Social Psychiatry and Psychiatric Epidemiology

, Volume 48, Issue 9, pp 1467–1479 | Cite as

Association of suicide rates, gun ownership, conservatism and individual suicide risk

Original Paper

Abstract

Objectives

The purpose of the study was to examine the association of suicide rates, firearm ownership, political conservatism, religious integration at the state level, and individual suicide risk. Social structural and social learning and social integration theories were theoretical frameworks employed. It was hypothesized that higher suicide rates, higher state firearm availability, and state conservatism elevate individual suicide risk.

Method

Data were pooled from the Multiple Cause of Death Files. Multilevel logistic regression models were fitted to all deaths occurring in 2000 through 2004 by suicide.

Results

The state suicide rate significantly elevated individual suicide risk (AOR = 1.042, CI = 1.037, 1.046). Firearm availability at the state level was associated with significantly higher odds of individual suicide (AOR = 1.004, CI = 1.003, 1.006). State political conservatism elevated the odds of individual suicides (AOR = 1.005, CI = 1.003, 1.007), while church membership at the state level reduced individual odds of suicide (AOR = 0.995, CI = 0.993, 0.996). The results held even after controlling for socioeconomic and demographic variables at the individual level.

Conclusion

It was concluded that the observed association between individual suicide odds and national suicide rates, and firearm ownership cannot be discounted. Future research ought to focus on integrating individual level data and contextual variables when testing for the impact of firearm ownership. Support was found for social learning and social integration theories.

Keywords

Suicides Firearm availability State suicide rate Conservatism Multilevel models 

Introduction

According to recent mortality data, in 2007 suicide was the 11th leading cause of death in the United States for all ages combined; it was the 7th leading cause of death for males, and the 15th leading cause of death for females (Centers for Disease Control and Prevention) [15]. More specifically, over 34,000 persons died of suicide, a figure that translates into incidents of 94 deaths per day, 1 death every 15 min, and nearly 11.3 suicides per 100,000 population [34]. Although firearms remain the most commonly used method of suicide among males and poisoning the most common among females, some epidemiologic studies show a growing tendency for females to use firearms in some jurisdictions [14, 39].

Between 1999 and 2007, there were 152,030 firearm suicides in the United States for all persons age 10 and above [15], representing an annual average rate of nearly 6.8 per 100,000 population. In the 8 year period, over 50 % of all suicides were gun related. In the same period, regional disparities in firearm suicides were also observed, with 15,854 (rate = 3.72) in the Northeast, 67,341 (rate = 8.31) in the South, 32,051 (rate = 6.3) in the Midwest, and 36,784 (rate = 7.23) in the West [15].

Numerous studies have been done to examine the relationship between gun availability and suicide, with many case-control designs finding a consistent relationship between presence of firearms in the home and elevated risk of completed suicides and parasuicides not only for the gun owner, but other members in the household [6, 8, 11, 46, 47, 52, 81]. Kellermann et al. [35] reported that ready availability of firearms was associated with an elevated suicide risk in the home. Kaplan et al. [32] observed that firearms were often used in male (58.1 %) and female (21.2 %) suicides. Another report [2] contends that the United States leads most countries in the developed world with regard to the percentage of suicides committed by firearms.

Although much has been written on the association of firearms and suicide, there are severe shortcomings in the literature. Almost all studies to date on completed suicides have been limited to specific geographic areas within the nation. For instance, the much cited Kellermann et al. [35] study on gun ownership and suicide was based on two counties, one in Washington State and the other in Tennessee. Callanan and Davis [14] used only one Ohio County to investigate the association of gender and suicide methods. Similarly, Kposowa and McElvain [39] employed only one southern California County to analyze the impact of gender and immigration on suicide methods. Likewise, although Kaplan et al. [32] study on firearms and suicide made use of the newly available National Violent Death Reporting System, the data set is not representative of the entire US population as only 17 states are covered. Shenassa et al. [67] investigation of firearm lethality relative to other methods was limited to Illinois. Limiting analysis to specific geographic areas could lead to potentially inexplicable findings. For example, in the Illinois based study, poisoning was the most common method of suicide among both men (59 %) and women (89 %), a finding that contradicts all previous US national mortality data. Another shortcoming in past research on firearms and suicide is that investigators have ignored contextual variables as contributing factors. Accordingly, most studies rely on individuals as the unit of analysis, and multilevel models are rare or non-existent. Yet, it is widely acknowledged in sociological science that the community in which people live can impact individual behavior [41, 71, 72, 84]. The present study extends previous research by employing a nationally representative sample of all 50 states and the District of Columbia. Second, the study employs multilevel modeling to investigate the influence of community (state) variables on individual suicide risk. Specifically the following questions are addressed: Does household firearm availability at the state level influence individual risk of suicide? Do state suicide rates influence individual suicidal behavior? Is there a link between state conservatism and individual suicide risk? Does religious integration at the state level affect suicide after controlling for family integration at the individual level?

Past research

The explanatory model: social structural and social learning theory

Social structural and social learning theory was originally proposed to explain deviance, and it contains differential association theory [76]. Sutherland [76] argued that an individual engages in crime or deviant behavior when there is an excess of definitions favorable to law violation over definitions unfavorable to law breaking. Given that in the present paper, suicide is not considered a criminal act, it is proposed that that suicide occurs when an individual lives in states where there is an excess of definitions favorable to suicide over definitions less favorable to committing suicide. By definitions, Sutherland [76] meant individual attitudes, meanings, or understandings that are attached to some behavior. Accordingly, the longer a person holds attitudes that proscribe deviant acts, the less likely that person will engage in them [4]. If an individual believes for example, that suicide is wrong, the more enduring this disposition, the lower the likelihood that he or she will commit suicide.

Differential association theory has not been employed often in the study of suicidality in sociology, given that it was originally developed for explaining crime. However, Akers [3] developed a much broader theory of deviance that combines key tenets of differential association and social learning theories: social structural and social learning theory. According to the theory, socio-structural factors can be expected to have indirect effects on individual behavior [4]. This is because contextual variables influence social learning variables that are inherent in differential association, definitions favorable or unfavorable to deviant behavior, along with imitation, all of which have direct consequences on a person’s conduct. As explained by Akers [3].

The social structural variables are indicators of the primary distal macro-level and meso-level causes of crime, while the social learning variables reflect the primary proximate causes of criminal behavior that mediate the relationship between social structure and crime rates…

Past studies have reported that suicide attempters have generally had more contact with role models that have committed suicide [7]. As Stack and Kposowa [73] observe, in communities with more persons that have died of suicide there could be greater interaction with individuals that have attitudes favorable to suicide. This is consistent with social learning theory. The present report recognizes the possibility of the complex relationship between macro structural factors and micro characteristics by modeling individuals as located in their states of residence, taking into account that some of the individual risk factors may be due to state differences.

