The impact of accession to the European Union on suicide rates: A cross-national time-series analysis

Abstract

Building upon prior research, this study examines the effects of European Union (EU) accession on suicide rates in the Eastern European countries that joined the EU in 2004 and 2007 using pooled cross-sectional time-series data that cover approximately 20 years (1990–2011). Results from fixed-effects regression analyses indicate that EU entry has no effect on total suicide rates and suicide rates among males, but has a negative effect on female suicide rates in the fully specified models. In addition, we find that EU entry also has a negative effect on the ratio of suicide rates to an aggregated indicator of lethal violence (homicide rates + suicide rates, or the suicide–homicide ratio) for the total population and for the female population. Consistent with previous research, we find some significant negative effects on suicide rates for economic growth and life expectancy at birth, and a positive effect for females. When interpreted with reference to the ‘stream analogy’ for understanding the two major forms of lethal violence (suicide and homicide), our findings suggest that the impact of any increase in the ‘flow’ of lethal violence associated with EU entry is likely to be manifested in an ‘outward’ rather than ‘inward’ direction for the nations in the sample. Our analyses also reaffirm previous research documenting appreciable gender differences in lethal violence.

Keywords

Accession to the European Union Eastern Europe European Union gender differences in suicide suicide

Introduction

Not long after the collapse of the Soviet regime, scholars and public health policy makers documented a sharp increase in suicide rates in post-Soviet countries. The levels of suicide almost doubled in the early 1990s in some transition countries, with particular increases among males (Brainerd, 2001; Mäkinen, 2000; Pray et al., 2013). In response to these observations, an extensive body of work has explored the connection between socioeconomic conditions and suicide rates (Brainerd, 2001; Kõlves et al., 2013). Drawing largely upon the Durkheimian framework, this research suggested that social, economic, and political changes brought about by the dissolution of the Soviet Union contributed to normative confusion and deregulation in post-Soviet nations. This, in turn, caused levels of suicide to rise (Kõlves et al., 2013; Minagawa, 2013; Pridemore et al., 2007).

While much research examines suicide rate changes in the wake of the collapse of the Soviet Regime, little is known about suicide trends in the later stages of transformation. After a period of social turmoil, many Eastern European nations experienced substantial economic growth and increased political stability (Curzio and Fortis, 2008; Harfst, 2000), and in 2004 and 2007, a combined total of 10 Eastern European nations joined the European Union (EU). Recent studies suggest that entry into the EU constituted a new form of relatively rapid transition for Eastern European nations that contributed to rising levels of criminal activity. Andresen (2009, 2011), for example, reported that EU entry was associated with increased levels of crime in Lithuania, which he attributed to social deregulation. Similarly, Piatkowska et al. (2016) found that entry into the EU increased the levels of homicide in 10 Eastern European nations that joined the EU in 2004 and 2007. Consistent with Andresen’s account, Piatkowska et al. concluded that EU accession could have triggered social deregulation of the existing systems in the new member states, which in turn caused homicide rates to rise.

This study seeks to expand the existing body of work by looking at the impact of EU accession on the rates of suicide in the 10 Eastern European nations that joined the EU in 2004 and 2007. Our analyses examine the extent to which EU entry is associated with suicide rates for the total population and for the sex-specific population. We also draw upon the conceptual framework developed in Unnithan et al. (1994), which uses a stream analogy to understand the determinants of lethal violence more generally, for example, the combined level of suicide and homicide. Informed by this framework, we consider the relationship between EU entry and the rates of suicide relative to the combined rates of lethal violence (suicides and homicides). Our analyses are based on a pooled cross-sectional time-series data set for the 10 countries that joined the EU in 2004 and 2007 during the 1990–2011 period.

The research settings: suicide in transition nations

As noted above, a fairly large share of the existing research on suicide rates in transition countries is guided by the Durkheimian framework (Kõlves et al., 2013; Minagawa, 2013; Pridemore et al., 2007). Durkheim (1951 [1897]) maintained that national suicide rates reflect the structural conditions of societies, wherein the processes of social and economic development produce long-term increases in national suicide rates and short-term fluctuations in suicide, which are indicative of sudden disruptions in the normal functioning of society. In his social deregulation theory, Durkheim argued that abrupt social and economic changes have a detrimental effect on the normative order of society: they create a state of anomie, thereby reducing the levels of social regulation in society. This, in turn, causes suicide rates to rise (Durkheim, 1951 [1897]).

Research on suicide in transition countries largely supports Durkheim’s social deregulation theory. The argument with regard to Eastern European nations is that social and economic changes brought about by the dissolution of the Soviet Union contributed to social deregulation, and consequently to rising levels of suicide (Kõlves et al., 2013; Minagawa, 2013). Pridemore et al. (2007), for example, found that suicide rates in post-Soviet Russia increased in the wake of the collapse of the Soviet Regime. The authors attributed this to the social disruptions and normative deregulation triggered by the Soviet collapse, thereby providing support for the Durkheimian thesis.

Outside of post-Soviet Russia, Minagawa (2013) built upon the Durkheimian framework to examine suicide trends in Eastern European nations between 1989 and 2006. The author found that suicide rates increased in the nations that underwent more radical structural changes in the years following the collapse of the Soviet Union. Brainerd (2001) and Mäkinen (2000) echoed these findings, reporting considerable changes in suicide rates in Eastern European countries during the period immediately following the Soviet collapse. Kõlves et al. (2013), meanwhile, showed that increases in Eastern European suicide rates in the transitional period were associated with the social and economic deregulation that resulted from the post-Soviet transition. They concluded that these findings are consistent with Durkheim’s thesis.

Research on suicide in transition nations also identifies a significant relationship between economic conditions and suicide rates. These results, however, are less consistent in terms of their compatibility with the Durkheimian hypothesis. In particular, such studies found suicide rates to be negatively associated with gross domestic product (GDP) (Kõlves et al., 2013), the growth rate (Minagawa, 2013), and economic development (Mäkinen, 2000), but positively related to unemployment (Kõlves et al., 2013; Pridemore, 2006). In addition, Pridemore (2006) reported a significant positive association between suicide rates and poverty when measured as the proportion of the regional population living below the poverty line in Russia, while he also found a negative and significant effect of urbanism on suicide rates.

There are also indications that suicide rates may be related to non-economic social and demographic variables that are often included as proxies for domestic integration and societal deregulation. Durkheim (1951 [1897]) argued that the family provides protection against suicide such that strong ties among family members contribute to lower suicide rates. In keeping with his thesis, Durkheim (1951 [1897]) observed that as suicide increased in France in the 1880s, birth rates went down. Recent research lends credibility for Durkheim’s thesis showing a negative effect of fertility – measured by the number of children per 1000 women aged 15–49 years who have children under age 5 – on suicide rates (Fernquist and Cutright, 1998). Studies also find suicide rates to be positively associated with divorce rates (Brainerd, 2001; Kõlves et al., 2013) and negatively associated with life expectancy at birth (Brainerd, 2001; Mäkinen, 2000) and with birth rates (Kõlves et al., 2013; Leenaars et al., 1993; Lester and Yang, 1992). These results are also in line with Durkheim’s thesis.

