THEORETICAL INTEGRATION: AN OVERVIEW
While causes of crime appear to be varied and diverse, (Braithwaite, 1989; Wilson &
Herrnstein, 1985) theories of crime traditionally involve only single factor explanatory
models (Liska, Krohn & Messner, 1989). Elliott (1985) argues that theoretical reliance on a
single explanatory variable to explain criminal behavior has resulted in theories that are
capable of explaining only a small percentage of the variance in crime or criminal behavior.
Some theorists argue that the only way in which to adequately account for the complexity of
such behavior and to increase explained variance is through theoretical integration (Elliott,
1985; Wellford, 1989). Theoretical integration is generally defined as “the act of combining
two or more sets of logically interrelated propositions into one larger set of interrelated
propositions, in order to provide a more comprehensive explanation of a particular
phenomenon” (Thornberry, 1989, p. 75).
Goals of Integration
There are generally three goals of theoretical integration. The first goal of integration is
theory reduction. Some criminologists argue that scientific progress has been retarded because
there are too many theories competing against one another in an effort to essentially explain the
same type of behavior (Barak, 1998). Consequently, an abundance of theories impedes their
development by diffusing research attention (Bernard & Snipes, 1996). Theory reduction is
proposed as one way to decrease the number of criminological theories, allowing researchers to
focus on a smaller number of theories. The second goal is to increase explained variance. As
previously stated, current theories are capable of explaining, at best, about 20% of variance in
criminal behavior (Elliott, 1985). While this might be just enough variance explained to keep the
theory alive, it is not enough to support the usefulness of the theory related to prediction, crime
prevention, and treatment.
One way that explained variance may be increased (allowing for its expanded use) is
through theoretical integration. The third goal of theoretical integration is theory
development through the clarification and expansion of existing propositions and
theoretical concepts.
Alternatives to Integration
Alternatives to theoretical integration include theory competition and theoretical
elaboration. Theoretical competition involves the pitting of two theories against each other
in an empirical test that determines which theory’s variables have the most explanatory
power (Hirschi, 1979; Paternoster & Bachman, 2001). The two theories that are compared
most often in this manner are social bonding and differential association. Overall,
differential association variables usually find more support when compared against social
bonding variables (Agnew, 1991; Allaird, Burton & Cullen, 2001; Costello & Vowell, 1998:
Matsueda & Heimer, 1987). The results of theory competition, however, are seldom
definitive. In addition, because each theory accounts for just enough variance to survive
outright rejection, rarely is one theory disregarded in favor of another (Liska et al., 1989).
Theoretical elaboration involves the expansion of a current theory with the end goal
of building a more comprehensive, and more well-developed theoretical model than was
proposed by the original theorist (Thornberry, 1989). This is typically done through the
clarification of original propositions, as well as the addition of new concepts that may or
may not be borrowed from existing theories or disciplines. Sampson and Laub’s (1993; Laub
and Sampson, 2003) Age-graded Social Control Theory is an example of such an approach.
Resurrecting the importance of social bonds, Sampson and Laub maintain Hirschi’s original
contention that delinquency occurs as a result of weakened or broken social bonds (Laub,
Sampson & Allen, 2001). Sampson and Laub (1993), however, expand on Hirschi’s (1969)
original Social Bonding Theory through the inclusion of an examination of the impact of
adult social bonds on offending, particularly related to persistence and desistance.
Therefore, they broaden the scope of the theory (which was originally proposed to explain
only continuity in juvenile offending) through the exploration of the manner in which social
bonds develop and change over the life course impacting criminal trajectories (Sampson &
Laub, 1993; Laub and Sampson, 2003). In doing so, Sampson and Laub (1993) incorporate
concepts from Social Disorganization, labeling and subcultural theories. While there are few
studies that have tested Sampson and Laub’s (1993) Age-graded Theory of Social Control,
research generally does find support for their arguments (Bouffard, 2003; Giordano,
Cernkovich & Holland, 2003; Horney, Osgood & Marshall, 1995; Wright & Cullen, 2004).
Advantages and Disadvantages of Theoretical Integration
Opponents of theoretical integration argue that, because of differences in underlying
philosophical assumptions, integration is not possible (Bernard, 1989; Hirschi, 1979, 1989).
For instance, strain, control, and learning theories are based on different (and arguably
incompatible) assumptions about human nature (Kornhauser, 1978). Strain theory proposes
that human nature is essentially good. Control theory, in contrast, is based on the
assumption that humanity is inherently antisocial, while learning theories center around the
notion that human nature is a blank slate and that individuals must learn to be either good
or bad. Because criminological theories were historically developed in direct opposition to
each other, opponents of integration, such as Hirschi (1979), argue that conflicting theories
cannot be integrated. Consequently, the only way to advance the field is through theoretical
competition or elaboration (Hirschi, 1979; Thornberry, 1989).
While research may provide support for integrated theories, some criminologists remain
unconvinced of the potential merits of theoretical integration. In addition, the complexity of
integrated theoretical models, which call for the use of advanced statistical methods as well as
an understanding of different disciplines (i.e., biology, psychology, and sociology), may be off-
putting for some. Proponents of theoretical integration, however, argue that criminologists’
staunch reliance on theoretical competition as the primary manner in which to advance the field
has only inhibited theoretical development (Bernard & Snipes, 1996; Elliott, 1985; Pearson &
Weiner, 1985).