The social structural and social learning theory has direct bearing on suicide research because it makes provision for integrating both individual and community (contextual) characteristics in explaining suicidal behavior. The social structural perspective suggests that while personal attributes (including depression, mental illness, marital status, alcoholism, etc.) may be proximate, structural conditions (though distal) influence individual suicide risk. This line of argument is consistent with recent alarms raised by Wray et al. [84] that “suicide continues to be framed and understood as a problem faced by individuals even when social and contextual factors are acknowledged.” With few exceptions [41, 50, 57, 71, 72, 79], sociologists have rarely combined individual and ecological variables in modeling suicidal behavior. Indeed, many have abandoned the study of suicide to psychiatrists, psychologists and epidemiologists. The latter initially placed heavy emphasis on biomedical models that emphasized individual risk factors in suicide causation, but they have in more recent decades sought to bring a more balanced approach that borrows heavily from sociological theories, concepts and approaches which emphasize context and societal characteristics [41, 84]. Based on social structural and social learning theory, the following hypotheses are tested in the present paper.H1: The higher the state suicide rate, the higher the individual suicide risk.H2: The higher the level of state firearm ownership the higher the individual suicide risk.H3: The higher the level of state conservatism, the higher the individual suicide risk.

Explanatory model: social integration theory

The social integration explanation of suicide has perhaps the longest tradition in sociology. Although there is debate as to whether the theory originated from Durkheim’s classic work le Suicide [23], there is ample evidence that he formalized the propositions and even hypotheses of the theory [42]. He argued that suicide rates tend to be high among social groups or in communities that show low levels of domestic integration, pointing out that “Suicide varies inversely with the degree of integration of the social groups of which the individual forms a part…. As collective force is one of the obstacles best calculated to restrain suicide, its weakening involves a development of suicide” ([23], p. 246). It appears that by referring to groups and individual, Durkheim introduced the notion of multilevel modeling nearly a century before the technique was applied to his theory. Given the data that he utilized as well as statistical limitations of the time, he perhaps did not test his theory in a much fuller manner. Social integration itself referred to the quantity and quality of ties that bind individuals to others, to groups, to their community, and to the broader society [9, 42, 59]. The social integration thesis is tested using three measures: marital status at the individual level, church membership and immigration at the state level. It is expected that the divorced, single, and widowed persons exhibit higher odds of suicide than the married. While divorce is a marker for family disintegration, Durkheim also argued that various religious denominations show different suicide rates. Denominational differences are subsumed here under religious integration. Church membership increases ties or bonds to groups (beginning with other church members), and to the broader community, and reduces suicide risk. At the macro level, it is conceivable that higher church membership in a state promotes greater community integration and thereby reduces suicide rates [9, 10]. It is hypothesized that the higher the church membership rate in a state, the lower the odds of individual suicides. The third measure of social integration is immigration. Higher immigration levels in a community might lead to greater anonymity stemming from faster population turnover [40]. Anonymity and rapid population change may reduce the intensity of community bonds, leaving individuals more vulnerable to suicide [25]. It is thus expected that the higher the state immigration rate, the higher the odds of individual suicide.

Explanatory model: political conservatism

In advanced societies, especially in the United States, individuals are apt to adopt political views that may be considered conservative, liberal, or middle of the road. Political parties in the US are also generally viewed as liberal (left leaning) and conservative (right leaning). Unfortunately, political ideology has not been used much in the study of suicide. It has been suggested, however, that individuals with generalized political liberalism are more likely to approve of suicide [73]. There is not much in the literature on the association between political ideology and completed suicides. An Australian study [58] reports that during conservative administrations the risk of suicide is elevated for both men and women. The pathway by which political ideology might influence suicide is unclear. It is plausible that during conservative administrations, there is a tendency to liberalize gun laws at state and even Federal levels, which might potentially increase suicide risk given increased gun purchases and availability, while during liberal administrations, gun laws may be tightened, an action that may reduce the number of firearms available. Some political parties might also emphasize different values, such as individualism, while others may stress the need for togetherness. The two studies cited above have produced contradictory findings with regard to the association between political ideology and suicide. While Kposowa and Stack [73] found that liberalism was significantly associated with suicide acceptability, Page et al. [58] observed that conservatism elevated suicide rates. The two studies are, however, not equivalent, since one concentrated on suicide approval, while the other focused on completed suicides. It should also be noted that the use of political conservatism in this study has nothing to do with electoral outcomes or war. Some research suggests that election outcomes may lead to despair, which in turn increases suicide risk for supporters of the defeated candidate [17]. Defeat at the ballot box may, in other words, have the effect of dividing a nation, generating political disintegration, and thereby leading to increase in suicide rates. This line of argument is consistent with Durkheim’s [23] contention that suicide rates decline in countries that are involved in wars because there is increase in political integration during conflict, a phenomenon that brings people together, binding them to their communities, to their nation, and increasing solidarity and nationalism. With elections, it is conceivable that people who expect to be excluded from society due to the loss of their candidate or party may experience psychological distress and suicidal tendencies. This stream of research has, however, produced inconsistent results [80]. As stated earlier, however, our position in this paper is not about war or electoral outcomes, but simply the degree of conservatism prevailing in a state. It is hypothesized that the more conservative a state the higher its individual suicide risk.

Methods

Analytical strategy

In multilevel data with a categorical response variable, it is desirable to use a hierarchical generalized linear modeling technique that incorporates a unique random effect into the equation for each higher level observation. This approach produces more robust standard errors than those produced via techniques that do not allow nested models [69]. In contextual analysis, dependence may arise because individuals within a given level 2 unit for example, the state, have shared experiences and expectations. Persons within a state may have more in common with their fellow state citizens than with individuals residing in other states. There may therefore be variations in outcome across states as well as variations among individuals within states. In multilevel models with binary outcomes, therefore, calculated standard errors must adjust for the variability across levels [62]. In the present analysis, the GLIMMIX procedure in SAS 9.3 [64] was used to estimate parameters.

Multilevel techniques have appeared only recently in suicide research, including that by Van Tubergen et al. [79], Stack and Kposowa [73], and Kposowa and D’Auria [41]. Tubergen et al. [79] and Kposowa and D’Auria [41] examined completed suicides, while Stack and Kposowa [73] studied suicide acceptability at the cross-national level. The mixed modeling approach recognizes the fact that most individual events do not occur in a vacuum, but are located in time and space.

Data

The data used in the study were derived from the national US Multiple Cause of Death Files (MCDF) for 2000 through 2004 [54]. The MCDF contains mortality statistics based on information from death certificate records of all deaths occurring in the United States. The information is received on computer data tapes coded by the States and provided to the National Center for Health Statistics (NCHS), through the Vital Statistics Cooperative Program, or are coded by NCHS staff from copies of the original certificates received from registration areas. NCHS receives the data for the MCD Files from the registration offices of all states, the District of Columbia, and New York City. Mortality data for the United States are limited to deaths occurring within the country to US residents and non-residents. Deaths occurring to US citizens outside the United States are not included in the MCDF [54]. A more detailed description of the MCDF has been presented elsewhere [54].

The current analysis focused on suicide deaths among non-Hispanic whites, non-Hispanic African Americans, non-Hispanic Native Americans (American Indians), non-Hispanic Asians, and Hispanics aged 18 years and above. The results presented are based on those adults 18 years old and above who died of suicide in their state of residence in the years 2000 (24, 279), 2001 (25, 315), 2002 (28, 944), 2003 (26, 202) and 2004 (26, 896).