Research also documents gender differences in suicide rates (Messner et al., 2006; Möller-Leimkühler, 2003; Pridemore, 2006). Durkheim (1951 [1897]) argued that females have lower suicide rates than males, and that male suicide rates are associated with higher divorce rates, while the opposite was apparently true for women.¹ Indeed, recent work indicates that Durkheim’s observation of gender-based differences has some merit, although there are inconsistencies in the research. Kõlves et al. (2013), for example, found alcohol consumption to be associated with suicide rates for females in Eastern European countries, but to be unrelated to suicide rate for males, while Brainerd (2001) reported a significant effect of gross national product (GNP) per capita on suicide rates only for males and consumption of alcohol to have a stronger effect on suicide rates for females in Eastern European countries.

The notion that rapid social and economic change leads to social deregulation and disruption was further utilized by Andresen (2009, 2011). Although the author did not explicitly refer to Durkheim’s social deregulation theory, he in essence extended this line of thought in his examination of the effects of accession to the EU on violent crime in Lithuania. Andresen argued that while EU enlargement does not yield as massive a degree of transition as the post-Soviet transformations, it nonetheless entailed rapid change for Eastern European nations. He posited that the social, economic, and political changes associated with EU accession lead to increased levels of crime because of the instability and social deregulation they trigger. His finding that EU accession was indeed associated with a one-time sharp increase in crime rates in Lithuania supported his claim. He also found that this increase was not offset by the overall decreasing trend in crime rates during the period under investigation (Andresen, 2011).

Andresen’s (2011) work was extended by Piatkowska et al. (2016) to examine the impact of EU accession on levels of homicide in 10 Eastern European nations that joined the EU in 2004 and 2007. Drawing on Andresen (2011), the researchers suggested that entry into the EU may have led to social and structural disruption in the new member states, which in turn could cause homicide levels to rise. To test this hypothesis, we employed pooled cross-sectional time-series data (1990–2011) for the full set of nations that joined the EU in 2004 and 2007. In line with Andresen’s findings, the results from fixed-effects analysis revealed that EU accession – measured as the EU dummy variable – had a positive effect on homicide rates. The authors concluded that EU entry could have triggered disruptive processes of democratization and adjustments to a market economy in the new member states, which in turn led to increased homicide rates.

The evidence presented provides a rationale for the hypothesis that entry into the EU can lead to social deregulation and consequently to increased levels of suicide. This is particularly true given prior research indicating that the collapse of the Soviet Union had detrimental effects on suicide rates in transition countries (Kõlves et al., 2013; Minagawa, 2013; Pridemore et al., 2007), and that entry into the EU increased levels of crime in Lithuania (Andresen, 2011) and in new EU member states (Piatkowska et al., 2016). These findings, considered in tandem with the Durkheimian framework, suggest that because accession to the EU led to social disruption, we may accordingly expect to find increases in suicide rates in the countries under investigation.

This outcome might be also anticipated given a wide array of social, economic, and political changes that were introduced in new member states along with EU accession. In order to fulfill the requirements stipulated by the Copenhagen entry criteria, new member states were required to meet the economic, political, and institutional conditions, which include but are not limited to ‘the existence of a functioning market economy’, ‘the capacity to cope with competitive pressures and market forces within the Union’, ‘stability of institutions guaranteeing democracy, the rule of law, human rights, and respect for and protection of minorities’ (European Commission, 2009, 2016). Andresen (2011) adds that ‘accession to the EU leads to the standardization and/or synchronization of laws that govern international trade in goods and services, the free movement of factors of production (capital and labour), common (economic) development policies, and a common currency’ (pp. 760–761). These changes provided grounds for the potential disruption of social organization in Lithuania, which in turn contributed to increased levels of violent crime (Andresen, 2011).

It is important to note that while the current research does not directly model suicide trajectories, it seeks to assess the impact of EU accession as a process that caused rapid change, and influenced suicide rates. Notably, economic and social development processes were already taking place in Eastern Europe prior to EU accession: part of gaining EU membership involved making significant economic, political, and social changes prior to accession. As such, by 2004 and 2007, pre-accession changes already contributed to more stable democratic governments and institutional frameworks and to more competitive economic environments (European Commission, 2009). Scholars generally agree, however, that accession further enhanced the economic growth of the new member states due to further liberalization of regulations related to labor, capital, and the market as well as to improvements in regulatory environments and the removal of technical and administrative barriers (European Commission, 2009; Šlosarčík, 2011; Yorgova, 2011). This suggests the potential for EU entry to promote social disruption and disorganization in new member states, consequently leading to increased suicide rates.

Alternative hypotheses of EU effects: the stream analogy

The arguments above about the impact of EU accession on suicide rates in the Eastern European nations that joined the EU in 2004 and 2007 are predicated on an implicit premise that the consequences of profound and rapid social change are isomorphic with respect to different forms of lethal violence. Social disruption and deregulation are expected to promote both suicide and homicide. In contrast, the analytic framework advanced by Unnithan et al. (1994) referred to above acknowledges shared originating causes of lethal violence, but it allows for varying manifestations of these causes in the form of suicide or homicide.

Unnithan et al. built upon the notion that homicide and suicide have long been considered to be mutually related (Unnithan and Whitt, 1992). Not only have rates for both suicide and homicide been regarded as indicators of social integration within the Durkheimian framework (Bills and Li, 2005), both phenomena have also been perceived as alternative forms of aggression within the frustration-regression model (Henry and Short, 1954; Unnithan and Whitt, 1992). Moreover, some prior research revealed that suicide rates are correlated with homicide rates (Kennedy et al., 1999), and that both homicide and suicide rates are influenced by some of the same predictors (Pridemore and Chamlin, 2006; Unnithan and Whitt, 1992).²

Unnithan et al. (1994) argued that homicide and suicide can be usefully conceptualized as two distinct channels of a single stream of lethal violence instigated by the same underlying causes. This phenomenon can be understood in terms of ‘force of production’ and ‘force of direction’. ‘Force of production’, the authors suggest, represents the social and cultural factors that generate lethal violence, and ‘force of direction’ refers to the social and cultural factors that determine the form of lethal violence. ‘Force of production’ is thus reflected in the overall lethal violence rates (LVRs; expressed as a sum of suicide and homicide rates), and ‘force of direction’ in the suicide–homicide ratio (SHR; expressed as suicide rates divided by the sum of suicide and homicide rates). For Unnithan et al. (1994), then, suicide and homicide represent different manifestations of lethal violence.

