|
In recent years, interest in strain theory has been revived with ever
increasing breadth. Many tests of strain theory remain true to the hypothesis
of earlier versions of strain theory (Merton 1938; Cohen 1955; Cloward
and Ohlin 1959, 1961) that structural strain is considered a cause of
crime/delinquency. Agnew's (1992) revision of strain theory into a more
general strain theory shifted the focus from social structural to social
psychological (Broidy 2001), thus alleviating much of the criticism plaguing
earlier versions. Agnew's greatest contribution from this revision of
strain theory has been an explication of the factors that condition the
strain-crime relationship. In addition to expanding the scope of sources
of strain, Agnew and others have attempted to increase the comprehensiveness
of other processes involved in strain theory. These expansions have provided
more specification on criminal motivation within the strain-crime relationship
(Agnew 1992), specification of types of strain (Agnew 2001), an examination
of gender differences (Broidy and Agnew 1997), and the consideration of
structural effects which condition the strain-crime relationship (Agnew
1999), even, life-course (Agne, 1997) and biological related aspects of
strain (Walsh 2000). Some of these theoretical "elaborations"
(Wagner and Berger 1985) or expansions of general strain theory have received
very limited or no empirical testing (e.g., structural, life-course, biological).
Often expansions of strain theory have been guided by statements and findings
made in previous studies (e.g., Agnew 1983, 1984, 1985). In his recent
theoretical presentation of a structural/macro version of general strain
theory, Agnew (1999: 128) argues that
community characteristics will have a significant direct effect on
individual crime after individual-level variables are controlled. Communities
also have an indirect effect on strain by influencing individual traits
and the individual's immediate social environment.
While the structural/macro version of general strain theory (Agnew 1999)
was not explicitly advanced as a multilevel explanation of the effect
of strain on crime, this statement raises the tantalizing possibility
that general strain theory may also be conceptualized and empirically
tested as a multilevel integrated theory. It is this possible expansion
of strain theory that the present study explores.
Initially developed as a micro-level social psychological theory, Agnew's
(1992) general strain theory (GST) hypothesizes that crime and delinquency
result from certain adaptations to strain. Agnew defines strain as "negative
or aversive relations with others" (Agnew 1992: 61). General strain
theory delineates three major types of strain that may lead to deviant
behavior (Agnew 1992: 59): failure to achieve positively valued goals,
removal of positively valued stimuli, and presentation of negative stimuli.
Agnew posits that an individual will experience at least one negative
emotion, negative affect, per experience of strain. These negative
emotions may span a broad spectrum ranging from depression to anxiety
to despair. However, Agnew argues that anger, one of the most potent reactive
emotions, producing a desire for retribution, may be key to strain-induced
deviance (Agnew 1992).
Whether or not negative affect leads to an illegitimate response depends
on individual coping strategies. Agnew describes three forms of coping
strategies: cognitive, emotional, and behavioral (Agnew 1992: 69). In
addition to coping strategies, Agnew discusses internal and external factors
that may condition the effects of strain. These conditioning factors range
from environmental variables and the nature of social support structures
to individual characteristics such as temperament, intelligence, and beliefs
(Agnew 1992: 70-73). The form of an individual's coping strategy conditioned
by environmental and personal factors directly affects how the individual
will adapt to strain.
Several studies have provided empirical support for the propositions Agnew
has set forth in general strain theory. A significant positive relationship
between various strain measures and delinquency has been reported (Agnew
1985, 1989, 2002; Agnew and Brezina 1997; Agnew and White 1992; Aseltine,
Gore, and Gordon 2000; Baron and Hartnagel 1997; Brezina 1998, 1999; Broidy
2001; Capowich, Mazerolle, and Piquero 2001; Eitle and Turner 2003; Hoffmann
and Cerbone 1999; Hoffmann and Miller 1998; Hoffmann and Su 1997; Maxwell
2001; Mazerolle 1998; Mazerolle, Burton, Cullen, Evans, and Payne 2000;
Mazerolle and Maahs 2000; Mazerolle and Piquero 1997, 1998; Mazerolle,
Piquero, and Capowich 2003; Paternoster and Mazerolle 1994; Piquero and
Sealock 2000).
On the other hand, empirical studies of the indirect relationship between
strain and delinquency, when mediated by negative affect, have been less
consistent. Strain has been significantly associated with anger or negative
affect (Agnew 1985; Agnew, Brezina, Wright, and Cullen 2002; Aseltine
et al. 2000; Bao, Haas, and Pi 2004; Benda and Corwyn 2001; Brezina 1996,
1998; Capowich et al. 2001; Hay 2003; Jang and Johnson 2003; Mazerolle
and Piquero 1997, 1998; Piquero and Sealock 2000), but the direction and
role of anger as a mediating variable on certain types of delinquency
is unclear. For example, some findings have suggested that anger may be
limited in its role as a mediator for the strain-delinquency relationship
to measures of violence or interpersonal aggression only, not acts of
non-violent behavior (e.g., property crimes) or substance use (Aseltine
et al. 2000; Piquero and Sealock 2000). Moreover, Mazerolle and associates
(2000) demonstrate that it is actually strain that mediates the relationship
between anger and delinquency. Another study conducted by Mazerolle and
associates (2003) suggests that differences in the types of anger (i.e.,
situational versus trait) may explain some of these inconsistencies. Other
studies (Aseltine et al. 2000; Bao et al. 2004; Broidy 2001; Hay 2003;
Piquero and Sealock 2000) have examined alternative measures of negative
affect, such as anxiety, depression, resentment, and guilt, and found
mixed results.
Empirical research examining forms of individual coping strategies, posited
to directly affect how the individual adapts to strain, have also lacked
empirical consistency. These studies include measures of conditioning
factors of the strain-delinquency relationship such as self control, self-esteem,
self-efficacy, delinquent peers, family communication, moral beliefs,
religiosity, and social support (Agnew and White 1992; Aseltine et al.
2000; Capowich et al. 2001; Eitle and Turner 2003; Hay 2003; Hoffmann
and Cerbone 1999; Hoffmann and Miller 1998; Jang and Johnson 2003; Mazerolle
and Maahs 2000; Mazerolle and Piquero 1997, 1998; Paternoster and Mazerolle
1994; Peter, LaGrange, and Silverman 2003; Piquero and Sealock 2004).
Although most tests of general strain theory have followed Agnew's (1992)
initial micro-level statement of the theory, Agnew has continued to elaborate
the general strain theoretical model. Recently, Agnew (1999) proposed
an expanded version of general strain theory that provides macro-social
implications for explaining crime (referred to as MST henceforth). In
this new theoretical elaboration, Agnew proposes a model that uses GST
to help explain differences in crime rates within differing communities.
Agnew argues that structural community characteristics (e.g., economic
deprivation, high inequality, etc.) lead both directly and indirectly
to high crime rates. While he acknowledges the ability of other theories
(e.g., social disorganization and subcultural deviance) to explain crime
rates and inference to a relationship between community differences in
crime and strain, he contends these theories have been lacking in their
explication of motivational processes of crime (Agnew 1999: 126). Therefore,
Agnew presents MST as a supplemental element to other macro-social theories
of crime; one that addresses the motivational aspect while acknowledging
other influences like social control and subcultural values (see social
disorganization and subcultural deviance theories) (Agnew 1999: 147).