Dependent variable

The MCD File is cross-sectional, and lacks population denominators. Therefore, an unmatched case–control design was employed in this study. The dependent variable was the odds of death by suicide (the cases), as opposed to deaths from natural causes (the controls). The data used do not permit a comparison of victims versus non-victims (those still alive), so the ‘non-victims’ in the analysis are those who died of natural causes. Deaths resulting from intentional self harm (suicides) were defined according to the Tenth Revision of the International Classification of Diseases [ICD-10] [83], with the cause codes X64 to X84, Y87.0 and US Clinical Modification code U03 [54]. Code U03 refers to suicides stemming from acts of terrorism, a category created by NCHS in response to the events of 11 September 2001 [55]. All suicides to persons age 18 and above (cases) were coded 1. Persons dying of natural causes (after eliminating homicides and accidents) were coded 0 and used as controls. Like the suicides, analysis was limited to individuals dying in their current state of residence. Due to serious technical difficulties encountered (computer memory problems, convergence problems, etc.) in analyzing all deaths from natural causes (nearly 10 million) a 20 % simple random sample was selected for each year studied. The deaths for all 5 years were then pooled into one concatenated file. The resulting deaths from natural causes per year (after sampling) were as follows: 369,023 (in 2000), 370,628 (in 2001), 429,099 (in 2002), 377,029 (in 2003), and 368,504 (in 2004). In all, there were 131,636 suicides, and 1,914,283 sampled deaths from natural sources. Thus, the effective sample size at the individual level comprised 2,045,919 cases.

Independent variables

State level variables (Level 2)

State suicide rate its inclusion was based on the reasoning that high suicides in a state increase not only awareness of suicide, but attitudes favorable to suicidal behavior, a position consistent with social learning theory. As Stack and Kposowa [73] explain, “In a community with a high suicide rate, there are proportionately more persons committing suicide and, hence, more cases that offer excuses for suicide.” It is also a marker for Sutherland’s [76] idea of definitions favorable to deviance. A high number of suicides in a state could arouse citizens’ awareness especially through the media. Over time, suicide may no longer be a phenomenon that is stigmatized, but may become an acceptable way to exit life. The variable was measured as the number of suicides per 100,000 resident population in 2000, and it was obtained from Health United States 2000 [56].

Firearm availability was measured as the percentage of households in each state that kept firearms at home. Firearms include shotguns, rifles, handguns, pistols, and semi-automatic weapons. The variable was obtained from the 2001 Behavioral Risk Factor Surveillance System (BRFSS), a nationally representative sample of the US population [16]. In the 2001 survey, respondents (n = 201,881) were asked the following question: “Are any firearms now kept in or around your home? Include those kept in a garage, outdoor storage area, car, truck, or other motor vehicle”. Responses to the question were tabulated for each state and the District of Columbia.

Conservatism the third level 2 variable was state conservatism which was measured as the percentage of votes cast for Republican candidate George W. Bush in the 2000 Presidential Election. The rationale is that the Republican Party is generally viewed as a conservative party with right leaning ideology. Thus, states in which plurality of votes are cast for a perceived conservative candidate are more likely to be conservative than states in which a higher percentage of votes is cast for a perceived liberal candidate. The variable was obtained from the Statistical Abstract of the United States [78].

Church Membership was a macro-level variable included to capture Durkheim’s idea of religious integration [9, 65, 66]. It was measured as the percent of the state population that were church adherents in 2000. The variable was obtained from the Statistical Abstract of the United States [78].

Immigration was a macro-level variable designed to measure anonymity, population turnover, and reduced community integration. It was measured as the percentage of a given state’s population that was foreign born in 2000. The variable was obtained from the Statistical Abstract of the United States [78].

Individual level variables

Marital status has figured prominently in many studies as a measure of individual integration, and it has been observed as a significant predictor of suicide and parasuicide [19, 22, 25, 29, 43]. The covariate was measured by a set of dummy variables, with 1 each for single, widowed, separated or divorced. Those married at the time of death constituted the reference category. There were 3,284 persons with missing marital status information across all sampled years. Sensitivity analysis was first performed with the missing category (MARMIS) coded 1. Given that the covariate category was not statistically significant, it was judged that absence of these 3,284 cases did not bias results, so they were deleted from the analysis.

Control variables

Owing to the fact that the data were pooled from deaths occurring in five different years, period effects might be present. It is conceivable that any single year is atypical and that the number of suicides may be affected by extraneous national and even international events. To control for period effects as well as possible trends in the data, analysis employed year of death as an independent variable. The variable was measured as a series of dummy variables. There was 1 each for 2001, 2002, 2003, and 2004. The year 2000 was selected as the reference category due to the fact that in the years following 2000 there were dramatic events taking place nationally and globally, including the attacks on New York City and Washington DC on 11 September 2001, and subsequent invasions of Afghanistan and Iraq by the United States. It is unknown how events of the post 2000 years affected the number of suicides within the country.

Studies in the US consistently show that there are substantial disparities in rates and risks of suicide by race/ethnicity [18, 31, 44, 48, 71, 72]. Racial differences in suicide have also been reported in South Africa, with whites experiencing much higher rates than Blacks, Coloureds or Asians in major South African cities [12]. The present report, therefore, controlled for race/ethnicity. The variable comprised non-Hispanic African Americans, non-Hispanic Native Americans, non-Hispanic Asians, and Hispanics. Each of these groups was coded 1, and non-Hispanic Whites were used as the reference category. Age at death was captured by leaving it in its interval form. For sex, females were the omitted group. Both age and sex were included as demographic control variables. The MCD File lacks many socioeconomic variables. Education, the only one collected in all states was used. The covariate was left in its format as highest grade completed. The data contained a large number of missing cases (43,697). Initial sensitivity analysis relied on calculating a dummy variable (EDUCMIS) with 1 for missing cases on education, and 0 for all others. The dummy variable was found to be statistically significant, suggesting that deleting these cases could lead to sample selection bias. The next task was to determine through Univariate procedures in SAS [65] whether the cases were missing at random or not. Results showed that cases on education were missing at random so the mean value (high school education) was substituted for persons with missing information. City size of residence was left in its ordinal format. Inclusion of city size is important for reassessing findings in some recent studies that rural areas have significantly higher suicide rates than urban areas [33, 61]. Place of residence referred to the decedent’s usual place of residence at the time of death. The variable was measured as 1 if the decedent resided in the central city of an MSA (metropolitan area), and 0 if otherwise.

Results

Descriptive findings

Table 1 presents data on 20 states with the highest suicide rates in the United States in 2000. Corresponding values on percent of households with firearms, percent of votes cast for George W. Bush in the 2000 election (used to index state conservatism), percent church adherents, and percent immigrants (foreign born) are also shown. As may be seen, the highest suicide rates in 2000 were observed for Alaska (28.8 per 100,000 population), followed by Nevada (25.3), New Mexico (22.9), and Montana (21.2). The state of Alaska also had one of the highest rates of firearm ownership, a figure that was second to only Wyoming (59.7). The states of Montana (57.7), South Dakota (56.6), and Idaho (55.3) also exhibited very high rates of gun ownership.
Table 1

Top 20 states with highest suicide rates, and scores on firearm ownership, conservatism, church membership, and immigration, United States 2000