A fairly large body of work has utilized the stream analogy (Batton, 2004; Chon, 2013; He et al., 2003; Unnithan and Whitt, 1992). Chon (2013), in particular, conducted cross-national analyses of the impact of economic development on the respective components of lethal violence – homicide rates, suicide rates, LVRs, and the SHR. The author reported a negative effect of GDP per capita on homicide rates, a null effect on suicide rates, a null effect on LVRs, and a positive effect on the SHR. Chon (2013) concluded that residents in nations with higher levels of economic development are less likely to commit ‘outward’ violence but are more likely to commit inward-directed violence as opposed to violence directed toward others.

Chon’s findings are particularly relevant to the current analyses given that the Eastern European nations in the sample are not highly developed nations, at least with respect to their counterparts in Western Europe and other advanced nations. Accordingly, it is possible that the disruptive processes associated with rapid social change depicted in the Durkheimian framework might have differential influences on suicide and homicide within the Eastern European context. Specifically, the disruption and deregulation stimulated by entry into the EU might have a strong effect on homicide rates – an outward expression of the ‘force of production’ – but much less of an impact on suicide rates – the inward direction of lethal violence. This reasoning implies that the positive effect of EU entry on homicide rates that was reported in prior work might not be transferable to suicide rates – there might be a weak or null association. A companion hypothesis also follows from this line of reasoning grounded in the stream analogy. Insofar as disruption and/or deregulation generated by EU entry increases the ‘flow’ of lethal violence, and insofar as this ‘flow’ tends to be directed more toward outward than to internal manifestations of lethal violence for the nations in the sample, EU entry is expected to exhibit a negative association with the SHR.

Data and methods

Following Piatkowska et al. (2016), this study uses pooled, cross-sectional time-series data for the 10 Eastern European nations that joined the EU in 2004 and 2007.³ These nations are the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia, and Slovenia, all of which joined the EU in 2004, as well as Bulgaria and Romania, which joined in 2007. Data in the resulting sample are from a span of approximately 20 years (1990–2011).

The central dependent variable is country-specific and year-specific age-standardized suicide and self-inflicted injury death rates per 100,000 of the population for men and women combined and for each sex of all ages. The data for these measures were taken from the World Health Organization (WHO) European Health for All Database (HFA-DB), which is a collection of major health statistics for the 53 WHO member states in the European Region (WHO, 2015). In HFA-DB, WHO employs a direct method to calculate age-standardized death rates, including suicide and self-inflicted injury death rates. The suicide and self-inflicted injury death rates data were available for 212 country/year combinations for this study. Univariate analysis revealed that the suicide rate is positively skewed. Following the work by Chon (2013), we applied a log transformation to the suicide rate to reduce the skewedness of this variable. With a natural logarithm, the mean of the suicide rate was 2.97, with a standard deviation of 0.44.

Following prior work (Kõlves et al., 2013; Minagawa, 2013), we incorporated several independent variables that are hypothesized to be related to suicide rates. To assess the impact of economic conditions on suicide rates, we included the GDP per capita (calculated in thousands of US dollars for the year 2000) and unemployment rates into the analysis. The former measure was obtained from the World Bank Development Indicators database (The World Bank, 2013), whereas the latter was taken from the WHO (2010) database. We also included the Gini Index as a measure of economic inequality. This measure is often employed as a proxy for economic integration (Kõlves et al., 2013). The data on this variable were taken from the World Income Inequality Database, which is offered by the United Nations World University–World Institute for Development Economics Research (UNU-WIDER) (2008). In keeping with prior research (Stamatel, 2009), the Gini Indexes utilized herein were those reported by nationally representative household surveys. These data were missing in a very few instances: in those cases, we selected this variable by choosing the Gini Index that created the most consistent series within a country (Stamatel, 2009).

We also incorporated demographic measures that are hypothesized to be associated with suicide rates. These measures are the divorce rate, the crude birth rate, and life expectancy at birth. The data on divorce rates were taken from the United Nations (UN) (1990–2011) Demographic Yearbook (various years), whereas the data on life expectancy at birth and crude birth were gleaned from the WHO (2010) database. The divorce rate and the crude birth rate are often employed as proxy variables for social deregulation and integration of a society (e.g. Kõlves et al., 2013), and life expectancy is used as a proxy for general pathogenic social stress (Mäkinen, 2000). Prior research shows alcohol consumption also to be related to suicide rates (Brainerd, 2001; Pridemore, 2006), prompting the inclusion of pure alcohol consumption (liters per capita for ages 15 and more) from the WHO (2010) mortality database as an additional measure.

For the independent variables outlined above, there were occasional instances of missing values for one or more nation-years. These variables exhibited a significant linear time trend most of the time, and when this was the case, the missing values were imputed based on the nation-specific time-trend regression. In the rare instances where the variables did not show a linear time trend and thus imputation using this method was not appropriate, the missing values were replaced with the country’s mean. A total of 26 missing values were imputed (five values for GDP per capita, one for alcohol consumption, eight for unemployment rate, and four each for Gini, divorce rates, and life expectancy at birth). This resulted in an overall sample size for the analyses of 212 – the maximum number of observations with observed suicide data.

Following the research work by Chon (2013), we also included the SHR as our second outcome variable. Drawing upon the logic of the stream analogy, Chon (2013) computed the SHR by dividing suicide rates by the sum of suicide and homicide rates. To address the issue of non-normality, he first converted the suicide and homicide rates into the natural logarithms. Given that the homicide rate and the suicide rate were both positively skewed in this study, we also applied a log transformation to these variables. We then calculated the SHR by dividing logged suicide rates by the sum of logged suicide and logged homicide rates.

We also incorporated several indicators of structural conditions purportedly related to the SHR. Our selection of these variables was determined by Chon’s study and by the availability of the reliable cross-national time-series data sources for these variables.⁴ These measures are the percentage of the population in urban areas, the percentage of females in the total population, the percentage of people aged 65 years or more, and the percentage of the population aged 20–34 years. The data on the percentage of the population aged 20–34 years were obtained from the Eurostat database (Eurostat, 2016). The data source for the remaining measures is the World Bank Development Indicators database (The World Bank, 2013). In contrast to the independent variables discussed above, no missing values were reported for these variables.

As mentioned above, the model specification utilized in Chon’s (2013) study provided the rationale for our selection of the independent variables purportedly associated with the SHR. It should be noted, however, that the structural covariates included in the research work by Chon diverge somewhat from the covariates often incorporated for suicide rates serving as a dependent variable. Accordingly, our model of SHR includes somewhat different independent variables from those presented in our model of suicide rates. These independent variables also diverge from the covariates included in models utilized in Piatkowska et al. (2016). This reasoning is grounded in the notion that SHR is conceptually distinct from either homicide rates or suicide rates, even though it comprises homicide and suicide rates in the strict mathematical sense. SHR has also been interpreted theoretically as a variable that might have determinants that differ from those of either homicide or suicide. Some of the factors that affect the ‘direction’ of lethal violence, which would be reflected in the SHR, are theorized to be different from factors that affect the ‘flow’ of lethal violence, which relate to both homicide and suicide (Unnithan et al., 1994).