Agnew suggests that variation in the propensity to commit crime within
disadvantaged communities depends on the "strainful" experiences
of individuals within these communities (Agnew 1999: 125). According to
MST, the variation in community crime/delinquency rates indirectly depends
on the levels of aggregate strain, aggregate negative affect/anger, and
other stressful community conditions (for a description of the sources
of strain, anger, and other conditioning variables within the community,
see Agnew 1999). Communities characterized as highly disadvantaged create
strain and anger by blocking community members' abilities to achieve positive
goals, creating a loss of positive stimuli, exposing members to negative
stimuli, and increasing overall relative deprivation (Agnew 1999:126-130).
Moreover, MST suggests that disadvantaged communities are more likely
to select and retain strained individuals and have higher levels of angry
individual interaction than communities less disadvantaged (i.e., interpersonal-friction
argument (Brezina et al. 2001)).
There have been very few tests of MST. Warner and Fowler (2003) recently
examined MST using neighborhood level data, defined by 1990 US Census
block groups, and aggregated individual surveys. Their findings showed
mixed support for the model. Specifically, their study found neighborhood
levels of disadvantage and stability significantly affected neighborhood
strain, and neighborhood strain was positively associated with neighborhood
violence. However, this relationship was not moderated by a conditioning
factor of neighborhood informal control. Hoffmann (2002) conducted a contextual,
multilevel analysis of differential association, social control, and strain
theories using 1990 US Census characteristics aggregated to the zip code
level and individual level data for tenth graders drawn from the National
Educational Longitudinal Study (NELS). Their results indicate strain,
as measured by individual negative life events and monetary strain, predicts
delinquency behavior among youths. Community (zip codes) characteristics
significantly affected delinquency; however, this relationship was not
mediated by individual-level variables. Brezina et al. (2001) have also
provided a multilevel test of MST using school-level and individual-level
data obtained from two waves of the Youth in Transition (YIT) data set.
They tested the effects of anger, school commitment, and values in favor
of aggression on aggressive/disruptive student behavior, controlling for
race, family stability, residential stability, socioeconomic status, and
school size. Their results provided partial support for a multilevel version
of general strain theory. School-level anger significantly, positively
affects student conflicts with peers, but not student aggressive behavior.
They observed that students were more likely to display aggressive behaviors
toward other students when the overall school anger level was high. Hoffman
and Ireland (2004) provide the last multilevel test of MST. They used
school-level and individual level data from the National Education Longitudinal
Study (NELS) data. Their test examined the conditioning effects of illegitimate
opportunity structures on the strain/stress-delinquency relationship.
They found that their measures of strain and stressful life events influenced
changes in both adolescent involvement in delinquency and in adolescent
self-concept over time. However, these relationships were uniform across
illegitimate opportunity structures, suggesting little to no evidence
of the multilevel conditioning effects implied by MST.
The study proposed here examines the efficacy of MST as a means to predict
individual differences in both strain and anger as outcomes of community-level
characteristics and to condition their influence on delinquency. Similar
to Brezina et al. (2001) and Hoffmann (2002), the proposed study utilizes
a multilevel approach. However, the proposed study includes measures of
both strain and anger.
THE PRESENT STUDY
Although Agnew's (1999) MST is modeled strictly at the macro-social level,
a multilevel approach to a general strain theory of crime is also tenable
(Brezina et al. 2001). Indeed, Agnew argues that:
community characteristics will have a significant direct effect on
individual crime after individual-level variables are controlled. Communities
also have an indirect effect on strain by influencing individual traits
and the individual's immediate social environment (Agnew 1999: 128).
In addition, Agnew states, "Crime rates are an aggregation of individual
criminal acts, so these [macro] theories essentially describe how community-level
variables affect individual criminal behavior." (Agnew 1999: 123).
Based on these statements, this study examines the degree to which community
characteristics influence individual levels of strain, negative affect,
and delinquency and whether the effects of strain and negative affect
on individual delinquency are more salient within communities characterized
by higher levels of social and economic disadvantage.
Similar to Agnew's (1999: 129) model of community differences and general
strain, Figure 1 predicts relationships between community characteristics,
strain, negative affect, and delinquency. Unlike Agnew's macro-level model,
strain, negative affect, and crime are measured at the individual level.
Figure 1 attempts to explain how disadvantaged communities interact with
an adolescent's ability to cope with strain. This exploratory model only
considers the motivation for individual crime, not differences in social
control and subcultural values; thus representing a conservative test
of a multilevel version of MST.
Figure 1. A Multilevel Model of Community Difference
and Individual Self-Reported Delinquency.
The model presented in Figure 1 poses one overall question: do the effects
of individual strain and negative affect on self-reported delinquency
vary by neighborhood context? Although not exclusive to the theoretical
tenants of MST, community characteristics may directly affect crime/delinquency.
Despite predictions from many venerable theories of crime for a relationship
between community or structural characteristics and individual crime (cf.
Durkheim 1951[1897]; Merton 1968; Shaw and McKay 1969; Colvin and Pauly
1983; Hagan, Gillis, and Simpson 1985; Akers 1998), empirical studies
of this relationship have been scarce. Although many of the findings have
been weak, empirical studies examining the relationship between structural
characteristics and individual delinquency suggest that there is a causal
link, both direct and indirect, between the community and the individual
(cf. Reiss and Rhodes 1961; Krohn, Lanza-Kaduce, and Akers 1984; Simcha-Fagan
and Schwartz 1986; Gardner and Shoemaker 1989; Rosenbaum and Lasley 1990;
Gottfredson, McNeil, and Gottfredson 1991; Sampson, Raudenbush, and Earls
1997; Cattarello 2000).
According to MST, community characteristics may also have indirect effects
on crime/delinquency. Similar to Agnew's (1999) MST argument, Figure 1
contends characteristics of disadvantaged communities (e.g., economic
inequality and racial inequality) contribute to levels of individual strain
and individual negative affect. Based on Agnew's GST (1992) assumption
that strain and negative affect are major sources of delinquent motivation,
individuals within these disadvantaged communities will be more likely
to be delinquent. Individual measures of strain may both directly and
indirectly lead to individual delinquency. Indirectly, the likelihood
that strain will lead to delinquency is mediated by feelings of negative
affect, specifically anger, among individuals. As discussed when referring
to GST (Agnew 1992), these theoretical micro-level effects of strain have
been supported empirically. On the other hand, empirical studies of the
indirect relationship between strain and delinquency when mediated by
negative affect have been less consistent. Although the findings regarding
the role of negative affect are contradictory, the proposed model reflects
the theoretical direction suggested by Agnew at both the micro-social
and macro-social levels.
Community characteristics may also indirectly affect individual delinquency
through negative affect alone. In his discussion of MST, Agnew (1999)
stated that disadvantaged communities are more likely to contain higher
concentrations of individuals experiencing negative affect/anger. This
increases the chance that angered individuals will come in contact with
other angered individuals (interpersonal-friction (see, Brezina et al.
2001)). Consequently, individual negative affect may increase individual
delinquency.
METHODOLOGY
The research reported here reflects a cross-sectional study examining
the causes and correlates of delinquency among high school students from
Largo, Florida. Participation in this study was contingent upon compliance
with passive parental consent procedures. In addition, students were informed
that participation in the study was completely voluntary and that all
information provided was confidential and anonymous. Students were surveyed
in various types of classes ranging from mainstream to emotionally handicapped
(EH) and gifted classes for grades 9 through 12. Overall, the response
rate was 79 percent (n=625) for the high school.
