What we know and what we need to know about developmental and life-course theories

Introduction

The core question of interest among criminologists centers on why people commit crime. The core question of interest among developmental/life-course (DLC) criminologists is why different people commit crime at different ages, why some offend at high rates and others at low rates, why some continue offending over the life-course and others desist in late adolescence, and what sorts of prevention and intervention efforts can be developed and implemented in order to, as Farrington and Welsh (2007) observed, save children from a life of crime.

In this essay, I consider what is known and what needs to be known about DLC theories and research. In so doing, I remark briefly on the period before DLC came about, and highlight some of the early criminological work, and then segue into the exciting period of the 1980s in criminological theory where intense and fruitful debates regarding age and crime and criminal careers help to spur DLC as it is known today. I then highlight some of the most prominent DLC theories, what they tell us about longitudinal offending patterns, and then I close with 11 challenges for DLC theory that must be overcome in order to further our understanding of why people begin to offend, why some continue offending, and why some desist from crime.

Background

The origins of the discipline of criminology may be located in the individual-focused biological and psychological theories of crime outlined by Lombroso, Goring, and others. The essence of these theories was that there was something innate in the person that led to criminal involvement. These theories, however, did not consider whether these individual factors were differentially related to offending at various points throughout the life-course. In this regard, their view of why persons committed crime was a purely static, unchanging one. At the turn of the 20th-century, however, criminology started to move away from these individual-centered theories of crime to consider the roles of environment (Shaw and McKay), structure (Agnew), and process (Hirschi, Akers). One notable exception was the work of Clifford Shaw, whose case study of Stanley, a Chicago delinquent, began to consider issues related to development and crime—but it fell short of espousing any sort of DLC view of why people engage in crime and why people stop engaging in crime.

Everything changed in 1983. Not only was this the year of monumental album releases (R.E.M.’s Murmur, The Police’s Synchronicity, U2’s War, Def Leppard’s Pyromania, Billy Idol’s Rebel Yell, Madonna’s self-titled record, and of course Mötley Crüe’s Shout at the Devil to name a few), but it was also the year that everything changed in criminology with the publication of Hirschi and Gottfredson’s (1983) article on age and crime. Three claims were advanced in this paper. First, the relationship between age and crime was invariant, and always had the same aggregate form across history, cultures, and demographic groups. Second, the correlates of crime were the same at all ages, implying that no correlate of crime interacts with age. Third, the age–crime relationship could not be accounted for by any known (at that time) variable or set of variables. These invariance, non-interactive, and inexplicability hypotheses were such that a purely static view of crime was all that was necessary and because of this the need for longitudinal data was a fruitless and expensive pursuit.

Simultaneously in 1983, a Panel on Research on Criminal Careers was convened by the National Academy of Sciences (NAS) and charged with evaluating the feasibility of predicting the future course of criminal careers, assessing the effects of prediction instruments in reducing crime through incapacitation, and reviewing the contribution of research on criminal careers to the development of fundamental knowledge about crime and criminals (Blumstein, Cohen, Roth, & Visher, 1986, p. x). Although the NAS report produced a number of important findings, one of the most important contributions was the articulation of the criminal career, which highlighted the fact that criminal careers were marked by involvement in a variety of dimensions including: onset, persistence, escalation, specialization, career length, co-offending, and desistance. In doing so, the Panel highlighted the differences that existed in the longitudinal patterning of criminal careers, as well as providing a focus on both between- and within-individual changes in criminal activity over time (Piquero, Farrington, & Blumstein, 2003, p. 377). Naturally, these articulations led to the possibility that the invariant aggregate relationship between age and crime outlined by Hirschi and Gottfredson may not hold at the individual-level much less be true for all offenders at all times and at all places (see also Sweeten, Piquero, & Steinberg). Thus, the Panel more than implied that theories may need to be revisited with respect to different patterns of offending within and between persons throughout the life-course.

Undoubtedly, Hirschi and Gottfredson’s age–crime article as well as the NAS report on criminal careers forced the field of criminology to seriously consider the age–crime relationship but moreover, how criminological theories, which were mainly static, could account for changes in the relationship between age and crime.