To test the alternative hypotheses outlined above, we employed fixed effects for nation and time. Country-specific fixed effects were included to control for unknown and/or unmeasured country characteristics that are constant over time. There may also be unknown and/or unmeasured over-time differences that impact all countries, however, and these were controlled for by way of the fixed effects for time. The test of equal years coefficients indicated that the hypothesis of equal coefficients should be rejected, both with EU dummy and structural covariates (χ²(19) = 68.69, p < 0.001) and without the covariates (χ²(20) = 79.90, p < 0.001) included in the models. A few dummies for years were significant, and taken together, were not equal to 0. This suggests that there are other macro-processes that affect change in suicide rates, motivating the inclusion of time fixed effects in the present models.

With panel data, it is critical to address nonstationarity because it may lead to faulty interpretations. The results of a Dickey-Fuller test for unit root revealed that suicide rates (for males and females combined and for each gender considered separately), GDP per capita, and life expectancy at birth were nonstationary. To account for this, we used a first difference for both the dependent and the independent variables. The EU dummy variable was also lagged 1 year because the effect of EU entry on suicide rates is not expected to take place immediately but rather to be observed in the next year and thereafter. The dummy variable indicating EU accession was coded as a dichotomous variable with a value of 1 for the year of accession and the years between then and the end of the time trend, and 0 otherwise. Finally, the results of the test for panel-level heteroscedasticity revealed that heteroscedasticity was present in the estimated models. To account for this, all models present heteroscedastic estimates of standard errors.⁵

We present two models for each of our dependent variables. The first model includes only the EU dummy variable and the nation fixed effects and the fixed effects for time; the second model introduces the control variables. In the standard causal modeling framework, control variables serve two distinct purposes – guarding against spurious associations and identifying mediating causal processes. Current theory and research do not provide unambiguous guidance about how EU entry might be related to the control variables. In some instances, the possibility of a mediating effect is plausible (e.g. disruption associated with EU entry could be manifested in increased alcohol consumption, which in turn is associated with higher suicide rates). In other instances, control variables are more readily conceptualized as potential confounding factors (e.g. life expectancy). We do not attempt to stipulate the causal role of control variables a priori, but we address this issue in the course of reporting the results.

Our estimated models are expressed in these two equations

Δ S_{i t} = β_{i, 0} + β_{t, 0} + γ E U_{i (t - m)} + ε_{i, t}

Δ S_{i t} = β_{i, 0} + β_{t, 0} + γ E U_{i (t - m)} + \sum_{k} β_{i, k} Δ X_{k, i, t} + ε_{i, t}

where S_it is the suicide rate for country i in year t, β_i,₀ is the fixed effect for country i, t is year, β_t,₀ is the fixed effect for year t, and EU_i_{(t − m)} is the EU dummy variable (1 = joined) for country in year t − m, where m is a lag from 1 to 3 years, and γ is the effect of EU entry. As noted below, we settled on a 1-year lag for the main analyses. Finally, X_k,_i is the kth independent variable for country i, β_i,_k is the coefficient on independent variable k for country i, and ε_i,t is stochastic error for country i in year t. The models outlined above were estimated with Generalized Least Squares in Stata 13.

As we discussed above, suicide rates and hypothesized covariates were all non-stationary and are therefore measured as first differences: ΔS_it = S_it − S_i,_t−1 and ΔX_it = X_it − X_i,_t−1. This is a change-change model. The rate (speed) of change in the suicide rate between consecutive years is affected by EU entry and by the rate (speed) of change in the independent variables between consecutive years. In addition, country fixed effects control for all between-country differences are constant over time, and year fixed effects control for all over-time differences are the same for all countries. Estimated parameters are, therefore, within-country effects that are assumed to be the same for all 10 countries that made the transition to the EU membership in 2004 and 2007.

Results

Table 1 displays descriptive statistics for the variables used in the study. The nations reported here have a mean suicide rate of 21.64 per 100,000 population, with a standard deviation of 10.12. The appreciable standard deviation reveals that there is a great deal of variation in national suicide rates. Of note is that the mean suicide rates for males are considerably higher than for females. For males, the mean suicide rate is 37.68 per 100,000 population, whereas for females, it is 7.92 per 100,000 population. These results are largely consistent with previous research (e.g. Kõlves et al., 2013).

Table 1.

Descriptive statistics for the pooled sample.

Variables	Mean	SD	Min.	Max.
Dependent variables
Suicide rates	21.64	10.12	8.58	49.14
Suicide rates – males	37.68	18.40	14.32	89.27
Suicide rates – females	7.92	3.84	1.89	18.74
Suicide–homicide ratio (SHR)	0.83	0.10	0.58	0.98
SHR – males	0.85	0.10	0.59	0.99
SHR – females	0.78	0.10	0.52	0.97
Independent variables
GDP per capita	4804.62	2604.93	1373.61	13,836.19
Gini	29.84	5.52	18	40.20
Divorce rate	2.21	0.92	0.72	5.53
Unemployment rate	9.94	4.30	0.1	19.60
Crude birth rate	10.30	1.49	7.64	15.38
Life expectancy at birth	72.69	2.63	65.49	80.57
Alcohol consumption (liters per capita)	10.72	2.28	3.62	16.15
Urban (%)	63.55	7.51	49.95	75.22
Age (20–34)	21.93	1.33	19.38	25.26
Age > 65	14.15	2.15	10.07	18.69
Female (%)	52.07	1.08	50.40	54.31

SD: standard deviation; SHR: suicide–homicide ratio; GDP: gross domestic product.

Table 1 also shows that countries greatly diverge with regard to predictor variables. For example, the mean GDP per capita in thousands of US dollars (stated for the year 2000) for all the nations included in the sample is 4804.62, with a standard deviation of 2604.93. Because this variable was positively skewed (1.11) with a kurtosis of 4.36, we applied a log transformation. With a natural logarithm, the mean GDP (ln) was 8.33, with a standard deviation of 0.55. GDP per capita was also highly correlated with life expectancy at birth (r = 0.795). While the log transformation of this measure did not substantially influence its effects on the suicide rates, it helped to reduce the issue of multicollinearity present in the model.⁶

Table 2 presents the average suicide rates and average suicide rates by gender in the countries under investigation. Consistent with prior work (Pray et al., 2013), we find that the three Baltic States of Estonia, Latvia, and Lithuania – as well as Hungary – have the highest suicide rates of all the selected countries, with rates ranging from an average of 39.03 per 100,000 population in Lithuania to 26.95 per 100,000 population in Estonia. Among the other countries reported herein, only Slovenia has an average of more than 20 suicides per 100,000 population. The results in Table 2 also reveal that males have considerably higher suicide rates than females in all the countries investigated here, which is largely in line with previous research on suicide rates in Eastern Europe (Kõlves et al., 2013; Pray et al., 2013).

Table 2.

Average suicide rates and suicide rates by gender and by country, 1990–2011.

Country	Suicide rates				Suicide rates – males				Suicide rates – females
Country	Mean	SD	Min.	Max.	Mean	SD	Min.	Max.	Mean	SD	Min.	Max.
Bulgaria	13.18	2.80	8.58	16.65	20.78	3.92	14.55	26.26	6.51	1.86	3.13	9.09
Czech Rep.	14.86	2.22	11.79	19.07	24.99	3.04	20.2	30.87	5.94	1.88	3.61	9.6
Estonia	26.95	8.99	14.29	41.7	48.55	16.17	26.65	75.46	9.32	3.52	4.46	14.79
Hungary	27.79	5.54	21.25	38.13	46.57	8.28	35.78	61.31	12.00	3.25	8.47	18.74
Latvia	28.22	7.97	17.53	42.58	51.13	14.04	32.6	76.09	9.52	3.23	4.01	15.44
Lithuania	39.03	7.62	27.16	49.14	70.79	13.95	48.39	89.27	12.18	2.47	8.25	16.64
Poland	14.72	0.65	12.94	15.78	26.06	1.21	23.04	28.26	4.34	0.46	3.4	5.12
Romania	11.89	1.09	9.37	13.65	20.31	2.22	14.32	23.69	4.12	0.52	3.17	4.97
Slovakia	13.07	1.74	10.21	16.56	23.47	2.85	18.53	28.66	3.84	0.92	1.89	5.88
Slovenia	25.39	4.57	17.17	32.26	42.11	7.79	28.75	54.43	10.68	2.34	6.09	13.86

SD: standard deviation.

The trends in suicide rates in the 10 Eastern European nations during the period under investigation are displayed in Figure 1 (see also Appendix 1). While suicide rates for all the nations represented in the sample slope downward over time, there are substantial differences among them in terms of both trajectory and magnitude. This is largely consistent with previous research (Kõlves et al., 2013; Mäkinen, 2000; Pray et al., 2013). Moreover, suicide rate trends in the nations investigated in the present research appear to have different functional forms: while some are approximately linear, others are not. Finally, the test of equal trend coefficients across nations indicated that the hypothesis of equal coefficients should be rejected (χ²(9) = 746.09, p < 0.001). Given these differences, modeling a single linear trend across the entire sample for this study would be inappropriate, as would specifying any particular trajectory, whether linear or not. Instead, we regarded these patterns of change over time as a consequence of more macro-level historical processes that affected all countries in the region. We therefore controlled for these complex macro-trends as time fixed effects which can include trajectories of virtually any form. As noted above, the data presented in Figure 1 show declining suicide rates for the majority of nations under investigation. These patterns, however, do not control for hypothesized covariates. It is possible that the downward trends in suicide rates might have been more pronounced than those presented in Figure 1, if entry into the EU had not taken place in the nations under investigation.

Figure 1.

Suicide rates.

Table 3 presents the results of regressions of the suicide rates on EU membership and the predictor variables included in this study. These regressions were estimated with nation and time entered as fixed effects. Models 1, 3, and 5 report only the EU dummy variable. The results reveal that EU accession exerts no overall effect for males and females combined and for males considered separately, but reaches borderline significance for suicide rates among females (Model 6: b = −0.081, p = 0.074). This finding is suggestive of gender differences in suicide rates. This finding also lends credibility to the hypothesis that any disruption and deregulation generated by EU entry had no effect on suicide rates – the inward direction of lethal violence.

Table 3.

Regressions of suicide rates on the EU and predictor variables.

	Suicide – total		Suicide – males		Suicide – females
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
EU	−0.008 (0.029)	−0.027 (0.027)	0.012 (0.032)	−0.012 (0.030)	−0.081^† (0.045)	−0.087* (0.044)
GDP per capita (ln)		−0.308** (0.098)		−0.302** (0.105)		−0.451* (0.176)
Gini		0.001 (0.002)		0.001 (0.002)		0.002 (0.003)
Divorce rate		0.018 (0.016)		0.006 (0.017)		0.081** (0.027)
Unemployment rate (×100)		0.003 (0.203)		−0.048 (0.226)		0.179 (0.335)
Crude birth rate		−0.015 (0.012)		−0.011 (0.013)		−0.019 (0.021)
Life expectancy		−0.058*** (0.010)		−066*** (0.011)		−0.023 (0.017)
Alcohol consumption (liters per capita)		−0.002 (0.005)		0.002 (0.005)		−0.004 (0.008)

EU: European Union; GDP: gross domestic product.

All models include nation fixed effects and fixed effects for time.

†

p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

Models 2, 4, and 6 in Table 3 add the predictor variables that are hypothesized to be related with suicide rates for both females and males combined and for each gender. For total suicide rates, the results reveal two noteworthy effects – a significant negative effect of logged GDP per capita and a significant negative effect of life expectancy at birth. These results are largely in line with prior research (Brainerd, 2001; Kõlves et al., 2013; Minagawa, 2013) and reaffirm earlier findings from cross-national research on suicide in Eastern European settings.⁷

The results in Table 3 also show that GDP per capita exerts significant effects on suicide rates for males and females considered separately (Models 4 and 6). The regression coefficient for each is significant and negative in sign, which accords with the findings for total suicide rates. In contrast, Models 4 and 6 report noteworthy differences in suicide rates between females and males. Specifically, life expectancy at birth has a significant and negative effect on suicide rates (Model 4: b = −0.066, p < 0.001) only for males, whereas the divorce rate exerts a significant and positive effect on suicide rates only for females (Model 6: b = 0.081, p < 0.01). These findings are consistent with a fairly large body of work that registers notable differences in suicide rates based on gender (Brainerd, 2001; Chandler and Tsai, 1993; Messner et al., 2006). Recall also that Durkheim (1951 [1897]) pointed out a gender gap in suicide rates, with males exhibiting higher levels of suicide than females. This study demonstrates the merit of this body of work by establishing differential suicide rates based on gender in Eastern European settings.

