MACRO MICRO THEORETICAL INTEGRATION
Multi-level integration includes the combination of macro- and micro-level theoretical
explanations. Macro-level, or aggregate level, theories “link social structural characteristics
to variations in the rates and distributions of crime” (Bernard & Snipes, 1996, p. 333). To
do this, macro-level theories have typically relied on variables drawn from geographic units
(i.e., nations, states, cities, or neighborhoods) to explain crime rates (Cattarello, 2000).
According to Bernard and Snipes (1996), these types of theories are founded on three
assumptions:
(1) Crime is said to be a response of individuals who are freely choosing
and whose choices are constrained and inspired by the immediate
environment (implying a causal relationship between immediate environment
and the actions of individuals within it).
(2) The immediate environment is said to be ‘structured’ in the sense that
its most important characteristics, in terms of their effect on the individual’s
responses, are causally related to the broader structural features of social
organization.
(3) Criminals are said to be ‘normal’ in that they are essentially similar to
noncriminals in the processes by which they interact with the immediate
environment and in the motives that direct their responses to that
environment. (p. 333)
Examples of macro-level theories include Classical Strain, Deterrence, Social
Disorganization, and Subcultural/Deviance theories.
Micro-level, or individual-level theories “link individual characteristics to the
probability that an individual will engage in criminal behaviors” (Bernard & Snipes, 1996, p.
335). Micro-level theories rely on individual characteristics to explain individual variation in
crime and delinquency (Cattarello, 2000). Micro-level theories, like macro-level theories,
are based on three primary assumptions:
(1) Differences in the probability of engaging in crime are explained by
differences that are uniquely attributed to the individual.
(2) The individual characteristics may be explained by
interactions with other people within the environment.
(3) Since crime is explained by individual characteristics, criminals
themselves are assumed to be different from noncriminals in some
measurable ways. (Bernard & Snipes, 1996, pp. 335-336)
Examples of individual theories include Social Control, General Theory of Crime, and Social
Learning theories.
Traditionally it has been argued that such theories contradict one another. As a
result, macro-micro theoretical integration will violate theoretical assumptions (Bernard &
Snipes, 1996; Hirschi, 1969). Bernard and Snipes (1996), however, argue that these are not
substantiated claims against macro-micro theoretical integration. Specifically, they argue
that macro- and micro-level theories do not present competing claims because they do not
ignore the possibility of variance at the other level (Bernard and Snipes, 1996). For
instance, macro-level theories operate on the assumption that there is a “normal
distribution of individual characteristics within a given structural situation” (Bernard &
Snipes, 1996, p. 339). They do not, however, “necessarily deny the existence of [individual]
differences or their possible relation to criminality” (Bernard & Snipes, 1996, p. 339). The
same can be said of micro-level theories. Thus, macro- and micro-level theories are not
incompatible, and thus, are conducive to theoretical integration.
In addition, each of these theories alone has weaknesses which may be overcome by
the integration of macro- and micro-level propositions. The principal weakness of macro-
level theories is their inattention to “personal motivation or the agency (volition) of the
individual offender” (Barak, 1998, p. 197). In contrast, the primary weakness of micro-level
theories is their inattention to “the context within which individuals are embedded and,
more specifically, the vulnerability of micro-level processes to local economic and social
conditions” (Bellair, Roscigno & McNulty, 2003, p. 25). Integrated macro-micro theories, in
contrast, “focus on both the individual and the structure plus on some kind of interaction
between the two” (Barak, 1998, p. 198). Integrated macro-micro theories are situated to
explain crime by examining the effect social structure has on individual characteristics and
subsequent individual action (Paternoster & Bachman, 2001). Macro-micro theoretical
integration differentiates the causal properties of structural and individual factors,
identifying mediating and moderating linkages between cross-level variables and their
relationship with crime and delinquency. Thus, according to Bernard and Snipes (1996),
It would seem possible to create a single theory of crime that incorporates
the structural conditions that are associated with higher crime rates, the
processes that explain why normal individuals who experience these
structural conditions are more likely to engage in crime, and the individual
characteristics that make it more or less likely that an individual will engage
in crime regardless of structural conditions. (p. 342)
Theoretical Examples
Over the last twenty years, a growing number of criminologists have advocated for
the integration of individual and structural approaches to theory construction and
elaboration (see for instance Sampson, 1991; Reiss, 1986; Tonry et al., 1991; Jensen &
Akers, 2003; Wikstrom, 2005). Macro-micro theoretical integration, according to Wikstrom
(2005) allows the field to advance by “break[ing] away from the common but unfruitful
division into individually or ecologically oriented explanations of crime involvement” (p.