The origins of DLC

The seeds of researchers’ interest in DLC took shape with the pioneering work of Sheldon and Eleanor Glueck, whose longitudinal studies of serious delinquents followed into adulthood helped to inform generations of researchers. But DLC, especially the heavy theorizing that is a hallmark of the framework today, can be traced to the work of researchers outside of criminology who had been thinking critically about developmental issues, though mainly in the childhood period of the life-course and primarily on direct (physical) and indirect aggression. Beginning in the late 1970s and into the 1980s, a handful of psychologists, but primarily Gerald Patterson at the Oregon Social Learning Center, Richard Tremblay with the Montreal Longitudinal Experimental Study, Rolf Loeber and colleagues, who were working on a planning study funded by National Institute of Mental Health (NIMH) focusing on the development of childhood aggression, and David Farrington, who was working closely with Donald West in the Cambridge Study in Delinquent Development, were laying the important groundwork for DLC. Along with the NAS report on criminal careers in the mid-80 s, with its organizing framework centered on describing and understanding the longitudinal patterning of criminal careers, this line of research helped to initiate the conversation around DLC issues, but the state of criminological theory in the late 1980s was, much as it was in the years prior, “in a state of theoretical paralysis” (Wellford, 1989, p. 119). These calm winds would soon turn into large gusts.

In 1990, Gottfredson and Hirschi’s A General Theory of Crime levied a strong set of critiques against criminological theory, the criminal careers framework, and longitudinal research designs more generally, and virtually forced researchers into demonstrating the need for (more complicated) theoretical frameworks and expensive data collections. After all, studying DLC requires longitudinal data because such data provide information about the development of offending over time, including data on ages of onset and desistance, frequency and seriousness of offending, duration of criminal careers, continuity or discontinuity of offending, specialization and escalation, as well as information about developmental sequences, effects of life events, and effects of risk and protective factors at different ages on offending at different ages (Farrington, 2013).

Not surprisingly, Gottfredson and Hirschi’s work was met with much criticism and debate, but there is no denying that it served (and continues to occupy) a critically important place in criminology. Soon after their work appeared, theoretical frameworks that were strongly cognizant of the accumulated knowledge about development more broadly continued to grow in psychology, but more importantly began to emerge in criminology. Although the 1990s were filled with many theoretical frameworks friendly to the DLC model, two of the most influential DLC-related theories are Terrie Moffitt’s (1993) developmental taxonomy of antisocial behavior and Sampson and Laub’s (1993) age-graded informal social control theory.

Moffitt’s theory challenges the notion that the aggregate age–crime curve is characteristic of all individual offenders. Instead, she contends that the aggregate relationship actually hides the variability in offending patterns of two distinct groups of offenders, each of whom offend for different reasons, at different rates, and in different manners at different periods of the life-course.

The first group of offenders, life-course-persistent (LCP), is made up of a very small number of persons, but they exhibit frequent, chronic, and aggressive antisocial behavior throughout the entire life-course. Because of the rarity of this group, and because of their extreme behavioral repertoire, Moffitt anticipates that the risk factors emerge in early childhood—and for some children in utero. Specifically, LCP’s are marked by neuropsychological impairments and tend to experience disadvantaged familial and neighborhood environments. This interaction is quite adverse, as children with extreme deficits tend to be born into families that do not have the resources necessary to help children overcome their problems. As a result, LCP’s experience problems in several life domains, including education and social relationships, and also manifest antisocial behavior early and throughout the life-course. Much of the behavior exhibited by LCP is anticipated to be serious, including person-oriented acts. The prospects for change among LCP’s is rather dim, as they experience failure in many other aspects of adulthood, most of which serve to compromise their chances of success in conventional activities.