Of the total 625 usable, completed surveys, 462 (74%) were able to be
geocoded (discussed below) for the multilevel analysis. In an effort to
improve the fit of the data, cases that contained missing values among
any of the items used to create the dependent variable were eliminated.
Thus, the sample size was further reduced to 430 adolescents for the study.
In the subset used in this analysis, the majority of the students described
themselves as white (82.3%). The rest of the respondents considered themselves
to be black (6.0%), Hispanic (4.2%), Asian (2.6%), or other (3.5%). The
geocoded sample was 45.6 percent male and 54.2 percent female. The ages
of the students ranged from 13 to 19, with the average age being 15.9
years old. Comparison of the geocoded versus non-geocoded high school
students revealed that there were no significant differences between the
two groups with regard to the variables employed in our analyses; however,
there were significant differences between the two groups with respect
to gender (Pearson  =8.346,
df=1, p=0.004) and race (Pearson
=7.572, df=1, p=0.006), such that the geocoded subset contained less males
and non-whites than the non-geocoded subset. These differences affect
the generalizability of the analyses, suggesting it is likely certain
demographic groups (males and non-whites) were excluded from the geocoded
sample.
Individual-Level Measures
The dependent variable is a summary measure of self-reported delinquency.
Students were asked how many times within the past 12 months they had
committed the following: (1) gone into or tried to go into a house to
steal something, (2) purposely damaged or destroyed property that did
not belong to you, (3) stolen another student's property worth $50 or
less, (4) stolen other things worth $50 or less, (5) stolen something
worth more than $50, (6) gone into or tried to go into a building to steal
something, (7) stolen or tried to steal a car or motorcycle, (8) hit someone
with the idea of hurting them, (9) attacked someone with a weapon, and
(10) used a weapon or force to get money or things from people. Responses
for each of these questions were summed to create an additive scale of
delinquency (mean=3.75, SD=25.14). A majority of the students (75.3%)
said they had committed zero of the ten delinquent acts within the past
12 months. However, because of marked skewness (15.30) and kurtosis (266.81)
in the delinquency scale, this variable was logarithmically transformed
(mean= -.58, SD= .78), with -1 being assigned to students reporting no
delinquent offenses prior to the log transformation (alpha=.39 for unlogged
delinquency scale; alpha=.70 for logged delinquency scale).
Table 1. Descriptive Statistics for Individual Level
and Community Level Variables.
Strain is measured using five composite variables that
comprise measures of the three types of strain (i.e., failure to achieve
positive goals (expectations versus achievements and just versus fair
outcomes), removal of positive stimuli, and presentation of negative stimuli).
Strain as the disjunction between aspirations and expectations, another
sub-category of strain as the failure to achieve positive goals, was not
included in the measures. Although classic strain theory contends that
the disjunction between aspirations and expectations is a form of strain
that influences delinquency, empirical studies have found little support
for this sub-category of strain (e.g., Voss 1966; Hirschi 1969; Liska
1971; Farnworth and Leiber 1989; Burton, Cullen, Evans, and Dunaway 1994).
Aspirations reflect distant goals whereas expectations refer to more immediate
goals. Since studies have shown that adolescents are more concerned with
immediate goals over distant goals (Hirschi 1969; Empey, Lubeck, and LaPorte
1971; Liska 1971; Quicker 1974; Farnworth and Leiber 1989; Burton et al.
1994), the present study included expectations and achievements rather
than aspirations and expectations.
To measure strain, students were asked a range of questions concerning
their expectations, feelings of inequality and relative deprivation, experience
of losses, and presence of negative stimuli. Nine items were used to represent
measures of strain as the failure to achieve positive goals. Specifically,
to measure strain as the disjunction between expectations and actual achievements,
students were asked to specify the degree to which they agreed or disagreed
with the following statements: (1) my teachers don't respect my opinions
as much as I would like, (2) people my age treat me like I'm still just
a kid, and (3) my parents don't respect my opinions as much as I would
like. Responses ranged from 1 = strongly disagree to 4 = strongly agree.
Measures of strain as the disjunction between just/fair outcomes and actual
outcomes were derived from responses to the following questions: (1) other
students get special favors from the teachers here that I don't get, (2)
compared to the rules my friends have to abide by, the rules my parents
set for me are unfairly strict, (3) among my group of friends, I think
I like them more than they like me, (4) even though I try hard, my grades
are never good enough, (5) even though I work hard, I never seem to have
enough money, and (6) no matter how responsible I try to be, my parents
don't trust me to do things on my own. These responses also ranged from
1 = strongly disagree to 4 = strongly agree. The first two items of the
just/fair outcomes represent notions of relative deprivation, in which
students compare their own situation to that of others; while the remaining
four items assess the degree of inequity in exchange relationships, where
students compare their "inputs" with the "outputs"
of their relationships with others.
All nine items measuring strain as the failure to achieve positive goals
were entered into a principal axis factor analysis, using mean substitution
(ranged from 0.7% to 2.3% missing among items). This analysis identified
a two-factor solution with eigenvalues greater than 1.0 among these 9
predictor variables-accounting for 44 percent of the variance. Loadings
for these two factors were moderate in size, ranging from 0.32 to 0.68.
These factors were Oblique rotated (factor correlation= -.492) for factor
clarity. Regression factor scores (Kim and Mueller 1978) of these two
Oblique rotated factors are included as predictor variables in the analyses.
The first factor reflects unfair teacher/peer relationships (alpha=.60),
while the second factor reflects unfair parent relationships (alpha=.69).
The second factor is negatively correlated with the first and suggests
strict parenting styles.
Strain as the removal of positive stimuli was measured by creating an
additive index of life losses among several items; however, due to low
bivariate correlations (generally r<.100 for excluded items) among
some of the items and multicollinearity issues (e.g., "changed schools"
with "moved", "divorce" with "parent moved out
or away"), only four items are used in this study (correlations ranged
from 0.113 to 0.303). Students were first asked whether or not the following
things happened to them within the past 12 months: (1) changed schools,
(2) parent moved out or away, (3) sibling moved out or away, and (4) lost
a friendship. Next the students were asked how much of a problem each
event was to them (1 = no problem, 2 = small, 3 = medium, or 4 = large).
The additive scale per item was computed by multiplying whether or not
each event had occurred (values 0 or 1) by the size of the problem (values
1 - 4).
The four items measuring strain as the removal of positive stimuli were
entered into a principal axis factor analysis, using mean substitution
(ranged from 1.4% to 2.1% missing among items). This analysis identified
a single factor solution with an eigenvalue greater than 1.0 among these
4 predictor variables-accounting for 39 percent of the variance. Loadings
for this unrotated single factor were moderate in size, ranging from 0.28
to 0.57 (alpha=.46).
Finally, strain as the presentation of negative stimuli was based on responses
to the following statements: (1) there are a lot of bullies at this school,
(2) my classmates do not like me, (3) I worry a lot about being beaten
up at school, (4) I worry a lot about being shot at school, (5) there
are a lot of strangers coming and going in my neighborhood who don't live
there, (6) I feel safe being inside my home at night, (7) I feel safe
being alone outside in my neighborhood at night, and (8) the people living
in my neighborhood take good care of the way the neighborhood looks. These
responses ranged from 1 = strongly agree to 4 = strongly disagree (responses
to the first four items were reverse coded).
