Vous êtes sur la page 1sur 37

Street Y outh, Unemployment, and Crime: Is It That Simple?

Using General Strain Theory to Untangle the Relationship1


Stephen W . Baron Queens University

de porter une plus grande attention aux Des chercheurs ont recommande mage et la criminalite . En particulier, ils ont sugge re variables reliant le cho `tent leur situation sur le marche du travail que la fac on dont les gens interpre ` laide des the action dans cette situation. A ories influe grandement sur leur re ne rales, la pre sente recherche porte sur le ro le joue par le des tensions ge mage dans le comportement criminel de 400 jeunes de la rue. Un e le ment cho ressant est la manie `re dont ces jeunes interpre `tent leurs expe riences de inte tations et ces expe riences influent toutes travail, et comment ces interpre sultats re ve `lent que les effets du deux sur le comportement criminel. Les re mage sur la criminalite sont principalement modifie s et mode re s par cho mage de pend dattributions externes dautres variables. En particulier, le cho `nent a ` la cole `re envers linemploi, ce qui engendre la criminalite . Leffet qui me mage sur le crime est mode re par le me contentement lie a ` direct du cho ` ` largent et a une recherche demploi minimale. La colere envers le chomage sulte aussi dinterpre tations subjectives ne gatives de la situation e conomire du travail. En outre, ces percepque et dun attachement incessant au marche tat, une diminution du gatives, labsence de soutien de la part de lE tions ne le social et labsence prolonge e de domicile augmentent la participation contro s criminelles. Cette participation est aussi encourage e par directe aux activite pris des peines encourues. On discute les pairs, les valeurs anormales et le me sultats obtenus et on offre dautres pistes de recherche. des re s : jeune de la rue, the ories des tensions ge ne rales, cho mage, Mots cle criminalite Researchers have called for greater attention to be paid to the variables linking unemployment to crime. In particular, it has been suggested that peoples interpretation of their labour market situation plays a large role in shaping their responses to it. Utilizing general strain theory, this research examines the role that unemployment plays in the criminal behaviour of 400 homeless street youths. Of particular interest is the way that these youths interpret their labour market experiences and how together these interpretations and

2008 CJCCJ/RCCJP doi:10.3138/cjccj.50.4.399

400

Revue canadienne de criminologie et de justice pe nale

juillet 2008

experiences influence criminal behaviour. Findings reveal that the effect of unemployment on crime is mediated and moderated primarily by other variables. In particular, unemployment is conditioned by external casual attributions that lead to anger over unemployment, which in turn leads to crime. The direct effect of unemployment on crime is moderated by monetary dissatisfaction and minimal employment searches. Anger over unemployment is also the result of negative subjective interpretations of economic situations and a continued attachment to the labour market. In addition, these negative subjective perceptions, the lack of state support, a decrease in social control, and prolonged homelessness lead to greater participation in criminal activities directly. Criminal involvement is also encouraged by peers, deviant values, and a lack of fear of punishment. Findings are discussed and suggestions for future research are offered. Keywords: street youth, general strain theory, unemployment, crime

Introduction
While earlier research reported an inconsistent or weak relationship between unemployment and crime (see Box 1987; Chiricos 1987), recent work has generally revealed that unemployment, or changes to the unemployment rate, are positively related to some types of criminal behaviour (see Bellair and Roscigno 2000; Britt 1997, 2001; Cantor and Land 2001; Carlson and Michalowski 1997; Chamlin and Cochran 2000; Crutchfield and Pitchford 1997; Greenberg 2001; Hale and Sabbagh 1991; Kleck and Chiricos 2002; Krivo and Peterson 2004; Land, Cantor, and Russell 1995; OBrien 2001; Paternoster and Bushway 2001; Witt, Clarke, and Fielding 1999). Most of this research, however, has utilized aggregate-level data linking unemployment rates to crime rates. Critics have argued that the link between unemployment and crime rests on a micro-theoretical foundation that cannot be captured utilizing aggregate-level data (Box 1987; Levitt 2001; Long and Witte 1981; Orsagh and Witt 1981). These scholars observe that the theoretical explanations for the link between unemployment and crime focus on subjective perceptions, interpretations, and emotional reactions to the experience of unemployment (see Box; Cantor and Land; Greenberg; Levitt). Levitt (388) notes that inferences from aggregate-level data may be misleading and argues that this research cannot capture the subtle predictions about a variety of possible behavioral channels through which the crime-unemployment nexus operates. Levitt (388) advocates the use of other approaches, including individual-level analysis, and notes that surprisingly,

Street Y outh, Unemployment, and Crime

401

however, there has been relatively little previous research on the subject utilizing these alternative strategies. The limited available individual-level research does suggest that there is a relationship between unemployment and crime (Bachman, OMalley, and Johnston 1978; Baron and Hartnagel 1997; Cernkovich, Giordano, and Rudolph 2000; Crowley 1984; Elliott 1994; Farrington, Gallagher, Morely, St Ledger, and West 1986; Hagan and McCarthy 1997; Hartnagel 1998; Hartnagel and Krahn 1989; Inciardi, Horowitz, and Pottieger 1993; Shavit and Rattner 1988; Thornberry and Christianson 1984). This research, however, tends to be limited by its focus on objective information about respondents employment status and history and, similar to aggregate-level work, tends to ignore the more subjective aspects of the theoretical foundations (although see Baron and Hartnagel; Cernkovich et al.; Hartnagel). Crutchfield and Pitchford (1997: 91) suggest, [At] a minimum we should be aware that simplistic notions about the relationship between unemployment and crime have limited utility. In response, researchers have been encouraged to pay greater attention to the variables theoretically thought to link unemployment and crime (Box 1987; Box and Hale 1985). Scholars note that the way people interpret their unemployment should influence the way they respond to unemployment (Box; Box and Hale; Hartnagel). The effect of unemployment may not be influenced only by actual financial problems but also by perceptions of economic difficulties and judgments of fairness in reward distribution. Further, responses to unemployment may depend on peoples employment histories, beliefs about the causes of their unemployment, their social circles, and the importance they place on employment. People respond with different emotions to their unemployment and these feelings can influence whether they engage in crime. Box and Hale note that because of these complexities one would expect only a sub-population of the unemployed to become involved in crime, and it is because of these various dimensions that attempts to establish monocausal connections have fallen short (see also Box). The following paper explores the link between unemployment and crime, using a sample of homeless street youth, focusing on the way that these young people interpret their unemployment experiences and how these interpretations lead to crime. Street youths are often unemployed, spend periods of time without shelter, and hang out on the street regularly (Baron 2004; Hagan and McCarthy 1997). The absence of legal financial resources to help house and feed

402

Canadian Journal of Criminology and Criminal Justice

July 2008

themselves, and the potential for alienation as a result of these experiences, leaves these youths at risk for criminal activity (Baron; Hagan and McCarthy), making them an ideal sample to explore the unemploymentcrime relationship.

Theoretical explanations of unemployment and crime


There is no shortage of theoretical explanations to help understand the link between unemployment and crime. Box (1987) notes that nearly every major criminological theory can be interpreted to support the link between the two. However, in light of the call by critics to examine a wide array of the more subjective components of the unemployment experience, one theory that is potentially fruitful in explaining the link is Agnews (1992, 2001, 2006) general strain theory (GST). This approach focuses attention on how objective experiences, subjective interpretations, and emotional reactions can all be linked to crime. Agnew emphasizes how negative experiences can lead to criminal behaviour. His theory outlines several categories of strain into which unemployment might fall, including goal blockage, failure to achieve positively valued goals, and presentation of negative stimuli. Agnew (2001) also distinguishes between objective strains and subjective strains. Objective strains are conditions thought to be disliked generally by most members of a given group (e.g., unemployment), while subjective strains are disliked by the individuals actually experiencing them (Agnew 2001, 2006). He notes that individuals can differ in their interpretation of objective strains and these may be influenced by values, resources, and life circumstances. Agnew observes that strain will more likely result in crime when it is viewed as unjust or when it involves the voluntary and intentional violation of a relevant justice norm (2001: 329) and when people believe their strain is undeserved (2001: 330). The impact of strain may also vary, depending upon its severity or magnitude, duration, recency, and centrality. Strains high or severe in magnitude decrease the perceived costs of criminal coping and increase the disposition for criminal coping. Strains that are of high frequency, longer in duration, and unresolved are expected to have a greater negative impact on an individual. Finally, strains that threaten core goals, needs, and values are seen as central and more likely to provoke a criminal response. Agnew (2001, 2006) outlines the major sources of economic strain that will have a significant impact on crime: unemployment, relative

