Abstract Using a random sample of 1,435 Ukrainian and Russian respondents, this study integrates the predictions of Agnew’s macro-level strain theory (MST) and general strain theory (GST). Specifically, it seeks to identify possible interactive effects of context—community-level strain, negative affect and religiosity—and individual strain-related variables on personal criminal involvement and depressive symptoms. Findings provide evidence of individual-level processes described by GST, revealing a relationship between personal strain and both criminal involvement and depression. However, community-level strains, anger and religiosity appear unrelated to individual behaviour, whether as direct predictors of crime or as moderators of the strain–crime relationship. The only statistically significant contextual effect uncovered by the study is the association between community disorder and depression. These findings highlight areas in need for further refinement in GST and MST, and they offer several insights into the cultural limitations of a different theoretical framework, the concentrated disadvantage paradigm. Originally, an individual-level explanation of criminal behaviour, general strain theory (GST) suggests that individuals unable to achieve desired goals or have experienced personal loss, negative treatment or other adverse events are likely to respond with anger and related emotions that elevate the probability of criminal coping (Agnew 1992; 2001; 2006; 2013). Over the years, the causal process articulated in GST has become complex due to the addition of multiple contingencies proposed to affect the linkages between strain, negative emotions and crime (e.g. Agnew 2001; 2006). This might explain why Agnew (1999) responded to accumulating theoretical thought and evidence linking social context to crime (e.g. Sampson et al. 1997; Sampson and Bartusch 1998) by advancing a macro-level strain theory (MST) rather than incorporating contextual influences into the arguments of GST. According to MST, disadvantaged neighbourhoods tend to have elevated crime rates because persons who experience significant strain are more likely to reside in disadvantaged neighbourhoods than in affluent ones, because neighbourhood disadvantage itself produces strain, and because the risk of exposure to strained individuals is highest in disadvantaged neighbourhoods. Although MST has been subjected to some empirical testing (e.g. Brezina et al. 2001; Warner and Fowler 2003), researchers have yet to draw explicit connections between the mechanisms described by GST and relevant neighbourhood characteristics and processes (but see Brezina et al. 2001; Op de Beeck et al. 2012). The current study uses data from two non-western sites, Lviv, Ukraine, and Nizhni Novgorod, Russia, to address this gap in the literature. Specifically, we aim to investigate the association between neighbourhood characteristics, collective strain/anger, individual strain/anger and individual criminal involvement and depression. We also examine the conditioning influences of neighbourhood-level strain-relevant processes and religiosity on associations between individual-level strain experiences/negative affect and criminal involvement. We argue that socio-economic conditions prevailing in Russia and Ukraine provide ideal contexts for detecting strain effects at macro and micro levels. Using these locales also helps us gauge the cultural generality of theorized relationships between neighbourhood processes, individual-level situational inducements and criminal behaviour. The empirical status of GST and MST GST proposes that adverse treatment and conditions (strain) contribute to criminal behaviour by increasing anger and other negative emotions such as depression (Agnew 1992; 2001; 2006). Over time, GST has integrated multiple elements and conditions, including characteristics of strain as well as individual qualities, perceptions and attitudes (e.g. religiosity, prior experience with strains, negative emotionality/low constraint) that elevate or attenuate the probability a particular person would perceive an event as stressful and respond with crime (Agnew 2001; 2006; 2013). Similar to other criminologists using the social disorganization paradigm (Shaw and McKay 1942; Sampson 2012), in MST, Agnew links neighbourhood processes to cross-neighbourhood variations in crime rates. However, unlike other perspectives that implicate absent community controls, persistent legal cynicism and trivialization of violence as mechanisms translating disadvantage into crime (e.g. Sampson and Wilson 1995; Sampson et al. 1997; Sampson and Bartusch 1998; see also Sampson 2012), MST claims collective strains increase crime in a given area (Agnew 1999; 2006). According to the theory, absolute and relative deprivation plaguing disorganized communities translates into multiple forms of strains, including those directly experienced by individuals (e.g. trouble finding employment or criminal victimization) and vicarious strains that influence their emotional state (e.g. witnessing crime in the community). Extreme levels of chronic strain exposure, combined with retention of socio-economically deprived and strained individuals, produce a significant number of people experiencing negative affect and motivated to engage in crime. In turn, interactions with other stressed residents in these often-overcrowded areas contribute to crime by increasing collective sentiments of anger and frustration. In sum, MST proposes that high residential mobility, absolute deprivation and other structural characteristics of disorganized communities indirectly affect neighbourhood crime rates via personal and vicarious collective strains and associated negative affect. At the same time, MST acknowledges that many contextual characteristics and situations associated with strain may also operate through social control and social learning mechanisms. Therefore, strain/negative affect is expected to mediate only part of the relationship between neighbourhood problems and crime. Overall, GST has accumulated a respectable body of supporting evidence. In the United States, straining experiences have been found to lead to anger and other negative emotions, and strain and negative emotions both have been found to predict various types of deviance, including self-harming behaviours, violence, property crime and general delinquency (e.g. Piquero and Sealock 2000; Hay and Meldrum 2010; Rebellon et al. 2012). Moderate associations between strains/negative affect and criminal coping also have been reported in studies conducted in Asia and some European countries (e.g. Bao et al. 2004; Froggio and Agnew 2007; Botchkovar et al. 2009; Cheung and Cheung 2010; Horton et al. 2012; Moon et al. 2012; Sigfusdottir et al. 2012; Botchkovar and Broidy 2013; Bao et al. 2014; Lin et al. 2014). By contrast, research incorporating key premises of MST has been scarce and predominantly US-based (but see Op de Beeck et al. 2012), drawing disproportionately on youth samples. An exception is provided by Warner and Fowler (2003), who found that community disadvantage amplified the positive effect of community strain on violence across 66 census blocks in a southern state. Applying MST/GST in a school context, Brezina and colleagues (2001) found evidence that collective, school-level, anger increases the likelihood of peer conflicts—but not aggressive behaviour—at both the school and individual levels. At the individual level, however, anger predicted misbehaviour. Another school-based study (Op de Beeck et al. 2012) reported mixed support for MST, finding aggregate negative affect and negative perceptions of the future to be important predictors of involvement in general delinquency and violent behaviour. Similar to results reported by Brezina et al., contextual- and individual-level strain/negative emotions were found to operate independently of one another. Finally, probing cross-level interactions between individual-level strains and community characteristics, Wareham et al. (2005) and Hoffmann (2003) reached discordant conclusions. Whereas Hoffmann found the association between strain and crime to be stronger in areas with high levels of male unemployment, Wareham et al. reported contextual effects to be immaterial for GST processes. Although criminal involvement is a possible outcome of stressful experiences, Agnew (1992; 2006) acknowledges that strains may also result in depression as final outcome. Whereas some individual-level studies explore strain and depression within the GST framework (e.g. Peck 2013), and multiple community-level studies have confirmed the link between neighbourhood disadvantage/disorder and mental health outcomes such as depression (e.g. Ross 2000; Latkin and Curry 2003; Haines et al. 2011), so far, no research has investigated strain and depression using the framework of MST. In sum, scarce research, limited applications and mixed findings make it difficult to ascertain the empirical status of MST and its connection to GST. Therefore, an integrative study outlining the relationship between MST and GST and investigating it empirically in a neighbourhood context is needed. Toward integrating GST and MST: individual strain in community context The strains faced by individuals living in disorganized communities are multiple and variable, including those leading to residency in these neighbourhoods as well as those accumulating as a result of living in impoverished areas. Earlier ethnographic work has documented common struggles of underprivileged youth in disadvantaged communities whose life prospects are predetermined by their social status (e.g. Willis 1977; MacLeod 2004). Agnew (1999) lists a number of conditions as sources of additional strain, such as neighbourhood incivilities and vicarious exposure to strain, suggesting that various manifestations of social disorder will be straining for community residents facing them on a daily basis. Community