All eight items measuring strain as the presentation of negative stimuli
were entered into a principal axis factor analysis, using mean substitution
(ranged from 0.7% to 1.9% missing among items). This analysis identified
a two-factor solution with eigenvalues greater than 1.0 among these 8
predictor variables-accounting for 55 percent of the variance. Loadings
for these two factors were modest in size, ranging from 0.47 to 0.87.
These factors were Oblique rotated (factor correlation= .442) for factor
clarity. Regression factor scores (Kim and Mueller 1978) of these two
Oblique rotated factors are included as predictor variables in the analyses.
The first factor reflects negative peer relationships (alpha=.71), while
the second factor reflects negative neighborhood conditions (alpha=.71).
Negative affect was the second construct examined at the
individual level. Students were asked to indicate how often the following
statements described them. (1) I feel annoyed when people don't notice
that I've done good work, (2) when I get mad, I say nasty things, (3)
it makes me very mad when I am criticized in front of others, (4) when
I get frustrated, I feel like hitting others, and (5) I feel furious when
I work hard but get a poor grade. These items were derived from the Spielberger
(1988) State-Trait Anger Expression Inventory (STAXI), which examines
anger as a personality trait that is situational. In addition, students
were asked how often they think the following: (1) it makes me mad when
people don't let me make my own decisions, (2) it makes me mad that others
are able to spend more money than I can, and (3) it makes me mad when
I don't get the respect from others that I deserve. These items also represent
situations that may lead to feelings of anger. These eight items of trait
anger appear to be more situational or reaction oriented (see, Mazerolle
and Piquero 1998). Similar to Baron's (2004) examination of strain and
anger (negative affect) on crime and drug use among homeless street youth,
we created an additive scale of the above eight items to measure negative
affect (alpha=.75). Although these items of anger do not represent all
forms of trait anger, the situational component of anger appears consistent
with general strain theory (Baron 2004: 469).
The mean, minimum, maximum, and standard deviation values for the composite
indexes of strain, negative affect/anger, and delinquency are reported
in Table 1. On average, the adolescents in this subset of the sample report
modest levels of strain and negative affect and low levels of delinquency.
Although the average number of total delinquent acts committed by these
adolescents is 4 acts per year, this is not an accurate interpretation
of the sample. In fact, examination of the data revealed 75 percent of
the subset reported committing zero delinquent acts. Obviously, the 11
percent of the sample that reported committing four or more delinquent
acts per year greatly affects the mean for the delinquency measure. This
was precisely why the delinquency scale was logarithmically transformed,
so as to reduce skewness and kurtosis.
Community-Level Data
In this study, community effects were defined by census block groups.
Block groups are subdivisions of census tracts containing between 250
and 550 housing units (U.S. Census Bureau 2000). Although most studies
of neighborhood effects utilize census data delimiting neighborhood by
census tracts, Agnew (1999: 124) suggests that his macro-social general
strain theory is better tested with data pertaining to smaller geographical
areas, which "are more homogeneous in terms of most of the independent
and intervening variables." Likewise, other researchers have advocated
for the use of smaller definitions of the community, such as block groups
(Brezina et al. 2001; Bursik 1989; Bursik and Grasmick 1993; Hoffmann
2002; Suttles 1972; Tienda 1991).
One purpose of the survey was to provide greater understanding of the
connection between the students and their surrounding environments. The
survey asked students to provide the street names of the intersection
closest to where they lived. Based on this information, block groups could
be attributed to each student. Among the 625 high school students who
successfully completed the survey, 462 (74%) students also provided a
street/cross-street location. Each address was then geocoded using ArcView
("ArcView GIS 3.2 [Computer software]," 1999).
Once the student's address was geocoded, each student was assigned a 2000
US census identification number. This number provided the tract and block
group number for that student's location. Due to incompatibilities in
geographical map projections, the street maps did not align perfectly
with the US Census tract and block group boundaries. This disparity, however,
was only problematic for students living on streets comprising the boundaries
between adjacent census tracts and block groups. Where individuals lived
on boundary streets or intersections (25% of the 462 students), they were
randomly assigned to one of the adjacent block groups. Students resided
in 108 different block group communities with an average of approximately
4 students per block group.
Measurement of Community Variables
Many of the measures of community characteristics reflect those mentioned
by Agnew in his derivation of MST (1999). Several of these have been empirically
tested in other multilevel studies (e.g., Avakame 1997; Cattarello 2000;
Gottfredson et al. 1991; Warner and Fowler 2003). Six characteristics
of disadvantaged communities were obtained from 2000 US Census block group
data. Each characteristic is a ratio level variable based on aggregate
measures.
The community measures include racial inequality, economic inequality,
education, family disruption, and residential mobility. Non-White was
defined as the proportion of minority population (i.e., black, Asian,
American Indian, or other) within each block group; for these block groups
blacks comprised the vast majority of non-white residents. Poverty
referred to the total number of people 15 years old or older with a ratio
of income to poverty level for 1999 of less than 1.00. For the 2000 census
data, the poverty threshold calculated by the Census Bureau for a family
of four was $17,029 (U.S. Census Bureau 2000), excluding monies received
by the family from members who were institutionalized, living in group
quarters in the military, or living in college dorms. Low education
referred to the proportion of persons within each block group with less
than a high school education or equivalent. Female- headed household
with children was constructed as the proportion of householders describing
themselves as female with no husband present and children under 18 years
old. Residential mobility represents the proportion of persons
within each block group that lived in a different residence four years
prior (1995). Non-home owners represents the proportion of persons
within each block group that did not own (i.e., rented, leased) their
residence.
The six items measuring community disadvantage were entered into a principal
axis factor analysis. This analysis identified a single factor solution
with an eigenvalue greater than 1.0 among these 6 predictor variables-accounting
for 59 percent of the variance. Loadings for this unrotated single factor
were moderately large in size, ranging from 0.58 to 0.88. Regression factor
scores (Kim and Mueller 1978) of the factor is included as a predictor
variable in some of the analyses (alpha=.80).
RESULTS
The purpose of this study was to examine a multilevel model of general
strain theory. This interest was influenced by the recent introduction
of a macro-level model of general strain theory (see Agnew 1999). The
question addressed by this multilevel model (see Figure 1) was the following:
do the relationships among strain, negative affect, and delinquency differ
significantly across communities? Hierarchical linear modeling (HLM) was
utilized to examine the effect of community differences on the relationships
among strain, negative affect, and delinquency. A two-level HLM was performed
on a path model using the Mplus version 3.1 (Muthén and Muthén
2004). This analysis proceeds in two stages: (1) tests the individual
level (within) model and (2) tests the community-level (between) model
(for detailed discussion of HLM see, Bryk and Raudenbush 1992; Wooldredge
et al. 2001).
Mplus is a versatile, multivariate statistical modeling program enabling
estimation of a variety of models for continuous and categorical observed
and latent variables. In these analyses, a
test is used to test the fit of the models to the data. Lack of significance
indicates an acceptable model fit. Mplus also provides a number of descriptive
fit measures to assess the closeness of fit of the model to the data.
Three fit indices were used to evaluate the model fit, using the following
criteria as indicating an adequate fit: (1) the comparative fit index
(CFI) (Bentler 1990), (2) the Tucker-Lewis coefficient (TLI) (Tucker and
Lewis 1973), and (3) root mean square error of approximation (RMSEA) (Byrne
2001). The typical range for both TLI and CFI is between 0 and 1 (although
TLI can exceed 1.0), with values greater than .95 indicating a good fit
(Browne and Cudeck 1993; Hu and Bentler 1999). For RMSEA, values at .05
or less indicate a close model fit, and values between .05 and .08 indicating
a mediocre model fit (Browne and Cudeck 1993).