Street Y outh, Unemployment, and Crime

403

deprivation, and monetary dissatisfaction. Unemployment may increase an individuals motivation to commit crime to overcome financial difficulties. Further, those on the margins of society are more likely to be monetarily dissatisfied and/or relatively deprived, living in poverty in the midst of plenty (Agnew 2006: 73; Agnew, Cullen, Burton, Evans, and Dunaway 1996), and these can both lead to crime. Unhappiness with ones current financial situation is seen as key because it is linked to the reality of the moment and not some abstract future often suggested by measures utilized in early work on strain theory, and this dissatisfaction is more likely to pressure or propel people into crime (see Cernkovich et al. 2000: 145). Relative deprivation is important because strain is seen as a function of the achievements of others in ones comparative reference group, as well as ones own failure to achieve. Pivotal here is the argument that the egalitarian ideology in North American society encourages individuals to compare themselves to those higher in the economic system, leading to resentment, frustration, and hostility (Agnew 2006; see also Passas 1997). Agnew (1992: 59) argues that strain can lead to various affective states, including the critical emotional reaction of anger. Anger is viewed as key because it is associated with feelings of power and stimulates desires for revenge and retaliation (Agnew 2006: 32). The level of anger should be closely linked to subjective strains in particular, although Agnew (2001) admits that similar levels of subjective strain can lead to different emotional responses or varying levels of the same emotion. Agnew suggests that anger is often experienced when individuals have difficulty satisfying goals. Further, it is a method of responding to perceived injustice when individuals feel they have the right to what others have (Agnew 2006: 33). Anger is most likely to result when people make external attributions, or blame the cause of their strain on others (see also Hoffman and Ireland 1995). Further, peoples beliefs or constitutive rules as well as ones peer group may also influence interpretations of strain and the type and level of emotional response that might evolve in response to strain (Bernard 1990; Hoffman and Ireland). Peer groups can also influence the types of attributions that people make, and this interaction can help determine if people will react to strain with anger (Hoffman and Ireland). An angry emotional reaction to strain leads actors to consider corrective action, which may include criminal activity. Whether one chooses criminal corrective action is said to be influenced by

404

Revue canadienne de criminologie et de justice pe nale

juillet 2008

a number of conditioning factors: attributions, deviant attitudes, and deviant peers. These conditioning factors either buffer one from strain or increase ones disposition towards criminal behaviour. Deviant peers can model and support criminal behaviour as responses to strain (see Hoffman and Ireland 1995). Criminal beliefs or regulative rules can influence options for appropriate behavioural responses to strain (Bernard 1990), and external attributions are said to make one more likely to seek revenge (Agnew 1992). Further, it may be the influence of a criminal subculture where the combination of beliefs and criminal peers can fuel criminal behaviour (Agnew et al. 1996; Cloward and Ohlin 1960; Cohen 1955). Agnew (2001, 2006) also argues that levels of social control can influence whether strain leads to crime. Certain strains are associated with low social control (e.g., unemployment). These types of strain are said to be more likely to lead to crime because the costs of engaging in crime are reduced. However, in the case where the unemployed person continues to actively search for work, holds high occupational commitment, and espouses a strong work ethic, there can be a continued source of social control (see Hartnagel 1998). These elements of social control may interact with unemployment, influencing whether one will engage in crime (Box 1987; Hartnagel). This suggests that it is in situations of low commitment where people have stopped looking for work, do not espouse a work ethic, and no longer believe work is important, that crime is most likely. Further, the impact of unemployment on crime should be greatest when conditioned by these characteristics of low social control. On the other hand, some of these factors may also be seen to generate strain through an alternative causal pathway (see Agnew 1992; Elliott, Ageton, and Cantor 1979; Hartnagel). That is, continuing to search for work, espousing a work ethic, and showing commitment to work while not being able to secure employment may directly or through their conditioning impact on unemployment create anger, which in turn can lead to crime. The perceived likelihood of formal punishment for crime must also be considered. Agnew (2006) notes that criminal coping is much more likely when individuals experiencing strain are in situations where the costs of engaging in crime are low and the benefits are high. Thus, the unemployed person who believes the odds of apprehension and punishment are minimal and the social and material rewards to be great, will view criminal behaviour as an attractive alternative. Further, it may be that the criminogenic effects of unemployment are

Street Y outh, Unemployment, and Crime

405

greater under conditions of reduced certainty and severity of punishment. Zimring and Hawkins (1973) stress that the maximum threat influence of formal sanctions is greater for those who have the most to lose if subjected to formal punishment. This suggests that the combination of unemployment and minimal perceived legal punishment is most likely to lead to crime. In sum, this literature suggests that unemployment as an objective experience may have only a small effect on crime. Instead, peoples attributions for their unemployment, their subjective interpretations of their conditions in terms of monetary dissatisfaction and relative deprivation, and their objective financial situation may lead to negative emotional reactions over unemployment. Ones commitment and attachment to the labour market might also increase the negative emotional response to unemployment. Further, peoples peer group and their own values regarding crime and appropriate responses to unemployment may influence whether they become angry about their unemployment. It may be the case that the effect of unemployment itself on anger is contingent on some of these other factors. The impact of unemployment on anger may be greater when people feel relatively deprived and monetarily dissatisfied, have few economic resources, attribute the cause of their unemployment to external sources, hold high levels of commitment to the labour market, hold deviant values, and share deviant peers the last of which shapes the direction of attributions. In turn, it is the anger over unemployment that might lead to crime, along with relative deprivation, monetary dissatisfaction, poor financial situation, external attributions, deviant peers, and deviant values. For some, the lack of employment commitment, displayed in work ethic and employment searches, along with the absence of fear of punishment may lead to crime. Similar to its impact on anger, the effect of unemployment on crime may be contingent on a host of factors. The link between unemployment and crime should be stronger when people feel deprived relative to others, are monetarily dissatisfied, have few objective financial resources, attribute the cause of their unemployment to external sources, lack social control and a fear of formal punishment, and possess deviant peer networks and deviant values, which interact to create a deviant subculture that supports crime.2 Past work on general strain theory has tended to focus on cumulative measures of strain, where measures of economic strain, if they

406

Canadian Journal of Criminology and Criminal Justice

July 2008

are included, are part of a battery of other negative life events. Agnew (2001, 2006) argues that these cumulative measures of strain mask the effects of the individual forms of strain and suggests that more work is needed to determine the effects of individual types of strain. Agnew (2001) also observes that there has been little research on certain key measures outlined in his theory, including those on goal blockage, and little work has utilized nonconventional samples (although see Baron 2004). Further, there has been little support in the general strain research for the conditioning effects outlined in the theory (see Agnew 2001, 2006). However, a few of the individual-level studies examining unemployment outside of the general strain perspective do suggest that peers (Hartnagel 1998), external attributions (Baron and Hartnagel 1997), and perceptual measures of deprivation (Baron 2006) can condition the impact of unemployment on crime. Further, the only study using the GST to explicitly include unemployment as a standalone measure found it was conditioned by deviant beliefs in explaining violent offending (Baron 2004). These studies, however, suffer because they tend to ignore either social control measures (Baron and Hartnagel; Baron 2004, 2006) or perceptual measures of poverty (Baron and Hartnagel; Hartnagel) in their analysis and thus are not able to provide a more complete test of the theory. Further, none of the research explores the direct and moderating role of the social control measures of job searching and work ethic to be examined here. It is also the case that the past research has not examined how the deviant subculture, the fear of punishment, and internal attributions for unemployment may condition the impact of unemployment, and no work has examined how peers may influence attributions. In addition, only one prior study focusing on unemployment has included a situational measure of anger (Baron and Hartnagel). However, in that research, which contained only males, no analysis was undertaken to determine how the factors outlined generated anger, and anger itself did not predict any of the offences examined. This also means that none of the previous research on unemployment has explored if the interaction effects predict anger. In fact, despite their centrality to the theory, only one piece of GST research has examined the impact of interaction effects on anger (see Aseltine, Gore, and Gordon 2000). Together this critique suggests that the strain perspective has yet to be fully tested and elaborated through an examination of socioeconomic strain.