211). Bernard and Snipes (1996) contend that such approaches:
seem both desirable and feasible. The effect of specific individual
differences on behavior may be magnified or attenuated depending on
the individual’s structural position. Incorporating structure as a
contextual variable may add additional variation to the individual-level
explanation of individual criminal behavior. (p. 343)
A logical starting place for the incorporation of individual and structural explanations of
crime would be with existing theories. As such, an overview of criminological theories that
have made, or provide the possibility for, cross-level arguments is presented below.
Differential Association/Social Learning Theory
One of the first cross-level efforts at theoretical integration was Sutherland’s (1947)
Differential Association Theory, which combined Social Disorganization and Conflict theory
with Differential Association concepts (Akers, 1989). Sutherland, in his fifth edition of
Differential Association Theory, proposed that a “person’s associations are determined in
the general context of social organization” (Sutherland & Cressey, 1955, p. 79). However,
tests of Sutherland’s theory have generally remained at the micro-level, ignoring
Sutherland’s suggestion that peer associations may vary as a result of contextual effects
(Matsueda, 1988; Matsueda & Anderson, 1998; Warr, 1996). Differential Association Theory
has also been heavily criticized for failing to explain why individuals have differential
associations (Kornhauser, 1978). In other words, “why persons have the associations they
have” (Reinarman & Fagan, 1988, p. 308). One exception is a study conducted by
Reinarman and Fagan (1988).
Utilizing a multi-level dataset, Reinarman and Fagan (1988) test Sutherland’s
proposition that differential associations vary largely due to social class. Data for their study
were collected in two manners. Individual-level data were collected from one wave of a
three-year longitudinal study of serious juvenile offenders in northern California. Drawing
upon Differential Association Theory, individual-level variables included measures of
associations with delinquent and non-delinquent peers; attitudes toward law and normative
order; and perceptions of norms and values of peers, parents, school environment, and
neighborhood (Reinarman & Fagan, 1998). In addition, Reinarman and Fagan (1998)
included variables drawn from Social Control Theory, including bonds to peers, family,
school, and conventionality of beliefs. Macro-level data were collected from the 1980 U.S.
Census for the residence of each juvenile offender surveyed. Structural-level variables
included percentage Black, unemployed, female-headed households below poverty, and
high density homes. Reinarman and Fagan (1988) did not find support for Sutherland’s
(1947) contention that the socio-economic status of the community in which an individual
lives impacts associations with delinquent peers among violent juvenile offenders. However,
despite having cross-level data, the researchers conducted their research using standard
linear regression. The use of multi-level statistical modeling, which would account for the
hierarchical nature of the dataset employed in Reinarman and Fagan’s (1998) study, may
have produced a different result.
Like Sutherland, Akers has also proposed a cross-level version of Social Learning
Theory. In his 1998 revision to Social Learning Theory, Akers offered a general theory of
crime (aptly named “Social Learning and Social Structure”) where social learning mediates
the relationship between social structure and individual behavior. To date, only Akers has
tested his Social Learning and Social Structure theory, finding initial support for the theory’s
hypotheses (Akers, 1998; Lee, Akers, & Borg, 2004). For instance, Lee, Akers and Borg
(2004) use structural equation modeling to test propositions drawn from Aker’s SSSL theory.
While the study finds support for Akers’ general statements, it is important to point out that
the study did not employ a multi-level dataset. Rather, all variables were measured at the
individual-level (including structural variables that were included as proxy measures of an
individual’s “differential location in the social structure”; Lee et al., 2004, p. 17).