Initial examination of bivariate relationships at the individual level
found statistically significant relationships among the five factors of
strain, negative affect, and delinquency (see Table 2). Most of the bivariate
relationships are significant and in the expected direction. However,
strain as unfair parent relationships is inversely related to the other
four measures of strain, negative affect, delinquency, and community disadvantage.
Interestingly, the bivariate relationship between the factor scores of
community disadvantage (attributed to individuals for the purposes of
data exploration, but not included in the HLM analysis) and negative affect
and delinquency are not significant and seem counterintuitive in their
direction.
Table 2. Zero-Order Correlation Matrix for Individual
and Community Level Variables.
The purpose of a multilevel model of general strain theory
(Figure 1) can best be addressed by three questions: (1) do individual
strain, negative affect, and delinquency vary within communities, (2)
do community characteristics explain any of this variation between communities,
and (3) what roles do strain and negative affect play when controlling
for the effects of community characteristics on individual delinquency?
These questions are arranged in order from the lowest level (individual)
to the highest level (community). A failure to significantly explain variance
in the model for any one of these questions prevents advancement to the
next level or question.
Do individual strain, negative affect, and delinquency vary within communities?
As seen in Figure 2, the HLM analysis indicated that individual strain,
negative affect, and delinquency do significantly vary within communities
(i.e., block groups). As Figure 2 shows, the fit indices indicated a good
fit of the model to the data: (1,
n=430) = 0.31, p=0.5781; CFI=1.000; TLI=1.072; and RMSEA=0.000.
Among high schoolers, strain as unfair teacher/peer relationships has
a significant positive effect on delinquency both directly and indirectly,
through negative affect. Strain as negative peer relationships has a significant
positive effect on negative affect. However, strain as unfair parent relationships
has a significant negative effect on negative affect. Negative affect
has a significant direct effect on delinquency. Neither strain as negative
neighborhood conditions nor as the removal of positive stimuli has a significant
effect on negative affect and delinquency. Consistent with GST (Agnew
1992), the individual-level model suggests that the experience of negative
affect, particularly anger, motivates individuals to cope with some forms
of strain through illegitimate means.
Figure 2. Individual-Level Model: Effects of Strain
and Negative Affect on Delinquency
(Log)-Unstandardized Estimates* (N=430).
* All paths are statistically significant (p<0.05); Standardized
Estimates in Parentheses.
The individual-level model significantly explained a portion
of the variance on delinquency, but would there be enough variance left
in the outcome measures to be accounted for by differences between the
108 block group communities? No, the community-level model did not fit
the data. The intraclass correlations for delinquency and negative affect
were very small: 0.024 and 0.013, respectively. Moreover, the average
cluster size (number of within-level cases) for the block groups was also
small (3.981). This suggests that the multilevel nature of the data could
be ignored for the endogenous variables, and justifies use of supplementary
analyses to examine the data contextually (Silver, Mulvey, and Swanson
2002).
SUPPLEMENTARY ANALYSES
Since it appears that the data were not able to support within-block group
analyses using HLM, we decided to perform ad hoc contextual analyses
examining the strain-anger-delinquency relationships comparing high schoolers
living in more disadvantaged communities to those living in less disadvantaged
communities. We divided the high school students into two groups based
on the factor scores for the six block group characteristics obtained
from the 2000 Census data. We used the median (-0.3313) for the community
disadvantage factor for the 108 block groups within which the 430 students
reported they lived. Students residing in block groups whose community
disadvantage factor score fell below the median were characterized as
being non-disadvantaged to affluent (n=199); and students residing in
block groups whose community disadvantage factor score fell above the
median were characterized as disadvantaged (n=231). ANOVA models (not
reported here) testing for differences in mean levels of strain, negative
affect, and delinquency between those residing in disadvantaged communities
and those from non-disadvantaged communities revealed only one statistically
significant difference; namely, those residing in the more disadvantaged
communities reported a lower mean level difference; namely, those residing
in the more disadvantaged communities reported a lower mean level of strain
as a product of negative neighborhood conditions than that reported by
students residing in the non-disadvantaged neighborhoods. Our supplementary
analyses now turn to an examination of separate structural equation models
for each of these two groups of students.
Figures 3 and 4 illustrate the group (1=non-disadvantaged; 2=disadvantaged)
structural equation model of strain, negative affect, and delinquency
measures for students heuristically defined as residing in non-disadvantaged
and disadvantaged communities respectively. The fit indices indicated
a good fit of the model to the data:
(2, n=430) = 1.60, p=0.4451; CFI=1.000; TLI=1.038; and RMSEA=0.000.
As seen in Figure 3, among high school students living in non-disadvantaged
areas, strain as unfair teacher/peer relationships has a significant positive
effect on negative affect. Strain as negative peer relationships has a
significant positive effect on negative affect. However, strain as unfair
parent relationships has a significant negative effect on negative affect
and a significant positive effect on delinquency. Negative affect has
a significant direct, positive effect on delinquency. Neither strain as
negative neighborhood conditions nor as the removal of positive stimuli
has a significant effect on negative affect or delinquency.
Figure 3. Effects of Strain and Negative Affect on
Delinquency (Log) for Non-Disadvantaged
Communities-Unstandardized Estimates* (N=199).
* All paths are statistically significant (p<0.05); Standardized
Estimates in Parentheses.
Figure 4 shows the structural equation model of strain, negative affect,
and delinquency measures for students heuristically defined as residing
in disadvantaged communities. Among high school students living in more
disadvantaged areas, strain as unfair teacher/peer relationships has a
significant positive effect on delinquency. Strain as the removal of positive
stimuli also has a significant positive effect on delinquency. Strain
as unfair parent relationships has a significant negative effect on negative
affect. However, strain as negative peer relationships and negative neighborhood
conditions and negative affect have no significant effects on other variables.
While these unstacked structural equation analyses produced two seemingly
different models for the strain, negative affect, and delinquency relationships,
a comparison of the parameter estimates generated revealed only one statistically
different effect; namely a more powerful effect of negative affect on
delinquency among those students who reside in the non-disadvantaged communities.
Figure 4. Effects of Strain and Negative Affect on
Delinquency (Log) for Disadvantaged
Communities-Unstandardized Estimates* (N=231).
* All paths are statistically significant (p<0.05); Standardized
Estimates in Parentheses.
DISCUSSION
The purpose of this study was to test a multilevel strain theory explanation
of adolescent delinquency. While general strain theory has been theorized
from both a micro-social (Agnew 1992) and macro-social (Agnew 1999) approach
and empirically upheld in the former (Agnew and White 1992; Paternoster
and Mazerolle 1994; Hoffmann and Su 1997; Brezina 1996, 1998; Mazerolle
1998; Mazerolle and Piquero 1998; Hoffmann and Cerbone 1999; Mazerolle
et al. 2000), multilevel tests of general strain theory have been limited.
To the best of our knowledge, there have only been three multilevel tests
of MST (Brezina et al. 2001; Hoffmann, 2002; Hoffmann and Ireland, 2004),
each providing partial support of MST. Since the impetus for examining
a multilevel model of strain was derived from Agnew's (1999) arguments
that community characteristics should directly affect individual crime,
a multilevel model, geared specifically toward a general strain theory
explanation of adolescent delinquency and neighborhood influence, was
examined.