strains and community-level negative emotions generated by these strains are expected to influence individual behaviour both directly, net of personal strains and the negative affect they produce, and as contingencies that sensitize individuals to their other strains. For instance, frequent contact with other strained and angry individuals may be a source of strain for many, and, consistent with the aggression displacement perspective (Dollard et al. 1939), interactions of frustrated residents with other angry people in the same neighbourhood may provoke angry emotional reactions to otherwise mundane sources of strain as well as increase the likelihood of responding to them with crime (see Anderson 1999 for a similar argument). In addition to their anger-generating and strain-sensitizing effects, contextual processes often tax individual coping resources in more subtle ways. For instance, high incidence of neighbourhood crime may strain individuals who directly experience or witness it, as well as indicate scarce social support available to the residents, an abundance of opportunities for criminal behaviour and the presence of crime-oriented groups and values conducive to crime—all important determinants of coping choices. Thus, exposure to neighbourhood strains and collective emotions generated by these strains potentially exacerbate the effects of personal strains/negative emotions on crime by generating negative affect, intensifying negative emotional reactions to unrelated strains and exhausting personal coping strategies. Although the range of neighbourhood conditions discussed in Agnew’s version of MST is limited to those that induce or exacerbate strains, it is well known that the majority of residents of disorganized communities refrain from crime despite inducements (e.g. Anderson 1999; Jones 2010). Accumulated research suggests that religiosity may prevent people from engaging in criminal behaviour, playing a role in shaping individual beliefs, helping individuals evaluate particular events as non-stressful, buffering against strains and being an important coping strategy for the strained (e.g. Tittle and Welch 1983; Junger and Polder 1993; Baier and Wright 2001; Jang and Johnson 2005; Johnson and Morris 2008). In addition, some literature suggests that religiosity may be an important contextual mechanism of informal control eliciting compliance from community residents (Stark et al. 1982; Ulmer and Harris 2013). Even in disadvantaged communities, strong religious beliefs shared among neighbourhood residents and communicated to each other at social gatherings or in church might generate conformity by reaffirming moral order, delivering threats of sanctions (imaginary or real) to those contemplating criminal response or offering a strong network of support to those seeking solutions to personal problems. Individual strain in community context in Russia and Ukraine The strain-laden conditions of life in Ukraine and Russia appear particularly appropriate for testing the key premises of MST and GST. Following the collapse of the Soviet Union in the early 1990s, both countries experienced lengthy periods of political and socio-economic instability that exposed much of their populations to a rapid breakdown in social norms, a rise in organized crime and corruption and skyrocketing poverty and inequality (Kalman 2002; Williams and Picarelli 2002; Gilinskiy 2006). Many polls from the last decade reflected significant concern over socio-economic stability. For instance, approximately 30 per cent of Russians polled about recent changes in economic status of their families reported worsening financial situation and only 5 per cent felt it had improved (FOM 2009). Both Russians and Ukrainians noted that 14–25 per cent of the people in their social circles had experienced panics and depression (Kamenchuk 2009). Income inequality has also changed the landscapes of Russian and Ukrainian cities. Once socio-economically indistinguishable, urban communities acquired new characteristics, with some increasingly perceived as cheaper, less prestigious and less desirable for living. In Russia, the percentage of population living in areas of concentrated disadvantage has doubled from 12.2 in 2005 to 24.4 in 2007 (Gusev 2009). Although GINI coefficients suggest near-parity with the United States, the consequences of inequality are dramatic for post-Soviet republics where economically disadvantaged populations receive little or no state assistance and, by western standards, often live in extreme poverty (Notzon et al. 1998; Round and Williams 2010). The arrival of the free market economy has also altered many values shared by Russians and Ukrainians. For instance, money and power as a means to achieve greater wealth were named key criteria for social success by the majority of Russian respondents in 2009 (Boikov 2010), and the emphasis on material success goals was found to be much more prominent in Russia and Ukraine than in the United States (Chamlin and Cochran 2007). A few years earlier, yet another study (ISRAS 2006) reported high emphasis on wealth as something to be achieved in life placed by the majority of respondents. This suggests that, like some other post-Soviet nations (e.g. Karstedt and Farrall 2006), Ukraine and Russia are clearly anomic societies (Messner and Rosenfeld 1994) and, for the majority of individuals whose life plans will never materialize, these societies are laden with strain on both micro and macro levels. On individual level, lack of confidence in the future likely has been a key stressor resulting in various types of negative emotions, ranging from anger to depression. On macro level, rapidly developing community segregation in both countries undoubtedly has increased the exposure of many Russians and Ukrainians to other frustrated individuals and community disorder. Although criminal coping is a likely response to these conditions, previous research reports only modest strain–crime links in Ukraine and Russia (e.g. Botchkovar et al. 2009; Botchkovar et al. 2013). This may be because overwhelming and sustained amounts of strain experienced by many Russians and Ukrainians have desensitized them to such an extent that community disadvantage and other aversive stimuli are more likely to result in non-angry emotional reactions such as sadness or depression than criminal coping (e.g. Gottlieb 1997). Consistent with the argument that chronic stressors may produce significant mental health issues (i.e. Pearlin 1989), a significant body of literature links neighbourhood deprivation/disorder to depressive disorders (e.g. Ross 2000; Ross et al. 2001; Latkin and Curry 2003; Haines et al. 2011). This relationship is commonly attributed to reports of crime, exposure to violence, evidence of disorder and social isolation experienced by neighbourhood residents (e.g. Silver et al. 2002; Stockdale et al. 2007; Kim 2010). Residents’ inability to control neighbourhood conditions associated with disorganization is often seen as the primary reason for poor health outcomes in disorganized neighbourhoods (Mirowski and Ross 1990). We address this causal alternative by investigating depressive symptoms, in addition to criminal involvement, as outcomes of exposure to strains at the individual and community levels. Hypotheses Overall, neighbourhood context may be instrumental in explaining how community residents experience and react to various types of strains. Based on the literature reviewed above, we hypothesize: 1. Community-level strain exposure is multi-faceted, consisting of neighbourhood-related negative events or processes (e.g. signs of decay, litter, drinking in public) and aggregated strains experienced by individuals (e.g. collective gloominess or anger). Each type of strain will exert an independent amplifying effect on individual criminal behaviour. MST appears to suggest that residents of impoverished communities are exposed to both personal and community-level strains. Although community-level strains may translate into individual strains, residents of disadvantaged communities are exposed to many strains that are unrelated to neighbourhood processes. In addition, the criminogenic effect of neighbourhood disadvantage may encompass mechanisms other than strain-generating characteristics, such as presentation of crime-conducive opportunities and deviant models of behaviour (see also Agnew 1999 for a similar argument). These possibilities lead us to hypothesize the following: 2. Individual strains and community-level strains have independent effects on personal criminal involvement. 3. The effect of community-level strains on personal criminal involvement is partially mediated by individual strains. Community-level disadvantage, including exposure to strains and collective negative affect, exacerbates the effects of personal strains/negative affect and exhausts personal coping resources, whereas strong emphasis on religion at the neighbourhood level provides alternative models of behaviour, deters individuals from committing crimes, imposes moral constraints and offers additional coping resources. Therefore, we hypothesize the following: 4. The positive relationship between individual strains/negative affect and criminal coping will be exacerbated by community-level strains/aggregated negative affect. 5. The positive effect of individual strains/negative affect on criminal behaviour will be mitigated by community-level religiosity. Finally, in light of US research on neighbourhood deprivation and depressive disorders and given the unique contextual characteristics of communities in Russia and Ukraine, we also examine depressive symptoms as outcomes of exposure to strains at the individual and community levels and hypothesize that: 6. Individual strains and community-level strains have independent effects on depressive symptoms. 7. The effect of community-level strains on depressive symptoms is partially mediated by individual strains. 