Life-Course, Developmental and Integrated Theories of Crime

The single most replicated finding in criminology is the age-crime curve: recorded offending rises sharply through early adolescence, peaks around the late teens, and then declines through the twenties and beyond, in almost every society and time period where it has been measured.

Travis Hirschi and Michael Gottfredson argued in the early 1980s that this curve is invariant across social groups and historical periods, and used that claim to argue that age itself, not any social variable, does the causal work, so theories should focus on stable individual differences established early in life rather than on anything that changes across the life course.

Sampson and Laub, and later Moffitt, pushed back on the invariance claim in different ways. Sampson and Laub accepted that the aggregate curve is real but argued it hides enormous individual variation.

Two people with near-identical backgrounds and identical adolescent offending records can diverge sharply by their thirties, one desisting and one persisting, and a theory that only explains the population-level curve cannot explain that divergence. Moffitt argued the aggregate curve is itself a statistical composite of two distinct groups with different causes, so a single explanation for the whole curve was the wrong target from the start.

Both responses required a different kind of data and a different theoretical unit. Explaining a snapshot difference between offenders and non-offenders needs cross-sectional comparison. Explaining why a trajectory bends needs longitudinal data on the same individuals over years or decades, plus a theory that can name the mechanism operating at each life stage. That shift in method and unit of analysis is what makes life-course and developmental criminology a distinct branch rather than a restatement of control or strain theory.

This is also where the branch gets its two-part name. Developmental criminology, closely associated with Moffitt, focuses on the origins of individual differences and the trajectories that follow from them, often drawing on developmental psychology.

Life-course criminology, closely associated with Sampson and Laub, focuses on how social institutions and life events, not just early traits, continue to shape behaviour well into adulthood. The two overlap heavily in practice, and this topic treats them together, but the emphasis differs: origin and trajectory-type on one side, change and social context on the other.

Sampson and Laub's age-graded theory of informal social control, set out fully in Crime in the Making (1993), extends Travis Hirschi's social bond theory across the whole life span. Hirschi's original 1969 theory explained delinquency through the strength of a child's or teenager's bonds to conventional institutions, principally attachment, commitment, involvement and belief.

Sampson and Laub kept the bonding logic but argued that the relevant institutions change with age: family and school bonds matter most in childhood and adolescence, while marriage, military service and employment bonds take over that stabilising function in adulthood. Weak bonds at any stage raise the likelihood of offending at that stage, and strong bonds can redirect even a heavily delinquent trajectory.

The theory rests on an unusual empirical foundation. Sheldon and Eleanor Glueck had assembled a longitudinal dataset in the 1940s on 500 delinquent boys and 500 matched non-delinquent boys from Boston, following them into their thirties, but the raw records had been effectively lost to the field for decades.

Sampson and Laub located the archived records, recoded them with modern statistical methods, and later traced many of the same men through official records and life-history interviews into their seventies, published as Shared Beginnings, Divergent Lives (2003). This gave them one of the longest individual-level offending datasets ever assembled, spanning childhood to old age for the same cohort.

Two findings from that data anchor the theory. First, childhood delinquency predicts adult crime only weakly once adult social bonds are accounted for, which argues against a purely trait-based account where early conduct problems mechanically determine adult outcomes.

Second, change is common: many heavily delinquent boys desisted by their thirties and forties, and the men who desisted were reliably the ones who had formed a strong marital bond or a stable job, while men with similar early records who lacked those bonds tended to persist. The theory therefore treats desistance as an ordinary, explainable outcome of changed social circumstances rather than as a mysterious exception to an otherwise stable criminal disposition.

A further contribution is the idea of cumulative disadvantage: early delinquency and the official response to it, such as school exclusion or juvenile confinement, can themselves weaken the very bonds that would otherwise support desistance, for example by damaging school and employment prospects. This creates a feedback loop in which early trouble makes later bonding harder to achieve, without making desistance impossible.

Sampson and Laub call the adult events that redirect a trajectory turning points, and the concept does specific theoretical work: a turning point is not just a positive life event, it is an event that restructures a person's daily routines, supervision and social attachments enough to change the balance of costs and rewards attached to offending. A wedding by itself is not the mechanism; a marriage that produces daily accountability to a spouse, altered peer associations and new obligations is.