Street Y outh, Unemployment, and Crime

407

Methods
The term street youth usually refers to youths who have run away or been expelled from their homes and/or who spend some or all of their time in public locations. For this study 400 respondents (265 males and 135 females) were identified on the basis of four sampling criteria: (1) participants must be aged 24 and under, (2) they must have left or finished school, (3) they must be currently unemployed, and (4) they have spent time without a fixed address or living in a shelter in the previous 12 months. The rationales for these criteria were (1) to cover the age range of those described as street youth (Baron 2004), (2) to eliminate those not eligible for full-time employment, and (3) to obtain a sample of serious at risk youth. Data collection Data were collected between May 2000 and August 2001 in a large western Canadian city with a population of approximately 2 million. The study was centred in and around the downtown business core of the city, bordered by the inner city and local skid row. The researcher situated himself in geographical areas known to be frequented by street youth and approached potential respondents, alerted them to the project, and screened them for study eligibility. Youths meeting the selection criteria were then provided with more information and invited to participate.3 Additional contacts were initiated by youths who had learned of the project and solicited interviews or through introductions offered by previous respondents. Youths were interviewed in fast-food restaurants, in front of storefront social services, in bus shelters and parks, and on the street. Face-to-face interviews provided the opportunity to gain some rapport with participants and to explain questions to ensure their understanding, increasing the quality of measurement and decreasing the odds of non-response (see Mosher, Miethe, and Phillips 2002). Interviews averaged an hour and 10 minutes in length, and respondents were provided 20 dollars in food coupons at a popular fastfood restaurant for participation. Since there is a limited amount of productive interview time with this population (Whitbeck and Hoyt 1999), lengthening the interview process may have stretched the parameters of valid responses.4 For this reason, some of the concepts lack multiple indicators.

408

Revue canadienne de criminologie et de justice pe nale

juillet 2008

The 400 youths who were interviewed had an average age of almost 20 years (x 19.90). The racial make-up of the sample was predominantly Caucasian (83%). Aboriginal youths made up the majority of the other respondents (12%).5 The average length of homelessness in the prior 12 months was close to 7 months (x 6.83). The average education level was approximately grade 10 (x 9.99), and most respondents had left school prior to completion of secondary education (79.5%; N 318). Measuring crime Information on a number of measures of criminal involvement and drug use was obtained via self-reports. Critics note that research incorporating a wide range of behaviours, including serious offences, that asks respondents to report on the actual not the relative number of times they have engaged in these behaviours, is more likely to uncover the conditions under which socioeconomic status is related to crime (Elliott and Ageton 1980; Hindelang, Hirschi, and Weiss 1979, 1981; Mosher et al. 2002). The respondents were asked how many times in the past year they had done the following: broken into a car, broken into a building, taken something worth less than $50, taken something worth more than $50, broken into a structure to sleep, stolen food, taken a car without permission of the owner, sold marijuana or other non-prescription drugs, used physical force to get money or things from another person, attacked someone with a weapon or fists injuring them so badly they probably needed a doctor, got into a fight, and taken part in a group fight. The raw scores of individual offences were aggregated across the range of offences to create measures of property crime, violent crime, and drug distribution. An analysis of the raw frequency distributions for the crime indices suggested a significant amount of variation as well as a high degree of skewness in the measures. To reduce the skewness in the measures, the index values were logged. Predictor variables Information was obtained in the interviews on the labour market and interpretive factors hypothesized to be important in explaining the criminal behaviour of street youth. Beginning with the potential important objective measures that might be associated with crime, the respondents length of unemployment was determined by asking, How many months in the last year were you out of work? To measure the length of time living on the street, respondents were asked

Street Y outh, Unemployment, and Crime

409

to indicate how many months in the last year they had been without a fixed address. A state-assistance variable was incorporated by a simple dummy variable indicating whether or not the respondent was receiving financial support from the government at the time of the research. Two interpretative measures of financial status were used. Monetary dissatisfaction was determined by asking respondents to agree or disagree with the statement Right now Im satisfied with how much money I have to live on (1 strongly agree; 4 strongly disagree) (see Agnew et al. 1996). Cantrils (1965) Self-Anchoring Striving Scale utilized by Walker and Mann (1987) was used to measure relative deprivation. Walker and Mann note that egoistic measures of relative social rank play an important role in determining behaviour and attitudes at an individual level, whereas fraternalistic measures of social rank explain behaviour at a group or social level. Their findings suggest egoistic measures tap the individual-level stress symptoms (including irritability) that can help explain the beliefs and actions of the deprived. Respondents were asked, On a scale from 1 to 10 where 1 is the worst possible rank in Canadian society and 10 is the best possible rank in Canadian society, where do you stand right now? The lower the nominated position on the ladder the greater the deprivation. This was then reverse-coded so that higher scores reflected greater feelings of deprivation. A number of measures were included to tap into the role of employment commitment. Respondents were asked how frequently they looked for a job (1 not looking for a job; 6 daily). Respondents work ethic was determined by four items: The people who dont succeed in life are just plain lazy; people who have failed at a job have usually not tried hard enough; if one works hard enough he is likely to make a good life for himself; a distaste for hard work usually reflects a weakness in character (alpha .62).6 A number of measures were also included to explore the role that attributions, values, peers, and emotions play in the criminal behaviour of street youth. First, to measure external attributions for unemployment the respondents were asked to indicate how influential the failure of private industry to offer enough jobs, the failure of government to create sufficient jobs, and the economic situation of the country were as antecedents to their joblessness (alpha .68).7 To measure internal attributions for unemployment, or perceptions of some sort of deficiency, respondents were asked how much

410

Canadian Journal of Criminology and Criminal Justice

July 2008

a lack of good education, lack of work experience, and lack of specific skills were associated with their lack of employment (alpha .68). To inspect the role of anger, respondents were asked to agree or disagree with the statement I feel angry about my unemployment (1 strongly disagree; 4 strongly agree). Deviant values were determined by asking respondents, How wrong to you think it is to break the law? (very wrong 1; not wrong at all 5). To determine the number of deviant peers, youths were asked, How many of your current friends have been picked up by the police? (1 none; 5 all). To examine the perceptions of sanctions, respondents were asked, If you were to break the law, how likely is it that you would be caught by the police? Five response options ranged from very unlikely ( 1) to very likely ( 5). To obtain the severity measure, respondents were asked, If you were caught breaking the law, by the police, taken to court, and punished, how much of problem would that create for your life? The three response options ranged from no problem at all ( 1) to a large problem ( 3). Job stability was calculated by summing all the jobs that a respondent had held over his or her life so far. This measure was logged to reduce skewness. Finally, the analysis controls for chronological age, gender (males 1; females 2), and race (white 1; minority 2). Table 1 provides a summary of the means and standard deviations of the predictor variables. A limitation of many of the independent variables is that they measure current emotional reactions, perceptions of deprivation, punishment risk, values, and associations that will then be correlated with past criminal behaviour. However, Agnew (1989) in his longitudinal analysis of strain found that adversity was a relatively stable variable, suggesting that asking about current perceptions of strain type variables should be sufficient. Further, Mosher et al. (2002) have noted that self-reported offending is relatively consistent and stable across time periods in panel research, suggesting that the current measures should be related to future crime in the short term. Nevertheless, caution should be exercised in interpreting the findings here. The literature reviewed above suggests that unemployment may have only a weak impact on anger and crime and that its strength may be moderated by a number of the variables described above. The GST perspective suggests that the effect of unemployment on anger and crime might be conditioned by relative deprivation,

Street Y outh, Unemployment, and Crime T able 1: Descriptive statistics for independent variables
M Age Gender Minority Unemployment Length No Fixed Address State Assistance Relative Deprivation Monetary Dissatisfaction Unemployment Internal Attribution Unemployment External Attribution Job Stability Job Search Work Ethic Certainty Severity Deviant Values Deviant Peers Situational Anger Property Crime Violent Crime Drug Sales
a

41 1

SD 2.61 1.34 0.37 3.01 3.80 0.48 2.36 78.00 0.96 0.91 1.04 1.92 0.53 1.26 0.76 1.22 1.11 0.82 2.26 (2188.99) 1.40 (223.35) 3.78 (13015.20)

Min. 13 ( 2.6) 1 (0.7) 1 (0.44) 1 ( 2.9) 1 (1.5) 1 (3.0) 1 ( 2.5) 1 ( 2.4) 1 ( 2.5) 1 ( 2.4) 0 (1.6) 1 (0.78) 1 (3.0) 1 (1.2) 1 (1.6) 1 (1.4) 1 ( 2.6) 1 (1.6) 0 0 0

