US vary considerably across states. The suicide
rate in Nevada (30.2), for example, is nearly thrice that in New York (10.4). The
District of Columbia is excluded, since it has essentially banned the possession of
handguns.
Table 1. Age-adjusted male suicide rates, by state, 2004
State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of Columbia*
Florida
Georgia
Hawaii
Idaho
Illinois
Deaths
444
118
694
291
2546
611
230
70
27
1800
773
90
191
819
Population
2180516
341960
2875320
1340696
17781851
2319950
1690101
400799
273235
8483460
4386229
627760
699441
6225442
Rate
20.6
33.8
24.9
22.0
15.2
26.6
13.5
17.2
9.3
20.3
19.0
14.0
28.9
13.4
3
We do not show average suicide rates over 1995-2004 because of the relative position of the states is
basically unchanged during the study period.
9
Indiana
563
3054027 18.9
Iowa
287
1448679 19.7
Kansas
291
1351179 21.8
Kentucky
464
2024358 22.9
Louisiana
437
2176259 20.8
Maine
137
637824 21.0
Maryland
403
2675138 15.3
Massachusetts
343
3114101 10.9
Michigan
875
4960895 17.8
Minnesota
427
2523241 16.8
Mississippi
295
1395815 21.8
Missouri
562
2799494 20.4
Montana
142
462398 30.6
Nebraska
132
861390 15.4
Nevada
342
1181165 30.2
New Hampshire
104
636504 16.4
New Jersey
482
4202733 11.4
New Mexico
291
931748 31.5
New York
972
9335736 10.4
North Carolina
796
4170787 19.5
North Dakota
63
318958 19.0
Ohio
1036
5570983 18.7
Oklahoma
395
1731819 23.3
Oregon
415
1775505 23.3
Pennsylvania
1130
5981910 18.5
Rhode Island
70
517179 13.1
South Carolina
391
2041647 19.4
South Dakota
94
385163 24.7
Tennessee
627
2880226 22.0
Texas
1803
11163013 17.4
Utah
310
1227347 28.4
Vermont
77
303996 24.7
Virginia
657
3651909 18.6
Washington
639
3073759 21.4
West Virginia
236
880805 26.4
Wisconsin
507
2732268 18.4
Wyoming
67
253954 27.5
United States
25,566
144,060,672 18.0
Note: *The District of Columbia is excluded from the remainder of the analysis because it
had virtually outlawed the possession of guns during the study period.
b. Independent variables
Firearms regulations
In order to maximize variation across states and over time in the measure of gun
control, we created three additive indices that reflect different intensities of firearms
regulations. The first index ––arguably is the most important in terms of suicide
10
prevention–– measures general prohibitions. It is the sum of two indicator variables
reflecting the presence of permit requirements and on firearm, purchases by minors.
This index thus varies between 0 and 2.
The second index measures prohibitions based on behavioral problems, some of
which have been identified as risk factors for suicide, but which are not likely to
actually affect suicide. This index is the sum of five indicators variables reflecting the
presence of bans on persons with mental health, alcohol, or drug problems, as well as
prohibitions on those with prior convictions for misdemeanors and domestic violence
offenses.
Our third and last index captures four types of prohibitions reflecting criminal
concerns. We include this variable primarily as a robustness check, since the
prohibitions captured are least likely to affect suicide. The index, varying between 0
and 4, is the sum of indicator variables measuring the presence of prohibitions against
“aliens”4, convicted felons, fugitives from justice, and those who committed serious
offenses as juveniles.
Gun ownership
Given the relationship between firearm regulations and firearm prevalence, as well as
that between firearm prevalence and suicide, it is necessary to control for gun
ownership in the alternative specification. Gun ownership is available every several
years from the CDC’s Behavioral Risk Factor Surveillance System, but there is no
annual data at the state level, and the available data only dates back to 2001. The most
commonly used proxies for gun ownership are the proportion of homicides and the
proportion of suicides committed with firearms (see, Lester, 1987; Lester, 1989;
Hemenway et al., 2001; Miller et al., 2002; Shenassa et al., 2006; Azrael et al., 2004).
These variables are combined to create an index called Cook’s index. However, given
that the dependent variable for this analysis is the total number of suicides, it was felt
that this proxy was inappropriate. As an alternative, the number of hunting licenses
per capita from the Fish and Wild Life Service5 was used as a control for gun
4
In some states, this prohibition refers to undocumented immigrants, while in others to individuals
who have “forsaken their allegiance to the United States.”