Frustrated by the inability of theoretical competition to rid the discipline of a large
number of theories that are largely incapable of adequately explaining crime and
delinquency, theorists contend that the integration of existing theories provides several
advantages over other methods of theory development (Bernard, 1990; Elliott, 1985;
Wellford, 1989). For instance, integrated theories provide a manner in which theorists can
piece together portions of existing theories that have been found to be related to crime
and delinquency, while at the same time disregarding portions of those theories that are
unrelated. Such exercises will ideally result in an integrated theoretical model capable of
explaining a greater portion of the variance that is left unexplained by separate theories
(Wellford, 1989).
Furthermore, advocates assert that theoretical integration will advance the
discipline by directing research interest and activity back to one of the original principles of
theory development, explanation of the dependent variable (Gibbons, 1994). Only through
integration of existing theories (including across levels of explanation) will criminologists
improve our understanding of the phenomenon we study, as well as improving the
predictive power of our theories (Elliott, 1985).
TYPES OF THEORETICAL INTEGRATION
As previously defined, theoretical integration involves the combination of two or
more theoretical propositions (Liska et al., 1989). Integration can take many different forms.
Generally, contemporary theoretical integration is either conceptual (i.e., integration of
theories with similar concepts) or propositional (i.e., integration of theories with differing
propositions). However, there has been a recent push towards the integration of cross-level
explanations of crime and delinquency (Short, 1989; Barak, 1998). The following section will
provide a more thorough discussion of the various types of theoretical integration, as well as
present examples of each1.
Propositional integration involves the formal process of integrating different
theoretical propositions (Liska et al., 1989). Through this process, a new separate theory is
created. Generally, there are three types of propositional integration: side-by-side, end-to-
end, and up-and-down (Hirschi, 1979).
Side-by-Side Integration. Side-by-side integration generally involves the integration
of partial theories to explain varied phenomena (Hirschi, 1979). For instance, different
theoretical propositions or concepts are selected to explain different types of criminal
behavior, such as different offense types (i.e., violence and property offending), or different
types of offenders within a general theoretical framework. According to Hirschi (1979), this
type of integration skirts the issue of differing theoretical assumptions while allowing the
theorist to increase the amount of variance explained. Typically, such an approach results
in a typological model, such as Moffitt’s (1993) typology of adolescent limited and life
course persistent offenders.
Moffitt (1993) proposes a developmental taxonomy that explains two primary
offending types, adolescent limited (AL) and life course persistent (LCP) offenders.
Adolescent limited offenders are described as adolescents who exhibit exaggerated
antisocial behavior which manifests only during adolescence and declines in early
adulthood (Moffitt, 1993). In contrast, LCP offenders are those whose antisocial behavior
manifests early and remains markedly stable over the individual’s life course (Moffitt,
1993). According to Moffitt (1993), different causal explanations are necessary to account
for etiological differences in offending types. Consequently, each theoretical explanation
incorporates different causal variables in its efforts to account for the described offending
types (Moffitt, 1993; Moffitt, Caspi, Rutter & Silva, 2001). Tests that have reviewed Moffitt’s
taxonomy have generally been supportive of her propositions (Moffitt, Caspi, Dickson, Silva
& Stanton, 1996; Piquero & Brezina, 2002; Tibbetts & Piquero, 1999).
End-to-End Integration. End-to-end integration entails the integration or reshuffling of
variables from differing theories so that the dependent variables of some theories become
the independent variables of the integrated theory (Hirschi, 1979). A prime example of this
type of integration is Thornberry’s (1987) Interactional Theory2. Thornberry integrates
control and learning theories in an end-to-end fashion. Thornberry (1987) speculates that
individuals with weak social bonds have a higher likelihood of associating with delinquent
peers, which increases their probability of engaging in delinquent behavior. Delinquent
behavior, in turn, further weakens attachments to social bonds and increases associations
with delinquent peers in a reciprocal manner. There are few tests of Thornberry’s theory;
however these studies have found tentative support for the theory’s propositions
(Thornberry, 1996; Thornberry, Gallant, Lizotte, Krohn, & Smith, 2003).
Up-and-Down Integration. Of the three types of propositional integration, up-and-
down is the rarest (Hirschi, 1979). Up-and-down integration involves the creation of an
abstract or general theory that encompasses multiple propositions from specific theories
(Liska et al., 1989). An example of such an approach is Cullen’s (1994) Social Support and
Coercion Theory. Rather than identifying concepts that can be absorbed by his theory,
Cullen (1994) creates a new theory based around a general concept, social support. Social
support, which can be either instrumental or expressive, refers to the ability of social groups
to meet the needs of its members (Colvin, Cullen & Vander Ven, 2002). According to Cullen
(1994), social support is a common theme running through multiple theoretical
perspectives. Hence, social support can act as a structuring concept which allows for the
development of a general theory of crime that explains crime and delinquency (Cullen,
1994; Cullen, Wright & Chamlin, 1999; Colvin et al., 2002). To date, there are few tests of
Cullen’s theory; however initial findings support it (Wright & Cullen, 2001; Wright, 1996).
Conceptual Integration
Conceptual integration is similar to up-and-down propositional integration in that it
identifies concepts that are similar between two or more theories (Akers & Sellers, 2004).
However, unlike propositional integration which maintains the original premises of each
theory, conceptual integration points out the similarity in theoretical concepts and then
absorbs the concepts of one theory into the concepts of the integrated theory (Bernard &
Snipes, 1996). For example, Akers (1998) argues that the concept of “belief” from Social
Bonding Theory can be absorbed by the concept of “definitions” in Social Learning
Theory. Akers (1998) goes on to contend that not only can Social Learning Theory take in
Social Bonding Theory, he boldly claims that the propositions and concepts situated by
social learning are capable of absorbing other theories including labeling, strain, conflict,
and deterrence.