Max. 24 (1.6) 2 (1.4) 2 (2.3) 12 (0.7) 12 (1.4) 2 (2.7) 10 (1.3) 4 (1.4) 5 (1.7) 5 (2.0) 6.22 (4.4) 6 (1.8) 4 (2.7) 5 (2.0) 3 (1.0) 5 (1.9) 5 (1.0) 4 (2.0) 10.17 (26206) 8.39 (4400) 12.17 (192000)

19.90a 1.35 1.16 9.80 6.83 1.63 6.98 2.91 3.40 3.18 1.61 2.51 2.59 2.45 2.21 2.70 3.88 2.34 5.46 (422.78)b 1.56 (25.10) 4.28 (4192.72)

The mean for the standardized variables is 0 with a standard deviation of 1. Ranges for standardized variables are in parentheses. The numbers inside parentheses are raw scores. Numbers outside parentheses are logged values of raw scores.

monetary dissatisfaction, state support, and homelessness, as well as attributions, peers, and values. Thus, interaction effects were created by multiplying unemployment with the relative deprivation, monetary dissatisfaction, state support, homelessness, external attribution, internal attribution, deviant peers, and deviant values measures. The theory also suggests that unemployment may have a stronger impact on crime when there is less commitment to work and more job instability. Therefore, three more interactions were created by multiplying unemployment with the work ethic, job searching, and job stability. The literature suggests that unemployment may have a greater impact when there is less fear of formal punishment. To explore this possibility, two more interactions were estimated by multiplying unemployment with the certainty and severity of punishment variables. Finally, the theory suggests that there may be interactions between peers and attributions and peers and values. Three additional interactions were created by multiplying deviant peers with

412

Revue canadienne de criminologie et de justice pe nale

juillet 2008

the external attribution, internal attribution, and deviant values variables. To decrease problems of multi-collinearity between the interaction variables and the lower-order variables from which they were created, all variables used in the interactions were first standardized, and their resultant z scores were used in the multiplication process (see Aiken and West 1991).

Results
The analysis proceeded by first entering the demographic control variables and unemployment in an OLS regression to predict anger over unemployment. In the next step the rest of the variables thought to predict anger were added to the equation. This allows us to see if the direct effect of unemployment on anger is mediated by these other variables. Interaction effects theorized to be linked to anger were then entered into the equation one at a time to examine their separate effects (see Paternoster and Mazerolle 1994). This procedure allows us to look at the impact of unemployment when it is conditioned by relative deprivation, monetary dissatisfaction, homelessness, state assistance, deviant peers, deviant values, external attributions, and internal attributions. Further, the interactions between deviant peers and the two forms of attribution are examined. The interactions involving work ethic, job search, and job instability are also included here since they may lead to anger at this stage. The unemployed may be more likely to become angry when they hold higher levels of work ethic, continue to search for employment, and have greater histories of job instability. The analysis then proceeds to predict the three different types of crime by first entering the demographic control variables and unemployment into the equations. It then moves to include all the lower-order variables, including anger, in the equations.8 Tests for interaction effects follow by entering the interaction variables one at a time to examine their separate effects as well the additional conditioning effects of certainty and severity on unemployment and the interacting role of peers and values. Predicting anger We begin by examining the link between the lower-order measures and anger. Table 2 shows that unemployment does not have a significant effect on anger when it is entered with only the control variables.

Street Youth, Unemployment, and Crime T able 2: OLS regression predicting situational anger
B (b )a SE Age Gender Minority Unemployment Length No Fixed Address State Assistance Relative Deprivation Monetary Dissatisfaction Unemployment Internal Attribution Unemployment External Attribution Job Stability Job Search Work Ethic Certainty Severity Deviant Values Deviant Peers Unemployment Unemployment Internal Attribution Unemployment Unemployment External Attribution B (b ) SE B (b ) SE B (b ) SE .070 (.057) .046 .009 (.007) .042 .052 (.042) .039 .017 (.014) .043 .067 (.056) .039 .028 (.023) .042 .091 (.075) .038 .211 (.174) .039 .204 (.167) .042 .072 (.059) .041 B (b ) SE

413

.125 (.102) .079 (.065) .091 (.074) .044 .045 .046 .023 (.019) .001 (.001) .010 (.018) .044 .042 .042 .071 (.058) .048 (.040) .050 (.041) .041 .039 .039 .056 (.046) .036 (.029) .051 (.042) .041 .043 .043 ^ .065 (.054) .059 (.049) .039 .039 ^ .029 (.024) .029 (.024) .041 .041 ^ .097 (.080) .106 (.087) .038 .038 .204 (.168) ^ .207 (.170) .039 .039 .201 (.165) ^ .213 (.174) .042 .042 ^ .072 (.059) .041 .085 (.070) .041

.078 (.064) .045 .004 (.003) .042 .049 (.040) .039 .034 (.028) .043 .062 (.051) .039 .030 (.024) .041 .096 (.080) .038 .209 (.171) .039 .212 (.174) .042 .066 (.054) .041

^ ^ ^ ^ ^ ^ ^ ^

.014 (.01 1) .010 (.008) .015 (.012) .008 (.007) .046 .045 .045 .046 .141 (.116) .137 (.113) .138 (.114) .145 (.119) .042 .042 .042 .042 .114 (.094) .117 (.097) .118 (.098) .1 17 (.097) .041 .041 .041 .041 .067 (.056) .071 (.059) .067 (.055) .074 (.061) .042 .041 .041 .042 .065 (.053) .073 (.060) .062 (.051) .058 (.048) .040 .041 .040 .040 .081 (.067) .081 (.067) .081 (.067) .083 (.068) .044 .043 .043 .043 .065 (.053) .058 (.048) .066 (.054) .060 (.049) .039 .038 .038 .039 ^ ^ .097 (.075) ^ .036

.087 (.071) .038

(Continued )

414

Canadian Journal of Criminology and Criminal Justice

July 2008

T able 2: Continued
B (b )a SE Peers Unemployment External Attribution R2 N

B (b ) SE ^

B (b ) SE ^

B (b ) SE ^

B (b ) SE .076 (.062) .037

.019 399

.233 391

.242 391

.240 391

.239 391

Sig .05 one-tailed test Sig .01 one-tailed test Sig .05 two-tailed test Sig .01 two-tailed test a B standardized coefficient; (b) unstandardized coefficient; SE standard error of unstandardized coefficient

The results do show, however, that older respondents tended to show greater anger over their unemployment. In the equation including the broader range of explanatory variables, we see that all the objective measures of poverty, unemployment, homelessness, and state assistance show no significant relationship with anger over unemployment. However, the results do suggest that the more interpretive components of poverty do have a relationship with this negative emotion. Those who feel more deprived relative to others and express more monetary dissatisfaction also express more anger over their unemployment. Further, it appears that those with a greater commitment to employment are more likely to express anger over their unemployment. Those who reported a stronger work ethic and reported looking for work more regularly also reported more anger. Moreover, contrary to expectations, it was those who felt that they lacked the proper qualifications to gain employment, or made an internal attribution for unemployment, that reported greater anger. External attributions, deviant peers, and deviant values had no impact on the level of anger reported by the youths, nor did gender or race, and the effect of age was mediated by the inclusion of all of the other variables. An examination of the interaction effects revealed three significant coefficients. First, length of unemployment was associated with lower levels of anger when youths expressed internal attributions for their unemployment. Second, the length of unemployment was associated with higher levels of anger when youths attributed the cause of their unemployment to external sources. Third, those

Street Y outh, Unemployment, and Crime

415

youths who had deviant peers and who made external attributions for their unemployment were more likely to report anger at their situation of unemployment. Thus, while not significant at the lower-order level, unemployment, deviant peers, and external attributions are important predictors of anger when included as part of interaction terms. In other words, these variables have an impact on anger when they are either conditioned by or condition other variables. Predicting Crime Model 1 in Table 3 reveals that unemployment has a direct effect on property crime when examined without the other explanatory variables. Model 2, however, suggests that this impact is mediated when all the other explanatory variables are added to the model. When the full models are examined, we see that property crime is associated
T able 3: OLS regression predicting property crime
B (b )a SE Age Gender Minority Unemployment Length No Fixed Address State Assistance Relative Deprivation Monetary Dissatisfaction Unemployment Internal Attribution Unemployment External Attribution Job Stability Job Search .086 (.195) .122 .045 (.102) .121 .015 (.033) .114 .141 (.318) .113 ^ ^ ^ ^ ^ ^ ^ ^ B (b ) SE .104 (.235) .111 .002 (.004) .118 .048 (.109) .109 .050 (.113) .120 .135 (.307) .111 .091 (.205) .115 .073 (.165) .108 .129 (.292) .112 .034 (.077) .119 .056 (.126) .115 .041 (.092) .128 .020 (.045) .118 B (b ) SE .090 (.203) .128 .008 (.018) .117 .058 (.131) .109 .051 (.116) .119 .123 (.278) .111 .091 (.206) .114 .080 (.181) .107 .137 (.310) .112 .035 (.079) .119 .057 (.130) .114 .039 (.088) .127 .018 (.042) .118