8. The positive relationship between individual strain and depressive symptoms will be exacerbated by community-level strains/aggregated negative affect. Data The data used in this study come from two cities, Nizhni Novgorod, Russia, and Lviv, Ukraine. They were collected in the summer of 2009 by reputable professional survey organizations with experience in various types of sociological research. A total of 700 respondents in Ukraine and 735 respondents in Russia were selected for paper-and-pencil face-to-face interviews using a multi-stage stratified random sampling. For census purposes, each city is divided into 6–8 districts of 60,000–250,000 residents. Because these city districts are too large to be defined as neighbourhoods and because there are no equivalents of the US census tracts or blocks, the professional survey organizations mapped each city into smaller (approximately 8,000 residents per community) and more homogenous neighbourhoods defined as having at least two parallel or perpendicular streets with a shared playground or similar area, a grocery store and access to public transportation. Those neighbourhoods have been historically established in each city and are recognized by residents as distinct residential areas, as indicated by the unique names with which they are identified. Yielding 60 micro-neighbourhoods in Lviv and 80 in Nizhni Novgorod, this procedure is consistent with the general definition of a neighbourhood as a geographical and social subsection of a larger community in which residents share a common sense of identity that persists over time (see Bursik and Grasmick 1993: 5–12). Twenty-one neighbourhoods in Nizhni Novgorod and 20 neighbourhoods in Lviv were randomly selected to ensure appropriate area coverage. Eligible apartments and houses then were randomly sampled from each neighbourhood (35 from each neighbourhood), and one adult respondent (18 years or older) whose birthday was closest to the date of the interview was selected from each household. The household replacement rate was about 65 per cent, with replacements made due to unavailability of respondents, refusal to participate, and commercial leasing of apartments. The random replacement rate is comparable to those of other household surveys conducted in Eastern Europe and the United States (e.g. Tittle et al. 2003; Kordos 2005). The instrument used in the survey was written in English, translated into Ukrainian and Russian, and then back-translated into English by translators fluent in these languages. All sensitive questions including those regarding criminal involvement were self-administered by respondents. After self-administering these questions on a separate piece of paper, each respondent used an envelope to place the answers in a basket with identical envelopes while sitting a few feet away from an interviewer to ensure confidentiality and increase data accuracy. The completion rate for each study variable was at least 98 per cent. Dependent variables Individual criminal involvement To estimate criminal involvement, we asked respondents how often they engaged in seven different criminal acts in the last five years and to gauge the projected likelihood that they would commit each of these acts in the future (see Appendix B). Each item offered five response options ranging from ‘never’ to ‘very likely’ (for projected offending) or ‘a very large number of times’ (for past offending), with higher scores indicating greater likelihood of misbehaviour. Response scores were summed to construct indices of past and projected offending (see Table 1 for alpha values and descriptive statistics). Table 1 Descriptive statistics Variable Mean Dtandard deviation Minimum Maximum α/r Dependent Criminal involvement 5.90 4.46 1 26 0.85 Depression 11.07 4.42 7 28 0.89 Predictors (Level 1 and Level 2) Anger Person 2.782 1.01 1 5 Community 0.50 1.54 3.91 Neighbourhood disorder Individually perceived 13.41 3.39 7 21 0.79 Community 1.99 8.51 17.37 Personal strain Personal 27.14 5.98 15 54 0.78 Community aggregated 3.20 18.14 32.91 Religiosity Person 0.00 1.84 −3.88 4.25 0.69 Community 0.95 −1.70 1.83 Controls (Level 1) Male 0.45 0.49 0 1 Age 42.73 16.67 18 91 Russian 0.49 0.50 0 1 Controls (Level 2) Neighbourhood turnover 22.52 5.31 11.40 32.91 Neighbourhood SES 2.44 0.404 1.71 3.20 0.54 Ukraine 0.49 0.50 0 1 Variable Mean Dtandard deviation Minimum Maximum α/r Dependent Criminal involvement 5.90 4.46 1 26 0.85 Depression 11.07 4.42 7 28 0.89 Predictors (Level 1 and Level 2) Anger Person 2.782 1.01 1 5 Community 0.50 1.54 3.91 Neighbourhood disorder Individually perceived 13.41 3.39 7 21 0.79 Community 1.99 8.51 17.37 Personal strain Personal 27.14 5.98 15 54 0.78 Community aggregated 3.20 18.14 32.91 Religiosity Person 0.00 1.84 −3.88 4.25 0.69 Community 0.95 −1.70 1.83 Controls (Level 1) Male 0.45 0.49 0 1 Age 42.73 16.67 18 91 Russian 0.49 0.50 0 1 Controls (Level 2) Neighbourhood turnover 22.52 5.31 11.40 32.91 Neighbourhood SES 2.44 0.404 1.71 3.20 0.54 Ukraine 0.49 0.50 0 1 View Large We opted for the self-projected measure of offending as the primary dependent variable because it helps establish causal order in a cross-sectional survey design and appears to possess greater validity and reliability as a measure of offending (e.g. Tittle et al. 2003). Projections of criminal behaviour have been found to provide reasonable estimates of actual individual involvement in crime (e.g. Murray and Erickson 1987; Albarracin et al. 2001; Pogarsky 2004). As expected, our data indicate that past and projected reports of criminal behaviour are highly correlated (r = 0.78). Nevertheless, as a precaution, we re-analyzed the data using past reports of misbehaviour as a dependent variable and found the same pattern of findings (see Appendix A). Depression To investigate the possibility that Russians and Ukrainians are more likely to respond to strain with depression than crime, we use a scale of depressive symptoms as an alternative outcome measure. This measure consists of seven items included in Radloff’s (1977) CES-D scale (see Appendix B). The four response options ranged from ‘never’ (1) to ‘often’ (4). Individual-level independent variables Individual perceptions of neighbourhood disorder. To tap neighbourhood-related strains, we use respondents’ reports of disorder in their own neighbourhoods, including police not patrolling the area, police irresponsive to calls, drinking in public, litter on sidewalks and streets, people using or selling syringes and youth setting mailboxes on fire or littering inside buildings or on the streets. Ranging from ‘not a problem at all’ (1) to ‘big problem’ (3), response scores were summed to create an index, with higher values indicating higher levels of strain. Individual-level personal strain. Personal strains are tapped through a series of questions about previous experiences of various life events ranging from a burglarized apartment to losing a valued job (Appendix B). Response options ranged from ‘never’ (1) to ‘a very large number of times’ (5). Item responses were summed to construct the index of individual-level strain, with higher values indicating higher levels of strain. Individual-level anger. To measure anger, the most salient type of negative affect considered in GST, we asked respondents how frequently they felt angry as a result of straining life events. Response options ranged from ‘never’ (1) to ‘very often’ (5). Individual-level religiosity. Religiosity is measured by commonly used survey items asking how religious respondents consider themselves to be and how often they go to church or attend religious services/meetings. Responses to these items, which were given on a five-point scale ranging from ‘not religious at all/never’ (1) to ‘very religious/very often’ (5), were summed to create an index of religiosity. Neighbourhood-level independent variables We use several neighbourhood-level predictors, which correspond to our individual-level measures: community-level neighbourhood disorder, aggregated personal strain, community-level anger and community-level religiosity. Consistent with the literature on random effects models, to construct those measures, we calculated the means individual-level predictors within neighbourhoods (Bell and Jones 2015; Clarke et al. 2015).1 The computed community-level variables reflect average levels of those predictors in neighbourhoods, with higher values representing higher levels of corresponding variables. Control variables We control for several antecedent and possibly confounding variables at the individual and neighbourhood levels. At the neighbourhood level, we include residential stability as the average length of residence reported by respondents in each neighbourhood. To control for possible differences between Ukraine and Russia, we also incorporate a dummy variable indicating the country in which the neighbourhood is located (1 = Ukraine). In addition, we control for neighbourhood socio-economic status (SES) by averaging individual-level SES indicators within neighbourhoods. Individual SES can be difficult to tap in Ukraine and Russia because true income is rarely reported by a significant proportion of citizens whose employment in multiple sectors of the economy often is hidden from authorities. Moreover, there is weak correspondence between incomes, no matter how measured, and education and occupation. We overcome these difficulties by focusing on respondents’ material deprivation, a particularly salient dimension of SES (Krieger et al. 1997) that we measure by combining consumption patterns for six important products and services into a cumulative standardized scale (Appendix B) (see Mayer and Jencks 1989 for a similar approach). At the individual level, we control for the usual socio-demographic correlates of crime, including gender (1 = male; 0 = female) and age in years. We also control for individual ethnicity (1 = Russian; 0 = other). We do not control for racial