5
www.fws.gov
11
ownership6. Hunting licenses per capita and firearm suicides as a proportion of
suicides were highly correlated (r =0.74, p-value <.05). Data on state gun regulations
was obtained from the Bureau of Justice Statistics.7
Data on state personal income (income) were obtained from the Bureau of Economic
Analysis and deflated by the consumer price index (CPI) extracted from the Bureau
of Labor Statistics (BLS). Unemployment rates (unemployment) also come from the
BLS. Data on per capita ethanol consumption of beer (beer), an estimate for the
amount of pure ethanol consumption per capita, was extracted from the NIIA
Surveillance Reports. Alcohol consumption and economic conditions have been
linked to suicide in a number of population level studies (Yang, 1992; Markowitz et
al., 2003) The percentage of people over 65 (psh65) years of age and the proportion
of the population which is non-Hispanic white (white) were obtained from the US
Census Bureau (2007). Table 2 reports summary statistics for the variables used in
regressions.
Table 2. Summary statistics
Variables
Dependent variables:
Male suicides, total
Male suicide, ages 15-24
Male suicide, ages 25-44
Male suicide, ages 45-64
Male suicide, ages > 65
Socioeconomic variables:
Percent with bachelor's degree
Real per capita income (log)
Unemployment rate
Beer consumption per capita
Percent non white
Percent 65 years or older
Gun supply:
Hunting licenses per capita
Firearm regulation:
General prohibitions(1)
Permit requirements
Ban on purchase by minors
General prohibitions index
N
Mean
Std. Dev.
Min
Max
500
500
500
500
500
484.09
70.95
191.06
134.20
93.36
476.87
62.67
181.23
134.80
102.79
49
3
18
7
1
500
500
500
500
500
500
24.65
10.20
4.81
1.27
0.78
0.14
4.67
0.15
1.17
0.20
0.15
0.11
12.70
9.82
2.20
0.73
0.26
0.05
38.70
10.65 Con formato: Italiano (Italia)
8.20
1.91
0.99
1.34
500
0.087
0.071
0.007
0.340
500
500
500
0.22
0.68
0.90
0.41
0.47
0.63
0
0
0
1
1
2
2939
421
1191
831
575
6
The model was also estimated using the firearm suicide proxy, and results were very similar. We do
not report them here for brevity.
7
http://www.ojp.usdoj.gov/bjs/
12
Behavioral prohibitions(2)
Mental health problem
Alcohol problem
Drug problem
Misdemeanor conviction
Domestic violence conviction
Behavioral prohibitions index
Criminal prohibitions(3)
Alien
Felony
Juvenile offense
Fugitive
Criminal prohibitions index
500
500
500
500
500
500
0.47
0.34
0.41
0.36
0.29
1.87
0.50
0.47
0.49
0.48
0.45
1.70
0
0
0
0
0
0
1
1
1
1
1
5
500
500
500
500
500
0.15
0.73
0.41
0.17
1.46
0.36
0.45
0.49
0.38
1.11
0
0
0
0
0
1
1
1
1
4
3. RESULTS
The regression results are reported in Table 3. The estimations include year and
individual state specific effects which are statistically significant in all regression
models. All coefficients have been put on an exponential scale; thus the parameters
(β) obtained from the negative binomial regression should be interpreted as incidence
rate ratios (henceforth, IRR). The IRR are obtained by exponentiation of the
regression coefficients, that is, exp (β). The expression 100*(exp (β)-1) is the
percentage change in the incidence or risk of suicide mortality for each unit increase
in the independent variable.
Table 3: Results of Negative Binomial Regression, all males, 1995-2004 (N=500).
Dependent variable is number of male suicides, exposure variable is male population
Model 1
State Fixed Effects
Year Fixed Effects
High School
graduates (%)
Beer consumption
.