Social Bonding and Age-Graded Theory of Informal Social Control
According to Hirschi (1969), the motivation to commit crime is constant across
individuals. As such, theorists need to ask why people refrain from committing crime (rather
than why they commit crime). Hirschi’s (1969) theory of Social Control is directly poised to
answer this question. In Causes of Delinquency, Hirschi (1969) states that what prevents
individuals from acting upon internal motivations to commit crime is informal social control.
Informal social control, according to Hirschi (1969) results from the development of social
bonds (defined as the tie between individual and society) through the process of
socialization. Hirschi hypothesizes that people with strong social bonds (which are comprised
of attachment, commitment, involvement, and belief) will conform and people with weak
social bonds will commit crime. Consistent with the theory’s premise, micro-level tests of the
theory have generally found support for Hirschi’s contentions (Agnew, 1985; Johnson, Jang,
De Li & Larson, 2000; Kempf, 1993; Krohn & Massey, 1980).
While Hirschi (1969) did not address the role structure or community-level effects
may have on social bonds, other researchers have brought attention to this possibility
(Cattarello, 2000; Matsueda & Heimer, 1987; Muftić, 2007; Stewart, 2003). For instance,
Matsueda and Heimer (1987) proposed that “broken homes, lower socioeconomic classes,
and high-crime neighborhoods should influence delinquency by impeding the formation of
strong attachments, commitments, involvements, and beliefs” (p. 828). Utilizing a single-
level statistical model where all variables were measured at the micro-level, Matsueda
and Heimer (1987) found support for their arguments that social bonds are impacted by
contextual effects, including neighborhoods.
Similarly, Cattarello (2000) examined the impact of neighborhood characteristics on
social bonds and peer associations. Cattarello (2000), however, constructed a multi-level
model where social disorganization variables were measured at the community-level and
social bonding and social learning variables were measured at the individual-level. A series
of HLM regressions found that social disorganization significantly influences associations
with delinquent peers increasing the likelihood of delinquency among adolescents.
However, Cattarello (2000) did not find social disorganization to significantly impact social
bonds.
Building upon Hirschi’s (1969) original writings on Social Control, Sampson and Laub
(1993; Laub & Sampson, 2003) have extended Social Control Theory to examine the impact of
social bonds on criminal motivation in adulthood. Specifically, they state that while adults who
were delinquent as juveniles have an increased likelihood of committing criminal acts as adults
(continuity), the formation of adult social bonds may decrease such likelihood (change).
Micro-level tests of the theory have generally supported Sampson and Laub’s theoretical
arguments (Bouffard, 2003; Horney, Osgood & Marshall, 1995; Laub & Sampson, 2003;
Uggen, 2000; Wright, Cullen & Williams, 2002).
In addition to expanding Social Control theory to include an analysis of adulthood,
Sampson and Laub (1993) propose a cross-level theoretical argument in their 1993 book,
Crime in the Making. Specifically, they hypothesize that adolescent delinquency can be
explained by examining the impact of structural characteristics, such as residential mobility,
socio-economic status, and family disruption, on informal social controls including the family
and the school. In their reanalysis of the Gluecks’ dataset, they find support for their
hypothesis that family process variables mediate the relationship between structural
characteristics and adolescent delinquency. Specifically, they found that “family process
mediated approximately 75% of the effect of structural background on delinquency”
(Sampson & Laub, 1993, p. 96). To date, there have been no other studies that have
examined Sampson and Laub’s cross-level arguments.
Power Control Theory
Hagan’s (1989) Power Control Theory is another example of a recent theory that
considers the necessity of cross-level theoretical explanation. In his effort to explain gender
differences in offending patterns, Hagan proposes that the relationship between gender and
delinquency is largely mediated by the interaction between class and social control (Bernard
& Snipes, 1996). Mainly, Hagan (1989) argues that the degree of control exercised within a
family varies based on the position of the family within the social-class structure of the
community in which it resides. Support for the theory, however, has been mixed (Hagan,
Simpson & Gillis, 1987; Jensen & Thompson, 1990; Morash & Chesney-Lind, 1991; Singer &
Levine, 1988). In addition, each of these tests was conducted using single-level datasets
(e.g., micro-level data). Interestingly, none of the aforementioned studies considered their
reliance on a single-level dataset as problematic in testing the cross-level propositions of
the theory.