composition because Russian and Ukrainian societies both are racially homogenous. Additional sensitivity analyses controlling for individual marital status, children in household and education produced similar findings. Analytical Strategy We use HLM 7 to estimate a series of two-level over-dispersed Poisson regression models assessing the effects of individual- and neighbourhood-level predictors on self-projected offending (Raudenbush and Bryk 2002). In addition to its appropriateness for positively skewed discrete outcome measures such as the ones used in our study, the advantages of this approach include efficient slope estimates and unbiased standard errors in nested data. All individual- and neighbourhood-level predictors were grand-mean centred prior to inclusion in the models. Slope-as-outcomes models were used to evaluate cross-level interactions between individual strain/anger and neighbourhood characteristics. We combine the two countries in the study in order to avoid sacrificing statistical power with site-specific analyses. Supplementary analyses executed separately for each country indicated a substantively similar pattern of findings across Russia and Ukraine. All models were checked for multi-collinearity prior to engaging in multilevel analyses. We use the EM algorithm to impute missing values (not more than 1 per cent for any survey item). Results Prior to assessing our key research hypotheses, we gauge the extent of cross-neighbourhood variation in individual offending by estimating an unconditional model without covariates (results are available upon request). The assessment of this model’s variance component and associated likelihood ratio chi-square test (τ00 = 0.11, P < 0.000) supports a multi-level modelling strategy, indicating that a significant portion of the variation in criminal involvement (16 per cent) occurs between neighbourhoods.2 Findings concerning our first hypothesis are reported in Table 2. We proposed that aggregated personal strains and strains associated with neighbourhood disorder would have independent effects on individual criminal involvement. The results reported in Model 1 provide mixed support for this hypothesis. While two contextual factors, community-level neighbourhood disorder strain and anger, significantly elevated the likelihood of individual offending (b = 0.039; incidence rate ratio [IRR] = 1.041 and b = 0.196; IRR = 1.379), aggregated personal strain failed to emerge as a significant predictor. Analyzed one at a time, the effects of these community-level predictors on criminal involvement remained unchanged. Table 2 Effects of individual- and neighbourhood-level predictors of projected criminal involvement Model 1 Model 2 Model 3 Model 4 b (SE) Event rate ratio b (SE) Event rate ratio b (SE) Event rate ratio b (SE) Event rate ratio Level 1 measures Male 0.193*** (0.032) 1.21 0.160*** (0.036) 1.173 0.159*** (0.037) 1.172 0.157*** (0.037) 1.170 Age 0.011*** (0.001) 1.011 0.011*** (0.001) 1.011 0.011*** (0.001) 1.011 0.011*** (0.001) 1.011 Russian 0.060 (0.080) 1.062 0.007 (0.089) 1.007 0.006 (0.088) 1.006 0.007 (0.009) 1.006 Anger 0.151*** (0.025) 1.163 0.157*** (0.025) 1.170 0.150*** (0.025) 1.161 Personal strain 0.026*** (0.003) 1.026 0.026*** (0.003) 1.026 0.026*** (0.002) 1.026 Perceived neighbourhood disorder 0.002 (0.008) 1.002 0.003 (0.008) 1.003 0.002 (0.008) 1.002 Religiosity −0.028 (0.015) 0.972 −0.029** (0.014) 0.970 −0.029** (0.014) 0.971 Level 2 measures Community anger 0.196* (0.107) 1.379 0.038 (0.111) 1.039 0.052 (0.121) 1.053 0.030 (0.112) 1.031 Community aggregated strain 0.028 (0.019) 1.028 0.002 (0.019) 1.002 0.001 (0.021) 1.000 0.000 (0.019) 1.000 Neighbourhood disorder 0.039** (0.020) 1.041 0.035 (0.021) 1.036 0.037 (0.023) 1.038 0.035 (0.022) 1.035 Community religiosity −0.030 (0.070) 0.971 0.003 (0.074) 1.003 −0.005 (0.081) 0.995 0.005 (0.079) 1.005 Population turnover −0.008 (0.005) 0.992 −0.007 (0.005) 0.993 −0.007 (0.005) 0.993 −0.008 (0.005) 0.992 Neighbourhood SES 0.194 (0.115) 1.215 0.179 (0.112) 1.196 0.152 (0.128) 1.163 0.151 (0.130) 1.163 Ukraine 0.321* (0.188) 1.379 0.274 (0.198) 1.314 0.284 (0.205) 1.328 0.288 (0.205) 1.334 Personal strain × community anger −0.003 (0.012) 0.997 Personal strain × community aggregated strain 0.001 (0.001) 1.001 Personal strain × community religiosity −0.003 (0.003) −0.996 Personal strain × neighbourhood disorder −0.003 (0.002) 0.997 Anger × community personal strain 0.011 (0.009) 1.011 Anger × community anger −0.102 (0.085) 0.903 Anger × community religiosity −0.006 (0.030) 0.993 Anger × neighbourhood disorder −0.015 (0.013) 0.985 Intercept 1.728*** (0.032) 5.572 1.691*** (0.035) 5.429 1.705*** (0.35) 5.502 1.692*** (0.035) 5.423 U0 (df) 0.047*** 0.046*** 0.050*** 0.048*** Model 1 Model 2 Model 3 Model 4 b (SE) Event rate ratio b (SE) Event rate ratio b (SE) Event rate ratio b (SE) Event rate ratio Level 1 measures Male 0.193*** (0.032) 1.21 0.160*** (0.036) 1.173 0.159*** (0.037) 1.172 0.157*** (0.037) 1.170 Age 0.011*** (0.001) 1.011 0.011*** (0.001) 1.011 0.011*** (0.001) 1.011 0.011*** (0.001) 1.011 Russian 0.060 (0.080) 1.062 0.007 (0.089) 1.007 0.006 (0.088) 1.006 0.007 (0.009) 1.006 Anger 0.151*** (0.025) 1.163 0.157*** (0.025) 1.170 0.150*** (0.025) 1.161 Personal strain 0.026*** (0.003) 1.026 0.026*** (0.003) 1.026 0.026*** (0.002) 1.026 Perceived neighbourhood disorder 0.002 (0.008) 1.002 0.003 (0.008) 1.003 0.002 (0.008) 1.002 Religiosity −0.028 (0.015) 0.972 −0.029** (0.014) 0.970 −0.029** (0.014) 0.971 Level 2 measures Community anger 0.196* (0.107) 1.379 0.038 (0.111) 1.039 0.052 (0.121) 1.053 0.030 (0.112) 1.031 Community aggregated strain 0.028 (0.019) 1.028 0.002 (0.019) 1.002 0.001 (0.021) 1.000 0.000 (0.019) 1.000 Neighbourhood disorder 0.039** (0.020) 1.041 0.035 (0.021) 1.036 0.037 (0.023) 1.038 0.035 (0.022) 1.035 Community religiosity −0.030 (0.070) 0.971 0.003 (0.074) 1.003 −0.005 (0.081) 0.995 0.005 (0.079) 1.005 Population turnover −0.008 (0.005) 0.992 −0.007 (0.005) 0.993 −0.007 (0.005) 0.993 −0.008 (0.005) 0.992 Neighbourhood SES 0.194 (0.115) 1.215 0.179 (0.112) 1.196 0.152 (0.128) 1.163 0.151 (0.130) 1.163 Ukraine 0.321* (0.188) 1.379 0.274 (0.198) 1.314 0.284 (0.205) 1.328 0.288 (0.205) 1.334 Personal strain × community anger −0.003 (0.012) 0.997 Personal strain × community aggregated strain 0.001 (0.001) 1.001 Personal strain × community religiosity −0.003 (0.003) −0.996 Personal strain × neighbourhood disorder −0.003 (0.002) 0.997 Anger × community personal strain 0.011 (0.009) 1.011 Anger × community anger −0.102 (0.085) 0.903 Anger × community religiosity −0.006 (0.030) 0.993 Anger × neighbourhood disorder −0.015 (0.013) 0.985 Intercept 1.728*** (0.032) 5.572 1.691*** (0.035) 5.429 1.705*** (0.35) 5.502 1.692*** (0.035) 5.423 U0 (df) 0.047*** 0.046*** 0.050*** 0.048*** SE, standard error. *P < 0.05, **P < 0.01, ***P < 0.001 (one-tailed tests). View Large Our findings lend modest support to our second hypothesis proposing that neighbourhood-level and person-level strains and emotions both are directly and independently associated with individual criminal behaviour. Neither of the two indicators of community-level strain nor aggregated anger is significantly associated with individual crime after the effect of individual-level strain is taken into account (Model 2). However, personal strains (b = 0.026; IRR = 1.026) and individual-level anger (b = 0.151; IRR = 1.163) both appear to be potent predictors of projected offending. Notably, the effect of personal strain on individual-level crime persists when individual anger is included in the model. Our third hypothesis suggested that the effects of neighbourhood-level strain would be partially mediated by individual strain. Model 2 shows that the effects of community-level disorder and anger on individual crime are fully attenuated when individual-level neighbourhood-related strains and anger are taken into account. However, although this result may indicate the presence of mediation, grand-mean centring averages within- and between-group effects in a way that potentially confounds the mediation effect (Zhang et al. 2009). To address this possibility, we repeated the analyses using group-mean centred variables (results available upon request). Group-mean centring removes all between-neighbourhood variation from the person-level variable, providing an individual-level estimate that is independent of neighbourhood effects. Group-mean centring also yields for each neighbourhood a mean of zero on the individual-level measure, guaranteeing that the inclusion of the individual-level measure in the regression model will not confound the effects of the cluster mean representing neighbourhood-level strain/anger (Kreft and de Leeuw 1998; Raudenbush and Bryk 2002). Using group-mean centring, we confirmed our initial finding that the effects of neighbourhood disorder and community anger on individual behaviour are not statistically significant when individual-level perceptions of disorder and anger variables are taken into account. Somewhat consistent with our third hypothesis, this suggests that individual-level disorder perceptions/anger mediate the effects of neighbourhood disorder and community anger, although mediation appears to be full rather than partial. Our fourth hypothesis is that individual-level strain/anger effects on criminal involvement would be amplified by negative affect (anger) and strains experienced at the community level. Contrary to expectations, none of the interactive relationships exhibited a statistically significant effect on individual crime involvement, regardless of whether analyzed in groups or separately from one another (Models 3 and 4).3 Our fifth hypothesis proposed an assuaging effect of neighbourhood religiosity on the relationship between individual strains/anger and crime. In our data, community-level religiosity failed to exhibit an independent effect on crime, and results reported in Models 3 and 4 provide no evidence