Unemployment rate
Model 2
Model 3
Model 4
Model 5
Model 6
Behavioral
prohibitions
Criminal
prohibitions
X
X
X
X
Fixed effects
only
Socioeconomic
variables
General
Prohibitions
General
prohibitions
and hunting
licenses per
capita
X
X
X
X
X
X
X
X
0.9985
0.9991
0.9991
0.9986
0.9984
[0.0017]
[0.0017]
[0.0017]
[0.0016]
[0.0017]
1.1067
1.1022
1.1021
1.1184
1.0988
[0.1062]
[0.1039]
[0.1037]
[0.1012]
[0.1011]
1.0181
[0.0075]
**
1.0185
[0.0076]
**
1.0185
[0.0075]
**
1.0165
[0.0074]
13
**
1.0179
[0.0073]
Log of median HH
income
0.6394
[0.1424]
3.5333
[1.5326]
1.1245
[0.0147]
Percent White
Percent > 65 years
General prohibitions
(1)
**
***
***
0.6433
[0.1418]
3.5115
[1.5247]
1.1232
[0.0153]
**
0.9440
[0.0093]
***
***
***
0.6428
[0.1429]
3.5152
[1.5171]
1.1232
[0.0153]
**
0.9438
[0.0099]
***
Behavioral
prohibitions index (2)
***
***
0.6376
[0.1338]
3.3638
[0.0145]
1.1166
[0.0145]
**
0.9946
[0.0030]
*
Criminal prohibitions
index (3)
***
1.0035
[0.0068]
Hunting Licenses Per
Capita
Log likelihood
Ln alpha
***
0.6375
[0.1435]
3.5250
[1.5327]
1.1261
[0.0157]
-2228.3
-7.0532
-2199.2
-7.6410
-7.6475
0.9692
[0.4548]
-2199.1
-7.6477
-2199.1
-7.6692
-2200.3
-7.6597
Notes:
(1) General prohibitions index: permit requirement, ban on purchase by minor.
Range 0-2
(2) Behavioral prohibitions index: mental health, alcohol problems (or intoxication), drug problems, domestic violence conviction, misdemeanor conviction.
Range: 0-5
(3) Criminal prohibitions index: alien, prior felony conviction, fugitive from justice, serious offense as a juvenile. Range:0-4
District of Columbia is excluded.
(4) Standard errors in brackets. Constant term included but not reported.
*=p<.10
** p<.05
***p<.01
The first two models show the effects of fixed state and year effects (Model 1), and
the fixed effects in addition to a set of socio-demographic variables namely
education, income, alcohol consumption, the proportion of the population over age
65, and the proportion of non-Hispanic white population (Model 2).
Model 3 introduces the index of general prohibitions – namely gun control
regulations which affect the largest number of people and which create general
barriers to entry. We find the general prohibition index to be statistically significant,
both individually and when we include our proxy for gun prevalence, hunting licenses
per capita, which enters insignificantly (Model 4).
The next model includes the second index of gun control measures, which aims to
capture firearm restrictions based on, behavioral issues such as a history of mental
14
health or alcohol/drug problems. While significant, the IRR is 0.9946 in this model
compared to 0.9440 in Model 3 that includes the general prohibition index; and the
coefficient is only significant at the 10 percent level. Model 6 includes our last index,
which captures gun control measures that are least unlikely to affect suicide. They are
designed rather to prevent criminal behavior. As expected, this variable does not enter
with a statistically significant coefficient.
Table 4 shows the effects of individual state level firearm restrictions on the
phenomenon of the suicide for particular age groups. In all models, we control for
gun prevalence with the number of hunting licenses per capita, and include the
standard covariates. We find that gun control measures do not affect all age groups
identically. For instance, a ban on firearm purchases by minors affects suicides
particularly among younger males, while restrictions on permits and waiting period
requirements have a more deterrent effect on for older males. Among behavior related
prohibitions that based on a history of mental illness prohibition is only significant for
males aged 25-44 years, and that against alcohol abusers only matters for 65 + males.
The drug and misdemeanor conviction bans do not enter significantly for any of the
age groups, and the prohibition linked to a history of domestic violence only affects
suicides among the middle aged. None of the criminal prohibitions enter significantly
for specific age groups, and are therefore omitted.
Table 4: Selected gun control measures, males by age, 19952004 (N=500).
Dependent variable is number of male suicides, exposure variable is male population, within age groups
General prohibitions
Permit requirement
.
Ban on minor purchase
Behavioral prohibitions
Mental health
.
Alcohol
.
Model 1
Model 2
Model 3
Model 4
Males 1524 years
Males 2544 years
Males 4564 years
Males 65+
0.9741
[0.0188]
0.8647
[0.0183]
0.8601
[0.0208]
0.9867
[0.0817]
1.2043
[0.0343]
0.8715
[0.0679]
0.9949
[0.0245]
1.0015
[0.0199]
***
*
0.9657
[0.0111]
1.0085
[0.0168]
***
***
***
0.9948
[0.0212]
0.9916
[0.0196]
0.8518
[0.0222]
1.0304
[0.0756]
1.0435
[0.0246]
0.9437
[0.0166]
15
***
***
Drugs
.