(Continued )

416

Revue canadienne de criminologie et de justice pe nale

juillet 2008

T able 3: Continued
B (b )a SE Work Ethic Certainty Severity Deviant Values Deviant Peers Anger Deviant Peers Deviant Values R2 N

B (b ) SE .051 (.115) .115 .149 (.338) .117 .081 (.183) .113 .21 1 (.478) .122 .149 (.334) .108 .019 (.043) .119 ^ .208 391

B (b ) SE .042 (.096) .115 .159 (.363) .117 .079 (.180) .112 .200 (.453) .122 .175 (.394) .111 .025 (.057) .116 .103 (.230) .110 .217 391

^ ^ ^ ^ ^ ^ ^ .027 399

Sig .05 one-tailed test Sig .01 one-tailed test Sig .05 two-tailed test Sig .01 two-tailed test a B standardized coefficient; (b) unstandardized coefficient; SE standard error of unstandardized coefficient

with longer periods of homelessness, lack of state assistance, and monetary dissatisfaction. Those who reported deviant values and deviant peers also reported more property offending. Further, those who indicated that they perceived the certainty of punishment to be low also indicated greater involvement in property offending. Anger over unemployment, however, was not significantly related to property offending. Tests for interactions revealed that property crime was more likely to take place when these youths were immersed in a deviant subculture. The interaction between deviant peers and deviant values was a significant predictor of property crime. An examination of the first model in Table 4 shows that unemployment had no direct effect on violent crime when entered into an equation with only the demographic variables. Instead, violence was associated with age and gender. The second model reveals, however, that anger over unemployment is associated with violent crime. This second model also reveals that relative deprivation and monetary dissatisfaction, rather than the more objective measures of poverty,

Street Y outh, Unemployment, and Crime T able 4: OLS regression predicting violence
B (b )a SE Age Gender Minority Unemployment Length No Fixed Address State Assistance Relative Deprivation Monetary Dissatisfaction Unemployment Internal Attribution Unemployment External Attribution Job Stability Job Search Work Ethic Certainty Severity Deviant Values Deviant Peers Anger Unemployment Monetary Dissatisfaction Peers Values .162 (.226) .074 .200 (.279) .074 .034 (.047) .070 .046 (.064) .069 ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ B (b ) SE .213 (.298) .079 .164 (.230) .073 .057 (.080) .067 .009 (.013) .074 .032 (.046) .068 .083 (.116) .071 .125 (.176) .066 .082 (.115) .069 .041 (.058) .074 .015 (.020) .071 .058 (.082) .079 .087 (.122) .073 .149 (.21 1) .071 .155 (.218) .072 .083 (.116) .070 .137 (.192) .075 .164 (.229) .067 .092 (.128) .073 ^ B (b ) SE .208 (.291) .078 .161 (.225) .064 .055 (.078) .067 .018 (.026) .074 .036 (.051) .068 .073 (.102) .071 .119 (.168) .066 .070 (.098) .069 .041 (.058) .073 .021 (.029) .071 .054 (.076) .078 .091 (.128) .073 .160 (.227) .071 .150 (.212) .072 .081 (.114) .069 .132 (.185) .075 .179 (.249) .067 .084 (.117) .073 .099 (.148) .071 ^ B (b ) SE

417

.199 (.279) .079 .170 (.238) .072 .066 (.093) .067 .008 (.011) .074 .020 (.028) .068 .083 (.112) .070 .132 (.186) .066 .090 (.126) .069 .040 (.057) .073 .013 (.018) .070 .057 (.079) .078 .088 (.124) .073 .141 (.199) .071 .165 (.233) .072 .081 (.114) .069 .126 (.177) .075 .190 (.265) .069 .086 (.120) .073 ^

.100 (.140) .068

(Continued )

418

Canadian Journal of Criminology and Criminal Justice

July 2008

T able 4: Continued
B (b )a SE R2 N

B (b ) SE .218 391

B (b ) SE .227 391

B (b ) SE .227 391

.046 399

Sig .05 one-tailed test Sig .01 one-tailed test Sig .05 two-tailed test Sig .01 two-tailed test a B standardized coefficient; (b) unstandardized coefficient; SE standard error of unstandardized coefficient

homelessness, and lack of state assistance, are related to violent crime, although additional analysis revealed that anger mediated the effect of state assistance. Further, having a strong work ethic and spending less time looking for work were related to violent offending. This model also reveals that those with deviant peers, deviant values, and lower perceptions of certainty and severity of punishment reported greater involvement in crime. Tests for interactions indicated two significant relationships. First, unemployment had a greater impact on violent crime at higher levels of monetary dissatisfaction. Second, deviant peers and deviant values interacted to predict violent crime. That is, having more deviant friends together with holding deviant values encourages involvement in a greater number of violent offences. The first model in Table 5 indicates that unemployment has a significant lower-order relationship with drug dealing, as does age. Model 2 shows that the addition of the other explanatory variables into the equation mediates these relationships. Further, Model 2 shows that youths reporting greater anger over their unemployment also reported more drug dealing. Those not on state assistance and those who expressed greater monetary dissatisfaction were also more likely to be involved in drug distribution. Reduced job-seeking behaviour was also related to dealing drugs. As with violent offences, lower perceptions of certainty and severity of punishment, along with having deviant peers and values contributed to selling illegal drugs. Tests for interactions revealed two significant relationships. Unemployment had a greater impact on drug dealing at lower levels of job search activity. Further, unemployment was associated with drug dealing at greater levels of monetary dissatisfaction. This suggests that drug dealing is more likely when unemployed individuals

Street Y outh, Unemployment, and Crime T able 5: OLS regression predicting drug sales
B (b )a SE Age Gender Minority Unemployment Length No Fixed Address State Assistance Relative Deprivation Monetary Dissatisfaction Unemployment Internal Attribution Unemployment External Attribution Job Stability Job Search Work Ethic Certainty Severity Deviant Values Deviant Peers Anger Unemployment Job Search Unemployment Monetary Dissatisfaction .106 (.402) .203 .055 (.207) .202 .041 (.156) .190 .124 (.470) .189 ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ B (b ) SE .069 (.263) .213 .062 (.235) .197 .000 (.001) .182 .059 (.225) .200 .040 (.154) .185 .091 (.345) .192 .046 (.174) .180 .094 (.356) .188 .009 (.035) .200 .086 (.324) .192 .022 (.082) .214 .192 (.732) .198 .028 (.109) .192 .140 (.534) .195 .090 (.343) .189 .162 (.616) .204 .156 (.592) .181 .120 (.454) .198 ^ ^ B (b ) SE .074 (.281) .212 .063 (.238) .196 .007 (.026) .181 .068 (.261) .200 .043 (.163) .184 .091 (.345) .191 .045 (.172) .179 .098 (.373) .187 .006 (.021) .199 .082 (.312) .191 .026 (.098) .213 .205 (.782) .198 .020 (.078) .191 .145 (.554) .194 .092 (.350) .188 .165 (.628) .203 .155 (.588) .180 .127 (.480) .198 .102 (.367) .169 ^ B (b ) SE

419

.074 (.280) .213 .059 (.223) .196 .001 (.078) .181 .050 (.192) .200 .037 (.140) .184 .082 (.310) .192 .041 (.155) .180 .083 (.315) .188 .009 (.036) .199 .080 (.302) .191 .018 (.066) .213 .196 (.746) .197 .039 (.147) .192 .136 (.520) .194 .088 (.337) .188 .157 (.598) .204 .170 (.642) .182 .112 (.426) .073 ^ .091 (.368) .192

(Continued )

420

Revue canadienne de criminologie et de justice pe nale

juillet 2008

T able 5: Continued
B (b )a SE R2 N

B (b ) SE .218 391

B (b ) SE .227 391

B (b ) SE .225 391

.034 400

Sig .05 one-tailed test Sig .05 two-tailed test Sig .01 one-tailed test Sig .01 two-tailed Test a B standardized coefficient; (b) unstandardized coefficient; SE standard error of unstandardized coefficient

have stopped looking for employment and when they are extremely unhappy with their monetary situation.