The main crux of this study focused on investigating the effects of strain
and negative affect on delinquency between community block groups. Hierarchical
linear modeling (HLM) (Bryk and Raudenbush 1992) was used to test individual
(within) and Census block group (between) effects of strain on delinquency
and indirect effects of strain on delinquency when mediated by negative
affect. Within block groups, the data support findings from previous studies
that strain has a significant positive effect on self-reported delinquency
and that the reported experience of negative affect, specifically anger,
served as a key motivator for such delinquency. However, between block
groups, there was no significant difference in the amount of explained
variance for delinquency. The HLM analysis suggested that community characteristics
do not significantly influence the process by which strain influences
delinquent behavior. Therefore, Agnew's contention that community characteristics,
when defined by smaller, more homogeneous areas (Agnew 1999: 124), significantly
influence strain's effect on delinquency went unfounded with the sample
tested.
Since the mere fact that communities were defined as separate block groups
did not necessarily mean that these communities differed in the characteristics
described by Agnew (1999) as more strain inducing, a supplementary contextual
analysis was performed with models examined and compared across similar
census block groups. Participants were assigned to one of two groups:
those below the median factor score for a community disadvantage variable
and those above the median community disadvantage factor score. Structural
equation models mimicking the individual level HLM model, predicting both
direct and indirect effects of strain on self-reported delinquency through
negative affect, were compared across the two heuristically defined community
groups (1=non-disadvantaged, 2=disadvantaged).
The models for non-disadvantaged communities revealed that strain has
both a significant positive direct effect on delinquency and a significant
positive indirect effect on delinquency, mediated by negative affect.
Among these less disadvantaged youths, the factor scores for strain as
unfair parent relationships was negatively related to negative affect
and positively related to delinquency. Although this contradicts what
general strain theory would predict, other studies of general strain theory
(Broidy 2001; Hay 2003) have revealed similar negative associations for
measures of blocked goals and unfair parental discipline, especially among
females. Since our subsample contains significantly more females than
the complete high school sample, perhaps this negative association is
reflective of gender differences in the strain-anger relationship. Future
studies should further explore this aspect. Within more disadvantaged
communities, strain does not appear to have an indirect effect on delinquency
through negative affect. However, measures of strain as the removal of
positive stimuli and unfair teacher/peer relationships was positively
related to delinquency among the more disadvantaged youths. Consistent
with the lesser disadvantaged group, youths living in more disadvantaged
communities were also experiencing strain as unfair parent relationships
that had a significant negative effect on negative affect.
Methodologically, this study emphasizes the importance of sample size
for HLM analysis and highlights problems associated with multilevel multivariate
analyses. One reason the HLM results were not significant between block
groups could have been due to the fact that the average sample sizes for
the within and between models were relatively small. It is recommended
that HLM analyses be conducted with either large samples at the within
level (usually greater than 20) and small samples at between level 2 (Muthén
and Muthén 2000). The data reflected in this study were relatively
small in the within level (average cluster size=3.98). Consequently, the
small sample size may have inflated the standard errors in the between
level analysis.
There are several theoretical implications of this study. In general,
the within model HLM results and supplemental analysis comparing communities
support micro-social general strain theory (Agnew 1992). Among adolescents,
higher levels of strain explain part of the variation in delinquent behavior.
The study also indicated strained individuals are more likely to express
high feelings of negative affect and adolescents experiencing higher levels
of negative affect, particularly anger, are more likely to be delinquent.
Unfortunately, the cross-sectional nature of the analysis prevented the
determination of a causal relationship among strain, negative affect,
and delinquency. The debate over whether anger serves as a mediator between
strain and delinquency (cf. Brezina 1996, 1998), or vice versa (cf. Mazerolle
et al. 2000), remains unanswered. Although this study failed to support
a multilevel model of general strain theory, such a theoretical advancement
of general strain should not be discounted. Instead it suggests that when
considering the effects of community differences on strain and ultimately
individual delinquency, other theoretical influences must be considered.
In fact, Agnew (1999: 147) stated that general strain theory should serve
as a supplemental explanation for crime. Future research should attempt
to include measures for strain in conjunction with those for theories
such as social control, differential association, social disorganization,
and subcultural deviance in one multilevel model of crime causation.
The multilevel model in this study presented a number of limitations.
These limitations can be divided into two groups, one relating to the
data and the other to the model. The most detrimental issue regarding
the data used in this study was sample size. Although the study began
with a robust sample size of 625 students, measurement errors (e.g., missing
data) resulting in incomplete measures led to substantial reduction in
the study sample (n=430). The final subset of the sample was comprised
of youths who were sparsely distributed spatially over 108 block groups.
In addition, the subset varied significantly from the original high school
sample along race and gender. This affects the generalizability of our
findings to other populations, and may have influenced the association
between strain as unfair parent relationships and negative affect. Moreover,
the models tested did not contain alternative measures of negative affect
(e.g., anxiety, depression, guilt, etc.), only trait anger, or other conditioning
variables. Future studies should attempt to incorporate more specific
measures of strain, measures for supplemental explanations of delinquent
motivation, and control measures.
CONCLUSION
Recently Agnew (1999) proposed an expanded version of general strain theory
that provides macro-social implication for explaining variation in crime
rates across differing communities. Agnew's macro-strain theory (MST)
was presented as a supplement to other macro-level theories of crime;
one that more completely addresses motivational aspects. According to
MST, variation in crime rates depends on the levels of aggregate strain,
aggregate negative affect/anger, and other stressful community conditions.
Communities characterized as highly disadvantaged create strain and anger
by blocking members' abilities to achieve positive goals, creating a loss
of positive stimuli, exposing members to negative stimuli, and increasing
overall relative deprivation. Moreover, disadvantaged communities are
also more likely to both select and retain strained individuals and to
produce interactions involving angry participants.
To date we have been able to identify only one macro-level test MST (Warner
and Fowler 2003) and three multi-level tests (Brezina et al. 2001; Hoffmann
2002; and Hoffmann and Ireland 2004). The results of these studies are
quite mixed. The purpose of the present study was to add to this emergent
research literature by providing an additional multi-level test. With
self-report survey data from 430 high school students we attempted to
answer three questions: (1) Does community context have any direct effects
on individual levels of strain, negative affect, and/or delinquency? (2)
Does community context have any indirect effects on delinquency through
its effect on strain and/or negative affect? (3) Does community context
condition the effects of strain and/or negative affect on delinquency?
Our initial analyses, based on hierarchical linear modeling (HLM), suggest
that the answer to each of these questions is in the negative. However,
the level-2 data were comprised of 108 block group communities for our
430 students, with an average of only 4 students per block group community.
Such a value is considered to be too small for HLM (Muthén and
Muthén 2000). Such small sample size tends to inflate the standard
errors in the between level analysis in HLM. Conversely, our supplementary
analyses of separate individual-level structural equation models for students
residing in disadvantaged and non-disadvantaged communities respectively
did produce evidence for the appearance of the conditioning effects of
community disadvantage on the relationships between strain, negative affect
and delinquency. More specifically, these data suggest different strain-negative
affect-delinquency models across levels of community disadvantage. However,
a comparison of the parameter estimates generated from these structural
equation models produced only one significantly different path, suggesting
quite strongly that Agnew's general strain theory may be invariant across
community types. Clearly additional multi-level tests of general strain
theory are in order.
ENDNOTES
1.