Marriage was the clearest turning point in the Glueck men's data, but only where it was what Sampson and Laub call a good marriage, meaning one with genuine attachment and mutual investment. A marriage entered into for convenience, or to a spouse who was also involved in offending, showed little desistance effect, which is consistent with the theory's emphasis on bond quality rather than bond existence.

Military service in the World War Two and Korean War era functioned as a similar turning point for many of the Glueck men, providing structure, a break from prior peer networks, and later the GI Bill's access to education and stable employment.

Sampson and Laub were careful to note this was historically specific: peacetime or combat-heavy service, or service without comparable post-service benefits, would not necessarily reproduce the effect, and later research on other cohorts has found more mixed results for military service as a turning point.

Stable employment works through a related mechanism: a steady job supplies routine, a stake in conformity through wages and reputation, and, often, a shift in daily peer contact away from prior co-offenders.

Sampson and Laub's later interview data added a further point that the original quantitative model could not capture well: desistance is rarely a single decisive break, and men often described a gradual process of drifting away from offending as new commitments accumulated, with occasional relapse, rather than an instant conversion at the moment of marriage or a new job.

The same pattern of institutional turning points shows up outside the United States: the Cambridge Study in Delinquent Development, David Farrington's prospective study of 411 London men born in 1953 and tracked from age eight into their sixties, found that marriage, stable employment and moving away from a delinquent peer network in London fostered desistance in much the same way as the Glueck men's data, evidence that the mechanism is not a quirk of one mid-century American city.

Terrie Moffitt's developmental taxonomy, published as Adolescence-Limited and Life-Course-Persistent Antisocial Behavior: A Developmental Taxonomy in Psychological Review in 1993, argues that the age-crime curve is a composite of two etiologically distinct groups rather than one process. This is the central point of difference from Sampson and Laub: Moffitt's model is primarily about explaining group differences in origin, while Sampson and Laub's is primarily about explaining change within a person across adulthood.

Life-course-persistent offenders are, in Moffitt's account, a small group whose antisocial behaviour appears very early, often by age three or four as biting and hitting, and continues in an age-appropriate form across the life span, from childhood conduct problems to adolescent delinquency to adult crime and related difficulties such as poor employment and relationship stability.

The proposed origin is an interaction between subtle neuropsychological deficits, particularly in verbal and executive function, and a disrupted or high-risk rearing environment. Neither factor alone is treated as sufficient; the theory is explicitly an interactionist one in which a vulnerable child in an adverse environment is caught in a self-perpetuating cycle, since a difficult temperament evokes harsher or less consistent parenting, which in turn worsens the child's self-regulation.

Adolescence-limited offenders are, by contrast, the majority of adolescent offenders, and their offending has no early conduct-problem history and typically stops by the early twenties.

Moffitt explains their offending through what she calls the maturity gap: modern adolescents reach physical and social maturity years before they gain access to adult privileges such as independent income, adult-sanctioned relationships and legal autonomy, and antisocial behaviour becomes a way of claiming adult status by mimicking the more visibly antisocial life-course-persistent peers. Because the underlying motivation is social mimicry rather than a stable deficit, this group's offending recedes once genuine adult roles become available.

The taxonomy has been influential precisely because it makes testable predictions that Moffitt and others followed up over subsequent decades, including in the Dunedin longitudinal study in New Zealand that Moffitt worked on directly: life-course-persistent offenders should show measurable neuropsychological differences in childhood, should offend across more varied crime types, and should show worse adult outcomes even outside crime, while adolescence-limited offenders should not show those childhood markers.

Later work, including from Moffitt's own team, refined the taxonomy further, for example identifying a low-level chronic group and questioning whether adolescence-limited offending is always harmless, since some in that group carry forward problems such as substance dependence or financial difficulty into adulthood even after crime itself stops.