that neighbourhood-level religiosity attenuates the effect of strain/anger. Finally, to investigate the possibility that Russians and Ukrainians respond to straining circumstances with non-angry emotional coping strategies than with crime (Hypotheses 6–8), we repeated the analyses presented above using depressive symptoms as a substitute outcome measure. Similar to criminal involvement, we found that 15 per cent of variation in depressive symptoms is across neighbourhoods. Table 3 reveals the effects of individual- and neighbourhood-level strain processes and religiosity on individual reports of depressive symptoms. We do not assess the interactive effects of individual anger and neighbourhood characteristics because depression, like anger, is theorized to be an outcome of strain (Agnew 1992). Estimating contextual effects on strain-relevant processes on depressive outcomes, Model 1 reveals that neighbourhood disorder is the only neighbourhood-level strain that significantly predicts individual depression is (b = 0.023; IRR = 1.026). Incorporating individual-level predictors related to strain, Model 2 shows that individual-level personal strain (b = 0.013; IRR = 1.013) predicts significantly increased depression among individuals, while neighbourhood disorder continues to predict significantly higher mean levels of depression (b = 0.028; IRR = 1.028). The latter effect is not mediated by individual-level strains, and it appears to be contextual rather than compositional because it remains statistically significant even in the presence of individual-level reports of neighbourhood disorder strain. Finally, contrary to Hypothesis 8, no evidence is observed of interactive relationships between individual-level strains and community strains/collective anger and religiosity. Thus, the individual-level effect of strains/collective anger on crime does not appear to be strengthened by high levels of neighbourhood strain/anger or weakened by a high degree of neighbourhood religiosity. Table 3 Effects of individual- and neighbourhood-level predictors of depression Model 1 Model 2 Model 3 b (SE) Event rate ratio b (SE) Event rate ratio b (SE) Event rate ratio Level 1 measures Male −0.108*** (0.018) 0.897 −0.101*** (0.019) 0.895 −0.111*** (0.019) 0.896 Age −0.005*** (0.0006) −0.005 −0.004*** (0.000) 0.995 −0.004*** (0.000) 0.996 Russian 0.079*** (0.032) 0.080 0.074*** (0.029) 1.080 0.077** (0.029) 1.077 Anger 0.028* (0.015) 1.030 0.028** (0.016) 1.028 Personal strain 0.013*** (0.003) 1.013 0.013*** (0.002) 1.014 Perceived neighbourhood disorder −0.029 (0.003) 0.997 −0.029 (0.003) 0.997 Level 2 measures Community anger 0.059 (0.051) 1.061 0.031 (0.050) 1.031 0.028 (0.049) 1.029 Community aggregated strain 0.012 (0.009) 1.012 −0.001 (0.009) 0.998 −0.001 (0.001) 0.998 Neighbourhood disorder 0.023** (0.010) 1.026 0.028** (0.010) 1.028 0.027** (0.011) 1.028 Population turnover −0.002 (0.004) 0.997 −0.002 (0.003) 0.997 −0.003 (0.003) 0.997 Ukraine 0.171** (0.071) 1.185 0.167** (0.071) 1.182 0.168** (0.071) 1.182 Neighbourhood SES −0.033 (0.072) 0.997 −0.034 (0.072) 0.996 −0.036 (0.074) 0.964 Personal strain × community anger 0.002 (0.004) 1.013 Personal strain × community aggregated strain −0.000 (0.001) 0.999 Personal strain × neighbourhood disorder 0.000 (0.001) 1.000 Intercept 2.395*** (0.020) 10.959 2.391*** (0.021) 10.924 2.393*** (0.021) 10.945 U0 0.016*** 0.015*** 0.016*** Model 1 Model 2 Model 3 b (SE) Event rate ratio b (SE) Event rate ratio b (SE) Event rate ratio Level 1 measures Male −0.108*** (0.018) 0.897 −0.101*** (0.019) 0.895 −0.111*** (0.019) 0.896 Age −0.005*** (0.0006) −0.005 −0.004*** (0.000) 0.995 −0.004*** (0.000) 0.996 Russian 0.079*** (0.032) 0.080 0.074*** (0.029) 1.080 0.077** (0.029) 1.077 Anger 0.028* (0.015) 1.030 0.028** (0.016) 1.028 Personal strain 0.013*** (0.003) 1.013 0.013*** (0.002) 1.014 Perceived neighbourhood disorder −0.029 (0.003) 0.997 −0.029 (0.003) 0.997 Level 2 measures Community anger 0.059 (0.051) 1.061 0.031 (0.050) 1.031 0.028 (0.049) 1.029 Community aggregated strain 0.012 (0.009) 1.012 −0.001 (0.009) 0.998 −0.001 (0.001) 0.998 Neighbourhood disorder 0.023** (0.010) 1.026 0.028** (0.010) 1.028 0.027** (0.011) 1.028 Population turnover −0.002 (0.004) 0.997 −0.002 (0.003) 0.997 −0.003 (0.003) 0.997 Ukraine 0.171** (0.071) 1.185 0.167** (0.071) 1.182 0.168** (0.071) 1.182 Neighbourhood SES −0.033 (0.072) 0.997 −0.034 (0.072) 0.996 −0.036 (0.074) 0.964 Personal strain × community anger 0.002 (0.004) 1.013 Personal strain × community aggregated strain −0.000 (0.001) 0.999 Personal strain × neighbourhood disorder 0.000 (0.001) 1.000 Intercept 2.395*** (0.020) 10.959 2.391*** (0.021) 10.924 2.393*** (0.021) 10.945 U0 0.016*** 0.015*** 0.016*** SE, standard error. *P < 0.05, **P < 0.01, ***P < 0.001 (one-tailed tests). View Large In sum, our results are largely supportive of GST but fail to confirm any contextual influences on projected offending, independently of or in interaction with individual-level processes. Individual strains also are consistently predictive of depression, even after controlling for contextual-level predictors. Although neither community-level SES nor population turnover, two indicators of social disorganization, consistently predicted individual criminal behaviour, our findings confirm a relationship between neighbourhood disorder and depressive symptoms among neighbourhood residents. Discussion In this study, we sought to integrate the key premises of Agnew’s MST with GST placing individual behaviour in community context. To do so, we explored direct and interactive effects of community- and individual-level strain processes and religiosity on two outcomes, personal criminal involvement and depressive symptoms. Several important findings emerge from our research. First, we found significant support for most individual-level processes described by GST, with individual-level strain and anger increasing the likelihood of criminal involvement. Whereas past research utilizing more general measures of strain revealed weak support for GST among Russians and Ukrainians (e.g. Botchkovar et al. 2013, but see Botchkovar and Broidy 2013), it appears that the specific straining experiences tapped in our research (i.e. personal and vicarious victimization experiences) are more criminogenic. Further, consistent with GST and supporting studies conducted in Russia and Ukraine (Botchkovar et al. 2009; 2013), our findings indicate that, while anger significantly predicts individual criminal involvement, it does not mediate the strain–crime relationship. This adds to a growing body of evidence casting doubt on the intervening role of anger in those associations (Hay and Evans 2006; Lin et al. 2011; Yun and Lee 2015). As with criminal behaviour, personal strains were found to predict depression, suggesting that Russians and Ukrainians respondents may react to straining circumstances not only with criminal adaptations but also with non-angry emotional coping strategies detrimental to their overall well-being.4 Another key finding is that community-level strains, anger and religiosity were immaterial for individual criminal behaviour, alone and interaction with individual-level strain/anger. The effects of strain associated with neighbourhood disorder and community anger were fully absorbed by individual strain exposure, indicating that observed neighbourhood-level effects likely are compositional (‘kinds of people’) rather than contextual (‘kinds of neighbourhoods’) as a result of a selection process through which people prone to committing crimes tend to settle in disorganized communities. We encountered similar difficulties explaining depression as he only community-level predictor associated with individual depression was neighbourhood disorder. In addition to providing mixed support for the integrated theoretical model of GST and MST, our findings may offer several insights into the cultural limitations of the concentrated disadvantage paradigm. Originating from Shaw and McKay’s (1942) theory, the notion of concentrated disadvantage has become almost synonymous with elevated crime rates in the American criminological literature (e.g. Sampson 2012). Criminologists have accepted the relationship as axiomatic and sought to ‘unpack’ it using both structural and subcultural explanations (e.g. Sampson and Bartusch 1998; Bellair 2000; Browning et al. 2004). Agnew’s MST similarly takes for granted the association between community disadvantage and crime assuming that exposure to strains and negative affect mediates the link between structural characteristics and criminal behaviour. Our findings show that the association between structural disadvantage and crime may be culture specific and limited to contexts similar to the United States. It is possible that a protracted history of community-level inequality is necessary to uncover strong enduring effects of neighbourhood deprivation on individual behaviour. Although evidence of growing community-level differences in welfare has been accumulating in Ukraine and Russia, these differences may not have existed long enough to be influential or to form a true ‘package' of concentrated disadvantage necessary to produce strong effects on individual behaviour. This is indirectly confirmed by the lack of a statistical association between neighbourhood SES and community disorder (results available upon request). Even more significant is the lack of importance of community-level strains and anger for criminal involvement. The Russians and Ukrainians in our study appear criminally insensitive to neighbourhood stressors, typically reacting with crime only to personal strains. This casts doubt on the potential of community strain processes to explain criminal involvement among residents. This is not to suggest, however, that macro-level strain processes are irrelevant for criminal behaviour in Russian and Ukrainian societies. For instance, the influence of the anomic processes