State

Suicide rate, 2000

% Households with firearms, 2000

% Votes cast for Bush, 2000

% Church adherents, 2000

% Immigrants, 2000

AK

28.8

57.8

58.6

34.3

5.9

NV

25.3

33.8

49.5

34.4

15.8

NM

22.9

34.8

47.8

58.2

8.2

MT

21.2

57.7

58.4

44.7

1.8

WY

21.0

59.7

67.8

46.7

2.3

AZ

19.5

31.1

51.0

39.9

12.8

OK

18.1

42.9

60.3

60.8

3.8

UT

17.9

43.9

66.8

74.7

7.1

OR

17.9

39.8

46.5

31.3

8.5

CO

17.9

34.7

50.7

39.5

8.6

AL

16.4

51.7

56.5

54.8

2.0

ID

16.3

55.3

67.2

48.5

5.0

WV

16.2

55.4

51.9

35.9

1.1

KY

16.1

47.7

56.5

53.4

2.0

AR

16.1

55.3

51.3

57.1

2.8

FL

16.0

24.5

48.8

41.1

16.7

TN

15.9

43.9

51.1

51.1

2.8

VT

15.8

42.0

40.7

39.1

3.8

SD

15.8

56.6

60.3

67.8

1.8

MO

15.7

41.7

50.4

51.7

2.7

Descriptive statistics of the covariates at both individual and state levels are shown in Table 2. The number of suicides constituted a little over 6 % of the sampled deaths. About 49 % of sampled deaths were male, while 51 % were female. Married individuals constituted 39.3 % of the sample, the divorced/separated, 11.9 %, single/never married, 10.6 %, and the widowed made up 38.2 %. The mean age was 73 (SD = 16.6), with the lowest age being 18 and the highest age 98. As for educational attainment, the mean was 11.6 (SD = 2.87). Most of the sample (78 %) resided in urban areas. For race/ethnicity, the distribution was as follows: 81.9 % of decedents were non-Hispanic whites, 11.3 % African Americans/Blacks, 1.6 % Asian/Pacific Islanders, 0.04 % Native Americans, and Hispanics made up 4.9 % of deaths. At level 2, the mean suicide rate was 14.7 (SD = 4.31). Means on firearm ownership (36.9; SD = 13.9), votes cast for Bush in 2000 (49.9, SD = 10.3), church adherence (49.9, SD = 10.6), and immigration (7.3, SD = 5.7) are also shown.
Table 2

Descriptive statistics of the variables

Variable

Mean

SD

Min

Max

Level 1 variables

 Suicides

0.064

   

 Sex

  Male

0.494

   

  Female

0.506

   

 Marital Status

    

  Married

0.393

   

  Divorced/Separated

0.119

   

  Single Never Married

0.106

   

  Widowed

0.382

   

 Age

73.32

16.6

18

98

 Education

11.65

2.87

0

17

 City Size of Residence

1.615

1.21

1

5

 Place of Residence

  Non-metropolitan

0.217

   

  Metropolitan

0.783

   

 Race/Ethnicity

  Non-Hispanic White

0.819

   

  Non-Hispanic African American

0.113

   

  Non-Hispanic Asian/PI

0.016

   

  Non-Hispanic Native American

0.004

   

  Hispanic

0.049

   

 Year of Death

  2000

0.192

   

  2001

0.193

   

  2002

0.224

   

  2003

0.197

   

  2004

0.193

   

Level 2 Variables

 Suicides per 1000,000, 2000

14.7

4.31

4.90

28.8

 Firearm ownership, 2000 (%)

36.9

13.9

3.80

59.7

 Votes cast for Bush, 2000 (%)

49.9

10.3

8.95

67.8

 Church adherents, 2000 (%)

49.9

10.6

31.3

74.7

 Immigration, 2000 (%)

7.3

5.7

1.10

26.2

For dummy variables, the mean refers to the proportion of persons in the category coded 1

Multivariate findings

The rest of the analysis focuses on examining the association between the odds of suicide deaths and the primary covariates of theoretical relevance in the study: the state suicide rate, firearm availability, and state conservatism. An effort was made, however, to determine (1) whether the log-odds of suicide vary across states, and (2) the effects of individual level characteristics on the log-odds of suicide. An empty (null) model containing no independent variable was estimated first. Results (not presented) showed that the covariance parameter estimate (0.00437; t = 26.7) was statistically significant, suggesting that there are differences in the log-odds of suicide across states that may not be captured by covariates included only at the individual level. The estimated average log-odds for that model was −2.606 (t = −257.2). With evidence of variation in the log-odds of suicide across states, the next task was to add explanatory variables. Covariates were added first at the individual level, and relevant results are presented in Table 3.
Table 3

Multilevel logistic regression results of the effects of individual characteristics on the odds of suicide, United States, 2000–2004 (N = 2,045,919)

Variables

β

t value

OR

95 % CI

Intercept

0.1548**

5.48

1.167

1.103, 1.236

Individual level (1)

 Marital status

  Married

Reference

 

1.000

Reference

  Divorce/separated

0.3282**

37.54

1.388

1.365, 1.412

  Single/never married

−0.1635**

−16.65

0.849

0.833, 0.866

  Widowed

0.3423**

27.97

1.408

1.375, 1.442

Control variables

 Sex

  Female

Reference

 

1.000

Reference

  Male

1.0762**

134.92

2.933

2.888, 2.979

 Race/ethnicity

  Black/African American

Reference

 

1.000

Reference

  White, non-Hispanic

1.3824**

100.55

3.984

3.879, 4.093

  Native American/Indian

0.9708**

24.16

2.640

2.440, 2.856

  Asian/Pacific islander

1.2893**

46.56

3.630

3.438, 3.832

  Hispanic

0.8569**

45.13

2.356

2.269, 2.445

 Place of residence

  Non-metropolitan

Reference

 

1.000

Reference

  Metropolitan

−0.1522**

−17.43

0.859

0.844, 0.874

 City size of residence

−0.0383**

−11.69

0.962

0.639, 0.968

 Age

−0.0905**

−372.56

0.913

0.913, 0.913

 Year of death

  2000

Reference

 

1.000

Reference

  2001

0.0302*

2.77

1.031

1.009, 1.053

  2002

−0.0567**

−5.38

0.945

0.925, 0.965

  2003

0.0489**

4.53

1.050

1.028, 1.073

  2004

0.0933**

8.68

1.098

1.098, 1.075

Covariance parameter

0.001877**

20.18

  

−2 Res log pseudo-likelihood

13993508

   

Generalized Chi-square

1895302

   

Gener. Chi-square/df

0.93

   

Suicides

131,636

   

Natural deaths

1,914,283

   

Total observations (level 1)

2,045,919

   

Total observations (level 2)

51

   

Max obs. per state

214,080

   

β unstandardized logistic coefficient, OR odds ratio, CI confidence interval

p = .05; ** p = .01

As may be seen, the individual level measure of social integration (marital status) was highly significant. Divorced/separated individuals had odds of suicide that were over 38 % higher than the odds of married persons (AOR = 1.388, CI = 1.365, 1.412). Widowed persons also had elevated suicide odds compared to the married (AOR = 1.408, CI = 1.375, 1.442). Single/never married individuals, however, experienced reduced odds of suicide compared to the married. Their odds of suicide were 15 % lower than those of married persons (AOR = 0.849, CI = 0.833, 0.866).