Misdemeanor conviction
Domestic violence conviction
0.9723
[0.0229]
1.0169
[0.0216]
0.9812
[0.0261]
0.9972
[0.0167]
0.9848
[0.0159]
1.0048
[0.0190]
1.0017
[0.0220]
0.9758
[0.0160]
0.9630
[0.0167]
**
Note: All models include state and year
fixed effects, as well as control variables
for the level of education, unemployment
rate, income per capita, and the percent of
Non-Hispanic white population. A proxy
for gun prevalence (hunting licenses per
capita) is also included. District of
Columbia is excluded. Standard errors in
brackets. Constant term included but not
reported.
*=p<.10
** p<.05
***p<.01
4. DISCUSSION
Access to lethal weapons is an important risk factor for suicide. Our study suggests
that general barriers to firearm access created through state regulation can have a
significant deterrent effect on male suicide rates in the United States. Permit
requirements and bans on sales to minors were the most effective of the regulations
analyzed. These findings have important implications for U.S. gun control policy,
which remains exceptionally heterogeneous across states. While all states except
Wyoming have banned sales of handguns to minors, twelve states still allow the sale
of long guns to minors. Furthermore, only twelve states currently require purchase
permits for firearms. While gun control remains a controversial issue both at the state
and federal level in the U.S., this analysis suggests that there are clear public health
benefits to restricting access to firearms through regulation.
6. References
Azrael D, Cook P, Miller M. 2004. State and local prevalence of firearms ownership
measurement, structure and trends. Journal of Quantitative Criminology 20: 43-62.
Beautrais A, Fergusson D, Horwood L. 2006. Firearms legislation and reductions in
firearm-related suicide deaths in New Zealand. Australian and New Zealand Journal
of Psychiatry 40: 253-259.
Cameron C, Trivedi P. 1998. Regression Analysis of Count Data. Econometric
Society Monograph No. 30, Cambridge University Press.
16
1.0086
[0.0241]
1.0062
[0.0293]
0.9700
[0.0185]
Centers for Disease Control and Prevention (CDC), National Center for Injury
Prevention
and
Control,
2009.
http://www.cdc.gov/ncipc/dvp/suicide/suicide_data_sheet.pdf [2 February 2009].
Cheung A, Dewa C. 2005. Current trends in youth suicide and firearm regulations.
Canadian Journal of Public Health 96: 131-135.
Clarke R, Jones P 1989. Suicide and increased availability of handguns in the United
States. Social Science and Medicine 28: 805-809.
Conner K, Zhong Y. 2003. State firearm laws and rates of suicide in men and women.
American Journal of Preventive Medicine 25: 320-324.
Cook P, Ludwig J. 1998. Defensive gun uses: New evidence from a National Survey.
Journal of Quantitative Criminology 14: 111-131.
Cook P, Moore M, Braga A. 2001.Gun Control, Working Paper Series
SANO1-14, Sanford Institute of Public Policy, Durham, NC.
Cook P, Ludwig J. 2006. Aiming for evidence based on gun policy. Journal of Policy
Analysis and Management 25: 691-735.
Cook P, Ludwig J. 2006. The social costs of gun ownership. Journal of Public
Economics 90: 379-391.
Cutler D, Glaeser E, Norberg K. 2001. Explaining the Rise in Youth Suicide, in
Gruber J, ed., Risky Behavior Among Youths: An Economic Analysis, University of
Chicago Press: Chicago.
Duggan M. 2003. Guns and Suicide: Correlation or Causation? In, in Evaluating Gun
Policy, Cook P, Ludwig J, eds., Brookings Institution Press: Washington.
Gallup
Poll,
2009.
Do
you
have
a
gun
in
http://www.gallup.com/poll/1645/Guns.aspx [5 November 2009]
your
home?
Grossman D, Mueller B, Riedy C, Dowd M, Villaveces A, Prodzinski J, Nakagawara
J, Howard J, Thiersch N, Harruf R. 2005. Gun storage practices and risk of youth
suicide and intentional firearm injuries. Journal of American Medical Association
293: 707-714.
Haw C, Sutton L, Simkin S, Gunnell D, Kapur N, Nowers M, Hawton K. 2004.
Suicide by gunshot in the United Kingdom: A review of the literature. Medicine,
Science and the Law 44: 295-310.
Hawton K, Fagg J, Simkin S, Harris L, Malmberg A. 1998. Methods used for suicide
by farmers in England and Wales. The contribution of availability and its relevance to
prevention. British Journal of Psychiatry 173: 320-324.
17
Hemenway D, Kennedy B, Kawachi R. 2001. Firearm prevalence and social capital.
Annals of Epidemiology 11: 484-490.
Kaplan M, Geling O, 1998. Firearm suicides and homicides in the United States:
Regional variations and patterns of gun ownership. Social Science and Medicine 46:
1127-1233.