described by institutional anomie theory (Messner et al. 2008) might be particularly salient for individual behaviour in Russia and Ukraine. The effects of institutional anomie may interact with neighbourhood-level processes described by MST (Agnew 1999), have direct effects on individual behaviour or operate indirectly through socialization effects. It is crucial, then, that future research incorporate these and other potential contextual level influences into studies of criminal behaviour in these countries. Our findings also point to the link between community stressors and mental health in both countries. They show that some neighbourhood-level strains such as neighbourhood disorder directly influence the likelihood of experiencing depression by individuals living in those residential communities. This result is in line with existing research in the US linking community disorder with depressive symptoms and anxiety (c.f. Ross 2000; Hill et al. 2005). Furthermore, the literature documents various coping techniques used by individuals to adapt to, survive and even thrive under the conditions of chronic strain (see Gottlieb 1997; Dunkel-Schetter and Dolbier 2011 for review). Thus, in social contexts where strain exposure is chronic and variable, environmental strains may lose their criminogenic potential and, instead, affect the emotional well-being of residents (e.g. Turner et al. 2013; see also Winlow et al. 2017: 178). More research is necessary to determine the potential of specific community-related strains to produce lingering effects on individual emotional states and behaviour. Although our study provides new information about the interlinkages between GST and MST, caution should be exercised given the nature of the data. First, while we employed various verification methods that appear successful in eliciting essential information, all survey respondents are vulnerable to possible telescoping, exaggeration and/or misreporting of information. Second, our research is cross-sectional. Thus, although projected rather than past crime was used to establish proper temporal ordering, and although substantive conclusions remained unchanged when we used past reports of crime as dependent variable, more research with longitudinal data is needed to confirm our findings. Finally, at least some neighbourhood boundaries, while based on city residents’ perceptions of their own neighbourhoods and carefully re-assessed by the organizations conducting the survey in both countries, could be drawn incorrectly, possibly affecting our findings. However, this limitation potentially applies to any neighbourhood-level studies that do not settle for the often-large community areas identified by local censuses or similar organizations (e.g. Sampson et al. 1997). Despite these limitations, our study significantly contributes to the GST/MST literature as well as to research linking community disadvantage in crime. Demonstrating support for GST, findings revealed strong links between individual strain/anger and criminal involvement and depression. Our attempt to integrate GST with MST was only partially successful because community-level strains did not interact with individual-level strain processes in a theoretically meaningful way. This may suggest that deleterious community effects on individual behaviour are a product of lengthy multi-generational exposure to structural inequality and disadvantage—a criterion that is not satisfied in post-Soviet countries. Thus, while our findings provide some support for the cultural generality of GST, they indicate that the US-based community disadvantage literature may be too narrowly focused and culturally limited. References Agnew R. (1992), ‘ Foundation for a General Strain Theory of Crime and Delinquency’, Criminology , 30: 47– 87. Google Scholar CrossRef Search ADS ——. (1999), ‘ A General Strain Theory of Community Differences in Crime Rates’, Journal of Research in Crime and Delinquency , 36: 123– 55. CrossRef Search ADS ——. (2001), ‘ Building on the Foundation of General Strain Theory: Specifying the Types of Strain Most Likely to Lead to Crime and Delinquency’, Journal of Research in Crime and Delinquency , 38: 319– 61. CrossRef Search ADS ——. (2006), Pressured into Crime. An Overview of General Strain Theory . Roxbury Publishing. ——. (2013), ‘ When Criminal Coping Is Likely: An Extension of General Strain Theory’, Deviant Behavior , 34: 653– 70. CrossRef Search ADS Agnew R. Matthews S. K. Bucher J. Welcher A. N. Keyes C. (2008), ‘ Socioeconomic Status, Economic Problems, and Delinquency’, Youth and Society , 40: 159– 81. Google Scholar CrossRef Search ADS Albarracin D. Johnson B. T. Fishbein M. Muellerleile P. A. (2001), ‘ Theories of Reasoned Action and Planned Behavior as Models of Condom Use: A Meta-analysis’, Psychological Bulletin , 17: 142– 61. Google Scholar CrossRef Search ADS Anderson E. (1999), Code of the Street: Decency, Violence, and the Moral Life of the Inner City . W.W. Norton and Company. Baier C. J. Wright B. R. E. (2001), ‘ If You Love Me, Keep My Commandments: A Meta-analysis of the Effect of Religion on Crime’, Journal of Research in Crime and Delinquency , 38: 3– 21. Google Scholar CrossRef Search ADS Bao W.-N. Haas A. Chen X. J. Pi Y. J. (2014), ‘ Repeated Strains, Social Control, Social Learning, and Delinquency: Testing an Integrated Model of General Strain Theory in China’, Youth & Society , 46: 402– 24. Google Scholar CrossRef Search ADS Bao W.-N. Haas A. Pi Y. (2004), ‘ Life Strain, Negative Emotions, and Delinquency: An Empirical Test of General Strain Theory in the People’s Republic of China’, International Journal of Offender Therapy and Comparative Criminology , 48: 281– 97. Google Scholar CrossRef Search ADS PubMed Bell A. J., and Jones K. (2015), ‘ Explaining Fixed Effects: Random Effects Modelling of Time-Series Cross-Sectional and Panel Data’, Political Science Research & Methods , 3: 133– 53. Google Scholar CrossRef Search ADS Bellair P. E. (2000), ‘ Informal Surveillance and Street Crime: A Complex Relationship’. Criminology , 38: 137– 69. Google Scholar CrossRef Search ADS Boikov V. E. (2010), ‘ Socio-political Aspirations of Russians: Their Contents and Routes Toward Their Fullfillment’, Socis , 6: 27– 35. Botchkovar E. V. Antonaccio O. Tittle C. R. (2013), ‘ Strain, Coping, and SES: Coping Histories and Present Choices’, Journal of Quantitative Criminology , 29: 217– 50. Google Scholar CrossRef Search ADS Botchkovar E. V. Broidy L. (2013), ‘ Accumulated Strain, Negative Emotions, and Crime: A Test of General Strain Theory in Russia’, Crime and Delinquency , 59: 837– 86. Google Scholar CrossRef Search ADS Botchkovar E. V. Tittle C. R. Antonaccio O. (2009), ‘ General Strain Theory: Additional Evidence Using Cross-cultural Data’, Criminology , 47: 801– 48. Google Scholar CrossRef Search ADS Browning C. R. Feinberg S. L. Dietz R. D. (2004), ‘ The Paradox of Social Organization: Networks, Collective efficacy, and Violent Crime in Urban Neighborhoods’, Social Forces , 83: 503– 34. Google Scholar CrossRef Search ADS Brezina T. Piquero A. R. Mazerolle P. (2001), ‘ Student Anger and Aggressive Behavior in School: An Initial Test of Agnew’s Macro-Level Strain Theory’, Journal of Research in Crime and Delinquency , 38: 362– 86. Google Scholar CrossRef Search ADS Bursik R. Grasmick H. (1993), Neighborhoods and Crime: The Dimensions of Effective Community Control . Lexington Books. Chamlin M. B. Cochran J. K. (2007), ‘ An Evaluation of the Assumptions That Underlie Institutional Anomie Theory’, Theoretical Criminology , 11: 39– 61. Google Scholar CrossRef Search ADS Cheung N. W. T. Cheung Y. W. (2010), ‘ Strain, Self-control, and Gender Differences in Delinquency Among Chinese Adolescents: Extending General Strain Theory’, Sociological Perspectives , 53: 321– 45. Google Scholar CrossRef Search ADS Clarke P. Crawford C. Steele F. Vignoles A. (2015), ‘ Revisiting Fixed- and Random-Effects Models: Some Considerations for Policy-Relevant Education Research’, Education Economics , 23: 259– 77. Google Scholar CrossRef Search ADS Dollard J. Doob L. Miller N. Mowrer O. Sears R. (1939), Frustration and Aggression . Yale University Press. Google Scholar CrossRef Search ADS Dunkel-Schetter C. Dolbier C. (2011), ‘ Resilience in the Context of Chronic Stress and Health in Adults’, Social and Personality Psychology Compass , 5: 634– 52. Google Scholar CrossRef Search ADS PubMed Fom: Fond “Obschestvennoe Mnenie”. ( 2009), available online at www.fom.ru. Froggio G. Agnew R. (2007), ‘ The Relationship Between Crime and “Objective” Versus “Subjective” Strains’, Journal of Criminal Justice , 35: 81– 7. Google Scholar CrossRef Search ADS Gilinskiy Y. (2006), ‘ Crime in Contemporary Russia’, European Journal of Criminology , 3: 259– 92. Google Scholar CrossRef Search ADS Gottlieb B. H. (1997), Coping with Chronic Stress . Plenum. Google Scholar CrossRef Search ADS Gusev A. B. (2009), Zhilischnoe neravenstvo rossiyskih domashnih hoziaystv: Estestvennyj put’ k segregatsii . Vestnik Universiteta (GUU)#25. Haines V. Beggs J. B. Hurlbert J. (2011), ‘ Neighborhood Disadvantage, Network Social Capital, and Depressive Symptoms’, Journal of Health and Social Behavior , 52: 58– 73. Google Scholar CrossRef Search ADS PubMed Hay C. Evans M. M. (2006), ‘ Violent Victimization and Involvement in Delinquency: Examining Predictions from General Strain Theory’, Journal of Criminal Justice , 34: 261– 74. Google Scholar CrossRef Search ADS Hay C. Meldrum R. (2010), ‘ Bullying Victimization and Adolescent Self-harm: Testing Hypotheses from General Strain Theory’, Journal of Youth and Adolescence , 39: 446– 59. Google Scholar CrossRef Search ADS PubMed Hill T. D. Ross C. E. Angel R. E. (2005), ‘ Neighborhood Disorder, Psychophysiological Distress, and Health’, Journal of Health and Social Behavior , 46: 170– 86. Google Scholar CrossRef Search ADS PubMed Hoffmann J. P. ( 2003), ‘ A Contextual Analysis of Differential Association, Social