Models 2 and 4 also show that the EU dummy variable is nonsignificant for total and male suicide rates, but is significant for suicide rates among females (Model 6: b = −0.087, p < 0.05). Comparisons across Models 1 and 2 for all three of the dependent reveal very little impact of introducing the control variables on the associations with EU entry. Substantive conclusions are essentially unchanged. Accordingly, the null effects for EU entry on total and male suicide rates cannot be attributed to any suppressor effect induced by the control variables in the model. Similarly, the control variables have little impact on the observed inverse association between EU entry and female suicide rates, serving neither as confounding variables nor mediating variables. These results imply that insofar as the changes associated with accession to the EU exert a genuine causal effect on suicide rates for females, the structural covariates commonly included in the suicide research do not account for this effect.

We considered the possibility that there might be ‘anticipatory’ disruptive effects of EU accession. Entry into the EU requires that nations set into motion specific policies that might entail the kinds of disruptions that would promote suicide from a Durkheimian perspective, such as free movement of capital, freedom of movement of workers, and policies in an area of freedom, justice, and security (Strategy Paper, 2001). To assess this possibility, we re-estimated models with the EU dummy variable lagged 2 and 3 years. The results of these analyses replicate the null effects of EU entry on total and male suicide rates reported in Table 3. In contrast, the effect of EU entry on suicide rates among females is not significant with the 2- or 3-year lagged measures.

Table 4 displays the results of regressions of the SHR on EU membership and the predictor variables included in this study. All regression analyses utilized the model specification with nation and time entered as fixed effects. Accordingly, Models 1, 3, and 5 report only the EU dummy variable, whereas Models 2, 4, and 6 add the predictor variables purportedly related with the SHR for both females and males combined and for each gender. The results in Model 1, 3, and 5 reveal that EU entry exhibits a significant and negative effect for males and females combined (Model 1: b = −0.031, p < 0.05) and for females considered separately (Model 5: b = −0.098, p < 0.05). The EU dummy variable is nonsignificant for males, which suggests gender differences in the SHR that are comparable to suicide rates.

Table 4.

Regressions of the suicide–homicide ratio (SHR) on the EU and predictor variables.

	SHR – total		SHR – males		SHR – females
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
EU	−0.031* (0.015)	−0.031* (0.015)	−0.020 (0.014)	−0.019 (0.015)	−0.098* (0.044)	−0.112* (0.046)
GDP per capita (ln)		−0.001 (0.047)		−0.012 (0.043)		−0.009 (0.150)
Gini		0.002 (0.001)		−0.001 (0.001)		0.001 (0.003)
Urban (%)		−0.006 (0.020)		−0.004 (0.019)		0.040 (0.055)
Age (20–34 years)		0.001 (0.009)		−0.002 (0.008)		0.036 (0.026)
Age > 65 years		0.014 (0.033)		0.014 (0.030)		0.009 (0.115)
Female (%)		−0.067 (0.088)		−0.079 (0.087)		−0.112 (0.249)

EU: European Union; GDP: gross domestic product.

All models include nation fixed effects and fixed effects for time.

†

p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

Models 2 and 6 also show that EU has a significant and negative effect on SHR for males and females combined (Model 2: b = −0.031, p < 0.05) and on the SHR among females (Model 6: b = −0.112, p < 0.05). This finding lends support for the hypothesis that the primary impact of any increase in the ‘flow’ of violence associated with EU entry is likely to be manifested in an ‘outward’ direction for the nations under investigation. Notably, the EU dummy variable remains nonsignificant for males, which again implies gender differences in the SHR.

Summary and conclusion

In an attempt to better understand the possible connection between EU accession and changes in suicide rates, this study examined suicide rates in 10 Eastern European nations. We also investigated the effect of EU entry on the SHR for the nations in the sample. Drawing upon prior literature, the Durkheimian framework, and the stream analogy, we proposed alternative hypotheses about the impact of entry into the EU on suicide rates in the 10 Eastern European countries that joined the EU in 2004 and 2007. In one hypothesis, suicide rates were expected to increase in the years following EU accession due to rapid change and attendant social deregulation. In the second, EU entry was expected to have a weak or null effect on suicide rates, if any increased ‘flow’ of disruptive forces associated with EU entry were manifested ‘outwardly’ rather than ‘inwardly’. This latter scenario also implies that entry into the EU would have a negative effect on the SHR.

Our analysis reveals that EU accession had no overall effect on total suicide rates and suicide rates among males in the Eastern European countries investigated. In the models with fixed time and nation effects, the dummy variable for EU membership is nonsignificant, with and without the control variables included in the models. This indicates that EU accession did not exacerbate suicide rates in new member states, as suggested by the Durkheimian (1951 [1897]) hypothesis. The results reveal, however, that the EU dummy variable has a negative effect of borderline significance on female suicide rates in the model with the EU variable alone, and a significant negative effect with the predictor variables included in the model. This finding reaffirms previous research pointing to gender differences in the correlates of suicide rates.

Our analysis also reveals that EU accession had a negative effect on the SHR for the total population and for females. These findings are consistent with the view that disruption and/or deregulation generated by EU entry increased the ‘flow’ of lethal violence, but this ‘flow’ has been directed more toward outward than to inward manifestations of lethal violence in the nations under investigation. Given the prior work of Chon (2013) reviewed above, one plausible explanation for the association between EU entry and the SHR is that Eastern European nations are less economically developed than their Western counterparts, and so entry into the EU increased homicidal aggression as opposed to aggression directed at oneself.

Turning to the structural covariates, our analyses reaffirm robust findings in the cross-national research on suicide rates based on other samples (Fernquist and Cutright, 1998; Neumayer, 2003). Specifically, we find that GDP per capita and life expectancy at birth have negative effects on total suicide rates. These results lend credibility for claims made by many scholars who argue that suicide rates are strongly influenced by socioeconomic factors (Brainerd, 2001; Mäkinen, 2000; Minagawa, 2013) and reaffirm the role of such predictors of suicide in Eastern European countries. At the same time, these covariates play no role in interpreting any connection between EU entry and suicide rates. Models with and without controls yield similar coefficients for the EU variable.⁸

Our results also show differences in the structural covariates of suicide rates based on gender. We find that the divorce rate has an impact on suicide rates for females, but not for males, while the opposite is true for life expectancy at birth. This again is largely in accord with previous research, which indicated the importance of gender in driving differences in suicide rates (Chandler and Tsai, 1993; Messner et al., 2006).

Our finding of a negative effect of EU accession on the SHR is in line with the findings presented in Piatkowska et al. (2016). Recall that they reported a positive association between EU accession and homicide rates in 10 Eastern European nations that joined the EU in 2004 and 2007. Increased homicide rates, combined with no increase in suicide rates, imply a reduced SHR. At the same time, our analyses reveal similar effects on both forms of lethal violence for measures of economic growth (GDP per capita). These findings reaffirm Bills and Li’s (2005) conclusion based on their cross-national analysis that there is likely to be ‘some kind of common link or connection between homicide and suicide …’ (p. 844).