Kapusta N, Etzersdorfer E, Krall C, Sonneck G. 2007. Firearm legislation reform in
the European Union: Impact on firearm availability, firearm suicide and homicide
rates in Austria. British Journal of Psychiatry 191: 253-257.
Kennedy P. 1998. A Guide to Econometrics. Blackwell: Oxford.
Killias M, van Kesteren J, Rindlisbacher M, 2001. Guns, violent crime and suicide in
21 countries. Canadian Journal of Criminology, October 2001: 429-448.
Kleck G, Kates D. 2001. Armed: New Perspectives on Gun Control.
Prometeus:Amherst.
Kleck G. 2004. Measures of gun ownership for macro level crime and violence
research. Journal of Research in Crime and Delinquency 41: 3-36.
Klieve H, Barnes M, De Leo D. 2009. Controlling firearms use in Australia: has the
1996 gun law reform produced the decrease in rates of suicide with this method?
Social Psychiatry and Psychiatric Epidemiology 44: 285-292.
Lambert M, Silva P. 1998. An update of the impact of gun control legislation on
suicide. Psychiatric Quarterly 69: 127-134.
Lester D. 1987. Availability of guns and the likelihood of suicide. Sociology and
Social Research 71: 287-288.
Lester D.1989. Gun ownership and suicide in the United States. Psychological
Medicine 19: 519-521.
Lester D. 1992. Why people kill themselves. Charles Thomas: Springfield.
Lester D, Clarke R. 1991. Note on suicide and increased availability of handgun in
the United States: The influence of firearm ownership on accidental deaths. Social
Science and Medicine 32: 1311-1313.
Lott J, Whitley J. 2000. Safe storage, gun laws: accidental deaths, suicides, and crime.
Working Paper # 237 Yale Law School.
Markowitz S, Chatterji P, Kaestner R. 2003. Estimating the impact of alcohol policies
on youth suicide. Journal of Mental Health Policy and Economics 6: 37-46.
Miller M, Azrael D, Hemenway D. 2001. Firearm availability and unintentional
firearm deaths. Accident Analysis and Prevention 33: 477-484.
18
Miller M, Azrael D, Hemenway D. 2002. Rates of household firearm ownership and
homicide across US Regions and States, 1988-1997. American Journal of Public
Health 92:1988-1993.
Miller M, Hemenway D, Azrael D. 2004. Firearms and suicide in the northeast. The
Journal of Trauma, Injury, Infection and Critical Care 57: 626-632.
Miller M, Azrael D, Hepburn L, Hemenway D, Lippman S. 2005. The association
between changes in household firearm ownership and rates of suicide in the United
States, 1981-2002. Injury Prevention 12: 178-182.
Miller M, Lippman S, Azrael D, Hemenway D. 2007. Household firearm ownership
and rates of suicide across the 50 United States. The Journal of Trauma, Injury,
Infection and Critical Care 62: 1029-1035.
Miller M, Hemenway D. 2008. Guns and suicide in the United States. The New
England Journal of Medicine 4: 989-991.
Open Society Institute, 2000. Gun control in the United States: A comparative survey
of
State
firearm
laws.
http://www.soros.org/initiatives/usprograms/focus/justice/articles_publications/public
ations/gun_report_20000401 [30 November 2009]
Ozanne-Smith J, Ashby K, Newstead S, Stathakis V, Clapperton A. 2004. Firearm
related deaths: The impact of regulatory reform. Injury Prevention 10: 280-286.
Price J, Thompson A, Dake J. 2004. Factors associated with state variations in
homicide, suicide, and unintentional firearm deaths. Journal of Community Health
29: 271-283.
Rosengart M, Cummings P, Nathens A, Heagerty P, Maier R, Rivara F. 2005. An
evaluation of state firearm regulations and homicide and suicide death rates. Injury
Prevention 11: 77-83.
Shennasa E, Dasklakis C, Buka S. 2006. Utility of indices of gun availability in the
community. Journal of Epidemiology and Community Health 60: 44-49.
Sumner S,Layde P, Guse C. 2008. Firearm death rates and association with level of
firearm purchase background check. American Journal of Preventive Medicine 35:16.
Wahlbeck K, Mäkinen M. (Eds). 2008. Prevention of Depression and Suicide.
Consensus
paper.
Luxembourg:
European
Communities.
http://ec.europa.eu/health/ph_determinants/life_style/mental/docs/Cyprus.pdf
[30
November 2009]
Yang B. 1992. The economy and suicide: A time-series study of the USA. American
Journal of Economics and Sociology 51: 87-9.
19