The rest of the covariates were in the expected direction typically found in the suicide literature. For instance, men had odds of suicide that were 2.9 times the odds of suicide for women (AOR = 2.933, CI = 2.888, 2.979). Significant racial disparities were observed in the log-odds of suicide. For instance, the odds of suicide for whites were 3.984 times higher than the odds of African Americans/Blacks. Native American, Asian/Pacific Islanders, and Hispanics also had much higher odds of suicide than African Americans. Individuals residing in metropolitan areas experienced lower odds of suicide than those in rural areas (AOR = 0.859, CI = 0.844, 0.874). Similarly, the larger the size of the city of residence, the lower the odds of suicide (AOR = 0.962, CI = 0.639, 0.968). Age was negatively associated with the odds of suicide (AOR = 0.913, CI = 0.913, 0.913). For year of death, with the exception of 2001, every year experienced higher odds of suicide than 2000. The random effect in the model, that is level 2 variance (\( \tau^{ 2}_{0} = {\text{var}}\left( {U_{0j} } \right) \)= 0.001877 was highly significant (t = 20.2). We turn next to the full model that incorporated level 2 covariates. A summary of results is presented in Table 4.
Table 4

Multilevel logistic regression results of the effects of national suicide rates, gun ownership and conservatism on individual suicide, United States, 2000–2004 (N = 2,045,919)

Variables

β

t value

OR

95 % CI

Intercept

−0.5714**

−9.32

0.571

0.499, 0.639

State level(2)

 Suicide rate, 2000

0.0408**

17.99

1.042

1.037, 1.046

  % Household with firearms 2000

0.0044**

4.82

1.004

1.003, 1.006

  % Votes for George Bush, 2000

0.0054**

5.85

1.005

1.003, 1.007

  % Church adherents, 2000

−0.0053**

−8.68

0.995

0.993, 0.996

  % Immigrants, 2000

0.0092**

7.11

1.009

1.006, 1.012

Individual level (1)

 Marital status

  Married

Reference

 

1.000

Reference

  Divorce/separated

0.3252**

37.17

1.384

1.361, 1.408

  Single/never married

−0.1584**

−16.12

0.853

0.837, 0.870

  Widowed

0.3423**

27.96

1.408

1.375, 1.442

Control variables

 Sex

  Female

Reference

 

1.000

Reference

  Male

1.0763**

134.84

2.934

2.888, 2.980

 Race/ethnicity

  Black/African American

Reference

 

1.000

Reference

  White, non-Hispanic

1.3733**

100.25

3.948

3.844, 4.056

  Native American/Indian

0.9187**

22.88

2.506

2.316, 2.711

  Asian/Pacific islander

1.2852**

46.70

3.615

3.426, 3.816

  Hispanic

0.8505**

44.80

2.341

2.255, 2.429

 Place of residence

  Non-metropolitan

Reference

 

1.000

Reference

  Metropolitan

−0.1343**

−15.27

0.874

0.859, 0.889

 City size of residence

−0.0393**

−12.04

0.961

0.955, 0.967

 Age

−0.0905**

−372.13

0.913

0.913, 0.914

 Year of death

  2000

Reference

 

1.000

Reference

  2001

0.0304*

2.79

1.031

1.009, 1.053

  2002

−0.0563**

−5.34

0.945

0.926, 0.965

  2003

0.0486**

4.50

1.050

1.028, 1.072

  2004

0.0931**

8.65

1.098

1.075, 1.121

 Covariance parameter

0.000519**

9.10

  

−2 Res log pseudo-likelihood

14007466

   

Generalized Chi-square

1900188

   

Gener. Chi-square/df

0.91

   

Suicides

131,636

   

Natural deaths

1,914,283

   

Total observations (level 1)

2,045,919

   

Total observations (level 2)

51

   

Max obs. per state

214,080

   

p = .05; ** p = .01

As may be observed in the table, the state suicide rate elevated individual risk of dying from suicide (AOR = 1.042, CI = 1.037−1.046). Specifically, the state suicide rate increased the log-odds of suicide death by 4.2 %. Firearm availability elevated individual suicide odds by 0.4 %. State conservatism was associated significantly with individual suicide (AOR = 1.005, CI = 1.003, 1.007). The higher the percentage of church adherents in a state, the lower the individual odds of suicide (AOR = 0.995, CI = 0.993, 0.996). Greater supply of immigrants in a state significantly increased individual suicide risk (AOR = 1.009, CI = 1.006−1.012).

Finally, the variance of the random state intercepts on the logit scale (\( \tau^{ 2}_{0} = {\text{var}}\left( {U_{0j} } \right) \)) = 0.000519 was statistically significant (t = 9.10), suggesting the existence of state differentials in individual odds of suicide. It should be noted, however, that adding level 2 variables substantially reduced across state variation in suicide odds by over 72 %, indicating that modeling the outcome variable only at the individual level would not have provided the full spectrum of risk factors for suicide.

Discussion

For sociological research in suicide to advance, investigators may need to make greater use of multilevel designs. Events which are frequently viewed as individual behavior may also be influenced by contextual factors. In the present study, state suicide rates, firearm availability, state conservatism, and the immigration rate significantly affected individual suicide behavior. How explain this finding? Consistent with social learning and differential association theories of deviant behavior [3, 4, 7, 77] persons living in states or localities with high suicide rates may have higher exposure to definitions favorable to suicide acceptance, and this in turn may increase their odds of committing suicide. Prevailing social, economic, and even political conditions in a state may further affect individual suicidal behavior by maintaining an environment in which people’s aspirations are thwarted and dreams of a better tomorrow are deferred [41].

Results showed that firearm availability at the state level is a significant risk factor for individual suicide. The finding is consistent with numerous studies at the individual level [e.g. 2, 6, 35, 53, 68]. Despite overwhelming evidence that firearms elevate suicide risk, policy makers and many American citizens remain opposed to gun control. Indeed, many believe rightly or wrongly that they have a constitutional right to bear arms [39]. Even modest efforts to reform gun laws are typically met with vehement opposition. There are also millions of Americans that continue to believe that keeping a gun at home protects them against intruders, even though research shows that when a gun is used in the home, it is often against household members in the commission of homicides or suicides [39]. Adding to the widespread misinformation is that powerful pro gun lobby groups, especially the National Rifle Association seem to have a strangle hold on legislators and US policy [5, 21, 24, 82]. A politician that calls for gun control may be targeted for removal from office in a future election by a gun lobby, such as the NRA [70, 74]. Given the state of paralysis with respect to gun control, the status quo is likely to continue and even worsen. The United States has one of the highest rates of gun ownership in the world, with Americans owning an estimated 283 million firearms, a figure that translates into slightly over 90 guns for every 100 people [26]. An estimated 46 % of US households have firearms [37, 60]. Deaths rates due to firearms are eight times higher in the United States than those in other OECD countries [15, 36]. Rates of suicide death, especially those attributable to firearms are likely to grow as minimal existing gun control laws are either repealed and new ones (e.g. ‘Stand your Ground,’ ‘Right to Carry’) are introduced into state legislatures and passed with relatively little attention to gun violence, including suicide. The United States is thus poised to remain a very armed and potentially dangerous nation for its inhabitants for years to come, though total suicide rates in the US do not seem to be much higher compared to other Western countries.

How might having a gun at home elevate suicide risk? What are possible pathways? Many studies show that of all suicide methods, firearms have the highest case fatality, implying that an individual that selects to use this technique has a very low chance of survival should the suicidal event unfold [8, 39]. Thus, one pathway is simply the fact that as opposed to other methods, firearms are efficient.