The second group offenders, adolescence-limited (AL), comprise the vast number of offenders who do not experience the injurious childhoods that that their LCP counterparts experience. For these relatively normal youth, offending is largely restricted to the adolescent time period. The primary reasons given for AL-offending include the peer social context and the maturity gap, or the recognition that adolescents physically resemble adults but are not afforded the privileges that come with adulthood. As a result of being “stuck” in this rather precarious social position, and because other peers are experiencing the same strains and pressures, AL’s engage in acts that are designed to help adolescents perceive that they are being freed from the grips of adolescence and make them feel like adults (alcohol and drug use, promiscuous sex, theft). The flirtation with delinquency, however, is short-lived as the majority of AL’s desist from offending just as soon as they gain the status, recognition, and privileges when they become adults. Only a few of these youth may continue to offend in adulthood, but only if they experience particular ensnaring events, such as a conviction or drug addiction, that catapults them into a persisting offending pattern. Much research has examined Moffitt’s developmental taxonomy, and reviews may be found in Moffitt (2006) and Piquero and Moffitt (2005).

Within criminology, Sampson and Laub’s age-graded informal social control theory stands as the case exemplar for a DLC theory that occupies a middle ground between Gottfredson and Hirschi’s purely static, non-developmental theory and Moffitt’s multi-group, developmental framework. For these authors, childhood factors are relevant for offending in adolescence, but less so for offending in adulthood, mainly because offending in adulthood is influenced by life events and life transitions. In their informal social control model, these authors pay close attention to the positive life events that may help deflect offenders from continuing their adolescent offending into adulthood. Relying on their analysis and extension of data collection from one of the classic Glueck studies, Sampson and Laub found that social bonds, including marriage—and especially good marriages—served to lower offending involvement into and throughout adulthood above and beyond the (little) influence of childhood risk factors. The argument underling their theoretical model is that social bonds like marriage and employment serve to increase one’s social capital. Conversely, weak social bonds and/or the severing of social bonds would lead to more or continued offending involvement.

Since the mid-80 s, there has been a significant amount of empirical research testing various aspects of these theories, as well as other DLC frameworks (Le Blanc & Loeber, 1998; Loeber & Le Blanc, 1990), and much of this research has been made possible by the advent and continuation of longitudinal studies throughout the world (Liberman, 2008). An excellent summary of the key findings from DLC research was articulated by Farrington (2003, pp. 223–224), and included: (1) the prevalence of offending peaks in the late teens; (2) the peak age of onset of offending is between 8 and 14, and desistance peaks between age 20 and 29; (3) an early age of onset predicts a longer criminal career; (4) there is marked continuity in offending and antisocial behavior from childhood through adolescence and adulthood; (5) a small number of offenders are responsible for most crimes; (6) versatility in the nature and mix of offenses, rather than specialized offending, is the norm; (7) criminal involvement is part of a general antisocial behavioral syndrome, with offenders exhibiting a variety of antisocial and criminal behaviors; (8) most offenses committed during adolescence are characterized by group involvement, while solo offending tends to characterize offending after age 20; (9) offending during adolescence is explained by a wide variety of factors, including utilitarian, excitement, anger, or boredom, while adult crime is primarily utilitarian; and (10) early offending is less serious in nature but there is a tendency for the severity of offending to increase over time.

Challenges

It is fair to say that theoretical and empirical research on DLC has generated an important number of insights with respect to the nature and patterning of criminal careers. But the limits of theory and data stand in the way of a more complete characterization of criminal activity over the life-course. Several contentious issues remain to be sorted out (Farrington, 2003), and a great many research questions remain to be addressed. With my closing remarks, I would like to highlight 11 of these for the research community to consider.

First, the jury is still out on the issue of different groups of offenders, whether they exist in the real world, whether they are useful heuristic, whether they follow the paths predicted for them, whether they are characterized by particular risk and protective factors, and so forth. Complicating matters is the fact that the number of groups seems variable across data sets, follow-up periods, and whether self-reported or official records serve as the basis for information on criminal activity (Piquero, 2008). The bottom line is whether the complexity associated with group-based theories is warranted or whether it is a distraction.

Second, the extent to which different risk factors at different ages for different people are needed to better understand the longitudinal patterning of criminal behavior is still marked by a lack of theory and empirical research. Work on this issue is not only of interest theoretically, but also policy-wise because it may help inform age-graded prevention and intervention efforts.