We acknowledge that this study is not without limitations. Following prior work (Kõlves et al., 2013; Minagawa, 2013), we included a limited number of predictor variables identified to be associated with suicide. It is certainly possible that some salient predictors have been excluded. Moreover, this study examined the effect of EU entry on suicide in the nations at large. Eastern European nations are not homogeneous (e.g. consider the large Russian minorities in Estonia and Latvia). Note also that due to data constraints, we imputed some values for independent variables included in this study. Furthermore, our analyses were constrained by a relatively short time period following EU enlargement. Societies can be highly adaptable, and the longer term consequences of entry into the EU might differ appreciably from those detected in the time frame under investigation.

With these caveats in mind, our analyses suggest some promising avenues for future research. A number of studies examined gender differences in suicide rates in Western countries (Krull and Trovato, 1994; Messner et al., 2006; Yang, 1992). However, little is known about the gender gap in suicide rates in Eastern European countries. It would therefore be informative to investigate whether the findings from Western nations hold true in Eastern European settings. The finding of a negative impact of EU entry on suicide rates and the SHR only for females is of special interest here because it suggests that the influence of EU membership in a society may be gender-driven in particular settings. An important task for further inquiry is to identify the features in the socio-historical context that can account for these gender differences. The precise nature of these processes needs to be theorized more fully and documented empirically in future research.

Another fruitful avenue for future studies is to explore the specific mechanisms that underlie the effect of EU accession on the various forms of lethal violence. What is it about EU accession that leads to increased homicidal aggression in Eastern European nations as opposed to aggression directed at oneself? A host of changes accompanied entry into the EU, and these may have rather complicated and subtle relationships with the manifestations of lethal violence. For example, antidepressant medications are more widely available in the EU market. It may be the case that the use of antidepressants as a means of coping has more of an inhibiting effect on inward rather than outward expressions of aggression, thereby increasing the ‘flow’ of violent impulses to homicide rather than suicide.

An additional important task for future research is to account for the impact of social inclusion/exclusion indicators on the relationship between EU accession and the homicide–suicide ratio. Recent research suggests that EU accession may have an effect on factors such as social welfare and social expenditures. Sevinç and Civan (2013), for example, found that EU accession had a positive impact on longevity and welfare growth rates, while Innamorati et al. (2010) found suicide rates for senior citizens to be negatively correlated with GDP per capita, urbanism, and public expenditure on health in the EU. Ferretti and Coluccia (2009), meanwhile, reported that suicide levels are high in EU nations that have high levels of at-risk-of-poverty rates and low healthcare expenditures. There is reason to consider, then, that EU accession influenced the levels of social inclusion/exclusion in Eastern European nations in complex ways, leading to increased outward violence as opposed to violence directed at oneself.

Another important task for future research is to account for the political predictors of suicide. Durkheim (1951 [1897]) argued that war and political crises reduce the level of suicide because they increase the level of social integration and collective sentiment in a society, whereas some studies showed that political factors such as war, political elections, and political movements reduce the level of suicide, primarily by means of reducing unemployment rates (see Stack, 2000a for a more detailed review of these studies). As Stack (2000b) pointed out, ‘very powerful social movements emanated from mass unemployment, and the movements for unionization and liberal politics may have channeled aggression from inner-directed to other-directed expressions’ (p. 150). Thus, it may be the case that EU accession, as a politically charged event, influenced the other-directed violence as opposed to violence directed at oneself. This would be consistent with Piatkowska et al.’s (2016) suggestion that an increase in homicide rates following EU accession could be triggered by the processes of democratization, which lead to social disruption and disorganization (see Karstedt, 2008; Karstedt and LaFree, 2006).

Finally, future studies could investigate the moderating effect of the availability of firearms on the relationship between EU accession and the homicide–suicide ratio. As Piatkowska et al. (2016) noted, ‘a growing body of work suggests that there has been an escalation of organized crime in the EU, and that escalation has been aided by the most recent EU enlargement’ (p. 159). The availability of firearms may have accompanied this escalation of organized crime, thereby increasing the levels of violence and especially homicidal violence.

Returning to our overarching objectives, the results of our analyses indicate that entry to the EU evidently has not led to increases in suicide rates for the Eastern European nations at large, and it has been associated with a reduction in suicide rates for females. EU entry has also been associated with a decrease in the ratio of suicides to the combined level of lethal violence (suicides plus homicides) for the population at large and for females. We encourage researchers to probe further into the consequences of the social changes induced by accession into the EU, not only for forms of lethal violence such as suicide and homicide but for other indicators of the quality of life as well.

Footnotes

Appendix

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Andresen

(2009) Crime in Lithuania: The impact of accession to the European Union. European Journal of Criminology 6(4): 337–369.

Andresen

(2011) The impact of accession to the European Union on violent crime in Lithuania. European Sociological Review 27(6): 759–771.

Batton

(2004) Gender differences in lethal violence: Historical trends in the relationship between homicide and suicide rates, 1960–2000. Justice Quarterly 21(3): 423–461.

Bills

(2005) Correlating homicide and suicide. International Journal of Epidemiology 34(4): 837–845.

Brainerd

(2001) Economic reform and mortality in the former Soviet Union: A study of the suicide epidemic in the 1990s. European Economic Review 45(4): 1007–1019.

Chandler

Tsai

(1993) Suicide in Japan and in the West: Evidence for Durkheim’s theory. International Journal of Comparative Sociology 94(3–4): 95–120.

Chon

(2013) Economic development, change of age distribution, and stream analogy of homicide and suicide: A cross-national assessment. Justice Quarterly 30(1): 169–193.

Curzio

Fortis

(eds) (2008) The EU and the Economies of the Eastern European Enlargement. Berlin: Physica-Verlag.

Durkheim

(1951 [1897]) Suicide: A Study in Sociology (trans. Spaulding

Simpson

). New York: The Free Press.

10.

European Commission (2009) Five years of an enlarged EU – Economic achievements and challenges. European Economy 1. Available at: http://ec.europa.eu/economy_finance (accessed 2 March 2014).

11.

European Commission (2016) Economic accession criteria. Available at: http://ec.europa.eu/economy_finance/international/enlargement/criteria/index_en.htm (accessed 6 August 2016).

12.

Eurostat (2016) Population on 1 January by age and sex [demo_pjan]. Available at: http://ec.europa.eu/eurostat/data/database (accessed 20 May 2016).

13.

Fernquist

Cutright

(1998) Societal integration and age standardized suicide rates in 21 developed countries, 1955–1989. Social Science Research 27(2): 109–127.

14.

Ferretti

Coluccia

(2009) Socio-economic factors and suicide rates in European Union countries. Legal Medicine 11: 92–94.

15.

Gelman

Hill

Yajima

(2012) Why we (usually) don’t have to worry about multiple comparisons. Journal of Research on Educational Effectiveness 5(2): 189–211.

16.

Harfst

(2000) Government stability in Central and Eastern Europe: The impact of parliaments and parties. Paper prepared for the ECPR joint session of workshops, Copenhagen, 14–19 April.

17.

Cao

Wells

. (2003) Force of production and direction: A test of an expanded model of suicide and homicide. Homicide Studies 7(1): 36–57.

18.

Henry

Short

(1954) Suicide and Homicide. Glencoe: Free Press.

19.

Innamorati

Tamburello

Lester

. (2010) Inequalities in suicide rates in the European Union’s elderly: Trends and impact of macro-socioeconomic factors between 1980 and 2006. Canadian Journal of Psychiatry 55(4): 229–238.

20.

Karstedt

(2008) Democratization and violence: European and international perspectives. In: Body-Gendrot

Spierenburg

(eds) Violence in Europe. New York: Springer, pp. 205–225.

21.

Karstedt

LaFree

(2006) Democracy, crime, and justice. Annals of the American Academy of Political and Social Science 605(1): 6–23.

22.

Kennedy

Iveson

Hill

(1999) Violence, homicide and suicide: Strong correlation and wide variation across districts. The British Journal of Psychiatry 175(5): 462–466.

23.

Kõlves

Milner

Vӓrnik

(2013) Suicide rates and socioeconomic factors in Eastern European countries after the collapse of the Soviet Union: Trends between 1990 and 2008. Sociology of Health & Illness 35(6): 956–970.

24.

Krull

Trovato

(1994) Quiet revolution and the sex differential in Quebec’s suicide rates: 1931–1986. Social Forces 72(4): 1121–1147.

25.

Leenaars

Yang

Lester

(1993) Effect of domestic and economic stress on suicide sates in Canada and the United States. Journal of Clinical Psychology 49(6): 918–921.

26.

Lehmann

(1995) Durkheim’s theories of deviance and suicide: A feminist reconsideration. American Journal of Sociology 100(4): 904–930.

27.

Lester

Yang

(1992) Fertility and suicide rates: A time series analysis in the United States. Journal of Biosocial Science 24(1): 97–101.

28.

Mäkinen

(2000) Eastern European transition and suicide mortality. Social Science & Medicine 51(9): 1405–1420.

29.

Messner

Bjarnason

Raffalovich

. (2006) Nonmarital fertility and the effects of divorce rates on youth suicide rates. Journal of Marriage and Family 68(4): 1105–1111.

30.

Minagawa

(2013) The social consequences of postcommunist structural change: An analysis of suicide trends in Eastern Europe. Social Forces 91(3): 1035–1056.

31.

Möller-Leimkühler

(2003) The gender gap in suicide and premature death or: Why are men so vulnerable? European Archives of Psychiatry and Clinical Neurosis 253(1): 1–8.

32.

Neumayer

(2003) Socioeconomic factors and suicide rates at large-unit aggregate levels: A comment. Urban Studies 40(13): 2769–2776.

33.

Piatkowska

Messner

Raffalovich

(2016) The impact of accession to the European Union on homicide rates in Eastern Europe. European Sociological Review 32(1): 151–161.

34.

Pray

Cohen

Mäkinen

. (2013) Suicide in Eastern Europe, the CIS, and the Baltic countries: Social and public health determinants. In: A foundation for designing interventions summary of a conference, Available at: http://www.suicidology.ee/public/files/Suicides%20in%20Eastern%20Europe%20RR-13-001-web.pdf

35.

Pridemore

(2006) Heavy drinking and suicide Russia. Social Forces 85(1): 413–430.

36.

Pridemore

Chamlin

(2006) A time-series analysis of the impact of heavy drinking on homicide and suicide mortality in Russia, 1956–2002. Addiction 101(12): 1719–1729.

37.

Pridemore

Chamlin

Cochran

(2007) An interrupted time-series analysis of Durkheim’s social deregulation thesis: The case of the Russian Federation. Justice Quarterly 24(2): 271–290.

38.

Sevinç

Civan

(2013) The effect of European Union membership on welfare. Review of European Studies 5(1): 65–81.

39.

Šlosarčík

(2011) The Czech Republic – Impacts of and experience with EU membership. Eastern Journal of European Studies 2(2): 21–30.

40.

Stack

(2000a) Suicide: A 15-Year review of the sociological literature part I: Cultural and economic factors. Suicide and Life-Threatening Behavior 30(2): 145–162.

41.

Stack

(2000b) Suicide: A 15-Year review of the sociological literature part II: Modernization and social integration perspectives. Suicide and Life-Threatening Behavior 30(2): 163–176.

42.

Stamatel

(2009) Correlates of national-level homicide variation in post-communist East-Central Europe. Social Forces 87(3): 1423–1448.

43.

Strategy Paper (2001) Making a success of enlargement. Report of the European Commission on the progress towards accession by each of the candidate countries. Available at: http://ec.europa.eu/enlargement/archives/key_documents/reports_2001_en.htm (assessed 22 September 2015).

44.

The World Bank (2013) World Development Indicators. Available at: http://data.worldbank.org/data-catalog/world-development-indicators (assessed 4 June 2013).

45.

United Nations (UN) (1990–2011) Demographic yearbook. United Nations Statistics Division. New York: UN. Available at: http://unstats.un.org/unsd/demographic/products/dyb/dyb2.htm

46.

United Nations World University–World Institute for Development Economics Research (UNU-WIDER) (2008) World Income Inequality Database. Version 2.0c, UNU-WIDER. Available at: https://www.wider.unu.edu/download/wiid-v30a (assessed 4 June 2013).

47.

Unnithan

Whitt

(1992) Inequality, economic development and lethal violence: A cross-national analysis of suicide and homicide. International Journal of Comparative Sociology 33(3–4): 182–196.

48.

Unnithan

Huff-Corzine

Corzine

. (1994) The Currents of Lethal Violence: An Integrated Model of Suicide and Homicide. Albany, NY: State University of New York Press.

49.

World Health Organization (WHO) (2010) European health for all database. WHO Regional Office for Europe, Copenhagen. Available at: http://www.euro.who.int/en/data-and-evidence/archive/european-health-for-all-database-hfa-db (assessed 4 June 2013).

50.

World Health Organization (WHO) (2015) European health for all database. WHO Regional Office for Europe, Copenhagen. Available at: http://www.euro.who.int/en/data-and-evidence/databases/european-health-for-all-database-hfa-db (assessed 22 September 2015).

51.

Yang

(1992) The economy and suicide: A time series study of the U.S.A. American Journal of Economics and Sociology 51(1): 87–99.

52.

Yorgova

(2011) Impacts of and experience with EU membership of the countries of Central, Eastern and South-Eastern Europe. Country study on Bulgaria. Eastern Journal of European Studies 2(2): 169–188.