Social integration theory received empirical support in the analysis. With regard to family integration, a consistent and notable finding is that divorce places a person at an especially high risk of becoming a suicide victim. How interpret the divorce findings. Durkheim [23] informed us over a century ago that a breakdown in social integration through divorce is a crisis and a profoundly stressful life event. This is now a well documented observation in mental health, health psychology, psychosomatic medicine, and family therapy [25, 41, 49]. Separated and divorced individuals have been found to be over-represented in the psychiatric patient population, for instance. Separated people are more likely to be involved in automobile accidents than non-separated people [43, 45]. Marital separation, as opposed to non-separation, is associated with higher rates of illness and disability as well as high death rates due to suicide, homicide, and other causes of mortality [51, 63]. Findings from this study confirm once again that divorce is a non-trivial contributing factor to suicide.

At the state level, the religious integration aspect of Durkheim’s theory was also supported in that church membership was associated with lower risk of suicide. This finding is consistent with many in the suicide literature [9, 71, 72, 73]. Church adherence may promote church attendance, which exposes an individual to religious beliefs, for example, about an afterlife. Suicide is proscribed in the three monotheistic religions: Judaism, Christianity, and Islam. In states with higher percentage of the population that belong to a church, it is plausible that religious views and doctrine about suicide are well known either through sacred texts, theology, or sermons, and adherents may be less likely to commit suicide.

The state immigration rate influenced individual suicide risk, another finding that supported the social integration perspective. Results on immigration should be interpreted with caution as there was no disaggregation by immigrants’ country of origin. Some studies show that immigrants from countries with low suicide rates tend to exhibit low suicide in the destination country, a process that could depress the overall suicide rate in the receiving nation [13, 27]. Similarly, subpopulations from nations with high suicide rates experience high suicides upon immigration, thereby inflating the overall suicide rate in the host society [75]. It has been found, however, that over time immigrant suicide rates tend to converge to those of citizens in the destination country [38, 40].

Political conservatism at the state level elevated individual suicide risk. There are not many studies linking political ideology to suicide, but the finding here is inconsistent with the few that have [1, 66, 73]. It is plausible that the inconsistency is due to measurement of conservatism. Previous studies often relied on the political ideology measurement found in the US General Social Survey [20] or the World Values Survey [30], and argued that generalized political liberalism promotes greater approval of suicide. Thus, while past studies used suicide approval, the present one employed completed suicides. It is often difficult, if not impossible to determine the political ideology of individuals that have died, due in large part to lack of data. By employing state conservatism, it was expected that we would tap into community as opposed to individual conservatism. The inconsistent findings provide an opportunity for future researchers to explore further whether it is liberalism that promotes suicide risk or conservatism.

Findings confirm that there are still significant rural–urban differences in suicide risk in that metropolitan areas experienced lower risks than non-metropolitan areas. Results are consistent with those reported in past research [33, 41, 44, 61]. Commenting on the on rural–urban differentials in suicide risk, Kposowa [44] argued that completed suicides may be influenced by emergency response services and the existence of nearby trauma centers. He argued that in localities with rapid emergency response, timely arrival at the scene of a suicide event could make the difference between life and death. In general, emergency response services as well as trauma centers are more likely to be concentrated in large urban centers than in smaller cities or rural areas [41]. It may well be that at some point in American history, urban areas had much higher suicide rates than rural areas, but with advances in technology, especially communication and transportation centered in metropolitan areas, there has emerged a gradual shift in suicide occurrence that disadvantages rural communities [28]. This line of argument is provided with further boost by a related finding that the greater the city size of residence, the lower the odds of individual suicide.

In general, support was found for social structural social learning theory in that national suicide rates and firearm ownership were associated with significantly higher odds of suicide. Evidence from the data analysis backed up all three hypotheses derived from the theory. Social integration theory was also supported in that divorce/separation at the individual level elevated suicide risk, and church membership at the state level reduced the odds of individual suicide. There are, however, some limitations that need to be pointed out for readers and future researchers. Ideally, we would have liked to have a measurement of firearm ownership at the individual level, but such information was simply not available in the data set used. Variables, such as religious attendance, immigration status of the decedent, or political views of individuals were also not available. Despite these limitations, however, it is hoped that findings from this study will be a catalyst for further looking at the link between firearms and suicide. As pointed out in the introduction, previous studies limited analysis of the association between firearms and suicide to specific geographic areas, e.g. two counties [35] or one county [14], or a limited number of states [52]. This study is the first to use a nationally representative sample to model the effect of firearm availability on suicide odds using a mixed modeling approach.