Discussion
This work has extended research on unemployment and crime by using the GST to help examine interpretive factors that enhance our understanding of the complex relationship between unemployment and crime. The results suggest that unemployment does have a direct effect on property crime and drug dealing that is mediated by other explanatory variables. The bulk of the impact of unemployment on crime is felt indirectly through the negative emotional reaction it generates in the form of anger, and more strongly directly where its impact on crime is conditioned by monetary dissatisfaction. This indirect relationship is complex in the sense that the anger over unemployment is not the direct result of the unemployment experience itself. Instead, anger emerges only among those who attribute the cause of their unemployment to external sources and who do not attribute their unemployment to personal deficiencies. Further, these important external attributions are influenced by ones peer group. Thus, without core attributional components and peer networks, unemployment would not lead to anger. But anger over unemployment is not only the result of not having a job. Anger over unemployment is also the result of subjective interpretations of economic circumstances, ones commitment to the labour market, and how one assigns blame for unemployment. Feelings of relative deprivation and monetary dissatisfaction along with high levels of job searching without success, and perceptions of deficient qualifications while holding fast to a strong work ethic, act together to generate

Street Y outh, Unemployment, and Crime

421

anger at the circumstances of prolonged unemployment. In turn, anger over unemployment predicts two out of the three types of crime examined (violent crime and drug dealing). The strong direct link between unemployment and crime is also complex. First, the findings suggest unemployment is more likely to have an impact on crime when people are unhappy with their monetary situation. Thus, the subjective interpretation of ones economic condition when unemployed is vital. This economic dissatisfaction is also a consistent predictor of crime by itself, reaching significance in all three equations. Further, relative deprivation another subjective measure of deprivation emerged as an important predictor of violent crime. However, unemployment also appears to have a greater impact on certain offences, such as drug dealing, when people have dropped out of the labour market. The direction of causality may be questioned here. It may be that continued failed job searches lead youths to search out work in what are seen as more lucrative deviant service sectors, or perhaps success in these alternative activities leads one to stop looking for work and to prolong the duration of unemployment. It also appears that more objective measures of deprivation associated with the unemployment experience rather than unemployment itself have a stronger direct link to crime. The lack of state support was associated with crime in two cases (property crime and drug dealing), and its effect was mediated by anger in the third (violent crime). This finding is consistent with past aggregate level work, which has demonstrated that state support can decrease crime (see DeFronzo 1983, 1996, 1997; Hannon and DeFronzo 1998). Furthermore, homelessness also emerged as a predictor of property offending. The findings also suggest that having a strong commitment to paid work that is frustrated through prolonged bouts of unemployment can lead to crime. Having a strong work ethic leads indirectly to violent and drug crime through its effect on anger and has a direct effect on violent crime. Further, those who searched actively for employment but were persistently rebuffed were more likely to be angry, and this led to violent crime and drug dealing. At the same time, those who have given up searching for employment are more likely to become involved in violent offending and drug dealing. The findings on job searches may be interpreted from a strain perspective or a control perspective, depending on where this variable appears significant in the causal process, and may also be a reflection of behavioural change

422

Canadian Journal of Criminology and Criminal Justice

July 2008

across time. It may be that, earlier in the unemployment experience, people continue to look for work and are angry and react with crime. Over time, however, they detach from the labour market and their behaviour is more understandable from a social control perspective. Or there may be two different types of people on the street who are unemployed: those who were attached to society and frustrated by their inability to succeed, and those who were never attached and commit crime as a result of their lack of social control (see Elliott et al. 1979). Thus, there may be more than one causal pathway. The findings also suggest that the lack of participation in the labour market combined with homelessness leaves youths to become absorbed in an environment where contact with and attachment to conventional society are decreased, leaving youths at even greater risk for involvement in criminal activities. The findings suggest that peers and values have an impact on all forms of crime, and these interact to predict violent and property offending. In other words, the street provides access to criminal peers who provide pressure for, and support in, carrying out criminal activities. It is likely that, as time progresses, street youth restrict their interactions to others who spend their time predominantly on the street peers who are seriously criminally inclined and who support and facilitate criminal pursuits. These friendship networks coalesce among youths with similar life situations who may be caught up in and supporting the street lifestyle. Further, these youths appear not to fear the threat of legal punishments. Living on the margins of society for long periods of time, lacking shelter, and being angry and dissatisfied about their economic condition leaves these youths fear of apprehension diminished and their perceptions of the severity of punishment undermined. It may be prior punishment experience as well as the experience of peers that contributes to this lack of fear (see Stafford and Warr 1993). Together, these perceptions of punishment allow these youths to engage in crime to satisfy their needs, although these perceptions do not influence the impact of unemployment on crime. A limitation of the current research is that these data were crosssectional by design. Thus, current emotional reactions, perceptions of deprivation, and punishment were correlated with past criminal behaviour. It may be that participation in crime increased anger over unemployment, drove up perceptions of deprivation, and strengthened monetary dissatisfaction, while decreasing perceptions

Street Youth, Unemployment, and Crime

423

regarding the odds of apprehension and the severity of punishment. Further, participation in crime could have led to increased association with criminal peers and the adoption of values supportive of criminal behaviour. While the longitudinal research that suggests that both strain and offending are relatively consistent and stable across time provides some reassurance for the results here (Agnew 1989; Mosher et al. 2002), future research is still required to verify these findings. Further, the results of this work cannot be generalized beyond the street youth population. Hagan and McCarthy (1997) note that research using street samples is important in that it can determine the causal processes linked to their criminal behaviour. The findings from these samples can then be utilized to make decisions about future research to verify findings in broader populations. They note that neither street or non-street samples by themselves will be able to provide the information required to understand crime. Thus, future work is required using broader or comparative samples to see how the findings here apply to others who are unemployed but not on the street. The use of the general strain perspective has allowed us to extend the previous research on street youth, unemployment, and crime in a number of ways. First, the theory pointed to the need to include a number of variables not included in prior analyses. The focus on subjective strains, like relative deprivation and monetary dissatisfaction, outlined in the general strain theory encouraged us to go beyond the traditional measures of financial status and length of unemployment. The theory also directed us to the importance of including social control variables like work ethic and actual job searching that go beyond determining lifetime job totals and a simple desire for employment. The theory also provided guidance on the need to include perceptions of punishment risk, which has been overlooked in past work. Further, while prior research stressed the importance of external attributions, it did not focus on the importance of internal attributions included here. Thus, the current research has been able to include a number of variables identified by general strain theory that have been overlooked in past research on street youth unemployment and crime, examine them together in an analysis to control for all their various influences, and determine that all of these variables play a central role in understanding the nature of unemployments relationship with crime.

424

Revue canadienne de criminologie et de justice pe nale

juillet 2008

Second, the current work, guided by GST, has explored a more complex causal path between street youth unemployment and crime than has previous research. While past research acknowledged the importance of anger, it did not explore its mediating role between unemployment and crime, and, unlike past work using situational measures of anger, was able to show the important relationship between this emotion and some forms of criminal activity. Further, the GST enabled the current work to explore for the first time how a wide range of subjective and objective strains, internal and external explanations for unemployment, and cultural and social supports can lead to anger over unemployment. Moreover, unlike other research, this allowed for an investigation of how these variables condition unemployments impact on anger and unemployments relationship with crime. The GST has also shed light into the important social influences of peers and the way they can sway causal attributions for unemployment and how this can generate anger. Likewise, the analysis advanced our understanding by directing us to the important interactive role of peers and culture in creating crime, allowing us to move beyond the traditional examination of their separate influences. In sum, by using general strain theory to identify additional important variables, establish a more complex causal process involving anger as a mediating variable, and outline a series of moderating effects not previously examined, this research has significantly broadened our understanding of the nature of the relationship between unemployment and crime in the street youth population. This research allows us to implicate the connection and interaction between objective labour market conditions and peoples perceptions of their experiences to their emotional reactions, which has not been accomplished previously. In turn this enables us to demonstrate how together these objective, subjective, and emotional factors foster criminal behaviour. This approach helps to explain why people, who at first glance appear to share the common experiences of unemployment, do not react in the same fashion in response to their conditions. There are variations in these adverse conditions that influence peoples interpretations of, and emotional reactions toward, their circumstances, and it is these differences and the way they are contextualized by actors that lead to criminal behaviour. Future work would do well to continue the exploration between the labour market and peoples reactions to it, utilizing both conventional and at risk populations. It might explore further why some actors