The taxonomy has also travelled outside the high-income, largely Western settings where it was first tested: the 2004 Pelotas Birth Cohort in Brazil, which has followed several thousand children from birth, found conduct-problem trajectories with the same shape, a small early-onset group whose problems persist, a larger group whose conduct problems appear later and fade, and a low-problem majority, suggesting the underlying developmental split is not confined to New Zealand or the United States.

By the 1970s, criminology had several well-developed but competing theoretical families: Robert Merton's strain tradition, Hirschi's control tradition, and Edwin Sutherland's differential association and its social learning descendants.

Each had accumulated real empirical support, and each also had well-documented gaps that its rivals filled. Strain theory struggled to explain why most strained individuals do not offend. Control theory struggled to explain the positive content of delinquent motivation, since a bond-deficit model says little about why offending takes the specific form it does. Social learning theory explained how delinquent conduct is transmitted through association but said less about why the initial exposure or motivation to offend arises.

Delbert Elliott, alongside Suzanne Ageton and Rachelle Canter, argued in a 1979 paper that these gaps were complementary rather than fatal, and that a model combining strain, weak conventional bonding and exposure to delinquent peer groups through a learning process would outperform any one theory alone.

Their integrated model proposed a causal sequence in which strain and weak conventional bonds increase the likelihood that a young person will affiliate with delinquent peers, and that peer group, through a social learning process, becomes the proximate cause of delinquent behaviour itself. Integration here means specifying how constructs from separate theories combine into a single causal chain, not merely listing risk factors side by side.

This was a genuine departure from the theory-competition norm that had dominated criminology, where researchers typically designed studies to show one theory outperforming another. Integration accepts that multiple mechanisms operate together and tries to model their combination and sequence explicitly, which raised a methodological bar: an integrated theory has to specify which variable causes which, in what order, not just that several variables correlate with offending.

The move toward integration coincided with, and partly motivated, the developmental turn covered earlier in this topic, since a model that specifies a causal sequence across strain, bonding and peer exposure naturally invites the further question of how that sequence unfolds differently at different ages, which is exactly the question Thornberry's interactional theory takes up next.

Terence Thornberry's interactional theory, set out in Toward an Interactional Theory of Delinquency in the journal Criminology in 1987, took integration a step further by making the causal relationships reciprocal rather than one-directional.

In Elliott's model, weak bonds and strain lead to delinquent peer association which leads to delinquency, in that order. Thornberry argued the arrows also run backward: delinquent behaviour itself weakens attachment to parents and school, and increases affiliation with delinquent peers, which then further increases delinquency in a self-reinforcing loop rather than a one-way chain.

Thornberry's second major addition was developmental: he argued the relative weight of each cause changes across adolescence. Family attachment matters most in early adolescence, school commitment and peer association become more influential in mid-adolescence, and broader commitment to conventional adult roles becomes more influential as a person moves toward early adulthood. A theory that fixes the same causal weights at every age, Thornberry argued, misses how the same young person's key vulnerabilities shift over a relatively short number of years.

This kind of contribution is generally described as theoretical elaboration rather than pure integration: rather than combining two named theories wholesale, Thornberry took control theory's bonding constructs and social learning theory's peer-transmission mechanism and rebuilt them into a new reciprocal, developmentally staged model, tested using the Rochester Youth Development Study, a longitudinal panel he directed. The reciprocal design has held up reasonably well empirically, particularly the finding that delinquency itself erodes bonds rather than only following from weak bonds.

Integration and elaboration both carry a recurring criticism: combining well-defined, independently falsifiable theories into one larger model can make the resulting theory harder to test as a whole, since a disconfirming result might reflect a failure in any one of several combined mechanisms rather than a clean rejection of the whole model.

Critics, including some working from a purer control-theory tradition, have also argued that integration sometimes papers over genuine theoretical disagreements, for instance between theories that treat motivation to offend as something requiring explanation and theories, like Hirschi's, that treat weak control as sufficient on its own.

Proponents respond that criminal behaviour is caused by multiple interacting mechanisms in the real world, so a model that captures that complexity, even at some cost to parsimony, is a better guide to both explanation and intervention than a simpler theory that is easier to test but leaves more variance unexplained.