References

  1. 1.
    Agnew R (1998) The approval of suicide: a social psychological model. Suicide Life Threat Behav 28:205–225PubMedGoogle Scholar
  2. 2.
    Ajdacic-Gross V, Weiss MG, Ring M, Hepp U, Bopp M, Gutzwiller F, Rössler W (2008) Methods of suicide: international suicide patterns derived from the WHO mortality database. Bull World Health Organ 86:726–732PubMedCrossRefGoogle Scholar
  3. 3.
    Akers R (1998) Social learning and social structure: a general theory of crime and deviance. Northeastern University Press, BostonGoogle Scholar
  4. 4.
    Akers R (2004) Criminological theories: introduction, evaluation, and application. Roxbury Publishing Company, Los AngelesGoogle Scholar
  5. 5.
    Anderson J (1996) Inside the NRA: armed and dangerous: an expose. BC, New Star Press, VancouverGoogle Scholar
  6. 6.
    Andrés AR, Hempstead K (2011) Gun control and suicide: the impact of state firearm regulations in the United States, 1995–2004. Health Policy 101:95–103CrossRefGoogle Scholar
  7. 7.
    Baller RD, Shin DJ, Richardson K (2005) An extension and test of Sutherland’s concept of differential social organization: the geographic clustering of Japanese suicide and homicide rates. Suicide Life Threat Behav 35:343–355PubMedCrossRefGoogle Scholar
  8. 8.
    Betz ME, Barber C, Miller M (2011) Suicidal behavior and firearm access: results from the second injury control and risk survey. Suicide Life Threat Behav 41:384–391PubMedCrossRefGoogle Scholar
  9. 9.
    Breault KD (1986) Suicide in America: a test of Durkheim’s theory of religious and family integration, 1933–1980. Am J Sociol 92:628–656CrossRefGoogle Scholar
  10. 10.
    Breault KD (1994) Was Durkheim right? A critical survey of the empirical literature on Le Suicide. In: Lester D (ed) Emile Durkheim Le Suicide, 100 years later. The Charles Press, Philadelphia, pp 11–29Google Scholar
  11. 11.
    Brent DA (2001) Firearms and suicide. Ann NY Acad Sci 932:225–782PubMedCrossRefGoogle Scholar
  12. 12.
    Burrows S, Laflamme L (2006) Suicide mortality in South Africa: a city level comparison across socio demographic groups. Soc Psychiatry Psychiatr Epidemiol 41:108–114Google Scholar
  13. 13.
    Burvill PW (1998) Migrant suicide rates in Australia and in country of birth. Psychol Med 28:201–208PubMedCrossRefGoogle Scholar
  14. 14.
    Callanan VJ, Davis MS (2011) Gender differences in suicide methods. Soc Psychiatry Psychiatr Epidemiol. doi:10.1007/s00127-011-0393-5 PubMedGoogle Scholar
  15. 15.
    Centers for Disease Control and Prevention (2012) Web-based Injury Statistics Query and Reporting System (WISQARS). National Center for Injury Prevention and Control, CDC (producer). Available from URL: www.cdc.gov/injury/wisqars/index.html. Online Accessed 20 Feb 2012
  16. 16.
    Centers for Disease Control and Prevention (2001) Behavioral Risk Factor Surveillance System, BRFSS 2001 Survey. Available online from : http://www.cdc.gov/brfss/technical_infodata/surveydata/2001.htm, Accessed 14 Mar 2012
  17. 17.
    Classen TJ, Dunn RA (2010) The politics of hope and despair: the effects of presidential election outcomes on suicide rates. Soc Sci Q 91:593–612PubMedCrossRefGoogle Scholar
  18. 18.
    Compton MT, Thompson NJ, Kaslow NJ (2005) Social environment factors associated with suicide attempt among low-income African Americans: the protective role of family relationships and social support. Soc Psychiatry Psychiatr Epidemiol 40:175–185Google Scholar
  19. 19.
    Cooper J, Husain N, Webb R, Waheed W, Kapur N, Guthrie E, Appleby L (2006) Self-harm in the UK. Soc Psychiatry Psychiatr Epidemiol 41:782–788Google Scholar
  20. 20.
    Davis JA, Smith TW, Marsden PV (2010). General Social Surveys 1972–2008 (Cumulative File). ICPSR25962-v2. Storrs, CT: roper center for public opinion researchGoogle Scholar
  21. 21.
    Davison OG (1998) Under Fire: The NRA and the Battle for Gun Control. University of Iowa Press, Iowa CityGoogle Scholar
  22. 22.
    Desaulniers J, Daigle MS (2008) Inter-regional variations in men’s attitudes, suicide rates and sociodemographics in Quebec (Canada). Soc Psychiatry Psychiatr Epidemiol 43:445–453Google Scholar
  23. 23.
    Durkheim E (1897/2002) Le Suicide: Étude de Sociologie. Presses Universitaires de France, ParisGoogle Scholar
  24. 24.
    Feldman R (2007) Ricochet: confessions of a Gun Lobbyist. Wiley, New YorkGoogle Scholar
  25. 25.
    Hawthorne G (2008) Perceived social isolation in a community sample: its prevalence and correlates with aspects of people’s lives. Soc Psychiatry Psychiatr Epidemiol 43:140–150PubMedCrossRefGoogle Scholar
  26. 26.
    Hepburn L, Miller M, Azrael D, Kemenway D (2007) The US gun stock: results from the 2004 national firearms survey. Inj Prev 13:15–19PubMedCrossRefGoogle Scholar
  27. 27.
    Hjern A, Allebeck P (2002) Suicide in first and second generation immigrants in Sweden: a comparative study. Soc Psychiatry Psychiatr Epidemiol 37:423–429PubMedCrossRefGoogle Scholar
  28. 28.
    Hirsch JK (2006) A review of the literature on rural suicide: risk and protective factors, incidence, and prevention. Crisis 27:189–199PubMedGoogle Scholar
  29. 29.
    Hunt IM, Kapur N, Robinson J, Shaw J, Flynn S, Bailey H, Meeehan J, Bickley H, Parsons R, Burns J, Amos T, Applby L (2006) Suicide within 12 months of mental health service contact in different age and diagnostic groups. Br J Psychiatry 188:135–142Google Scholar
  30. 30.
    Inglehart R (2004) World values survey and European values surveys, 1999–2001. Inter University Consortium for Political and Social Research, Ann ArborGoogle Scholar
  31. 31.
    Kachur SP, Potter LB, James SP, Powell KE (1995) Suicide in the United States 1980–1992. Centers for Disease Control and Prevention, AtlantaGoogle Scholar
  32. 32.
    Kaplan MS, McFarland BH, Huguet N (2009) Characteristics of adult male and female firearm suicide decedents: findings from the national violent death reporting system. Inj Prev 15:322–327PubMedCrossRefGoogle Scholar
  33. 33.
    Kapusta ND, Zorman A, Etzersdorfer A, Ponocny-Seliger E, Jandl-Jager E, Sonneck G (2008) Rural-urban differences in Austrian suicides. Soc Psychiatry Psychiatr Epidemiol 43:311–318PubMedCrossRefGoogle Scholar
  34. 34.
    Karch D, Dahberg L, Patel N, Davis T, Logan J, Hill H, Ortega L (2009) Surveillance for violent deaths—national violent death reporting System, 16 States, 2006. MMWR 58:1–44PubMedGoogle Scholar
  35. 35.
    Kellermann AL, Rivara FP, Somes G, Reay DT, Francisco J, Banton JG, Prodzinski J, Fligner C, Hackman BB (1992) Suicide in the home in relation to gun ownership. N Engl J Med 327:467–472PubMedCrossRefGoogle Scholar
  36. 36.
    Kellermann AL, Waeckerle JF (1998) Preventing firearm injuries. Ann Emerg Med 32:77–79PubMedCrossRefGoogle Scholar
  37. 37.
    Kleck G (1997) Targeting guns and their control. Aldine de Gruyter, New YorkGoogle Scholar
  38. 38.
    Kliewer EV, Ward RH (1988) Convergence of immigrant suicide rates to those in the destination country. Am J Epidemiol 127:640–653PubMedGoogle Scholar
  39. 39.
    Kposowa AJ, McElvain JP (2006) Gender, place and method of suicide. Soc Psychiatry Psychiatr Epidemiol 41:435–443PubMedCrossRefGoogle Scholar
  40. 40.
    Kposowa AJ, McElvain JP, Breault KD (2008) Immigration and suicide: the role of marital status, duration of residence, and social integration. Arch Sui Res 12:82–92CrossRefGoogle Scholar
  41. 41.
    Kposowa AJ, D’Auria S (2010) Association of temporal factors and suicides in the United States 2000 to 2004. Soc Psychiatry Psychiatr Epidemiol 45:433–445PubMedCrossRefGoogle Scholar