This research was supported by a grant from the National Institute of
Justice - Office of Community Oriented Policing (#12-21-047-L0). The views
expressed do not necessarily reflect those of the NIJ. back
REFERENCES
Agnew, Robert. 1983. "Social class and Success Goals: An Examination
of Relative and Absolute Aspirations." Sociological Quarterly
24:435-452.
------. 1984. "Goal Achievement and Delinquency." Sociological
and Social Research 68:435-451.
------. 1985. "A Revised Strain Theory of Delinquency." Social
Forces 64:151-167.
------. 1989. "A Longitudinal Test of Revised Strain Theory."
Journal of Quantitative Criminology 5:373-387.
------. 1992. "Foundation for a general Strain Theory of Crime
and delinquency." Criminology 30:47-87.
--. 1997. "Stability and Change in Crime over the Life Course: A
Strain Theory Explanation." Pp. 101-132 in Developmental Theories
of Crime and Delinquency, Advances in Criminological Theory, Vol. 7,
edited by T. P. Thornberry. New Brunswick, NJ: Transaction.
------. 1999. "A General Strain Theory of Community Differences
in Crime Rates." Journal of Research in Crime and Delinquency
36:123-155.
------. 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 38:319-361.
------. 2002. "Experienced, Vicarious, and Anticipated Strain: An
Exploratory Study on Physical Victimization and Delinquency." Justice
Quarterly 19:603-632.
Agnew, Robert and Timothy Brezina. 1997. "Relational Problems with
Peers, Gender and Delinquency." Youth & Society 29:84-111.
Agnew, Robert, Timothy Brezina, John Paul Wright, and Francis T. Cullen.
2002. "Strain, Personality Traits, and Delinquency: Extending General
Strain Theory." Criminology 40:43-71.
Agnew, Robert and Helen RaskinWhite. 1992. "An Empirical Test of
General Strain Theory." Criminology 30:475-499.
Akers, Ronald L. 1998. Social Learning and Social Structure: A General
Theory of Crime and Deviance. Boston, MA: Northeastern University
Press.
ArcView GIS 3.2 [Computer software]. (1999). Environmental Systems Research
Institute, Inc.
Aseltine, R.H., S. Gore, and J. Gordon. 2000. "Life Stress, Anger
and Anxiety, and Delinquency: An Empirical Test of General Strain Theory."
Journal of Health and Social Behavior 41:256-275.
Avakame, Edam. F. 1997. "Urban Homicide." Homicide Studies
1:338-358.
Bao, W.-N., Haas, A., and 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 48:281-297.
Baron, Stephen W. 2004. "General Strain, Street Youth and Crime:
A Test of Agnew's Revised Theory." Criminology 42:457-484.
Baron, Stephen W. and Timothy F. Hartnagel. 1997. "Attributions,
Affect, and Crime: Street Youths' Reactions to Unemployment." Criminology
35:409-434.
Benda, Brent B. and R. F. Corwyn. 2002. "The Effect of Abuse in
Childhood and in Adolescence on Violence among Adolescents." Youth
& Society 33:339-365.
Bentler, P.M. 1990. "Comparative Fit Indexes in Structural Models."
Psychological Bulletin 107:238-246.
Brezina, Timothy. 1996. "Adapting to Strain: An Examination of Delinquent
Coping Responses." Criminology 24:39-60.
------. 1998. "Adolescent Maltreatment and Delinquency: The Question
of Interviewing Process." Journal of Research in Crime and Delinquency
26:71-99.
------. 1999. "Teenage Violence toward Parents as an Adaptation
to Family Strain." Youth and Society 30:416-444.
Brezina, Timothy, Alex R. Piquero, and Paul Mazerolle. 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 38:362-386.
Broidy, Lisa M. 2001. "A Test of General Strain Theory." Criminology
39:9-35.
Broidy, Lisa and Robert Agnew. 1997. "Gender and Crime: A General
Strain Theory Perspective." Journal of Research in Crime and Delinquency
34:275-306.
Browne, M.W. and R. Cudeck. 1993. "Alternative Ways of Assessing
Model Fit." Pp. 445-455 in Testing Structural Equation Models,
edited by K. A. Bollen andJ. S. Long. Newbury Park, CA: Sage.
Bryk, Anthony S., and Stephen W. Raudenbush. 1992. Hierarchical Linear
Models. London: Sage.
Bursik, Robert J., Jr. 1989. "Political Decision Making and Ecological
Models of Delinquency: Conflict and Consensus." Pp. 105-117 in Theoretical
Integration in the Study of Deviance and Crime: Problems and Prospects,
edited by S. F. Messner, M. D. Krohn and A. E. Liska. Albany, NY: State
University of New York Press.
Bursik, Robert J., Jr., and Harold G. Grasmick. 1993. Neighborhoods
and Crime. New York: Lexington Books.
Burton, Velmer, Francis Cullen, David Evans, and R. Greg Dunaway. 1994.
"Reconsidering Strain Theory: Operationalization, Rival Theories,
and Adult Criminality." Journal of Quantitative Criminology
10:213-239.
Byrne, B.M. 2001. Structural Equation Modeling with AMOS: Basic Concepts,
Applications, and Programming. Mahwah, NJ: Lawrence Erlbaum Associates.
Capowich, George E., Paul Mazerolle, and Alex Piquero. (2001). "General
Strain Theory, Situational Anger, and Social Networks: An Assessment of
Conditioning Influences." Journal of Criminal Justice 29:445-461.
Cattarello, Anne. M. 2000. "Community-level Influences on Individuals'
Social Bonds, Peer Associations, and Delinquency: A Multilevel Analysis."
Justice Quarterly 17:33-60.
Cloward, Richard and Lloyd Ohlin. 1959. "Illegitimate Means, Anomie,
and Delinquent Behavior." American Sociological Review 24:164-177.
------. 1961. Delinquency and Opportunity. Glencoe, IL: Free Press.
Cohen, Albert K. 1955. Delinquent Boys. Glencoe, IL: Free Press.
Colvin, Mark and John Pauly. 1983. "A Critique of Criminology: Toward
an Integrated Structural-Marxist Theory of Delinquency Production."
American Journal of Sociology 89:513-551.
Durkheim, Emile. 1951 [1897]. Suicide (J. A. Spaulding & G.
Simpson, Trans.). New York: Free
Press.
Eitle, David and R. Jay Turner. 2003. "Stress Exposure, Race, and
Young Adult Male Crime." Sociological Quarterly 44:243-269.
Empey, Lamar T., S. G. Lubeck, and R. L. LaPorte. 1971. Explaining
Delinquency. Boston, MA: Heath Lexington Books.
Farnworth, Margaret and Michael J. Leiber. 1989. "Strain Theory
Revisited: Economic Goals, Educational Means, and Delinquency." American
Sociological Review 54:263-274.
Gardner, L., and Donald J. Shoemaker. 1989. "Social Bonding and
Delinquency: A Comparative Analysis." Sociological Quarterly
30:481-500.
Gottfredson, Denise C., Richard J. McNeil, III, and Gary D. Gottfredson.
1991. "Social Area Influences on Delinquency: A Multilevel Analysis."
Journal of Research in Crime and Delinquency 28:197-226.
Hagan, John, A. R. Gillis, and John Simpson. 1985. "The Class Structure
of Gender and Delinquency: Toward a Power-Control Theory." American
Journal of Sociology 90:1151-1178.
Hay, Carter. 2003. "Family Strain, Gender, and Delinquency."