Street Youth, Unemployment, and Crime

425

react with anger and dissatisfaction, perceive injustice, or blame others for their conditions. It would also seem appropriate to examine differences between certain groups and their reactions to unemployment. For example, Agnew (2006) has outlined how race, gender, and age are related to forms of strain that will affect crime. He notes that there may be racial differences in exposure to strains conducive to crime, such as chronic unemployment and work in the secondary labour market. He argues, however, that the relationship between race and crime will be diminished if we take into account some of the socio-economic factors examined in this study. Nevertheless, he suggests that we must also take into account other strains conducive to crime not included here, including discrimination in the labour market and other sites, which he argues increase the likelihood that strains will be attributed to the deliberate acts of others, strengthening perceptions that the strain is unjust. The findings here showed that being of minority status was not associated with levels of offending, but there may still be differences in emotional reactions, economic dissatisfaction and perceptions of deprivation, and the way that these affect crime may vary across groups. Agnew (2006; see also Broidy and Agnew 1997) has also outlined how males and females may place different emphases on economic strains, have dissimilar emotional reactions in response to strain, and be differentially exposed to moderating factors like peers and values that can lead to criminal coping strategies. He also argues, however, that the causal process for males and females will be similar if they experience strain similarly, respond with like emotions, and encounter comparable conditioning factors. The results here show that gender was associated with levels of violent crime but not anger, property offending, or drug dealing. The effect, however, was not mediated to any large degree when other variables were added to the analysis, suggesting that something beyond strain explains the relationship between gender and violence. Baron (2007), working with a street youth sample, found few gender differences in exposure to economic strain, perceptions of deprivation, or emotional reactions in response to the strain, and discovered that these variables tended to predict crime similarly for both males and females. He did find, nevertheless, that there were gender differences in the impact of some of the interaction effects. Agnew (2006) has also suggested that adolescents are subject to more strain than adults, have fewer coping mechanisms available to deal with their strain in a non-criminal manner, and are more likely to

426

Canadian Journal of Criminology and Criminal Justice

July 2008

respond to strain with crime. He also outlines, however, that adults, particularly those in the urban underclass, experience a great deal of subjective and objective economic strain that is conducive to their involvement in crime. Evidence suggests that while employment can reduce criminal earnings of all age groups (Uggen and Thompson 2003) it may be more important as youths age and begin to move into their twenties (see Uggen 2000). The findings here suggest younger street youths were more associated with violent crime, but older youths were more likely to be angry about their unemployment and to deal drugs, and these later relationships were mediated when other variables were entered into the models. This suggests that the effect of unemployment may vary with age, and further work should explore how age affects emotional reactions and interpretations leading to criminal responses. Strategies to explore the differential impact of these demographic factors would include gathering large enough samples to ensure an adequate number of cases to examine the range of variables explored here, together in a comparative multivariate analysis where differences could be assessed through coefficient comparison tests (see Paternoster, Brame, Mazerolle, and Piquero 1998). Altogether, this type of integrative and comparative work would allow researchers to map commonalities and to establish the differences between these groups of homeless and their responses to conditions of unemployment.

Notes
1. The author acknowledges the financial support of the Social Sciences and Humanities Research Council of Canada and the Queens University Chancellors Research Award. 2. Agnew (2001) has also outlined how other background experiences of homeless youth, including victimization in the home, may affect their behaviour. However, Hagan and McCarthy (1997; see also McCarthy and Hagan 1992) found that while these background and developmental variables were vital in understanding why these youths are on the street, their effect on crime was mediated by measures of current economic adversity that had the direct effects on crime. Since the focus of this research is on current economic adversity we have not included background factors, recognizing that they are important in understanding how youths end up on the street.

Street Y outh, Unemployment, and Crime

427

3. Those who agreed were supplied with informed consent forms outlining study goals and their rights within the interview. Subjects were told they were not obliged to answer any of the questions and could withdraw from the interview at any time. None of the youths exercised this power. 4. Some readers may be concerned about the reliability and validity of data gathered in this population through the use of surveys. Hagan and McCarthy (1997), using a similar approach, provide evidence that street youth provide reliable and valid responses to the types of questions used here (see also Calsyn, Allen, Morse, Smith, and Tempelhoof 1993). 5. Aboriginals are drastically over-represented in the sample. According to Peters and Murphy (1993) only about one per cent of the youths in the city schools are Native. We cannot determine, however, if the percentage of minority youth is representative of those on the street, since traditional social science techniques of sampling from an exhaustive list of eligible participants where various parameters are already known is impossible. 6. Scales were created by summing the response scores across items and then dividing the scores by the number of items. Cronbachs alpha was then calculated to determine the validity of each scale. 7. As a general rule, the higher the alpha coefficient the better. Carmines and Zellers (1979) suggest that reliabilities for widely used scales should not be below .80. They do admit, however, that the alpha is a conservative test for reliability. Garrett (1966) point out that the size of the reliability coefficient required depends on the size and variability of the group. They argue that in some cases a reliability coefficient need be no higher than .50 or .60. The more heterogeneous the group the higher the reliability coefficient needed. The measure just reported and those that follow are between .62 and .68 and thus should be explored with some caution. However, the sample, while certainly containing variation, is atypical and probably somewhat more homogenous than other populations sampled in social sciences. They are restricted by age, housing status, and other study-imposed parameters. Thus, we believe that with caution we can make descriptive explorations and initial predictive modelling using these measures. 8. The GST also outlines that anger may mediate the impact of strain. The equations predicting crime were first run without anger included, to examine if any of the variables that significantly predicted anger significantly predicted the dependent variables. The equations were then run with anger included, to see if the impact of any of these variables

428

Revue canadienne de criminologie et de justice pe nale

juillet 2008

was reduced to non-significance. This proved not to be the case in any of the three models. However, in the model predicting violent crime, state assistance predicted crime when anger was not in the model and became non-significant when anger was added to the model.

References
Agnew, Robert 1989 A longitudinal test of revised strain theory. Journal of Quantitative Criminology 5: 373387. Agnew, Robert 1992 Foundation for a general strain theory of crime and delinquency. Criminology 30: 4766. Agnew, Robert 1997 The nature and determinants of strain: Another look at Durkheim and Merton. In Nikos Passas and Robert Agnew (eds.), The Future of Anomie Theory. Boston, MA: Northeastern University Press. Agnew, Robert 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: 31961. Agnew, Robert 2006 Pressured into Crime. Los Angeles: Roxbury. Agnew, Robert, Francis T. Cullen, Velmer S. Burton, T. David Evans, and R. Gregory Dunaway 1996 A new test of classic strain theory. Justice Quarterly 13: 681704. Aiken, Leona S. and Stephen G. West 1991 Multiple Regression: Testing Newbury Park, CA: Sage.

and

Interpreting

Interactions.

Aseltine, Robert H., Susan Gore and Jennifer Gordon 2000 Life stress, anger, anxiety, and delinquency: An empirical test of general strain theory. Journal of Health and Social Behavior 41: 256275. Bachman, Jerald G., Patrick OMalley and Jerome Johnston 1978 Youth in Transition, Vol. 6, Adolescence to Adulthood. Ann Arbour: Institute of Social Research.

Street Y outh, Unemployment, and Crime

429

Baron, Stephen W. 2004 General strain, street youth, and crime: A test of Agnews revised theory. Criminology 42: 457483. Baron, Stephen W. 2006 Street youth, strain theory, and crime. Journal of Criminal Justice 34: 209223. Baron, Stephen W. 2007 Street youth, gender, financial strain, and crime: Exploring Broidy and Agnews extension to general strain theory. Deviant Behavior 28: 273302. Baron, Stephen W. and Timothy F. Hartnagel 1997 Attributions, affect and crime: Street youths reactions to unemployment. Criminology 35: 409434. Bellair, Paul. E. and Vincent J. Roscigno 2000 Local labor-market opportunity and adolescent delinquency. Social Forces 78: 15091538. Bernard, Thomas 1990 Angry aggression among the truly disadvantaged. Criminology 28: 7395. Box, Steven 1987 Recession, Crime and Punishment. Basingstoke, U.K.: Macmillan. Box, Steven and Chris Hale 1985 Unemployment, imprisonment and prison overcrowding. Contemporary Crises 9: 209228. Britt, Chester L. 1997 Reconsidering the unemployment and crime relationship: Variation by age group and historical period. Journal of Quantitative Criminology 13: 405428. Britt, Chester L. 2001 Testing theory and the analysis of time series data. Journal of Quantitative Criminology 17: 343357. Broidy, Lisa M. and Robert Agnew 1997 Gender and crime: A general strain theory perspective. Journal of Research in Crime and Delinquency 38: 362386.