  42. 42.
    Kposowa AJ, Glyniadaki A (2012) Mental health and suicide: an ecological hierarchical analysis of US counties. In: Nazrul H, David A Swanson (eds) Opportunities and challenges for applied demography in the 21st century. Springer, New York, p 289–308Google Scholar
  43. 43.
    Kposowa AJ (2000) Marital status and suicide in the National Longitudinal Mortality Study. J Epidemiol Community Health 54:254–261PubMedCrossRefGoogle Scholar
  44. 44.
    Kposowa AJ (2009) Psychiatrist availability, social disintegration and suicide deaths in US counties, 1990–1995. J Community Psychol 37:73–87CrossRefGoogle Scholar
  45. 45.
    Kposowa AJ, Adams M (1998) Motor vehicle crash fatalities: the effects of race and marital status. Appl Behav Sci Rev 6:66–91Google Scholar
  46. 46.
    Kung HC, Pearson JL, Liu X (2003) Risk factors for male and female suicide decendants ages 15–64 in the United States: results from the 1993 National Mortality Followback Survey. Soc Psychiatry Psychiatr Epidemiol 38:419–426PubMedCrossRefGoogle Scholar
  47. 47.
    Kung HC, Pearson JL, Wei R (2005) Substance use, firearm availability, depressive symptoms, and mental health service utilization among white and African American suicide decedents aged 15–64 years. Ann Epidemiol 15:614–621PubMedCrossRefGoogle Scholar
  48. 48.
    Kung HC, Liu X, Juon HS (1998) Risk factors for suicide in Caucasians and in African Americans: a matched case-control study. Soc Psychiatry Psychiatr Epidemiol 33:155–161PubMedCrossRefGoogle Scholar
  49. 49.
    Lönnquist JK (2000) Psychiatric aspects of suicidal behaviour. In: Hawton K, van Heeringen K (eds) International handbook of suicide and attempted suicide. Wiley, New York, p 107–120Google Scholar
  50. 50.
    Maimon D, Kuhl DC (2008) Social control and youth suicidality: situating Durkheim’s ideas in a multilevel framework. Am Sociol Rev 73:921–943CrossRefGoogle Scholar
  51. 51.
    Miech R, Pampel F, Kim J, Rodgers RG (2011) The enduring association between education and mortality: the role of widening and narrowing disparities. Am Sociol Rev 76:913–934CrossRefGoogle Scholar
  52. 52.
    Miller M, Azrael D, Hemenway D (2002) Household firearm ownership and suicide rates in the United States. Epidemiology 13:517–524PubMedCrossRefGoogle Scholar
  53. 53.
    Miller M, Azrael D, Hemenway D (2001) Firearm availability and unintentional firearm deaths. Accid Anal Prev 33:477–484PubMedCrossRefGoogle Scholar
  54. 54.
    National Center for Health Statistics (2008). Multiple cause of death files, 2000–2004 (Computer file). Available at http://www.cdc.gov/nchs. Accessed 2 Feb 2009
  55. 55.
    National Center for Health Statistics (2012) Classification of death and injury resulting from terrorism. Available at (http://www.cdc.gov/nchs/icd/terrorism_code.htm). Accessed 15 Dec 2012
  56. 56.
    National Center for Health Statistics (2001) Health United States 2000. NCHS, HyattsvilleGoogle Scholar
  57. 57.
    Norström T (1995) The impact of alcohol, divorce, and unemployment on suicide: a multilevel analysis. Soc Forces 74:293–314Google Scholar
  58. 58.
    Page A, Morrell S, Taylor R (2002) Suicide and political regime in New South Wales and Australia during the 20th century. J Epidemiol Community Health 56:766–772PubMedCrossRefGoogle Scholar
  59. 59.
    Pampel FC (2007) Sociological lives and ideas: an introduction to the classical theorists. Worth Publishers, New YorkGoogle Scholar
  60. 60.
    Price JH, Thompson AJ, Dake JA (2004) Factors associated with state variations in homicide, suicide, and unintentional firearm deaths. J Community Psychol 29:271–283Google Scholar
  61. 61.
    Qin P (2005) Suicide risk in relation to level of urbanicity: a population-based linkage study. Int J Epidemiol 34:846–852PubMedCrossRefGoogle Scholar
  62. 62.
    Raudenbush SW, Bryk AS (2002) Hierarchical linear models: applications and data analysis methods. Sage Publications, Thousand OaksGoogle Scholar
  63. 63.
    Rogers RG, Hummer RA, Nam CB (2000) Living and dying in the USA: behavioral, health and social differentials in US adult mortality. Academic Press, New YorkGoogle Scholar
  64. 64.
    Institute SAS (2012) SAS/STATS: the GLIMMIX procedure [computer software]. SAS Institute, Inc., CaryGoogle Scholar
  65. 65.
    SAS Institute (2012b) SAS/STATS: the univariate procedure (computer software). SAS Institute Inc, CaryGoogle Scholar
  66. 66.
    Sawyer D, Sobal J (1987) Public attitudes toward suicide: demographic and ideological correlates. Public Opin Q 51:92–101Google Scholar
  67. 67.
    Shenassa ED, Catlin SN, Buka SL (2003) Lethality of firearms relative to other suicide methods: a population based study. J Epidemiol Community Health 57:120–124PubMedCrossRefGoogle Scholar
  68. 68.
    Slater GY (2011) The missing piece: a sociological autopsy of firearm suicide in the United States. Suicide Life Threat Behav 41:474–490PubMedCrossRefGoogle Scholar
  69. 69.
    Snijders TAB, Bosker RJ (2012) Multilevel analysis. Sage, Los AngelesGoogle Scholar
  70. 70.
    Spitzer RJ (2011) The politics of gun control. Paradigm Publishers, New YorkGoogle Scholar
  71. 71.
    Stack S, Kposowa AJ (2011) The effect of survivalism-self-expressionism culture on black male suicide acceptability: a cross-national analysis. Soc Sci Med 72:1211–1218PubMedCrossRefGoogle Scholar
  72. 72.
    Stack S, Kposowa AJ (2011) Religion and suicide acceptability: a cross national analysis. J Sci Study Relig 50:289–306PubMedCrossRefGoogle Scholar
  73. 73.
    Stack S, Kposowa AJ (2008) The association of suicide rates with individual-level suicide attitudes: a cross-national analysis. Soc Sci Q 89:39–59CrossRefGoogle Scholar
  74. 74.
    Sugarman J (1992) National Rifle Association: money, firepower and fear. National Press Books, WashingtonGoogle Scholar
  75. 75.
    Sundaram V, Qin P, Zollner L (2006) Suicide risk among persons with foreign background in Denmark. Suicide Life Threat Behav 36:481–489PubMedCrossRefGoogle Scholar
  76. 76.
    Sutherland EH (1947) Principles of criminology, 4th edn. J.B. Lippincott, PhiladelphiaGoogle Scholar
  77. 77.
    Sutherland EH, Cressey DR, Luckenbill DF (1992) Principles of criminology. General Hall, New YorkGoogle Scholar
  78. 78.
    US Bureau of the Census (2001) Statistical sbstract of the United States. US Government Printing Press, WashingtonGoogle Scholar
  79. 79.
    Van Tubergen F, te Grotenhuis M, Ultee W (2005) Denomination, religious context, and suicide: neo-Durkheimian multilevel explanations tested with individual and contextual data. Am J Sociol 111:797–823CrossRefGoogle Scholar
  80. 80.
    Tubergen Van, Ultee W (2006) Political integration, war and suicide: the Dutch paradox? Int Sociol 21:221–236CrossRefGoogle Scholar
  81. 81.
    Wiebde DJ (2003) Homicide and suicide risks associated with firearms in the home: a national case–control study. Ann Emerg Med 41:771–782CrossRefGoogle Scholar
  82. 82.
    Winkler A (2011) Gunfight: the battle over the right to bear arms in America. WW Norton and Co, New YorkGoogle Scholar
  83. 83.
    World Health Organization (2005) International statistical classification of diseases and health related problems (The) ICD-10, 2nd edn. World Health Organization, GenevaGoogle Scholar
  84. 84.
    Wray M, Colen C, Pescosolido B (2011) The sociology of suicide. Ann Rev Sociol 37:505–528CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  1. 1.Department of SociologyUniversity of California, RiversideRiversideUSA

Personalised recommendations