Third, the role of biology and genetics with respect to antisocial behavior over the life-course still merits theoretical and empirical scrutiny. There is little denying that this matters, but the problem has been understanding why it matters and why it matters in some but not all contexts. The lack of data on these factors has hampered in-depth work and remains a high priority.

Fourth, longitudinal studies in criminology have mainly been focused on individual, familial, and peer-related issues and has paid less attention to the role of neighborhood contexts. With a few important exceptions (Sampson, Morenoff, & Raudenbush, 2005), we know very little about how and why context matters and whether contextual effects are more salient at different periods of the life-course (Piquero, Moffitt, & Lawton, 2005). This remains a very important question, but one that is largely devoid in the theoretical literature.

Fifth, with a few exceptions (Moffitt, 1994), DLC theory has not tended to devote much attention to demographic similarities and differences. Yet, there is more than ample research documenting important demographic variation in antisocial and criminal behavior over the life-course. Further understanding the patterns of similarity and dissimilarity is critical if criminology is to provide ample explanation of offending over the life-course for all persons.

Sixth, an important but less studied question revolves around prediction, and especially the prediction of adult offending from childhood risk factors (Farrington, Piquero, & Jennings, 2013; Laub, & Sampson, 2003; Piquero, Farrington, & Blumstein, 2007). Clearly, theories need to be able to predict the future course of offending, and such information is of obvious relevance for designing programs.

Seventh, most DLC theory, and certainly almost all DLC-related empirical research, has prioritized understanding more general patterns of criminal involvement and devoted less attention to situational factors and/or the decision-making process of offenders within certain situations. Yet, we know from other, non-DLC research that these two are important considerations in understanding crime (Horney, 2006; Piquero & Moffitt, 2005). Integrating situations and offender decision-making patterns into DLC would be very useful (see Wikström, Oberwittler, Treiber, & Hardie, 2012).

Eighth, DLC research and the longitudinal studies that form their playing field have started to consider the potential effects and consequences of incarceration for subsequent criminal careers. This line of work has tended to find either null or adverse effects of imprisonment on subsequent offending, with almost no studies finding any deterrent effect associated with incarceration (see Nagin, Cullen, & Jonson, 2009; Piquero & Blumstein, 2007). Continued investigation of the effect of incarceration on criminal offending in general, and then across individual and demographic characteristics in particular, would be welcomed.

Ninth, themes gleaned from qualitative narratives of offenders have provided some rich detail regarding the reasons underlying stability and change in offending careers (Laub & Sampson, 2003; Maruna, 2001). There are simply some insights that can be obtained from these interviews that cannot be obtained from traditional survey formats, and while such work is both difficult and time-consuming, the payoffs are large and may also lead to significant theoretical developments.

Tenth, DLC theories and subsequent research have paid close attention to the effect of life events in altering subsequent offending trajectories (Horney, Osgood, & Marshall, 1995; Laub, Nagin, & Sampson, 1998; Nagin, Pagani, Tremblay, & Vitaro, 2003), but there has not been much investigation of life events and transitions above and beyond the most traditional ones like marriage, employment, and children. Even further, there has been even less attention devoted to understanding how individuals interpret—and give meaning to—life events and transitions. Additional quantitative and qualitative research on life events remains important.

Finally, like most criminological research, the vast majority of DLC theorizing and research has been limited to comparisons between white and black offenders and also tended to rely on mainly general population or community samples. Data constraints have precluded more detailed analysis of Hispanic and Asian populations, as well as the potentially unique experiences and risk/protective factors that may be related to the offending patterns of such persons (i.e. assimilation, acculturation, etc.). Recent work on differences in longitudinal offending patterns across immigrant and native-born (US) offenders has been insightful, and largely dispelled the myth that immigrants offend at higher rates and more seriously than non-immigrants (Bersani, Loughran, & Piquero, 2014), but more work remains to be done. Also, many longitudinal studies rely on general or community samples, such that little DLC research has been conducted on serious (youthful) offenders followed into adulthood (for an important exception see Mulvey et al., 2004). As these offenders represent an important policy-relevant sample, much more consideration should be given there.