430

Canadian Journal of Criminology and Criminal Justice

July 2008

Calsyn, Robert, Gary Allen, Gary Morse, Ruth Smith and Betty Tempelhoof 1993 Can you trust self-report data provided by homeless mentally ill individuals? Evaluation Review 17: 353366. Cantor, David and Kenneth C. Land 2001 Unemployment and crime rate fluctuations: A comment on Greenberg. Journal of Quantitative Criminology 17: 329342. Cantril, Hadley 1965 The Pattern of Human Concerns. New Brunswick, NJ: Rutgers University Press. Carlson, Susan M. and Raymond J. Michalowski 1997 Crime, unemployment, and social structures of accumulation: An inquiry into historical contingency. Justice Quarterly 14: 209241. Carmines, Edward G. and Richard A. Zeller 1979 Reliability and validity assessment. Beverly Hills: Sage. Cernkovich, Stephen A., Peggy C. Giordano and Jennifer Rudolph 2000 Race crime and the American dream. Journal of Research in Crime and Delinquency 37: 13170. Chamlin, Mitchell B. and John K. Cochran 2000 Unemployment, economic theory, and property crime: A note on measurement. Journal of Quantitative Criminology 16: 433455. Chiricos, Theodore 1987 Rates of crime and unemployment. Social Problems 34: 187212. Cloward, Richard and Lloyd Ohlin 1960 Delinquency and Opportunity. New York: Free Press of Glencoe. Cohen, Albert 1955 Delinquent Boys. New York: Free Press of Glencoe. Crowley, Joan E. 1984 Delinquency and employment. In Micheal Borus (ed.), Youth in the Labor Market. Kalamazoo, MI: Upjohn Institute for Employment Research. Crutchfield, Robert D. and Susan R. Pitchford 1997 Work and crime: The effects of labor market stratification. Social Forces 76: 93118.

Street Youth, Unemployment, and Crime

431

DeFronzo, James 1983 Economic assistance to impoverished Americans. Criminology 21: 119136. DeFronzo, James 1996 Welfare and burglary. Crime and Delinquency 42: 223230. DeFronzo, James 1997 Welfare and homicide. Journal of Research in Crime and Delinquency 34: 395406. Elliott, Delbert S. 1994 Serious violent offenders: Onset, developmental course, and termination. Criminology 32: 122. Elliott, Delbert S. and Susan Ageton 1980 Reconciling race and class differences in self-reported and official estimates of delinquency. American Sociological Review 45: 95110. Elliott, Delbert S., Susan S. Ageton and Rachel J. Cantor 1979 An integrated theoretical perspective on delinquent behavior. Journal of Research in Crime and Delinquency 16: 327. Farrington, David P., Bernard Gallagher, Lynda Morely, Raymond J. St Ledger and Donald J. West 1986 Unemployment, school leaving, and crime. British Journal of Criminology 26: 335356. Garrett, Henry E. 1966 Statistics in psychology and education. Toronto: Longmans, Green. Greenberg, David F. 2001 Time series analysis of crime rates. Journal of Quantitative Criminology 17: 291327. Hagan, John and Bill McCarthy 1997 Mean Streets. Cambridge University Press: Cambridge. Hale, Chris and Dima Sabbagh 1991 Testing the relationship between unemployment and crime: A methodological comment and empirical analysis using time series data from England and Wales. Journal of Research in Crime and Delinquency 28: 400417.

432

Revue canadienne de criminologie et de justice pe nale

juillet 2008

Hannon, Lance and James DeFronzo 1998 Welfare and property crime. Justice Quarterly 15: 273288. Hartnagel, Timothy F. 1998 Labour-market problems and crime in the transition from school to work. Canadian Review of Sociology and Anthropology 35: 436459. Hartnagel, Timothy F. and Harvey Krahn 1989 High school dropouts, labour market success, and criminal behavior. Youth and Society 20: 416444. Hindelang, Michael J., Travis Hirschi and Joseph G. Weiss 1979 Correlates of delinquency: The illusion of the discrepancy between self-report and official measures. American Sociological Review 44: 9951014. Hindelang, Michael J., Travis Hirschi and Joseph G. Weiss 1981 Measuring Delinquency. Beverly Hills, CA: Sage. Hoffman, John. P. and Timothy Ireland 1995 Cloward and Ohlins strain theory reexamined: An elaborated theoretical model. In Freda Adler and William S. Laufer (eds.), Advances in Criminological Theory. Vol. 6, The Legacy of Anomie Theory. New Brunswick, NJ: Transaction. Inciardi, James A., Ruth Horowitz and A.E. Pottieger 1993 Street Kids, Street Drugs, Street Crime. Belmont, CA: Wadsworth. Kleck, Gary and Ted Chiricos 2002 Unemployment and property crime: A target specific assessment of opportunity and motivation as mediating factors. Criminology 40: 649680. Krivo, Lauren J. and Ruth D. Peterson 2004 Labor market conditions and violent crime among youth and adults. Sociological Perspectives 47: 485505. Land, Kenneth C., David Cantor and Stephen T. Russell 1995 Unemployment and crime rate fluctuations in the postWorld War II United States. In John Hagan and Ruth D. Peterson (eds.), Crime and Inequality. Stanford: Stanford University Press. Levitt, Steven D. 2001 Alternative strategies for identifying the link between unemployment and crime. Journal of Quantitative Criminology 17: 377390.

Street Youth, Unemployment, and Crime

433

Long, Sharon K. and Ann D. Witte 1981 Current economic trends. In Kevin N. Wright (ed.), Crime and Criminal Justice in a Declining Economy. Cambridge, MA: Oelgeschlager, Gunn. and Hain. McCarthy, Bill and John Hagan 1992 Mean streets: The theoretical significance of situational delinquency among homeless youths. American Journal of Sociology 98: 597627. Mosher, Clayton J., Terrence D. Miethe and Dretha Phillips 2002 The Mismeasure of Crime. Thousand Oaks, CA: Sage. OBrien, Robert M. 2001 Theory, operationalization, identification, and the interpretation of different differences in time series models. Journal of Quantitative Criminology 17: 359375. Orsagh, Thomas and Ann Dryden White 1981 Economic status and crime. Journal of Criminal Law and Criminology 72: 10551071. Passas, Nikos 1997 Anomie, reference groups, and relative deprivation. In Nikos Passas and Robert Agnew (eds.), The Future of Anomie Theory. Boston, MA: Northeastern University Press. Paternoster, Raymond, Robert Brame, Paul Mazerolle and Alex Piquero 1998 Using the correct statistical test for the equality of regression coefficients. Criminology 36: 859866. Paternoster, Raymond and Shawn D. Bushway 2001 Theoretical and empirical work on the relationship between unemployment and crime. Journal of Quantitative Criminology 17: 391407. Paternoster, Raymond and Paul Mazerolle 1994 General strain theory and delinquency. Journal of Research in Crime and Delinquency 31: 235263. Peters, Larry and Aileen Murphy 1993 Adolescent Health Survey: Report for the Greater Vancouver Region of British Columbia. Vancouver, BC: McCreary Centre Society. Shavit, Yossi and Arye Ratner 1988 Age and crime and the early life course. American Journal of Sociology 93: 14571470.

434

Canadian Journal of Criminology and Criminal Justice

July 2008

Stafford, Mark C. and Mark Warr 1993 A reconceptualization of general and specific deterrence. Journal of Research in Crime and Delinquency 30: 123135. Thornberry, Terrence P. and R.L. Christenson 1984 Unemployment and criminal involvement. American Sociological Review 49: 398411. Uggen, Christopher 2000 Work as a turning point in the life course of criminals: A duration model of age, employment, and recidivism. American Sociological Review 67: 529546. Uggen, Christopher and Melissa Thompson 2003 The socioeconomic determinants of ill-gotten gains: Within-person changes in drug use and illegal earnings. American Journal of Sociology 109: 146185. Walker, Ian and Leon Mann 1987 Unemployment, relative deprivation, and social protest. Personality and Social Psychology Bulletin 13: 27583. Whitbeck, Les B. and Dan R. Hoyt 1999 Nowhere to Grow. Hawthorne, NY: de Gruyter. Witt, Robert, Alan Clarke and Nigel Fielding 1999 Crime and economic activity. British Journal of Criminology 39: 391400. Zimring, Frank E. and G.J. Hawkins 1973 Deterrence: The Legal Threat in Crime Control. Chicago: University of Chicago Press.

Vous aimerez peut-être aussi