Sociological Perspectives 46:107-135.
Hirschi, Travis. 1969. Causes of Delinquency. Berkeley: University
of California Press.
Hoffmann, John P. 2002. "A Contextual Analysis of Differential Association,
Social Control, and Strain Theories of Delinquency." Social Forces
81:753-785.
Hoffmann, John P. and Felicia Gray Cerbone. 1999. "Stressful Life
Events and Delinquency Escalation in Early Adolescence." Criminology
37:343-373.
Hoffmann, John P. and Timothy O. Ireland. 2004. "Strain and Opportunity
Structures." Journal of Quantitative Criminology 20:263-292.
Hoffmann, John P. and Alan S. Miller. 1998. "A Latent Variable Analysis
of General Strain Theory." Journal of Quantitative Criminology
14:83-110.
Hoffmann, John P. and S. Susan Su. 1997. "The Conditional Effects
of Stress on Delinquency and Drug Use: A Strain Theory Assessment of Sex
Differences." Journal of Research in Crime and Delinquency
34:275-306.
Hu, L.-T. and P. M. Bentler, P.M. 1999. "Cutoff Criteria for Fit
Indexes in Covariance Structure Analysis: Conventional Criteria Versus
New Alternatives." Structural Equation Modeling: A Multidisciplinary
Journal 6:1-55.
Jang, Sung Joon and Byron R.Johnson. 2003. "Strain, Negative Emotions,
and Deviant Coping among African Americans: A Test of General Strain Theory."
Journal of Quantitative Criminology 19:79-105.
Kim, J. and C. E. Mueller. 1978. Factor Analysis: Statistical Methods
and Practical Issues. Beverly Hills, CA: Sage.
Krohn, Marvin D., Lonn Lanza-Kaduce, and Ronald L. Akers. 1984. "Community
Context and Theories of Deviant Behavior: An Examination of Social Learning
and Social Bonding Theories." Sociological Quarterly 25: 353-371.
Liska, Allen E. 1971. "Aspirations, Expectations, and Delinquency:
Stress and Additive Models." Sociological Quarterly 12:99-107.
Maxwell, S.R. 2001. "A Focus on Familial Strain: Antisocial Behavior
and Delinquency in Filipino Society." Sociological Inquiry
71:265-292.
Mazerolle, Paul. 1998. "Gender, General Strain, and Delinquency
: An Empirical Examination." Justice Quarterly 15:65-91.
Mazerolle, Paul, Velmer Burton, Francis T. Cullen, T. David Evans, and
Gary L. Payne. 2000. "Strain, Anger, and Delinquent Adaptations:
Specifying General Strain Theory." Journal of Criminal Justice
28:89-101.
Mazerolle, Paul and Alex Piquero. 1997. "Violent Responses to Strain:
An Examination of Conditioning Influences." Violence and Victims
12:323-343.
------. 1998. "Linking Exposure to Strain with Error: An Investigation
Deviant Adaptations." Journal of Criminal Justice 26:195-211.
Mazerolle, Paul, Alex Piquero, and George E. Capowich. 2003. "Examining
the Links between Strain, Situational and Dispositional Anger, and Crime:
Further Specifying and Testing General Strain Theory." Youth &
Society 35:131-157.
Mazerolle, Paul and Jeff Maahs. 2000. "General Strain and Delinquency:
An Alternative Examination of Conditioning Influences." Justice
Quarterly 17:753-778.
Merton, Robert K. 1938. "Social Structure and Anomie." American
Sociological Review 3:672-682.
------. 1968. Social Theory and Social Structure. New York: Free
Press.
Muthén, L.K. and B. O Muthén. 2000. Mplus User's Guide.
Second Edition. Los Angeles, CA: Muthén & Muthén.
------. 2004. Mplus User's Guide. Third Edition. Los Angeles,
CA: Muthén & Muthén.
Paternoster, Raymond and Paul Mazerolle. 1994. "General Strain Theory
and Delinquency: A Replication and Extension." Journal of Research
in Crime and Delinquency, 31:235-263.
Peter, T., T. C. LaGrange, and R. A. Silverman. 2003. "Investigating
the Interdependence of Strain and Self-Control." Canadian Journal
of Criminology and Criminal Justice 45:431-464.
Piquero, Nicole Leeper and Miriam D. Sealock. 2000. "Generalizing
General Strain Theory: An Examination of an Offending Population."
Justice Quarterly 17:449-484.
------. 2004. "Gender and General Strain Theory: A Preliminary Test
of Broidy and Agnew's Gender/GST Hypotheses." Justice Quarterly
21:125-158.
Quicker, J. C. 1974. "The Effect of Goal Discrepancy on Delinquency."
Social Problems 22:76-86.
Reiss, Albert J., Jr. and A. L. Rhodes. 1961. "The Distribution
of Juvenile Delinquency in the Social Class Structure." American
Sociological Review 26:720-732.
Rosenbaum, Jill Leslie and James R. Lasley. 1990. "School, Community
Context, and Delinquency: Rethinking the Gender Gap." Justice
Quarterly 7:493-513.
Sampson, Robert J., Stephen W. Raudenbush, and Felton Earls. 1997. "Neighborhoods
and Violent Crime: A Multilevel Study of Collective Efficacy."
Science 277:918-924.
Shaw, Clifford and Henry D. McKay. 1969. Juvenile Delinquency and
Urban Areas (Rev. ed.). Chicago, IL: University of Chicago Press.
Silver, Eric, E. P. Mulvey, and J. W. Swanson. 2002. "Neighborhood
Structural Characteristics and Mental Disorder: Faris and Dunham Revisited."
Social Science & Medicine 55:1457-1470.
Simcha-Fagan, Ora and Joseph E. Schwartz. 1986. "Neighborhood and
Delinquency: An Assessment of Contextual Effects." Criminology
24:667-699.
Spielberger, Charles D. 1988. Manual for the State-Trait Anger Expression
Inventory. Research edition. Odessa, FL: Psychological Assessment Resources,
Inc.
Suttles, G. D. 1972. The Social Construction of Communities. Chicago,
IL: The University of Chicago Press.
Tienda, M. 1991. "Poor People and Poor Places: Deciphering Neighborhood
Effects on Poverty Outcomes." Pp. xxx-xxx in Macro-Micro Linkages
in Sociology, edited by J. Huber. Newbury Park, CA: Sage.
Tucker, L.R. and C. Lewis, C. 1973. "A Reliability Coefficient for
Maximum Likelihood Factor Analysis." Psychometrika 38:1-10.
U.S. Census Bureau. 2000. 2000 Census of Population and Housing, Summary
Tape File 3A.
Voss, Harwin L. 1966. "Socio-Economic Status and Reported Delinquent
Behavior." Social Problems 13:314-324.
Wagner, D.G. and J. Berger. 1985. "Do Sociological Theories Grow?"
American Journal of Sociology 90:697-728.
Walsh, Anthony. 2000. "Behavior Genetics and Anomie/Strain Theory."
Criminology 38:1075-1107.
Warner, Barbara D. and S. K. Fowler. 2003. "Strain and Violence:
Testing a General Strain Theory Model of Community Violence." Journal
of Criminal Justice 31:511-521.
Wooldredge, John, Timothy Griffin, and Travis Pratt. 2001. "Considering
Hierarchical Models for Research on Inmate Behavior: Predicting Misconduct
with Multilevel Data." Justice Quarterly 18:203-231.
|