Control, and Strain Theories of Delinquency’, Social Forces , 81: 751– 85. Horton R. Rice S. K. Piquero N. L. Piquero A. (2012), ‘ On the Variability of Anger Cross-Culturally: An Assessment of General Strain Theory’s Primary Mediator’, Deviant Behavior , 33: 260– 81. Google Scholar CrossRef Search ADS Hox J. (2010), Multilevel Analysis: Techniques and Applications . Routledge. ISRAS (Institut Sotsiologii Rossiyskoy Akademii Nauk) (2006), ‘Russian Identity from a Sociological Standpoint’, available online at http://www.isras.ru/analytical_report_Ident0.html. Jang S. R. Johnson B. R. (2005), ‘ Gender, Religiosity, and Reactions to Strain Among African Americans’, The Sociological Quarterly , 46: 323– 57. Google Scholar CrossRef Search ADS Johnson M. C. Morris R. G. (2008), ‘ The Moderating Effects of Religiosity on the Relationship Between Stressful Life Events and Delinquent Behavior’, Journal of Criminal Justice , 36, 486– 93. Google Scholar CrossRef Search ADS PubMed Jones N. (2010), Between Good and Ghetto: African American Girls and Inner-City Violence . Rutgers University Press. Junger M. Polder W. (1993), ‘ Religiosity, Religious Climate, and Delinquency Among Ethnic Groups in the Netherlands’, British Journal of Criminology , 33: 416– 35. Google Scholar CrossRef Search ADS Kalman A. (2002), ‘Organized Economic Crime and Corruption in Ukraine. Final Report for the National Institute of Justice’, available online at http://www.ojp.usdoj.gov/nij/international/programs/ukraine.html. Kamenchuk O. (2009), ‘Nastroeniya rossiyan: Griadet li sotsialno-politicheskiy vzvryv?’, available online at http://shabrov.info/elbrus/kamen.pdf. Karstedt S. Farrall S. (2006), ‘ The Moral Economy of Everyday Crime: Markets, Consumers and Citizens’, British Journal of Criminology 46: 1011– 1036. Google Scholar CrossRef Search ADS Kim J. (2010), ‘ Neighborhood Disadvantage and Mental Health: The Role of Neighborhood Disorder and Social Relationships’, Social Science Research , 39: 260– 71. Google Scholar CrossRef Search ADS Krieger N. Williams D. Moss N. E. (1997), ‘ Measuring Social Class in US Public Health Research: Concepts, Methodologies, and Guidelines’, Annual Review Public Health , 18: 341– 78. Google Scholar CrossRef Search ADS Kreft G. G. De Leeuw J. (1998), Introducing Multilevel Modeling, Number 1 of Introducing Statistical Methods . Sage. Google Scholar CrossRef Search ADS Kordos J. (2005), ‘Household Surveys in Transition Countries’, in Household Sample Surveys in Developing and Transition Countries . United Nations, available online at http://unstats.un.org/unsd/hhsurveys/pdf/Household_surveys.pdf. Latkin C. A. Curry A. D. (2003), ‘ Stressful Neighborhoods and Depression: A Prospective Study of the impact of Neighborhood Disorder’, Journal of Health and Social Behavior , 44: 34– 44. Google Scholar CrossRef Search ADS PubMed Lin W. H. Cochran J. K. Mieczkowski T. (2011), ‘ Direct and Vicarious Violent Victimization and Juvenile Delinquency: An Application of General Strain Theory’, Sociological Inquiry , 81: 195– 222. Google Scholar CrossRef Search ADS PubMed Lin W. H. Dembo R. Sellers C. S. Cochran J. Mieczkowski T. (2014), ‘ Strain, Negative Emotions, and Juvenile Delinquency: The United States Versus Taiwan’, International Journal of Offender Therapy and Comparative Criminology , 58: 412– 34. Google Scholar CrossRef Search ADS PubMed Maas C. J. M. Hox J. J. (2005), ‘ Sufficient Sample Sizes for Multilevel Modeling’, Methodology: European Journal of Research Methods for the Behavioral and Social Sciences , 1: 86– 92. Google Scholar CrossRef Search ADS MacLeod J. (2004), Ain’t No Makin’ It: Aspirations and Attainment in a Low-Income Neighborhood . Westview Press. Mayer S. E. Jencks C. (1989), ‘ Growing up in Poor Neighborhoods: How Much Does It Matter?’, Science , 243: 1441– 5. Google Scholar CrossRef Search ADS PubMed Messner S. F. Rosenfeld R. ( 1994), Crime and the American Dream . Wadsworth. Messner S. F. Thome H. Rosenfeld R. (2008), ‘ Institutions, Anomie, and Violent Crime: Clarifying and Elaborating Institutional Anomie Theory’, International Journal of Conflict and Violence , 2: 163– 81. Mirowski J. Ross C. E. (1990), ‘ Control or Defense? Depression and the Sense of Control Over Good and Bad Outcomes’, Journal of Health and Social Behavior , 31: 71– 86. Google Scholar CrossRef Search ADS PubMed Moon B. Morash M. McCluskey J. D. (2012), ‘ General Strain Theory and School Bullying: An Empirical Test in South Korea’, Crime & Delinquency , 58: 827– 55. Google Scholar CrossRef Search ADS Murray G. F. Erickson P. G. (1987), ‘ Cross-sectional Versus Longitudinal Research: An Empirical Comparison of Projected and Subsequent Criminality’, Social Science Research , 16: 107– 18. Google Scholar CrossRef Search ADS Notzon F. C. Komarov Y. M. Ermakov S. P. Sempos C. T. Marks J. S. Sempos E. V. (1998), ‘ Causes of Declining Life Expectancy in Russia’, Journal of the American Medical Association , 279: 793– 800. Google Scholar CrossRef Search ADS PubMed Op de Beeck H. Pauwels L. Put J. (2012), ‘ Schools, Strain and Offending: Testing General Strain Theory in a School Context’, European Journal of Criminology , 9: 52– 72. Google Scholar CrossRef Search ADS Pearlin L. I. (1989), ‘ The Sociological Study of Stress’, Journal of Health and Social Behavior , 30: 241– 56. Google Scholar CrossRef Search ADS PubMed Peck J. (2013), ‘ Examining Race and Ethnicity in the Context of General Strain Theory, Depression, and Delinquency’, Deviant Behavior , 34: 706– 26. Google Scholar CrossRef Search ADS Piquero N. L. Sealock M. D. (2000), ‘ Generalizing General Strain Theory: An Examination of an Offending Population’, Justice Quarterly , 17: 449– 89. Google Scholar CrossRef Search ADS Pogarsky G. (2004), ‘ Projected Offending and Contemporaneous Rule-Violation: Implications for Heterotypic Continuity’, Criminology , 42: 111– 35. Google Scholar CrossRef Search ADS Radloff L. S. (1977), ‘ The CES-D Scale: A Self-report Depression Scale for Research in the General Population’, Applied Psychological Measurement , 1: 385– 401. Google Scholar CrossRef Search ADS Raudenbush S. W. Bryk A. S. (2002), Hierarchical Linear Models: Applications and Data Analysis Methods , 2nd edn. Sage. Raudenbush S. W. Liu X. (2000), ‘ Statistical Power and Optimal Design for Multisite Randomized Trials’, Psychological Methods , 5: 199– 213. Google Scholar CrossRef Search ADS PubMed Rebellon C. J. Manasse M. E. Van Gundy K. T. Cohn E. S. (2012), ‘ Perceived Injustice and Delinquency: A Test of General Strain Theory’, Journal of Criminal Justice , 40: 230– 37. Google Scholar CrossRef Search ADS Ross C. E. (2000), ‘ Neighborhood Disadvantage and Adult depression’, Journal of Health and Social Behavior , 41: 177– 87. Google Scholar CrossRef Search ADS Ross C. E. Mirowsky J. Pribesh S. (2001), ‘ Powerlessness and the Amplification of Threat: Neighborhood Disadvantage, Disorder, and Mistrust’, American Sociological Review , 66: 568– 91. Google Scholar CrossRef Search ADS Round J. Williams C. ( 2010), ‘ Coping with the Social Costs of ‘Transition’: Everyday Life in Post-Soviet Russia and Ukraine’, European Urban and Regional Studies , 17: 183– 96. Sampson R. J. (2012), Great American City: Chicago and the Enduring Neighborhood Effect . Chicago University Press. Google Scholar CrossRef Search ADS Sampson R. J. Bartusch D. J. (1998), ‘ Legal Cynicism and (Subcultural?) Tolerance of Deviance: The Neighborhood Context of Racial Differences’, Law and Society Review , 32: 777– 804. Google Scholar CrossRef Search ADS Sampson R. J. Raudenbush S. W. Earls F. (1997), ‘ Neighborhoods and Violent Crime: A Multilevel Study of Collective Efficacy’, Science , 277: 918– 24. Google Scholar CrossRef Search ADS PubMed Sampson R. J. Wilson W. J. (1995), ‘Toward a Theory of Race, Crime, and Urban Inequality’, in J. Hagan R. D. Peterson, eds, Crime and Inequality , 37–54. Stanford University Press. Shaw C. R. McKay H. D. (1942), Juvenile Delinquency in Urban Areas . University of Chicago Press. Snijders T. A. B. Bosker R. J. (1993), ‘ Standard Errors and Sample Sizes for Two-Level Research’, Journal of Educational Statistics , 18: 237– 59. Google Scholar CrossRef Search ADS Sigfusdottir I.-D. Kristjansson A. L. Agnew R. (2012), ‘ A Comparative Analysis of General Strain Theory’, Journal of Criminal Justice , 40: 117– 27. Google Scholar CrossRef Search ADS Silver E. Mulvey E. P. Swanson J. W. (2002), ‘ Neighborhood Structural Characteristics and Mental Disorder: Faris and Dunham Revisited’, Social Science & Medicine , 55: 1457– 70. Google Scholar CrossRef Search ADS Stark R. Kent L. Doyle D. P. (1982), ‘ Religion and Delinquency: The Ecology of a “Lost” Relationship’, Journal of Research in Crime and Delinquency , 19: 4– 24. Google Scholar CrossRef Search ADS Stockdale S. E. Wells K. B. Tang L. Belin T. R. Zhang L. Sherbourne C. D. (2007), ‘ The Importance of Social Context: Neighborhood Stressors, Stress-Buffering Mechanisms, and Alcohol, Drug, and Mental Health Disorders’, Social Science & Medicine , 65: 1867– 81. Google Scholar CrossRef Search ADS Tittle C. R. Ward D. A. Grasmick H. G. (2003), ‘ Self-control and Crime/Deviance: Cognitive vs. Behavioral Measures’, Journal of Quantitative Criminology , 19: 333– 65. Google Scholar CrossRef Search ADS Tittle C. R. Welch M. R. (1983), ‘ Religiosity and Deviance: Toward a Contingency Theory of Constraining Effects’, Social Forces , 61: 653– 82. Google Scholar CrossRef Search ADS Turner H. A. Finkelhor D. Hamby S. Shattuck A. (2013), ‘ Family Structure, Victimization, and Mental Health in a Nationally Representative Sample of Children and Youth’, Social Science & Medicine , 87: 39– 51. Google Scholar CrossRef Search ADS Ulmer J. T. Harris C. (2013), ‘ The Effects of Religious Contexts on Black, Latino, and White Violence Rates’, The Sociological Quarterly , 54: 610– 46. Google Scholar CrossRef Search ADS PubMed Wareham J. Cochran J. K. Dembo R. Sellers C. S. (2005), ‘ Community, Strain, and Delinquency: A Test of a Multi-Level Model of General Strain Theory’, Western Criminology Review , 6: 117– 33. Warner B. D. Fowler S. K. (2003), ‘ Strain and Violence: Testing a General Strain Theory Model of Community Violence’, Journal of Criminal Justice , 31: 511−– 21. Google Scholar CrossRef Search ADS Williams P. Picarelli J. (2002), ‘Organized Crime in Ukraine: Challenge and Response. Final Report for the National Institute of Justice’, available online at http://www.ojp.usdoj.gov/nij/international/programs/ukraine.html. Willis P. (1977), Learning to Labor: How Working Class Kids Get Working Class Jobs . Columbia University Press. Winlow S. Hall S. Treadwell J. (2017), The Rise of the Right: The English Defence League and the Transformation of Working Class Politics . Policy Press. Yun I. Lee C. H. (2015), ‘ Hazardous Alcohol Use Among South Korean Police Officers: Examining Predictions From General Strain Theory’, International Journal of Law, Crime and Justice , 43: 194– 213. Google Scholar CrossRef Search ADS Zhang Z. Zyphur M. J. Preacher K. J. (2009), ‘ Testing Multilevel Mediation Using Hierarchical Linear Models: Problems and Solutions’, Organizational Research Methods , 12: 695– 719. Google Scholar CrossRef Search ADS Footnotes 1 The aggregate neighbourhood score is calculated from 35 randomly selected individuals per neighbourhood. This number of cases is sufficient to produce reliable aggregate scores, given the homogeneity within micro-neighbourhoods. We also weigh in on the issue of statistical power in our study. Research suggests that our study has a sufficient number of neighbourhoods—and individuals interviewed in each neighbourhood—to reliably detect at least medium-sized and, under most conditions, even smaller effects of neighbourhood characteristics on individual outcomes (Snijders and Bosker 1993; Raudenbush and Liu 2000; Maas and Hox 2005; Hox 2010). All models were also re-analyzed with step-wise introduction of predictors to avoid overloading with no change in results. 2 We investigated whether our key predictors demonstrate sufficient variation between neighbourhoods and find that to be the case. Specifically, between-neighbourhood variation in anger was 23 per cent, neighbourhood disorder—33 per cent, cumulative strain—33 per cent, residential mobility—11 per cent and religiosity—26 per cent. 3 All findings were confirmed in a series of analyses based on ordinary least squares models with natural logged dependent variables. 4 One important consideration in any study seeking to assess the relationship between strains and criminal behaviour is capturing strain. Our study follows the recommendations provided by Agnew (2006) and measures strain using a list of items ranging from death of someone close to victimization experiences. However, conceptual overlaps may still exist. For instance, a measure of neighbourhood-level SES aggregated from individual responses could be conceptualized as another measure of strain. Of note, in this case is the explicit theoretical distinction Agnew make between SES and strain (e.g. Agnew et al. 2008) as well as his insistence that such a relationship would typically be non-linear and modest in size. In our research, community-level SES was not predictive of criminal behaviour. We also investigated the relationship between SES and crime on individual level but failed to detect a consistent association between them across all models (results are available upon request). Appendix A. Sensitivity analyses. Effects of individual- and neighbourhood-level predictors on past offending behavioura b (SE) Level 1 measures Anger 0.285*** (0.047) Personal strain 0.035*** (0.005) Perceived neighbourhood disorder 0.002 (0.010) Religiosity −0.063*** (0.021) Level 2 measures Community anger 0.042 (0.148) Community aggregated strain −0.001 (0.028) Neighbourhood disorder 0.068 (0.042) Community religiosity −0.029 (0.141) Personal strain × community anger 0.001 (0.013) Personal strain × community aggregated strain 0.000 (0.000) Personal strain × community religiosity −0.001 (0.004) Personal strain × neighbourhood disorder 0.001 (0.002) b (SE) Level 1 measures Anger 0.285*** (0.047) Personal strain 0.035*** (0.005) Perceived neighbourhood disorder 0.002 (0.010) Religiosity −0.063*** (0.021) Level 2 measures Community anger 0.042 (0.148) Community aggregated strain −0.001 (0.028) Neighbourhood disorder 0.068 (0.042) Community religiosity −0.029 (0.141) Personal strain × community anger 0.001 (0.013) Personal strain × community aggregated strain 0.000 (0.000) Personal strain × community religiosity −0.001 (0.004) Personal strain × neighbourhood disorder 0.001 (0.002) aThe model contains all individual- and neighbourhood-level control variables included in main analyses. SE, standard error. ***P < 0.001 (one-tailed test). View Large Appendix B. Measures of strain, non-criminal coping strategies and criminal probability/ self-reported criminal behaviour used in analyses Strain How often in the last six months have you experienced some of the following? 1. Your apartment was burglarized 2. Some of your belongings were taken from you without your permission 3. You were physically attacked or threatened with violence by a stranger 4. You were physically attacked or threatened with violence by somebody you know 5. You were robbed of something by force (purse snatched, mugged etc.) on the street 6. Your property was purposefully destroyed or vandalized by someone (car damaged, windows broken etc). 7. You have been sexually harassed or abused. 8. You have suffered from a serious or prolonged illness. 9. You have gotten into conflict with a friend, partner or a family member 10. You have had serious money issues. 11. You have had to take a job that you particularly disliked. 12. You have broken up with a close friend or intimate partner. 13. You have lost a job you valued 14. Someone you cared about died 15. You have been unhappy with the place you are living (or your conditions of living?) Depression After each statement, please indicate the step on the stairs that is closest to how often the event I describe has happened to you during the past week. 1. Felt you just could not get going 2. Felt sad 3. Felt depressed 4. Had trouble getting to sleep or staying asleep 5. Felt everything was an effort 6. Felt you couldn’t shake the blues 7. Had trouble keeping your mind on what you were doing Projected criminal behaviour How likely would you be to commit (each of the following acts) if you had a strong desire or need and the opportunity to do it? 1. Take money or property from others worth less than $5. 2. Take money or property from others worth more than $5 but less than $50. 3. Take money or property from others worth $50 or more. 4. Hit another person on purpose in an emotional outburst. 5. Physically harm another person on purpose. 6. Use violence or threat of violence to accomplish some personal goal. 7. Intentionally avoid paying for something (movies, bus ride etc.) SES What can you afford on your income? 1. Can you afford the groceries that you want? 2. Can you afford medications that you need? 3. Can you afford clothes that you need? 4. Can you renovate your apartment/house? 5. Can you travel abroad for leisure? 6. Can you afford to buy an apartment or house? Strain How often in the last six months have you experienced some of the following? 1. Your apartment was burglarized 2. Some of your belongings were taken from you without your permission 3. You were physically attacked or threatened with violence by a stranger 4. You were physically attacked or threatened with violence by somebody you know 5. You were robbed of something by force (purse snatched, mugged etc.) on the street 6. Your property was purposefully destroyed or vandalized by someone (car damaged, windows broken etc). 7. You have been sexually harassed or abused. 8. You have suffered from a serious or prolonged illness. 9. You have gotten into conflict with a friend, partner or a family member 10. You have had serious money issues. 11. You have had to take a job that you particularly disliked. 12. You have broken up with a close friend or intimate partner. 13. You have lost a job you valued 14. Someone you cared about died 15. You have been unhappy with the place you are living (or your conditions of living?) Depression After each statement, please indicate the step on the stairs that is closest to how often the event I describe has happened to you during the past week. 1. Felt you just could not get going 2. Felt sad 3. Felt depressed 4. Had trouble getting to sleep or staying asleep 5. Felt everything was an effort 6. Felt you couldn’t shake the blues 7. Had trouble keeping your mind on what you were doing Projected criminal behaviour How likely would you be to commit (each of the following acts) if you had a strong desire or need and the opportunity to do it? 1. Take money or property from others worth less than $5. 2. Take money or property from others worth more than $5 but less than $50. 3. Take money or property from others worth $50 or more. 4. Hit another person on purpose in an emotional outburst. 5. Physically harm another person on purpose. 6. Use violence or threat of violence to accomplish some personal goal. 7. Intentionally avoid paying for something (movies, bus ride etc.) SES What can you afford on your income? 1. Can you afford the groceries that you want? 2. Can you afford medications that you need? 3. Can you afford clothes that you need? 4. Can you renovate your apartment/house? 5. Can you travel abroad for leisure? 6. Can you afford to buy an apartment or house? View Large © The Author(s) 2017. Published by Oxford University Press on behalf of the Centre for Crime and Justice Studies (ISTD). All rights reserved. For permissions, please e-mail: firstname.lastname@example.org
The British Journal of Criminology – Oxford University Press
Published: Mar 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera