ArticlePDF Available

Self-Control through Emerging Adulthood: Instability, Multidimensionality, and Criminological Significance


Abstract and Figures

This study assesses self-control theory’s (Gottfredson & Hirschi 1990) stability postulate. We advance research on self-control stability in three ways. First, we extend the study of stability beyond high school, estimating group-based trajectory models (GBTM) of self-reports of self-control from age 10 to 25. Second, drawing in part on advances in developmental psychology and social neuroscience, especially the dual systems model of risk taking (e.g., Steinberg 2008), we investigate whether two distinct personality traits—impulsivity and sensation seeking—often conflated in measures of self-control, exhibit divergent developmental patterns. Finding that they do, we also estimate multitrajectory models to identify latent classes of co-occurring developmental patterns for these two traits. We supplement GBTM stability analyses with hierarchical linear models and reliable variance estimates. Lastly, using fixed effects models, we explore whether the observed within-individual changes in the global self-control measure as well as impulsivity and sensation seeking are associated with within-individual changes in crime net of overall age trends. We examine these ideas using five waves of data from a sample of several hundred African American adolescents from the Family and Community Health Study (FACHS). Results suggest that self-control is unstable, that distinct patterns of development exist for impulsivity and sensation seeking, and that these changes are uniquely consequential for crime. We conclude by comparing our findings to extant research and discussing the implications for self-control theory.
Content may be subject to copyright.
Arizona State University
KEYWORDS: self-control, impulsivity, sensation seeking, crime, developmental trajec-
tories, personality
This study assesses self-control theory’s stability postulate. We advance research on
self-control stability in three ways. First, we extend the study of stability beyond high
school, estimating GBTMs of self-control from ages 10 to 25. Second, drawing on
advances in developmental psychology and social neuroscience, especially the dual
systems model of risk taking, we investigate whether two distinct personality traits—
impulsivity and sensation seeking—often conflated in measures of self-control, exhibit
divergent developmental patterns. Finding that they do, we estimate multitrajectory
models to identify latent classes of co-occurring developmental patterns. We supple-
ment GBTM stability analyses with hierarchical linear models and reliable variance
estimates. Lastly, using fixed effects models, we explore whether the observed within-
individual changes are associated with changes in crime net of overall age trends. These
ideas are tested using five waves of data from the Family and Community Health Study.
Results suggest that self-control is unstable, that distinct patterns of development exist
for impulsivity and sensation seeking, and that these changes are uniquely consequen-
tial for crime. We conclude by comparing our findings with extant research and dis-
cussing the implications for self-control theory.
Over the past two decades, an incredible amount of research has examined the verac-
ity of Gottfredson and Hirschi’s (1990) self-control theory (hereafter, SCT), especially
its central tenet that self-control is the major source of individual differences in crimi-
nal propensity. This large body of work shows that various measures of self-control are
Additional supporting information can be found in the listing for this article in the Wiley Online
Library at
This research uses data from the FACHS, a project designed by Ron Simons, Frederick Gib-
bons, and Carolyn Cutrona, and funded by grants from the National Institute of Mental Health
(MH48165, MH62669), the Centers for Disease Control (029136–02), the National Institute on
Drug Abuse (DA021898), and the National Institute on Alcohol Abuse and Alcoholism. An ear-
lier version of this article was presented at the 2012 annual meetings of the American Society
of Criminology in Chicago, IL. The authors thank Kara Hannula, Tanja Link, Travis Pratt, and
three anonymous reviewers for insightful comments on earlier drafts of the article, and a particular
thanks goes to D. Wayne Osgood for his many valuable suggestions and guidance, all of which sig-
nificantly improved the article. Direct all correspondence to Callie H. Burt, School of Criminology
and Criminal Justice, Arizona State University, 411 N. Central Ave, Ste. 600, Phoenix, AZ 85004
C2014 American Society of Criminology doi: 10.1111/1745-9125.12045
CRIMINOLOGY Volume 52 Number 3 450–487 2014 450
strongly related to offending (for reviews, see Burt, 2014; Goode, 2008; Schulz, 2006).
Indeed, the size of the self-control–crime link renders it “one of the strongest known cor-
relates of crime” (Pratt and Cullen, 2000: 952).
Given the impressive relationship between self-control and offending, scholars have
increasingly focused on testing other aspects of SCT. In particular, a growing number of
studies has examined what is perhaps the most controversial aspect of the theory—the sta-
bility proposition. SCT proposes that low self-control is the natural state of humankind
and that effective parenting during the first decade of life is needed to inculcate self-
control. After childhood, between-individual levels of self-control are fixed. According
to SCT, either children develop self-control during these formative years, or they do not,
in which case, they will suffer from low self-control and its many negative manifestations
across the life course. In the words of Hirschi and Gottfredson (2001: 90): “The differ-
ences observed at ages 8 to 10 tend to persist. . . . Good children remain good. Not so
good children remain a source of concern to their parents, teachers, and eventually to the
criminal justice system.”
Notably, SCT’s strict relative stability proposition contrasts with more dynamic views
of self-control and analogous traits emerging from theory and research in psychol-
ogy. For example, models of personality recognize both within- and between-individual
changes over the life span (e.g., Caspi and Roberts, 2001; Helson, Jones, and Kwan, 2002;
Roberts and Mrozcek, 2008) as well as a different normative (mean-level) developmental
timetable for self-control and related concepts and traits (e.g., Casey, Getz, and Galvan,
2008; Steinberg, 2008; Vaidya et al., 2002). Moreover, the corollary proposition that indi-
vidual differences in criminality remain stable over time is in stark contrast to other crim-
inological theories, which identify both within- and between-individual change in crimi-
nality after childhood (Hirschi, 1969; Sampson and Laub, 1993; Simons and Burt, 2011).
As such, the SCT stability proposition has sparked debate, and an increasing number of
studies has been conducted to examine its accuracy.
At present, at least 14 studies have examined some aspect of the stability of self-control
from the perspective of SCT. In general, these studies have found that self-control is only
moderately stable, with findings of approximate rank-order stability for a majority of the
sample as well as considerable change for a notable minority (e.g., Burt, Simons, and Si-
mons, 2006; Hay and Forrest, 2006; Ray et al., 2013; Turner and Piquero, 2002; Winfree
et al., 2006). Although these studies have greatly advanced our knowledge of self-control
stability, they all have certain shortcomings. In particular, due to data limitations, exist-
ing studies have examined the stability of self-control only through the high-school years
(e.g., Hay and Forrest, 2006; Higgins et al., 2009; Ray et al., 2013). Although Gottfred-
son and Hirschi (1990) proposed that relative levels of self-control are fixed after early
childhood, it may be the case that self-control rankings stabilize after adolescence, a pos-
sibility consistent with some psychological views of personality (e.g., McCrae and Costa,
1990). The present study seeks to fill this gap in our knowledge by examining self-control
stability from the age SCT proposes relative stability (10) through emerging adulthood
An additional limitation of current tests of stability concerns the informant for self-
control measures. Two of the more rigorous tests of the stability of self-control are
based on other-reported (mother and teacher) measures of self-control (e.g., Hay and
Forrest, 2006; Na and Paternoster, 2012). Although each has strengths and weaknesses
(see Meldrum et al., 2013), other reports of self-control may paint a very different por-
trait of stability than self-reported measures, given the reality that adolescents and young
adults spend increasing time unsupervised and away from their homes and outside of
school. Mothers may be unaware of youths’ changes (or the magnitude of changes)
in self-control during the adolescent years, and therefore, tests of stability with other-
reported measures might be biased toward stability. Teachers generally report on self-
control based on behavior in one, albeit important, domain. The present study adds to
this knowledge base by using self-reports of self-control.
Finally, studies testing the stability postulate have examined the stability of global mea-
sures of self-control. There is, however, an ongoing debate about the unidimensional-
ity of measures of self-control within criminology (e.g., Burt and Simons, 2013; DeLisi,
Hochstetler, and Murphy, 2003; Marcus, 2004). Although Gottfredson and Hirschi (1990)
argued that their central concept is unidimensional, the most common measures of self-
control—which build explicitly on Gottfredson and Hirschi’s (1990) description of the
“elements of self-control”—appear to be multidimensional (e.g., Arneklev et al., 1993;
DeLisi, Hochstetler, and Murphy, 2003; Longshore, Turner, and Stein, 1996; Wood,
Pfefferbaum, and Arneklev, 1993; but see Arneklev, Grasmick, and Bursik, 1999; Piquero
and Rosay, 1998). Furthermore, evidence from a voluminous literature on personality
suggests that the elements of self-control that form the basis of the most common mea-
sures, such as the Grasmick scale (Grasmick et al., 1993), are associated with different
(and independent) facets of personality (e.g., Marcus, 2004; Morizot and Le Blanc, 2005).
Adding to this debate, recent advances in developmental psychology and social neuro-
science have underscored the importance of distinguishing between two traits—sensation
seeking and impulsivity—that are conflated in most existing measures of self-control (e.g.,
Grasmick et al., 1993). Although both sensation seeking and impulsivity influence risk be-
haviors, such as crime, a wealth of evidence from personality research suggests these are
distinct traits with different manifestations and developmental patterns (e.g., Duckworth
and Kern, 2011; Gullo et al., 2011). Moreover, recent work suggests that these two traits
have different neurobiological underpinnings as well as different normative maturational
timetables (e.g., Casey, Getz, and Galvan, 2008; Paus, 2005; see review in Steinberg, 2008).
Given this evidence that common measures of self-control are multidimensional, it is pos-
sible that the different constituent traits develop differently and that these differences are
obscured in global measures of self-control.
Thus, in the present study, after finding that the global measure of self-control we
employ is indeed multidimensional, cohering into impulsivity and sensation-seeking con-
structs, we examine the possibility that these two traits develop differently. Using group-
based trajectory models (GBTMs) as a statistical summarizing device, we first investigate
the developmental patterns of sensation seeking and impulsivity separately. In addition,
we examine co-occurring trajectories for the two traits. In these multitrajectory models,
each group is defined by its trajectory on both dimensions, which allows us to examine
common co-occurring developmental patterns. Our focus in these models is whether the
co-occurring patterns are different for the two traits within groups, which would provide
further evidence that examining the stability of global self-control conceals distinct trajec-
tories for impulsivity and sensation seeking. We supplement the GBTMs with additional
stability analyses, namely hierarchical linear models (HLMs) and reliable stable variance
estimates, to provide additional evidence on stability under different model assumptions.
Finally, the present study examines whether the observed changes in self-control are
consequential for offending. Surprisingly few studies have examined whether this is the
case. To our knowledge, only two studies (Burt, Simons, and Simons, 2006; Hay et al.,
2010) have examined whether changes in self-control or its elements are associated with
changes in offending over time. We seek to add to this work by investigating whether
within-individual changes in global self-control as well the disaggregated impulsivity and
sensation-seeking components are associated with changes in offending net of overall age
trends using fixed-effects models.
In sum, grounded in the idea that the trait (or traits) captured in common attitudi-
nal measures of self-control is important for crime, the present study seeks to advance
knowledge on the stability and unidimensionality of this construct and the criminogenic
consequences of observed instability. We explore these issues using data from the Family
and Community Health Study, a multisite panel study of African American families. In
the following pages, we briefly review SCT, paying particular attention to criminological
research on stability and multidimensionality. Next, we discuss theory and evidence from
personality and developmental psychology, which have quite different views of the cohe-
sion between “traits composing low self-control” (Gottfredson and Hirschi, 1990: 97) and
their stability. This is followed by a discussion of a new model of risk taking, the dual sys-
tems model, which is undergirded by recent advances in neurobiology. Finally, we discuss
the criminological significance of changes in self-control and its composite traits.
Boldly proclaiming an explanation of all crime at all times, Gottfredson and Hirschi
(1990) presented their self-control theory, also known as the general theory of crime.1
SCT is “a theory built on the idea that the decision to commit crime is governed by its
short-term, immediate benefits, without consideration of long-term costs” (p. 33). SCT
adopts the classical view of human nature and action, assuming that humans are hedo-
nistic and rational, and that given the immediate gratification it provides, motivation for
crime is ubiquitous. According to the theory, low self-control, defined as “the tendency
of individuals to pursue short-term gratification without consideration of the long-term
consequences of their acts” (p. 177), is the primary individual characteristic accounting
for variation in offending.2
According to SCT, self-control is learned through early childhood socialization, espe-
cially from the family. Specifically, SCT proposes that effective parenting inculcates self-
control in children. After the formative years, parenting as well as other social factors
should have no influence on self-control. According to SCT, “all that is required to re-
duce the crime problem to manageable proportions is to teach people early in life that
1. Unless otherwise noted, citations in this section refer to Gottfredson and Hirschi’s (1990) mono-
2. It is worth noting that SCT’s conceptualization of self-control as consideration of future conse-
quences is distinct both from its use in common parlance as well as in much scholarly work. For
example, in psychology, self-control often refers to either the capacity or the exercise of control
over the self both to stop oneself from engaging in tempting acts that go against a goal or to force
oneself to engage in acts that work toward a valued goal (e.g., Baumeister, Vohs, and Tice, 2007).
they will be better off in the long run if they pay attention to the eventual consequences
of their current behavior” (Hirschi and Gottfredson, 2003: 289).
The basic thrust of the theory, then, is that the absence of self-control leads to crime.
Empirical assessments of this central proposition are plentiful, and research has con-
sistently shown an inverse relationship between measures of self-control and crimi-
nal/deviant behavior (e.g., Burt, 2014; Pratt and Cullen, 2000; Schulz, 2006). Consistent
evidence of a robust relationship between self-control and offending has stimulated re-
search into other facets of the theory, including the theory’s strict stability proposition.
One of the most contentious aspects of SCT, and one that distinguishes it most sharply
from other social theories of crime, is its stability proposition. In effect, this postulate as-
serts that self-control rankings between individuals remain stable after 8 to 10 years of
age. After this time, although a cohort’s overall level of self-control likely increases as so-
cialization continues throughout life (absolute instability through normative maturation),
between-individual levels of self-control are fixed.
Although research assessing SCT’s stability proposition was slow to emerge, the num-
ber, breadth, and methodological complexity of studies has grown in recent years. Early
studies invariably used two time periods and rank-order stability correlations (i.e., Pear-
son correlations). In general, these studies showed that self-control, measured in a variety
of ways, was moderately stable, with correlations as high as r=.8 (e.g., Beaver et al.,
2008) but usually in the range of r=.4 to .5 (e.g., Arneklev et al., 1993; Burt, Simons, and
Simons, 2006; Turner and Piquero, 2002). Notably, SCT predicts stability correlations ap-
proaching (but not achieving due to measurement error) unity. Thus, rank-order stability
correlations do not support SCT’s strict stability postulate, as they suggest only moder-
ate stability of self-control across time. Other simple but informative stability tests, such
as quartile and percentile comparisons, confirm the findings of moderate stability. More-
over, these comparisons indicate that although most individuals evidenced approximate
between-individual stability, there were considerable changes for a portion of the sam-
ples, with individuals moving from among the lowest to the highest levels of self-control,
and vice versa (e.g., Burt, Simons, and Simons, 2006; Hay et al., 2010).
More recently, building on the findings from these earlier works, scholars have under-
taken more sophisticated analyses of self-control stability. A number of these studies have
utilized GBTMs. Notably, GBTMs are conservative tests of SCT’s stability proposition,
as they ignore all rank-order instability within groups, which in some cases include more
than 1,500 individuals (e.g., Hay and Forrest, 2006).
In the first GBTM study of self-control, Hay and Forrest (2006) estimated GBTMs of
mother reports of self-control from 7 to 15 years of age with 2-year intervals. Their results
showed that roughly 84 percent of the sample fit into one of four trajectories displaying
approximate absolute stability over this time period. However, approximately 5 percent
of the sample belonged to one of two unstable trajectory groups, which changed in both
absolute level and ranking (crossed other groups). Similar findings were reported by Hig-
gins et al. (2009) using yearly intervals when the youth were 12–16 years of age, and Ray
et al. (2013) using yearly assessments from grades 7 through 10. Overall, findings from
GBTMs suggest that most sampled individuals belong to groups displaying approximate
stability—at least between groups—with a sizable minority fitting in trajectory groups that
continue to increase or decrease after 10 years of age, thereby producing some marked
shifts in rank-order stability.
In another recent study testing the stability of self-control over multiple time periods
post-childhood, Na and Paternoster (2012) complemented GBTMs of stability with alter-
native models and added to the stability tests a consideration of measurement error and
invariance. One critique of tests of stability up to this point had been that rather than be-
ing “real” or substantively important, observed changes in measures of self-control could
be due to measurement error. Na and Paternoster (2012) analyzed data from youth as-
sessed at yearly intervals from grades 6 to 12 using behavioral measures of self-control
gleaned from teachers. The results from their growth models were inconsistent with SCT’s
stability proposition, as they revealed substantial instability in the growth pattern of self-
control. Importantly, Na and Paternoster’s (2012: 454) analyses also showed that the
changes in self-control over time are “not an artifact of statistical methods, measurement
error, or model specification.”
In sum, research explicitly testing SCT has not supported the strict stability postulate.3
In general, this work indicates that self-control rankings are roughly stable for many, but
marked changes occur for a notable minority. Although existing studies have greatly con-
tributed to our understanding of self-control stability, there are at least two ways research
can go beyond these studies. First, studies have not examined self-control stability past
the tenth grade (or roughly 16 years of age); thus, it is unknown whether self-control be-
comes more stable or remains malleable into emerging adulthood. We contribute to this
gap in knowledge by examining the stability of self-control from late childhood to early
adulthood (10–25 years of age). In addition, research needs to examine the development
of various dimensions of self-control, an issue to which we now turn.
Although Gottfredson and Hirschi (1990) implied that self-control is unidimensional,
they described “the nature of self-control” in ways that connote multidimensionality. In
particular, in their section on the “elements of self-control,” they stated that individu-
als with low self-control can be characterized as “impulsive, insensitive, physical (as op-
posed to mental), risk-taking, short-sighted, and nonverbal” (p. 90). For better or worse,
most common measures of self-control are based on this description of the elements of
self-control, including the most popular Grasmick scale (Grasmick et al., 1993).4,5 As a
3. Research on Baumeister and colleagues’ (Baumeister, Heatherton, and Tice, 1994; Baumeister,
Vohs, and Tice, 2007) strength model of self-control provides clear evidence that self-control is not
stable in the short or long term. However, this body of work is not discussed in this article because
its conceptualization of self-control is distinct from that of SCT. As such, the two theories and their
research are focusing on different constructs (see footnote 2).
4. More recently, Hirschi (2004: 542) argued against the “elementary approach” toward measuring
self-control and SCT’s foray into personality psychology, ultimately rejecting this approach and the
measures stemming from it. He then proposed a new conceptualization and operationalization of
self-control. In our view, this new measure is inconsistent with SCT. (For a brief discussion of this
point, see the online supporting information.) As such, we do not discuss Hirschi’s (2004) revised
conception any further.
5. Additional supporting information can be found in the listing for this article in the Wiley Online
Library at
consequence, the construct (or constructs) captured in the Grasmick scale is one of the
strongest correlates of crime.
Tests of the unidimensionality of the Grasmick scale and similar ones are mixed. Al-
though some studies found that such scales are unidimensional (Arneklev, Grasmick, and
Bursik, 1999; Gibson and Wright, 2001; Piquero and Rosay, 1998; Vazsonyi et al., 2001),
other studies examining the factor structure of these measures have identified multiple
factors (e.g., Arneklev et al., 1993; Burt and Simons, 2013; DeLisi, Hochstetler, and Mur-
phy, 2003; Longshore, Turner, and Stein, 1996; Vazsonyi et al., 2001). For example, sev-
eral factor analytic studies have shown that the six dimensions of self-control operational-
ized in the Grasmick scale neither coalesce onto a latent global trait, nor do constituent
items load solely (or at all) on the subscale to which they are identified (e.g., DeLisi,
Hochstetler, and Murphy, 2003; Longshore, Stein, and Turner, 1998). In general, these re-
sults indicate that the elements of self-control do not combine in a linear fashion to form
a latent, unidimensional construct (e.g., Hirschi, 2004; Piquero, MacIntosh, and Hickman,
Additionally, research has shown that the various elements of self-control differen-
tially predict crime and different types of crime (e.g., Arneklev et al., 1993; Greenberg,
Tamarelli, and Kelley, 2002; Pfefferbaum and Wood, 1994). Arneklev et al. (1993), for
example, found that risk seeking and impulsivity were consistently moderately associ-
ated with imprudent behaviors (smoking, drinking, and gambling); two elements, on the
other hand, were not even significantly correlated with imprudent behaviors. Moreover,
Arneklev et al. (1993) found that the risk-seeking component was more strongly related to
imprudent behavior than was the global self-control scale. Similar findings were reported
by Wood, Pfefferbaum, and Arneklev (1993: 124), who concluded that “the composite
self-control measure deserves to be treated in a multidimensional fashion” and suggested
the disaggregation of self-control into its constituent elements. In sum, criminological re-
search on the dimensionality of attitudinal measures of self-control suggests the wisdom
of disaggregating at least some elements or traits combined in self-control measures.
Personality refers to relatively “characteristic patterns of thinking, feeling, and behav-
ing” (Miller and Lynam, 2001: 765). As Van Gelder and de Vries (2012: 639) noted, “the
self-control concept essentially implies a personality trait as it refers to stable individual
differences in the propensity to act, think, and feel in certain ways.” More specifically,
viewed apart from the logic of the theory, Gottfredson and Hirschi’s (1990) description
of the nature of self-control could be interpreted as creating a criminal personality typol-
ogy, which they happened to label self-control, but other terms are equally if not more
appropriate, such as conscientiousness (which they considered) or antisocial personal-
ity. In arriving at the depiction of this self-control personality typology, Gottfredson and
Hirschi (1990) identified the traits of self-control by “inferring from the nature of crime
what people who refrain from [and commit] criminal behavior are like” (p. 109). As such,
personality research is apposite for SCT.
Although the exact structure of personality remains debated, the most well-validated
structural model of personality is McCrae and Costa’s (1990) Five Factor Model (FFM),
which consists of five broad, higher order factors called domains, including agreeable-
ness, conscientiousness, neuroticism, extraversion, and openness to experience, each of
which is composed of six subfactors called facets. These domains are conceptualized
as largely independent, if not perfectly orthogonal. Importantly, included in the list of
six elements of self-control are those that belong to each of the five different domains
(Marcus, 2004). For example, risk seeking would fall under the excitement-seeking facet
of the extraversion domain; insensitivity would fall under the agreeableness domain, and
impulsivity is a facet of the neuroticism domain. Moreover, this pattern—that the var-
ious elements of self-control correspond to traits associated with independent dimen-
sions of personality—is replicated when other validated models of personality are used,
such as Tellegen’s (1985) three-factor model and Ashton et al.’s (2009) HEXACO model
(Marcus, 2004; Van Gelder and de Vries, 2012). Thus, models of personality contradict
Gottfredson and Hirschi’s (1990) assertions that the elements of self-control come to-
gether in the same people to compose a unidimensional personality trait (Hirschi, 2004;
Marcus, 2004).
Developmental research on personality also provides a counterpart to SCT’s stabil-
ity proposition. Over the past two decades, longitudinal studies, which have assessed
whether and how individuals change in personality, evinced non-negligible personality
change across the life course (e.g., Caspi and Roberts, 2001; Helson, Jones, and Kwan,
2002; Morizot and LeBlanc, 2005). With some exceptions (such as sensation seeking),
the common pattern of change identified has been referred to as “normative psycholog-
ical maturation,” such that, on average, older individuals show better conscientiousness
and agreeableness and less neuroticism, including better impulse control, extraversion,
and openness, than younger individuals (e.g., Helson, Jones, and Kwan, 2002; McCrae
et al., 1999). Importantly, in addition to this normative pattern, studies have shown that
there is considerable interindividual variability in intraindividual change (e.g., Morizot
and LeBlanc, 2005; Roberts and Mroczek, 2008; Vaidya et al., 2002). In sum, personality
research provides a counterpart to SCT’s arguments that the elements (traits) of self-
control tend to come together in the same people and comprise a stable construct.
More pointed evidence contradicting the unidimensionality and stability arguments of
SCT can be found in recent psychological research on the personality correlates of risk
taking. A wealth of research identifies impulsivity and sensation seeking as important per-
sonality traits strongly related to risky, reckless behavior (e.g., Miller and Lynam, 2001;
Romer, 2010; Romero, Luengo, and Sobral, 2001; Steinberg, 2008). Importantly, how-
ever, research demonstrates that impulsivity and sensation seeking are conceptually and
empirically distinct (e.g., Quinn and Harden, 2012; Steinberg et al., 2008). Impulsivity,
which like self-control has been defined in various ways, is generally described as “a ten-
dency to act rapidly without deliberation or consideration” (Pickering and Gray, 1999:
278) or “lacking cognitive control over behavior” (Romer, 2010: 266), which begets risk
taking via hasty decision making. Sensation seeking, on the other hand, is a motivational
tendency defined as a preference for intense and novel stimuli given the tendency to ex-
perience such sensations as pleasurable (Forbes and Dahl, 2010; Roth, Hammelstein, and
Brahler, 2007; Zuckerman, 1994). To be sure, although these traits may come together
in some individuals (e.g., impulsive sensation seekers), they are independent, such that
individuals can be high on sensation seeking and low on impulsivity and vice versa. More-
over, these traits can have different manifestations and consequences. Behaviors driven
by sensation seeking can involve considerable advance planning (e.g., a trip to sky dive),
and impulsive acts may be unrelated to sensation seeking (rashly deciding to quit a job)
(Steinberg et al., 2008). Focusing on criminal behavior, impulsivity can lead to crime
because individuals make rash decisions to engage in crime, whereas sensation seeking
can lead to crime because individuals are motivated by the sensations or potential re-
wards (Burt and Simons, 2013).
Dual Systems Model of Risk Taking
Both theoretical and measurement overlap have hampered research into the indepen-
dent contributions of impulsivity and sensation seeking to involvement in risky and ille-
gal behaviors (e.g., Burt and Simons, 2013; Steinberg et al., 2008). Prominent measures
of impulsivity generally include items assessing sensation seeking and vice versa (White-
side and Lynam, 2001), and the most common measures of self-control combine the two
constructs (e.g., Grasmick et al., 1993). However, recent dramatic advances in develop-
mental neuroscience have stimulated research in this domain (see Steinberg, 2008). This
work, which identifies different neurobiological underpinnings for sensation seeking and
impulsivity as well as different normative maturational timetables, has contributed to the
emergence of theories that link brain maturation in adolescence to increased risk taking
(Casey, Getz, and Galvan, 2008; Dahl, 2004; Nelson et al., 2005; Steinberg, 2008).
The most prominent of these theories is known as the dual systems model of risk taking
(Steinberg, 2008; Steinberg et al., 2008). According to this model, risk taking is a func-
tion of two distinct yet interacting neurobiological systems, the cognitive control and
socioemotional systems, which develop according to different normative timetables,
achieving structural and functional maturity at different ages (Somerville and Casey, 2010;
Steinberg et al., 2008). The cognitive control system, which includes the prefrontal cortex
and is crucial for impulse control, decision making, and emotional regulation (Miller and
Cohen, 2001), develops linearly, gradually maturing through adolescence and beyond,
thereby permitting more advanced emotion regulation and decreased impulsivity (Casey,
Galvan, and Hare, 2005; Nelson et al., 2005). The subcortical socioemotional system, or
incentive processing system, which governs sensation seeking, is driven by frontostriatal
reward circuits and is responsive to novelty, emotion, and reward (e.g., Cardinal et al.,
2002; Galvan et al., 2006; Harden and Tucker-Drob, 2011). This system undergoes remod-
eling during adolescence (around puberty), resulting in a rapid increase in dopaminergic
activity, which is presumed to lead to a heightened sensitivity to rewards and, thus, in-
creased sensation seeking (Chambers, Taylor, and Potenza, 2003; Steinberg et al., 2008).
The central premise of the dual systems model is that the adolescent increase in risky be-
haviors such as crime results from the temporal gap between the maturation of the two
systems. This gap produces a normative state of heightened sensitivity to rewards due to
the climax of sensation seeking around 15–17 years of age, combined with an insufficiently
matured cognitive control system to assess risks accurately and govern behavior (Casey,
Jones, and Sommerville, 2011; Steinberg, 2008).
Early evidence supporting the dual systems model primarily came from neurobiologi-
cal research (see Steinberg et al., 2008; Paus, 2005 for reviews); however, behavioral re-
search supporting the model has begun to accumulate. Both cross-sectional (Steinberg
et al., 2008, 2009; Vaidya et al., 2010) and longitudinal studies (Harden and Tucker-Drob,
2011; Quinn and Harden, 2012) have indicated that the normative pattern of impulsiv-
ity is one of a linear decline from childhood into the 30s. In contrast, research indicates
that sensation seeking rises to a peak in mid-adolescence (15–17 years of age), before de-
clining into adulthood (see also Cauffman et al., 2010; Harden and Tucker-Drob, 2011;
Romer and Hennessy, 2007). Notably, this age–sensation-seeking curve strongly resem-
bles the ubiquitous age–crime curve (e.g., Farrington, Loeber, and Jolliffe, 2008; Hirschi
and Gottfredson, 1983). Additionally, the few studies that have examined interindividual
differences in patterns of intraindividual change have revealed significant individual dif-
ferences around the population trends in impulsivity and sensation seeking (e.g., Harden
and Tucker-Drob, 2011; Lynne-Landsman et al., 2011). For our purposes, more impor-
tant than the specifics of this nascent line of work is the evidence that impulsivity and
sensation seeking have different neurobiological underpinnings as well as distinct nor-
mative developmental patterns and that considerable individual variation exists around
these normative trajectories.
In sum, several relatively recent advances in criminology, personality and developmen-
tal psychology, and neuroscience inform our study, providing a counterpart to SCT’s
stability proposition. First, evidence suggests that the common measures of self-control
are multidimensional, cohering into several distinct personality traits. Second, psycho-
logical research highlights the importance of both sensation seeking and impulsivity, and
it demonstrates that these are separate, albeit related, personality traits that contribute
independently to risk taking and antisocial behavior. Moreover, developmental studies
suggest that impulsivity and sensation seeking develop along different timetables, with
sensation seeking peaking around 16 years of age and declining thereafter, and impul-
sivity declining linearly after late childhood. Importantly, both of these normative devel-
opmental trajectories are different from the SCT model. Finally, as with other forms of
personality (e.g., Helson, Jones, and Kwan, 2002; Roberts and DelVecchio, 2000), against
the backdrop of normative maturational change, there is evidence of substantial between-
individual differences in changes in impulsivity and sensation seeking, with studies indi-
cating substantial variability in individual trajectories of these two constructs from late
childhood to early adolescence (Harden and Tucker-Drob, 2011; Quinn and Harden,
Given the likely multidimensionality of self-control implied by both criminological and
personality research, we begin by documenting that our global measure of self-control
is multidimensional. After verifying that it is, cohering into separate impulsivity and
sensation-seeking constructs, we examine the possibility that these two traits have differ-
ent patterns of development. In addition, we estimate multitrajectory models to summa-
rize the different developmental types—or latent trajectory classes—of these two traits.
Here, trajectories for sensation seeking and impulsivity are estimated simultaneously and
latent developmental classes are identified based on common patterns of development
(Nagin and Odgers, 2010). This approach allows us to identify clusters of individuals who
share similar patterns of stability and change among the different dimensions, patterns
that are collapsed when using global measures of self-control. In addition, we describe sta-
bility and change using stability correlations corrected for attenuation and HLMs. These
different models are intended to provide a fuller portrait of longitudinal patterns of sta-
bility and change under different model assumptions.
Inherent in the stability proposition is the idea that because self-control is equivalent
to criminality, the relative stability of self-control implies the relative stability of individ-
ual propensity to crime and deviance across the life course (Gottfredson and Hirschi,
1990). Although the age–crime curve reveals decreased offending after the peak in
adolescence, Gottfredson and Hirschi (1990; Hirschi and Gottfredson, 1983) argued that
this ubiquitous age effect is due to the inexorable aging of the organism and is not ex-
plicable by social factors or levels of self-control. However, insofar as self-control is a
major cause of crime and is not fixed after childhood, we would expect that changes in
self-control produce changes in offending. Findings from the two studies that have exam-
ined this possibility are consistent with this hypothesis. Burt, Simons, and Simons (2006)
and Hay et al. (2010) found that improvements in self-control were significantly associ-
ated with decreases in offending, net of earlier levels of self-control. We extend this work
by estimating fixed-effects models to examine the criminological significance of changes
in self-control, impulsivity, and sensation seeking above and beyond age trends. In doing
so, we investigate whether the effects of changes in impulsivity and sensation seeking are
equally associated with changes in crime. As noted, prior research provides evidence that
these different facets of self-control are not equally predictive of offending (e.g., Arneklev
et al., 1993; Burt and Simons, 2013; Wood, Pfefferbaum, and Arneklev, 1993).
The present study contributes to research on SCT’s stability proposition in several
ways. First, we fill a gap in our understanding of the stability of self-control after the
high-school years by following a sample of youth from late childhood (10–12 years of
age) to the mid-20s with five assessments of self-control. Using self-reported measures,
we examine the stability and developmental patterns of self-control from late childhood
to emerging adulthood using several different models. In addition, we assess develop-
mental patterns of impulsivity and sensation seeking with a focus on whether divergent
patterns of development exist that are masked in stability tests of self-control. Finally, we
assay the criminological significance of these changes.
We explore these ideas using data from the Family and Community Health Study
(FACHS), an ongoing investigation of health and psychosocial development among a
sample of African American families. Notably, although the youth have been interviewed
at six different occasions, the wave 3 interviews did not include measures of self-control.
Thus, we use data from waves 1, 2, 4, 5, and 6. Importantly, the methods we use (GBTM
and HLM) can accommodate the missing wave. The missing wave 3, of course, means that
the lag between waves 2 and 4 is twice as long as the others, which will have the effect of
smoothing the data over this time period, thereby upwardly biasing stability and favor-
ing SCT’s stability postulate. The timing of this gap makes our study ill-suited for testing
crucial predictions of the dual systems model. As such, this study is designed to test the
stability and dimensionality arguments of SCT drawing on insights from the dual systems
model, rather than as a rigorous test of the predictions of the dual systems model.
The FACHS is an ongoing study of development among a panel of African Ameri-
can families, half from Iowa and half from Georgia (see Gibbons et al., 2004; Simons
et al., 2002). The FACHS was designed to analyze the particular risks and resources that
disrupt or promote African American family functioning and youth development in var-
ious contexts. The study families resided in neighborhoods that varied considerably on
demographic characteristics, such as racial composition and economic level. A total of
889 African American families participated in the first wave of the FACHS. Each family
included a fifth-grade target youth at wave 1. Fifty-four percent of the youth were female.
The mean family income at wave 1 was approximately $29,500. In general, the sample
was representative of the African American populations of the communities from which
participants were recruited (Cutrona et al., 2000).
Of the 889 target youth who participated in the first wave of the study, 779 (88 per-
cent), 767 (86 percent), 714 (80 percent), 689 (78 percent), and 699 (79 percent) were
re-interviewed in waves 2 through 6, respectively. Data collection was completed for the
waves in 1998, 2001, 2004, 2007, 2009, and 2011. The mean ages of the target youths were
10.5 (range: 10–12), 12.6 (12–14), 15.7 (15–17), 18.8 (18–20), 21.6 (20–23), and 23.5 (22–25)
at waves 1 through 6, respectively. These data allow for the examination of the stability
of self-control after the time at which Gottfredson and Hirschi (1990) argued that self-
control should be fixed until young adulthood. Notably, because our analyses focus on
changes in self-control and crime, we require individuals to have participated in at least
three waves with valid self-control data to be included in our sample. This reduces our
sample size to 775 youths and 3,580 person-waves.
The measurement of self-control has been an ongoing source of debate (see Marcus,
2004; Piquero, 2008).6Much of this debate has centered on the use of cognitive/attitudinal
versus behavioral measures, with Gottfredson and Hirschi (1990) preferring the latter and
others who see tautology as an issue preferring the former (e.g., Akers, 1991; Meier, 1995).
Both have been used frequently in past research, and fortunately, the research has shown
that they are significantly correlated and similarly predict crime and deviance (e.g., Evans
et al., 1997; Pratt and Cullen, 2000; Tittle, Ward, and Grasmick, 2003). This study uses a
self-reported measure of self-control consisting of both attitudinal/cognitive items (e.g.,
“You usually think before you act”) and behavioral items (“When you have to wait in
line, you do it patiently”).
Fourteen items consistent with Gottfredson and Hirschi’s (1990) nominal definition of
self-control were identified in the data. The items were gleaned from two instruments.
Four items were taken from Eysenck and Eysenck’s (1978) personality inventory, and ten
items were culled from Kendall and Wilcox’s (1979) measure of self-control. For all items,
response categories range from 1 (“Not at all true”) to 3 (“Very true”). To create the
global low self-control measure, four items from Kendall and Wilcox’s (1979) inventory
of self-control were reverse coded so that high scores indicate lower self-control; then all
6. In the present study, we leave aside the question as to whether existing measures of self-control
capture the variable at the heart of SCT (for an informative discussion see Marcus, 2004). Instead,
we build on the research showing that the concept (or concepts) captured in existing measures
of self-control (such as the Grasmick scale) has proven to be a strong predictor of crime and,
therefore, assessing whether this construct and its components are stable is an important task.
items were averaged (mean α=.73) and standardized across person-waves. This measure
has been used in several prior studies (e.g., Burt, Simons, and Simons, 2006). A list of the
self-control items is presented in appendix A in the online supporting information.
Theory, past research, and factor analyses directed the formation of the self-control
subscales. Results from the factor analyses using waves 1 and 6 as examples are presented
in the online supporting information. Briefly, across all of the waves, the items from the
Eysenck and Eysenck (1978) inventory cohered into a separate factor assessing sensation
seeking. The Kendall and Wilcox (1979) items loaded on a single factor across all waves,
with item loadings greater than or equal to .45 at wave 6; in the other waves, all item
loadings were greater than .40, excepting one item. This item, however, varied across the
waves and was in the .37 to .39 range. Consistent with the two-dimensional conceptual-
ization in personality theory as well some past work testing SCT (e.g., Burt and Simons,
2013; Winfree et al., 2006), we separated the global scale into two dimensions, which we
label impulsivity and sensation seeking.7The 10-item measure of impulsivity captures the
extent to which respondents seek immediate gratification, act on impulse, and fail to en-
gage in forethought or persist in tasks with delayed rewards (mean α=.70). Sensation
seeking is measured with the four items from Eysenck and Eysenck’s (1978) personal-
ity inventory that assess whether respondents enjoy taking risks and are bored by a life
without danger. High scores indicate higher sensation seeking (mean, α=.65). Both sub-
scales are standardized across person-waves. Summary statistics and stability correlations
for self-control, impulsivity, and sensation seeking are displayed in table 1.
The measure of offending was generated using youth self-reports at each wave and con-
sists of the number of different illegal acts (of 11) respondents committed in the past year,
such as binge drinking, shoplifting, aggravated assault, marijuana use, vandalism, theft,
and assault with a weapon. Previous research has shown that offending variety scales are
preferable to frequency and dichotomous scales, which largely reflect variation in the least
serious offenses (Sweeten, 2012). The items were culled from the Diagnostic Interview
Schedule for Children, Version 4 (DISC-IV; American Psychiatric Association, 1994), in
waves 1 through 4 and from the measure created by Elliott, Huizinga, and Menard (1986)
in waves 5 and 6. Although the wording and exact content of these scales varies across
these two forms, most of the items were comparable. Prior to the creation of the scales,
the items that were not analogous or available across the two instruments were identified
and dropped from the scales (see appendix B in the online supporting information for the
list of items).
7. Like self-control, the label impulsivity has been applied to constructs that are quite diverse (e.g.,
Whiteside and Lynam, 2001). In general, there are broad and narrow uses of impulsivity, with
the latter generally being components of the former (Lynam and Miller, 2004). The Grasmick
et al. (1993) operationalization of impulsivity is a narrow version of the construct capturing aspects
of premeditation. Consonant with recent work on the dual systems model (e.g., Steinberg et al.,
2008), we use impulsivity here in its broader sense to include aspects of premeditation as well as
urgency and perseverance, all of which have been identified as elements of broad impulsivity (e.g.,
Whiteside and Lynam, 2001) and are captured in the most popular measure of impulsivity, the
Barratt Impulsiveness Scale (BIS-11; Patton, Stanford, and Barratt, 1995).
Table 1. Stability Correlations for Self-Control, Impulsivity, Sensation
Seeking, and Crime
Variable 12456MeanSD
1. Low self-controlW1 .29 .21 1.02
2. Low self-controlW2 .44 .34 .23 1.03
3. Low self-controlW4 .27 .40 .29 .03 .93
4. Low self-controlW5 .22 .36 .55 .32 .07 .95
5. Low self-controlW6 .23 .39 .54 .63 .31 .36 .95
6. ImpulsivityW1 .27 .24 1.05
7. ImpulsivityW2 .45 .27 .31 1.01
8. ImpulsivityW4 .28 .37 .17 .03 .90
9. ImpulsivityW5 .18 .30 .50 .23 .22 .90
10. ImpulsivityW6 .21 .31 .51 .61 .23 .34 .97
11. Sensation seekingW1 .19 .05 .95
12. Sensation seekingW2 .31 .32 .04 1.00
13. Sensation seekingW4 .17 .35 .33 .02 1.08
14. Sensation seekingW5 .18 .35 .54 .34 .10 .98
15. Sensation seekingW6 .18 .34 .47 .54 .32 .09 .98
16. CrimeW1 .75 1.36
17. CrimeW2 .34 — 1.12 1.60
18. CrimeW4 .17 .26 1.86 2.23
19. CrimeW5 .11 .17 .17 .70 1.50
20. CrimeW6 .08 .13 .20 .34 .45 1.23
NOTES: Italicized correlations on the diagonal are the correlation between the row variable and the wave-
matched crime variable. For all correlations, p<.01 (two-tailed); nranges from 775 to 631 using listwise deletion
for each correlation.
ABBREVIATION:SD=standard deviation.
The resulting variety scale gives one point for every type of delinquency committed by
the respondents in the past year. The mean Kuder–Richardson coefficient of reliability
(KR20; Kuder and Richardson, 1937), designed to assess the reliability of dichotomously
scored scales, was .84 across the six waves. Notably, as shown in table 1, the stability
correlations for crime over time are similar, supporting the commonality of the measure
across waves.
In the trajectory models, we incorporate age (measured in years) as the time variable
as well as squared or cubed age terms where they improve model fit. Age (measured in
months) is also included as a control in the fixed-effects models along with a measure of
age squared to examine possible nonlinear age effects.
The analysis proceeds in a series of steps. In the first, we examine self-control with
GBTMs, a semiparametric method for identifying similar patterns of development (Jones,
Nagin, and Roeder, 2001). GBTMs can be thought of as a useful statistical device for sum-
marizing what is likely a continuous distribution of individual trajectories into groups
based on similar developmental patterns (Nagin and Odgers, 2010). Importantly, this
model does not assume or imply that the population is composed of distinct taxonomies.
Instead, the trajectory groups are summary descriptions that provide more information
than that for a single average pathway for a population but less information than the
unwieldy depiction of individual trajectories. “Like contour lines on a topographical map,
the groups are not literal entities, but rather approximations of distinctive regions on the
surface” (Nagin and Odgers, 2010: 117).
We use a multistage process to determine the best model (number of groups and shape
of trajectories). Specifically, we utilized Bayesian information criterion (BIC) scores, av-
erage posterior probabilities of group membership (avePP), and odds of correct classifi-
cation (OCC) as well as whether adding a group revealed substantively different longitu-
dinal patterns to determine the number of groups to present (Nagin, 2005). Details of the
model fitting process are presented in the online supporting information.
Using this method, we explore developmental trajectories of low self-control as well as
its two dimensions: impulsivity and sensation seeking. SCT would predict that a GBTM
should uncover groups whose trajectories are roughly parallel with little-to-no intersec-
tion. Intersecting groups, indicative of groups (and thus individuals) changing in their
relative rankings and levels of self-control, would be inconsistent with SCT’s relative sta-
bility hypothesis.
To assess whether there are subgroups in the data displaying distinct and divergent
co-occurring patterns of development for impulsivity and sensation seeking, we em-
ploy a multitrajectory group-based model (Jones and Nagin, 2007). Here, each group
is characterized by their trajectories on both traits, with a focus on differences in de-
velopmental trajectories for impulsivity and sensation seeking within groups. If this
analysis reveals groups with discordant levels and patterns of impulsivity and sensa-
tion seeking, this provides evidence for the developmental distinctiveness of the two
In addition, we supplement these GBTMs with two additional models of stability.
Specifically, we present the stability correlations corrected for attenuation (dividing the
square root by the product of the reliabilities), which are squared to provide an in-
dex of the proportion of reliable variance in the constructs that is stable versus chang-
ing. We also estimate HLMs, which estimate a single growth trajectory for the average
individual and permit one to examine the extent of individual variation around mean
levels and growth rates, assess the reliability of the estimates of the growth parame-
ters, and evaluate the correlation between status (level at a given age) and change.
Additionally, HLMs provide an estimate of rho, or the intraclass correlation, which
is the proportion of the variance in each construct that is attributable to between-
versus within-individual variation around the person-specific mean. This proportion
would be close to 1 if the rank-order stability assumption holds. These different mod-
els of stability are intended to complement the GBTMs under different model assump-
tions and provide additional information about the extent of stability and change over
Last, we examine the criminogenic consequences of changes in self-control using fixed-
effects models. First, we use the global measure of low self-control to assess whether
within-individual changes in self-control are associated with concurrent changes in crime.
Next, we estimate a second fixed-effects model assessing the independent effects of
changes in impulsivity and sensation seeking on changes in crime. This serves as a fur-
ther test of the discriminant validity of the separate constructs. Because our dependent
variable is a count variable with overdispersion, we employ an unconditional fixed-effects
negative binomial model (Allison and Waterman, 2002). These models use person-year
Figure 1. Group-Based Trajectory Model of Low Self-Control
(Six Groups)
Group 1 (28.8%) Group 2 (27.8%) Group 3 (16.0%) Group 4 (16.1%) Group 5 (8.0%) Group 6 (3.2%)
Low self-control
data incorporating dummy variables for all individuals (except one) after dropping those
individuals who do not engage in any crime across all six waves (n=67) as these cases
do not contribute to the likelihood function and their dummy variables do not converge
(Allison, 2009). We also test whether changes in sensation seeking are more strongly as-
sociated with changes in offending than changes in impulsivity by comparing coefficients
with a likelihood ratio test.
For the global low self-control measure, we selected a six-group model, which is dis-
played in figure 1.9For the groups at the extremes of the self-control distribution, there
appears to be little change. The highest self-control group, located at the bottom of figure
8. For interested readers, we display the one-group models for each of the three measures in appendix
C in the online supporting information. These are all estimated separately but combined into one
figure to facilitate comparisons and save space. Compared with the six-group models we present,
one can clearly see that significant heterogeneity in developmental patterns is collapsed in the one-
group models thereby supporting the use of GBTMs.
9. To guard against reification of groups, we refrain from attaching labels to them as has often been
done in past work. Furthermore, we must emphasize that these group-based models, much like
regression analysis, are post hoc analyses and do not imply ex ante identification of group mem-
Figure 2. Group-Based Trajectory Model of Impulsivity (Six Groups)
Group 1 (32.1%) Group 2 (32.7%) Group 3 (18.3%) Group 4 (10.5%) Group 5 (4.9%) Group 6 (1.6%)
1 and comprising 28.8 percent of the sample, and the lowest self-control group, mak-
ing up 3.2 percent of the sample, do not cross any other groups over this 13-year span.
However, for the remaining four groups, we observe a considerable degree of between-
group instability in self-control, which implies considerable rank-order instability, as a
result of different developmental patterns. Notably, approximately 68 percent of the
sample belongs to crossing trajectories of increasing (32.1 percent) or decreasing (35.8
percent) self-control across the study period. Not unexpectedly, given the longer ob-
servation period and use of self-reports, our GBTMs show more instability than prior
Next, we examined trajectories of impulsivity, which are displayed in figure 2. We again
selected a model with six groups. These trajectories are similar to that for self-control,
which is not unexpected given the overlap in these measures. As before, the groups at
the upper and lower extremes do not cross any other groups, whereas the middle groups
cross each other. Group 5, composed of 5 percent of the sample, becomes more impul-
sive throughout the survey period, intersecting with groups 3 (18 percent) and 4 (11 per-
cent), which both become significantly less impulsive. Group 3 transitions from being
near the group with the highest impulsivity at 11 years of age to nearly the lowest at
24 years of age. Overall, approximately 35 percent of the sample belongs to one of the
10. We assessed the overall level of between-group stability by calculating the chance that any two
random individuals cross (ignoring within-group instability). This allowed us to compare the level
of stability found in our models with that from prior GBTM studies. See the online supporting
information for more details and the specific results of these analytic comparisons.
Figure 3. Group-Based Trajectory Model of Sensation Seeking
(Six Groups)
Group 1 (47.2%) Group 2 (22.5%) Group 3 (17.0%) Group 4 (4.3%) Group 5 (4.0%) Group 6 (4.9%)
Sensation seeking
relatively stable groups at the high and low end, while most of the sample belongs to
groups of increasing (5 percent), decreasing (29 percent), or curvilinear increasing then
decreasing (33 percent) trajectories of impulsivity. Thus, we again find considerable rela-
tive and absolute instability.
The optimal sensation-seeking model, presented in figure 3, also consists of six groups
but is markedly different from the previous two. Group 1, comprising nearly half of the
sample, exhibits consistently low levels of sensation seeking and does not cross any other
groups. The other groups, however, exhibit noticeable changes in sensation seeking and
a considerable degree of between- and within-group instability. Unlike the prior models,
there is no stable high group that does not cross others. Instead, group 5 (4.0 percent)
peaks around 15 years of age and then dramatically declines in sensation seeking, whereas
group 6 (4.9 percent) ramps up slowly throughout the teens to peak in sensation seeking
around 20 years of age.
It is clear from the juxtaposition of longitudinal trajectories of sensation seeking and
impulsivity that these traits have divergent patterns of development. The trajectories of
sensation seeking are much more curvilinear and display larger changes than the typical
patterns for impulsivity. At the same time, by 17 years of age and continuing to the end
of the survey, 64 percent of the sample (groups 1 and 3) has very low and stable levels of
sensation seeking. For impulsivity, on the other hand, three groups making up 61 percent
of the sample exhibit steady declines throughout the study period.
Although separate GBTMs for impulsivity and sensation seeking indicate that these
traits develop differently from childhood to young adulthood, they do not show how
they develop together within individuals. To identify typical co-occurring patterns of
development for these two traits, we use multitrajectory models (Jones and Nagin,
2007). Here, we again present a six-group model, which is shown in figure 4. We shade
Figure 4. Multitrajectory Group-Based Model of Impulsivity and
Sensation Seeking (Six Groups)
11 14 17 20 23
Age Age
Group 1 (45.4%) -1
11 14 17 20 23
Group 2 (18.1%)
11 14 17 20 23
Group 3 (18.2%) -1
11 14 17 20 23
Group 4 (8.6%)
11 14 17 20 23
Group 5 (6.1%) -1
11 14 17 20 23
Group 6 (3.6%)
Impulsivity Sensation Seeking
insignificant differences (p>.05) between levels of sensation seeking and impulsivity
within the groups in figure 4.11
Two findings stand out from these results. First, there are marked differences between
groups in the co-occurring patterns of development. Second, and most important for our
11. Differences between the impulsivity and sensation-seeking trajectories at each age were calculated
with a Wald test of the equality of coefficients using the %trajtest macro described by Jones and
Nagin (2007).
purposes, there are also significant within-group differences in both the intercepts and
slopes in impulsivity and sensation seeking, providing clear evidence that divergent lev-
els and patterns of change for these two traits characterize common co-occurring pat-
terns over the study period. Indeed, within-group mean levels of impulsivity and sensa-
tion seeking differ for more than half of the sample at all ages and for the entire sample
at some ages. Additionally, the slopes of impulsivity and sensation seeking differ within
groups for 70 percent of the sample from 11 to 15 years of age and for 88 percent of the
sample thereafter.
Not surprisingly, the biggest group, composed of 45 percent of the sample, resembles
the biggest groups for the separate models. This group displays low stable levels of sen-
sation seeking and low decreasing levels of impulsivity throughout the age range. This
group, and to a lesser extent group 6, are most consistent with the arguments of SCT. Even
so, the slopes for sensation seeking and impulsivity are significantly different throughout
the age range for group 1, for example, and the absolute levels of sensation seeking and
impulsivity are significantly different from each other except for 16–18 years of age where
they cross. The remaining groups, which comprise approximately half of the sample, are
inconsistent with SCT, as there are significant differences in the levels and developmental
patterns for these two traits. Group 5, for example, is characterized by a steady decrease
in impulsivity but increases in sensation seeking that peaks around 18 years of age, fol-
lowed by a decline but still a high level at 24 years of age. Overall, this multitrajectory
model reveals that there are distinct, and sometimes markedly different, co-occurring tra-
jectory patterns for much of the sample. These different patterns are obscured when using
a global measure of self-control.
Reliable Stable Variance Estimates
Figure 5 displays estimates of the reliable stable variance in self-control, impulsivity,
and sensation seeking between various waves. These estimates can be interpreted as
the proportion of reliable stable variance in the constructs between the waves in ques-
tion (which are listed in parentheses by the point estimates). The most notable fea-
ture of these estimates, common to all three constructs, is that they rapidly decline as
a function of the distance between the observations. In general, after 5 years, the pro-
portion of reliable stability drops below .5, and after 10 years, as can be seen with the
point at the lower right of all three figures, it is typically below .10. This finding indi-
cates that only 10 percent of the reliable variance in self-control, impulsivity, and sensa-
tion seeking is stable from wave 1, when the youth were 10–12 years of age, to wave 6
taken a little more than a decade later. Clearly, these traits are not fixed at 10 years of
Second, the results displayed in figure 5 show that, for all three constructs, stability
tends to be higher in later years, indicating that these traits tend to become more stable
with age. For example, the reliable stability between waves 5 and 6, when respondents
are, on average, 22 and 24 years old, is considerably higher than the stability between
waves 1 and 2, also a span of 2 years, but much earlier at 11 and 13 years of age. Even so,
the proportion of reliable variation that is stable versus changing between waves 5 and 6
when stability is at its zenith is .68 for self-control and impulsivity, and .59 for sensation
seeking. Thus, after accounting for measurement error, the stability estimates in emerging
Figure 5. Estimates of the Proportion of Reliable Stable Variance by
Years Between Measurements
Proportion reliable variance
(4,5) (4,6)
(2,5) (2,6)
(1,5) (1,6)
(a) Self-control
Proportion reliable variance
(4,5) (4,6)
(2,5) (2,6)
(1,5) (1,6)
(b) Impulsivity
Proportion reliable variance
(5,6) (4,5)
(2,5) (2,6)
(1,5) (1,6)
(c) Sensation seeking
Table 2. Hierarchical Linear Models of Global Self-Control, Impulsivity,
and Sensation Seeking (n=775 Respondents and 3,580
Low Self-Control Impulsivity Sensation Seeking
Fixed Effects Coeff. SE Coeff. SE Coeff. SE
Intercept (mean initial level), β00 .079 .032.066 .031 .008 .025
Age (mean growth rate), β01 .209 .017.253 .017.004 .016
Age squared, β02 .081 .018.074 .018
Random Effects SD SE SD SE SD SE
Intercept, r0i .636 .021.607 .020.606 .021
Growth rate, r1i .306 .017.334 .017.255 .020
Level 1 error, eti .678 .011.675 .011.760 .012
Rho .520 .513 .428
Reliability of Estimates
Intercept, π0i .788 .773 .729
Growth rate, π1i .483 .528 .342
Correlations Between Intercept and Growth Rate
12 years of age .425.526.109
18 years of age .110 .041 .405
22 years of age .446.434.631
ABBREVIATIONS:SD=standard deviation; SE =standard error.
p<.05 (two-tailed tests).
adulthood are still far from unity. Although there are some minor differences in estimates
of reliable stability between impulsivity and sensation seeking, they are less notable than
the similarities, indicating that differences between the two are less about the magni-
tude of rank-order stability and more about the nature of the changes during the study
Hierarchical Linear Models
Next, we estimated HLM models examining the development of self-control, impul-
sivity, and sensation seeking over time. Table 2 displays results from random-coefficient
models that allow individuals to vary both in their initial levels (r0i) and in their rates
of change (r1i). (See the online supporting information for more details on the HLM
models.) SCT predicts differences in intercepts, as individuals have different levels of
self-control, but little variation around the average growth rate that reflects the ongoing
socialization of individuals (insignificant r1i).
Turning first to the quadratic low self-control model in table 2, the findings are again
inconsistent with SCT’s stability proposition. First, the significant age-squared parameter
indicates a curvilinear trajectory wherein low self-control continues to increase past 10
years of age, before beginning a steady decline. Importantly, however, there is significant
variation in both the levels and growth rates, indicating that individuals are not developing
in a relatively homogeneous fashion (r1i =.306 p<.001).
We also can see the relationship between levels and slopes at different ages by cen-
tering our data at different ages; we estimated models centering the data at 12, 18,
and 22 years of age. As shown in the table, where we present correlations (instead of
covariances for ease of interpretation), the signs and magnitudes of the intercept and
growth rate correlations are highly dependent on age. When low self-control is on the
rise at 12 years of age, lower levels (higher self-control) predict steeper growth and
vice versa. On the other hand, at 22 years of age, when low self-control is consider-
ably lower, on average, higher levels (lower self-control) are associated with steeper
negative slopes and hence greater declines. Notably, these patterns are consistent with
those observed in the GBTMs. Rather than the portrait of lock-step parallel change
painted by SCT, these age-specific correlations suggest a complex longitudinal distribu-
tion full of crossed paths. The reliability estimate (π1i) for the growth rates in self-control
is .483, indicating that 48.3 percent of observed variation in growth rates reflects true
Focusing on the impulsivity and sensation-seeking models, one can see a stark contrast
between the growth patterns in these two traits. First, the mean growth trajectory for sen-
sation seeking is flat as evidenced by the insignificant age coefficient and the absence of
a quadratic age effect, which was dropped given its insignificance. In contrast, the mean
growth trajectory for impulsivity displays a significant curvilinear pattern, with increases
up to 13 years of age, followed by a steady, significant decline. Not unexpected, but incon-
sistent with SCT, for both traits there is significant variation around both the intercepts
and growth parameters, and the relationship between the two varies at different ages. The
reliability estimates for the growth parameters indicate that the proportion of population
variance in the observed rate of change that is variance in the true rate of change for sen-
sation seeking is lower than that for impulsivity (π1i =.34 vs. .53), likely reflecting less
precision due to more curvilinear individual curves (see the GBTMs).
Finally, we attend to the rho estimates. Recall that rho is the proportion of the vari-
ance in each construct that is attributable to between- versus within-individual variation
around the person-specific mean and would be close to 1 if the rank-order stability as-
sumption holds. The rho from these models are far from unity—.52, .51, and .43 for global
self-control, impulsivity, and sensation seeking, respectively—providing further evidence
against rank-order stability.
Altogether, then, the findings from these supplementary models of stability provide
convergent evidence of the instability of self-control both between and within individuals.
Moreover, these findings also support the evidence from the GBTMs that distinct patterns
of development exist for impulsivity and sensation seeking that are masked when these
two traits are combined in a global measure of self-control. Evidence is clear that, at least
among this population, self-control, impulsivity, and sensation seeking are not fixed at
10 years of age and that many individuals have different levels and patterns of changes
in impulsivity and sensation seeking. We next consider the criminological significance of
these changes.
12. Singer and Willett (2003) noted that a disadvantage of reliability as a gauge of measurement quality
is that it confounds the effect of within-person precision with the effect of between-person hetero-
geneity in true change. Thus, reliability tends toward 0 when individual precision is poor or when
interindividual heterogeneity in true change is small. Reliability can be improved in two ways:
greater precision at the individual level (e.g., more observations) or increased between-individual
variability in true change.
Table 3. Unconditional Negative Binomial Fixed Effects Models
Predicting Self-Reported Offending (Waves 1, 2, 4, 5, and 6)
Model 1 Model 2
Independent VariablesabSE %βbbSE %βb
Global Low Self-Control .85∗∗∗ .07 29.9
Impulsivity .42∗∗∗ .07 15.3
Sensation Seeking .44∗∗∗ .05 21.9
(Std.) Age .45 .29 56.5 .43 .29 53.4
Age-squared .38∗∗∗ .07 21.0 .37∗∗∗ .07 21.0
∗∗∗p<.001; ∗∗ p<.01; n =746 (3335 person-waves).
aWith the exception of age, the independent variables were not standardized.
b%βindicates the percent change in the expected count of the outcome for a standard deviation increase in the
predictor, holding all other variables constant (i.e., 100[Exp(βSDx)-1]).
Note: Dummy variables for all respondents (less 1 for comparison) were included to estimate the effects of
changes net of stable individual characteristics. In addition, dummy variables for each wave less one were also
included as controls. The coefficients for these variables are not shown for brevity. Standard errors were adjusted
for overdispersion using the vce(opg) option in Stata.
The results displayed in table 3 indicate that the within-individual changes are con-
sequential for changes in offending. The results of these models indicate the effect of
a within-individual change in the independent variable on the within-individual change
in crime, net of all time-stable individual characteristics and other variables in the model
(i.e., age and wave of survey). Model 1 reveals that changes in self-control are significantly
associated with changes in crime, net of age and stable individual characteristics. Illustra-
tively, a standard deviation increase in low self-control increases the expected count of
crime by almost 30 percent. In model 2, we examine the contribution of within-individual
changes in impulsivity and sensation seeking to changes in crime, and the results show
that changes in each are associated with within-individual changes in offending, net of the
other, age, and stable individual characteristics. A standard deviation increase in impul-
sivity increases the expected count of offending by roughly 15 percent, whereas a standard
deviation increase in sensation seeking increases the expected count of crime by almost
22 percent. We also tested whether the slightly larger coefficient for sensation seeking is
significantly different from that of impulsivity. Wald tests indicate that the coefficients are
not significantly different in size from one another (χ2=.06; p>.80).
In sum, these results suggest that the changes in global self-control as well as the
changes in the separate constructs of impulsivity and sensation seeking observed over
adolescence into young adulthood are not merely measurement error or irrelevant fluc-
tuations. Instead, these changes are significantly associated with changes in crime. Fur-
thermore, changes in both sensation seeking and impulsivity are uniquely associated with
changes in crime.
Over the past decades, scholarly attention to Gottfredson and Hirschi’s (1990) SCT has
remained incredibly high. Having established that self-control is indeed a strong correlate
of crime, attention has shifted to other aspects of SCT, such as the stability proposition,
which asserts that, after childhood, between-individual levels of self-control are fixed. The
current study has focused on this stability proposition and two corollary ones, gleaning
insights from prior work on SCT as well as theory and research from developmental and
personality psychology and social neuroscience. Analyses investigating the stability of
self-control and the criminological significance of observed instability were conducted
with longitudinal data on African American youth followed from 10 to 25 years of age.
Several important findings emerged from our analyses, and these are discussed next. This
discussion is followed by a consideration of the implications of these findings for SCT and
suggestions for future work.
Building on several recent studies that have examined GBTMs of self-control from
childhood up to the 10th grade, we extended the study of self-control stability into the
mid-20s. Consistent with prior research, we found notable between-group instability in
self-control. Even using a highly conservative method of assessing rank-order stability,
SCT’s strict stability postulate was not supported. Interestingly, our GBTMs showed more
between-group instability than prior research (e.g., Hay and Forrest, 2006; Ray et al.,
2013), which is likely due to both our longer follow-up period as well as our use of self-
reports of self-control (vs. mother or teacher reports). Thus, consonant with psychological
research on self-regulation and related personality constructs (e.g., Helson et al., 2002;
Morizot and Le Blanc, 2003; Raffaeli, Crockett, and Shen, 2005), these findings suggest
that between-individual levels of self-control are not fixed at 8–10 years of age followed
by a homogeneous pattern of normative maturation.
Our supplementary models provided convergent evidence that self-control is not sta-
ble between or within individuals after 10 years of age but rather continues to change
into emerging adulthood. For example, the proportion of reliable stability between waves
1 and 6 and was roughly .10 for self-control. This rate is remarkably low given that this
is after the time that SCT argues self-control is stabilized. These estimates also provided
unique information. For example, the reliable stability estimates revealed that the sta-
bility in these constructs decreases rapidly as the time between measurements increased.
This finding is consistent with personality research that has shown that time is positively
associated with trait instability (e.g., Roberts and Mroczek, 2008). Such evidence contra-
dicts “set point” models that view personality traits as having genetically determined set
points that people may stray from once in a while but to which they drift back over time
(e.g., Headey and Wearing, 1989; Lykken and Tellegen, 1996).
Additionally, consistent with predictions from the dual systems model, stability was
much lower in adolescence than emerging adulthood. These findings, which are also
consonant with general research on personality development (e.g., Caspi, Roberts and
Shiner, 2005; Fraley and Roberts, 2005), have potentially important implications for re-
search on stability. Studies that span 5 years or less will see only the first part of the
picture, where reliable stability is above .5, running the risk of overestimating the overall
stability of these constructs. On the other hand, studies that focus solely on adolescence
may yield lower stability estimates than those using adult samples. Of course, both pat-
terns are inconsistent with SCT.
In terms of theoretical implications, these findings are consonant with the verity that
adolescence is a period of dramatic biological, behavioral, and social changes that involve
a general reorientation of social behavior as individuals move from the pre-reproductive
to the reproductive stage (Forbes and Dahl, 2010). As we have discussed, these biological
changes include substantial restructuring of cortical regions and connections undergird-
ing impulsivity and sensation seeking (e.g., Steinberg, 2008). Scholars have suggested that
neural regions undergoing dramatic restructuring are particularly open to environmental
input (Anderson, 2003; Blakemore and Choudhury, 2006), and therefore, the extensive
remodeling and restructuring of these brain regions in adolescence creates a sensitive
period for change (Ellis et al., 2012). From this perspective, environmental influences
operate in concert with neurobiological changes to create a period of heightened devel-
opmental plasticity—a window of both vulnerability and opportunity—in impulsivity and
sensation seeking in adolescence (Anderson, 2003). Of course, such ideas only make sense
from a position recognizing the developmental plasticity of these traits.
Negative evidence for SCT’s stability proposition also emerged from the HLM models,
which revealed both age-related and between-individual differences in rates of change
during the study period. Overall, our findings, combined with existing research using sev-
eral different samples, measures, and methods, evince that neither between- nor within-
individual levels of self-control are stable or fixed after late childhood (e.g., Burt, Simons,
and Simons, 2006; Hay and Forrest, 2006; Na and Paternoster, 2012; Ray et al., 2013).
Moreover, it is also evident that these changes are not merely due to measurement error
or irrelevant fluctuations (Burt, Simons, and Simons, 2006; Na and Paternoster, 2012). In
our view, sufficient evidence now exists to adjudge the stability proposition as false. It
no longer makes sense to talk about self-control as if it is impervious to change after 10
years of age, as it is clear this is not the case. Additionally, we submit that going forward
it will be more useful to move beyond a concentration on rank-order stability, given that
what is important for one’s likelihood of offending is not one’s ranking but one’s level of
The second focus of our study was to examine the stability and developmental tra-
jectories of two key “elements of self-control”—impulsivity and sensation seeking. We
expected that impulsivity and sensation seeking would demonstrate different develop-
mental patterns given evidence from various domains. Criminological research suggests
that commonly used measures of self-control are not unidimensional and that these dif-
ferent dimensions are differentially predictive of crime (e.g., Arneklev, Grasmick, and
Bursik, 1999; Pfefferbaum and Wood, 1994). Research on the structure of personality
and the cognitive traits underlying risky behaviors has distinguished between sensation
seeking and impulsivity, two dimensions that are conflated in most existing measures of
self-control in criminology (e.g., Burt and Simons, 2013; Marcus, 2004). Furthermore, re-
cent advances in developmental neuroscience suggest that different neurobiological sys-
tems underlie the distinct constructs of sensation seeking and impulsivity and that these
systems normatively develop along different maturational timetables (e.g., Casey, Getz,
and Galvan, 2008; Steinberg, 2008). Consonant with these bodies of work, we identified
separate sensation-seeking and impulsivity factors, and these two constructs displayed
markedly different patterns of development. These GBTMs revealed that divergent pat-
terns of stability and change exist for these distinct dimensions, patterns that are conflated
in global measures of self-control. The HLMs confirmed this conclusion, as they clearly
showed different average growth trajectories as well as age–growth rate correlations for
the two traits.
We also assessed whether latent classes of development for these two dimensions
were identifiable by estimating multitrajectory models. Here, groups were defined by
co-occurring trajectories of impulsivity and sensation seeking. For much of the sample
across most of the study period, significant differences existed in the levels and patterns
of change within groups. Overall, slightly more than half of the sample was classified into
groups where the levels or patterns of change in the two traits were dramatically different.
Given that these differences are collapsed in global measures of self-control, these mod-
els further underscore the importance of disaggregating these constructs when examining
patterns of development.
Thus, our findings combine with prior studies to challenge SCT’s contention that the
various “elements of self-control” come together in the same people to comprise a stable
construct. The present study suggests the need to disaggregate at least two traits that
are conflated in many measures of self-control. Although our results suggest that the
“elements of self-control” come together for some individuals at some points in the life
course, it is also apparent that they do not come together for most. In our view, evidence
from criminological, psychological, and neurological studies convincingly suggests that
the most common conceptualization and operationalization of self-control based on its
“elements” is misguided and should be discarded (see also Arneklev et al., 1993; DeLisi,
Hochstetler, and Murphy, 2003; Pfefferbaum and Wood, 1994). Alternatively, the theo-
retical model could be reformulated such that the argument becomes: Those who are high
on all (or several) of the elements are most likely to engage in crime, and those who are
low are least likely, recognizing that for many these traits do not coalesce. At present,
SCT’s unidimensionality argument provides parsimony at the expense of accuracy.
Finally, given the observed instability, we investigated whether these alterations matter
for crime. Our results revealed that within-individual changes in the global measure of
self-control are significantly associated with within-individual changes in offending net
of all time-stable individual characteristics and age. The more novel finding is that the
component dimensions of sensation seeking and impulsivity are independently associated
with within-individual changes in offending. This finding provides further evidence that
these changes are real (as opposed to measurement error) and consequential for criminal
Although we believe this study makes an important contribution to the body of re-
search on SCT, it is not without limitations. First, the sample consists of African Amer-
ican youth living in various communities in Iowa and Georgia at the first wave of data
collection (scattered across 34 states in wave 6). We assume that the processes we exam-
ine are not limited to this sample or these contexts, an assumption bolstered by the fact
that our findings are similar to those reported in prior studies. Nonetheless, it is important
that future research replicate these findings using different samples.
Another caveat is related to our measure of self-control. Here, we have ignored the
question of whether existing measures of self-control, including the Grasmick-type scales,
appropriately capture Gottfredson and Hirschi’s conception of self-control (“the ten-
dency to consider long-term consequences”). Instead, we have asked whether such op-
erationalizations of self-control, which have been identified as among the strongest cor-
relates of crime, are stable and multidimensional. It may be the case, however, that these
measures do not accurately capture the construct at the heart of SCT, which would of
course vitiate the implications of these findings for the theory. For example, despite be-
ing incorporated in self-control measures in the risk-seeking element, sensation seeking
is a motivational factor (e.g., Forbes and Dahl, 2010), and Gottfredson and Hirschi were
adamant that motivations have no place in their theory of crime because the potential
gains from crimes are universally desirable. “The theory requires that crime be under-
stood without reference to motives or benefits” (Hirschi and Gottfredson, 2008: 221).
Our measures of impulsivity and sensation seeking are also imperfect. There is an
ongoing debate among scholars regarding the nature and measurement of impulsivity
in personality psychology including research on its (multi)dimensionality (Smith et al.,
2007; Whiteside and Lynam, 2001). Although our measures of these two traits are con-
sistent with much psychological work on risk taking, including tests of the dual systems
models (e.g., Quinn, Stappenbeck, and Fromme, 2011; Steinberg et al., 2008), further
work is needed that uses different measures of these traits. We view this effort as a first
step in demonstrating the value of these alternative models in developmental and per-
sonality psychology, and we hope it is used as a springboard for further research on these
constructs and their effects.
Additionally, we should note that our findings are based on self-reports. Although we
view the use of self-reports of self-control in the present study as a strength, given the age
of the youth and the fact that most prior studies of stability have used other reports of
self-control, some scholars including Hirschi and Gottfredson (1983) argued that survey
responses will be less valid for those with low self-control (see Piquero, Macintosh, and
Hickman, 2000). To the extent that low self-control biases survey responses, our results
will be affected.
Notably, we only follow individuals to 25 years of age, and it is possible that levels
of self-control become more stable as individuals age. Importantly, the preponderance
of theory and research on personality trait stability indicates that although there is con-
sistency and, in some cases, increased stability over time, personality is never fixed but
remains malleable over the life course (e.g., Caspi and Roberts, 2001; Morizot and Le
Blanc, 2005; Roberts and Mroczek, 2008). For example, in their study of personality de-
velopment, Scollon and Diener (2006: 1162) found that “individuals over age 30 exhibited
just as much change as those under 30, thus refuting the idea that personality becomes ‘set
like plaster by age 30’ or that development slows down after young adulthood.”
A final limitation is that the FACHS does not include self-control items in wave 3; thus,
our analyses do not include observations between 15 and 17 years of age. As we have
noted, this will have the effect of smoothing over instability over this age period, thereby
making our trajectory models more conservative. Moreover, this gap covers the period
when we would expect to see the peak in sensation seeking and the widest maturity gap
between sensation seeking and impulse control, which makes our study inappropriate for
testing key developmental predictions from the dual systems model. We hope that our
findings stimulate further criminological research into the veracity of the dual systems
model and its application to criminal offending, especially the crime peak in adolescence.
Despite these limitations, we believe this investigation contributes in important ways
to research on SCT. It is hoped that future research explores these issues with different
samples and different measures, and that it follows individuals further into adulthood. In
general, our findings provide contrary evidence for a number of facets of SCT, findings
that are consistent with both prior research in criminology and elsewhere, and suggest
several avenues for future research.
First, we hope that future research moves away from the question of whether these
constructs are stable, as evidence clearly indicates they are not, and focuses instead on
when,how,orwhy these traits change. Specifically, we hope studies identify contextual
factors that effect change, in both the short and the long term, as well as factors that may
relate to individual differences in the degree to which these traits are responsive to envi-
ronmental influences. In our view, it is important to examine both the factors that shape
initial levels and trajectories of these traits as well as the factors that effect change in
impulsivity and sensation seeking after childhood. Evidence that adolescence and young
adulthood are crucial periods for personality change underscores the import of investigat-
ing the causal mechanisms responsible for personality change over these periods in the life
course. Because adolescence is characterized by significant biological and social changes,
it will be difficult but valuable to disentangle the effects of social context from biological
maturation while attending to their interactions. Additionally, given the observed differ-
ences in the developmental patterns of sensation seeking and impulsivity, we hope future
research explores contributors to these distinct patterns observed among the majority of
the sample. Several studies have begun to engage in these issues and have linked social
factors, such as alterations in peer affiliations, parenting practices, and school attachment,
to changes in self-control (e.g., Burt, Simons, and Simons, 2006; Meldrum, 2008; Ray et al.,
2013). This work might beneficially link to research on personality development, which
indicates that individuals demonstrate unique trait trajectories over the life course and
that the various changes appear to be the result of the interaction among life experiences,
their timing, and individual characteristics (e.g., Caspi, Roberts, and Shiner, 2005; Helson,
Jones, and Kwan, 2002; Roberts and Mrozcek, 2008; Vaidya et al., 2002).
Future work might also explore sex/gender differences in the development and predic-
tors of changes in these two traits. At present, there is a dearth of knowledge on how gen-
der dynamics, as main effects or in interaction with biological ones (such as hormones),
influence the development of these traits over time. Evolutionary-developmental argu-
ments emphasize the different evolutionary forces that favor different personality charac-
teristics for males and females, including the influence of sexual selection on risk-taking
behaviors (e.g., Ellis et al., 2012; James et al., 2012). Consistent with these arguments,
studies have identified sex differences in adolescent brain regions that relate to impulsiv-
ity and sensation seeking (e.g., Lenroot and Giedd, 2010), and applications of sexual selec-
tion theory to risk behaviors identify sex-specific pathways (e.g., James et al., 2012). Given
well-documented sex differences in impulsivity and sensation seeking as well as involve-
ment in reckless behavior (e.g., Byrnes, Miller, and Schafer, 1999; Campbell, Muncer,
and Bibel, 2001; Simons and Burt, 2011), further research investigating how such biologi-
cal factors act in concert with social influences to shape differences in risk taking between
males and females is needed.
Finally, our results indicated that within-individual changes in sensation seeking and
impulsivity independently predict changes in crime, which is consistent with prior re-
search that has shown that the various elements of self-control are differentially related to
different (criminal) outcomes. This finding points to the potential value of further explor-
ing differences in the manifestations of various within-individual co-occurring patterns of
these two traits. Different combinations of these traits may be associated with different
offenses or the same offenses but for different reasons. Such information may be of use
for designing interventions that work with individuals’ propensities to decrease reckless
behaviors instead of efforts tailored toward a one-model-fits-all approach. For example,
offering alternative (not reckless) options for nonimpulsive sensation seekers (e.g., Group
2 in mid-adolescence) might reduce involvement in reckless pursuits for thrills, whereas
such interventions may have no influence on the reckless behaviors of impulsive, low
sensation seekers (Group 4). In sum, the present study raises many new questions, and
we hope that future work tackles some of these issues.
“As the limits of the general theory are revealed—as occurs with all prominent
theories—a rigid fidelity to the original statement of the general theory is likely to ensure
its staleness if not decline . . . it appears that the time has come for the general theory to
broaden its horizons so as to confront criminological realities that now rest beyond its
boundaries” (Cullen et al., 2008: 74).
Gottfredson and Hirschi’s (1990) SCT changed the course of criminological research.
With its parsimony and clearly stated propositions as well as its tone and bold statements,
SCT was able to steer criminological attention to the relevance of internal controls in
crime causation, as earlier theorists had been unable to do before. SCT also rightly em-
phasized the potential of the family to influence levels of self-control while underscoring
the reality that some people suffer from the negative consequences of low self-control
throughout the life course.
SCT, however, has gotten some things wrong. One of these is the stability of self-
control. Evidence clearly contradicts the SCT proposition that those who were not for-
tunate enough to be effectively parented and, therefore, failed to develop self-control are
doomed to a life of myopic decisions and negative consequences thereof. We hope this
evidence is used to revise SCT, perhaps by modifying the strict stability postulate to rec-
ognize that while there is consistency in self-control over time, there is also considerable
change that is complex and ongoing as individuals adapt to their social environments. An-
other important target for revision is the SCT’s developmental timetable for self-control,
given clear evidence that the neurobiological systems that govern impulsivity and sen-
sation seeking undergo significant remodeling and refining in adolescence and are not
yet “mature” at the end of the second decade of life. As an alternative to refining SCT,
criminologists might construct a new theory to take its place, drawing on its valuable in-
sights while correcting misguided or overly rigid positions, or adopt existing theories from
outside the field, such as the dual systems model, that emphasize the role of lack of pre-
meditation and hasty decision making in the genesis of reckless behaviors. In our view,
whatever form they take, such efforts should incorporate the many significant advances
in developmental psychology and neuroscience that have the potential to greatly enhance
our understanding of mechanisms involved in the etiology of crime, including the crime
peak in mid-adolescence, and suggest new methods of crime prevention and intervention.
Given the power of self-control (and its constituent traits) in predicting offending, we
believe such efforts will be worthwhile.
Akers, Ronald L. 1991. Self-control as a general theory of crime. Journal of Quantitative
Criminology 7:201–11.
Allison, Paul D. 2009. Fixed Effects Regression Models. Thousand Oaks, CA: Sage.
Allison, Paul D., and Richard P. Waterman. 2002. Fixed-effects negative binomial regres-
sion models. Sociological Methodology 32:247–65.
American Psychiatric Association. 1994. Diagnostic and Statistical Manual of Mental Dis-
orders. Washington, DC: American Psychiatric Association.
Anderson, Susan L. 2003. Trajectories of brain development: Point of vulnerability or
window of opportunity. Neuroscience and Biobehavioral Reviews 27:3–18.
Arneklev, Bruce J., Harold G. Grasmick, and Robert J. Bursik, Jr. 1999. Evaluating the
dimensionality and invariance of “low self-control.” Journal of Quantitative Criminol-
ogy 15:307–31.
Arneklev, Bruce J., Harold G. Grasmick, Charles R. Tittle, and Robert J. Bursik. 1993.
Low self-control and imprudent behavior. Journal of Quantitative Criminology 9:225–
Ashton, Michael C., Kibeon Lee, Lewis R. Goldberg, and Reinout E. de Vries. 2009.
Higher-order factors of personality: Do they exist? Personality and Social Psychology
Review 13:79–91.
Baumeister, Roy F. Todd F. Heatherton, and Dianne M. Tice. 1994. Losing Control: How
and Why People Fail at Self-Regulation. San Diego, CA: Academic Press.
Baumeister, Roy F., Kathleen D. Vohs, and Dianne M. Tice. 2007. The strength model of
self-control. Current Directions in Psychological Science 16:351–5.
Beaver, Kevin M., Matt DeLisi, Michael G. Vaughn, John Paul Wright, and Brian B.
Boutwell. 2008. The relationship between self-control and language: Evidence of a
shared etiological pathway. Criminology 46:201–32.
Blakemore, Sarah-Jayne, and Suparna Choudhury. 2006. Development of the adolescent
brain: Implications for executive function and social cognition. Journal of Child Psy-
chology and Psychiatry 47:296–312.
Burt, Callie H. 2014. Self-control and crime: A sociological perspective. In The Nurture
versus Biosocial Debate in Criminology, eds. Kevin M. Beaver, J. C. Barnes, and Brian
B. Boutwell. Thousand Oaks, CA: Sage.
Burt, Callie H., and Ronald L. Simons. 2013. Self-control, thrill seeking, and crime: Moti-
vation matters. Criminal Justice and Behavior 40:1326–48.
Burt, Callie H., Ronald L. Simons, and Leslie G. Simons. 2006. A longitudinal test of the
effects of parenting and the stability of self-control: Negative evidence for the general
theory of crime. Criminology 44:353–96.
Byrnes, James P., David C. Miller, and William D. Schafer. 1999. Gender differences in
risk-taking: A meta-analysis. Psychological Bulletin 125:367–83.
Campbell, Anne, Steven Muncer, and Daniel Bibel. 2001. Women and crime: An evolu-
tionary approach. Aggression and Violent Behavior 5:481–97.
Cardinal, Rudolf N., John A. Parkinson, Jeremy Hall, and Barry J. Everitt. 2002. Emotion
and motivation: The role of the amygdala, ventral striatum, and prefrontal cortex.
Neuroscience & Biobehavioral Reviews 26:321–52.
Casey, B. J., Adriana Galvan, and Todd A. Hare. 2005. Changes in cerebral functional
brain development and its relation to cognitive development. Biological Psychology
Casey, B. J., Sarah Getz, and Adriana Galvan. 2008. The adolescent brain. Developmental
Review 28:62–77.
Casey, B. J., Rebecca M. Jones, and Leah H. Sommerville. 2011. Braking and accelerating
of the adolescent brain. Journal of Research on Adolescence 21:21–33.
Caspi, Avshalom, and Brent W. Roberts. 2001. Personality development across the life
span: The argument for change and continuity. Psychological Inquiry 12:49–66.
Caspi, Avshalom, Brent W. Roberts, and Rebecca L. Shiner. 2005. Personality develop-
ment: Stability and change. Annual Review of Psychology 56:453–84.
Cauffman, Elizabeth, Elizabeth P. Shulman, Laurence Steinberg, Eric Claus, Marie T.
Banich, Sandra Graham, and Jennifer Woolard. 2010. Age differences in affective de-
cision making as indexed by performance on the Iowa Gambling Task. Developmental
Psychology 46:193–207.
Chambers, R. Andrew, Jane R. Taylor, and Marc N. Potenza. 2003. Developmental neu-
rocircuitry of motivation in adolescence: A critical period of addiction vulnerability.
The American Journal of Psychiatry 160: 1041–52.
Cullen, Francis T., James D. Unnever, John Paul Wright, and Kevin M. Beaver. 2008.
Parenting and self-control. In Out of Control: Assessing the General Theory of Crime,
ed. E. Goode. Stanford, CA: Stanford University Press.
Cutrona, Carolyn, Daniel Russell, Robert Hessling, P. Adama Brown, and Velma Murry.
2000. Direct and moderating effects of community context on the psychological well-
being of African American women. Journal of Personality and Social Psychology
Dahl, Ronald. 2004. Adolescent brain development: A period of vulnerabilities and op-
portunities. Annals of the New York Academy of Sciences 1021:1–22.
DeLisi, Matt, Andy Hochstetler, and Daniel S. Murphy. 2003. Self-control behind bars:
A validation study of the Grasmick et al. scale. Justice Quarterly 20:241–63.
Duckworth, Angela Lee, and Margaret L. Kern. 2011. A meta-analysis of the convergent
validity of self-control measures. Journal of Research in Personality 45:259–68.
Elliott, Delbert S., David Huizinga, and Scott Menard. 1986. Multiple Problem Youth:
Delinquency, Substance Use, and Mental Health Problems. New York: Springer-
Ellis, Bruce J., Marco Del Guidice, Thomas J. Dishion, Aurelio Jos´
e Figueredo, Peter
Gray, Vladas Griskevicius, Patricia H. Hawley, W. Jake Jacobs, Jen´
ee James, An-
thony A. Volk, and David Sloan Wilson. 2012. The evolutionary basis of risky adoles-
cent behavior: Implications for science, policy, and practice. Developmental Psychol-
ogy 48: 598–623.
Evans, T. David, Francis T. Cullen, Velmer S. Burton, Jr., R. Gregory Dunaway, and
Michael L. Benson. 1997. The social consequences of self-control: Testing the general
theory of crime. Criminology 35:475–504.
Eysenck, Sybil, and Hans J. Eysenck. 1978. Impulsiveness and venturesomeness: Their
position in a dimensional system of personality description. Psychological Reports
Farrington, David P., Rolf Loeber, and D. Jolliffe. 2008. The age-crime curve in reported
offending. In Violence and Serious Theft: Development and Prediction from Childhood
to Adulthood, eds. Rolf Loeber, David P. Farrington, Magda Stouthamer-Loeber, and
Helene Raskin. New York: Routledge/Taylor and Francis.
Forbes, Erika E., and Ronald E. Dahl. 2010. Pubertal development and behavior: Hor-
monal activation of social and motivational tendencies. Brain and Cognition 72: 66–72.
Fraley, Chris R., and Brent W. Roberts. 2005. Patterns of continuity: A dynamic model
of conceptualizing the stability of individual differences in psychological constructs
across the life course. Psychological Review 112:60–74.
Galvan, Adriana, Todd A. Hare, Cindy E. Parra, Jackie Penn, Henning Voss, Gary
Glover, and B. J. Casey. 2006. Earlier development of the accumbens relative to
orbitofrontal cortex might underlie risk-taking behavior in adolescents. Journal of
Neuroscience 26:6885–92.
Gibbons, Frederick, Meg Gerrard, Michael Cleveland, Thomas Wills, and Gene Brody.
2004. Perceived discrimination and substance use in African American parents and
their children: A panel study. Journal of Personality and Social Psychology 86:517–29.
Gibson, Chris, and John Wright. 2001. Low self-control and co-worker delinquency: A
research note. Journal of Criminal Justice 29:483–92.
Goode, Erich. 2008. Out of Control: Assessing the General Theory of Crime. Stanford,
CA: Stanford University Press.
Gottfredson, Michael R., and Travis Hirschi. 1990. A General Theory of Crime. Palo Alto,
CA: Stanford University Press.
Grasmick, Harold, Charles Tittle, Robert Bursik, and Bruce Arneklev. 1993. Testing the
core empirical implications of Gottfredson and Hirschi’s general theory of crime.
Journal of Research in Crime and Delinquency 30:5–29.
Greenberg, David F., Robin Tamarelli, and Margaret S. Kelley. 2002. The generality of
the self-control theory of crime. In Advances in Criminological Theory, vol. 10: Crime
and Social Organization, eds. Elin J. Waring and David L. Weisburd. New Brunswick,
NJ: Transaction.
Gullo, Matthew J., Elizabeth Ward, Sharon Dawe, Jane Powell, and Chris J. Jackson.
2011. Support for a two-factor model of impulsivity and hazardous substance use in
British and Australian young adults. Journal of Research in Personality 45:10–18.
Harden, K. Paige, and Elliot M. Tucker-Drob. 2011. Individual differences in the devel-
opment of sensation seeking and impulsivity during adolescence: Further evidence for
a dual systems model. Developmental Psychology 47:739–46.
Hay, Carter, and Walter Forrest. 2006. The development of self-control: Examining self-
control’s stability thesis. Criminology 44:739–74.
Hay, Carter, Ryan Meldrum, Walter Forrest, and Emily Ciaravolo. 2010. Stability and
change in risk seeking: Investigating the effects of an intervention program. Youth
Violence and Juvenile Justice 8:91–106.
Headey, Bruce, and Alexander Wearing. 1989. Personality, life events, and subjective
well-being: Toward a dynamic equilibrium model. Journal of Personality and Social
Psychology 57:731–9.
Helson, Ravenna, Constance Jones, and Virginia S. Y. Kwan. 2002. Personality change
over 40 years of adulthood: Hierarchical linear modeling analyses of two longitudinal
samples. Journal of Personality and Social Psychology 83:752–66.
Higgins, George E., Wesley G. Jennings, Richard Tewksbury, and Chris L. Gibson. 2009.
Exploring the link between low self-control and violent victimization trajectories in
adolescents. Criminal Justice and Behavior 36:1070–84.
Hirschi, Travis. 1969. Causes of Delinquency. Berkeley: University of California Press.
Hirschi, Travis. 2004. Self-control and crime. In Handbook of Self-Regulation, eds. Roy
F. Baumesiter and Kathleen D. Vohs. New Brunswick, NJ: Transaction.
Hirschi, Travis, and Michael R. Gottfredson. 1983. Age and the explanation of crime.
American Journal of Sociology 89:552–611.
Hirschi, Travis, and Michael R. Gottfredson. 2001. Self-control theory. In Explaining
Criminals and Crime, eds. Raymond Paternoster and Ronet Bachman. Los Angeles,
CA: Roxbury Press.
Hirschi, Travis, and Michael R. Gottfredson. 2003. Punishment of children from the
perspective of control theory. In Control Theories of Crime and Delinquency, eds.
Chester L. Britt and Michael R. Gottfredson. New Brunswick, NJ: Transaction.
Hirschi, Travis, and Michael R. Gottfredson. 2008. Critiquing the critics: The authors
respond. In Out of Control: Assessing the General Theory of Crime, ed. Erich Goode.
Stanford, CA: Stanford University Press.
James, Jenee, Bruce J. Ellis, Gabriel L. Schlomer, and Judy Garber. 2012. Sex-specific
pathways to early puberty, sexual debut, and sexual risk taking: Tests of an integrated
evolutionary-developmental model. Developmental Psychology 48:687–702.
Jones, Bobby L., and Daniel S. Nagin. 2007. Advances in group-based trajectory modeling
and an SAS procedure for estimating them. Sociological Methods & Research 35:542–
Jones, Bobby L., Daniel S. Nagin, and Kathryn Roeder. 2001. A SAS procedure based
on mixture models for estimating developmental trajectories. Sociological Methods &
Research 29:374–93.
Kendall, Philip C., and Lance E. Wilcox. 1979. Self-control in children: Development of a
rating scale. Journal of Consulting and Clinical Psychology 47:1020–9.
Kuder, G. F., and M. W. Richardson. 1937. The theory of the estimation of test reliability.
Psychometrika 2:151–60.
Lenroot, Rhoshel K., and Jay N. Giedd. 2010. Sex differences in the adolescent brain.
Brain and Cognition 72:46–55.
Longshore, Douglas, Judith A. Stein, and Susan Turner. 1998. Reliability and validity of
a self-control measure: Rejoinder. Criminology 36:175–82.
Longshore, Douglas, Susan Turner, and Judith A. Stein. 1996. Self-control in a criminal
sample: An examination of construct validity. Criminology 34:209–28.
Lykken, David, and Auke Tellegen. 1996. Happiness is a stochastic phenomenon. Psy-
chological Science 7:186–9.
Lynam, Donald R., and Joshua D. Miller. 2004. Personality pathways to impulsive behav-
ior and their relations to deviance: Results from three samples. Journal of Quantitative
Criminology 20:319–41.
Lynne-Landsman, Sarah D., Julia A. Graber, Tracy R. Nichols, and Gilbert J. Botvin.
2011. Is sensation seeking a stable trait or does it change over time? Journal of Youth
and Adolescence 40: 48–58.
Marcus, Bernd. 2004. Self-control in the general theory of crime: Theoretical implications
of a measurement problem. Theoretical Criminology 8:33–55.
McCrae, Robert R., and Paul T. Costa, Jr. 1990. Personality in Adulthood.NewYork:
Guilford Press.
McCrae, Robert R., Paul T. Costa, Jr., Margarida Pedroso de Lima, Antonia Simoes,
Fritz Ostendorf, Alois Angleitner, et al. 1999. Age differences in personality across
the adult life span: Parallels in five cultures. Developmental Psychology 35:466–77.
Meier, Robert F. 1995. Book review. Social Forces 73:1627–9.
Meldrum, Ryan C. 2008. Beyond parenting: An examination of the etiology of self-
control. Journal of Criminal Justice 36:244–51.
Meldrum, Ryan C., Jacob T. N. Young, Callie H. Burt, and Alex R. Piquero. 2013. Ma-
ternal versus adolescent reports of self-control: Implications for testing the general
theory of crime. Journal of Criminal Justice 41:24–32.
Miller, Earl K., and Jonathan D. Cohen. 2001. An integrative theory of prefrontal cortex
function. Annual Review of Neuroscience 24:167–202.
Miller, Joshua D., and Donald Lynam. 2001. Structural models of personality and their
relation to antisocial behavior: A meta-analytic review. Criminology 39:765–98.
Morizot, Julien, and Marc Le Blanc. 2005. Searching for a developmental typology of
personality and its relations to antisocial behavior: A longitudinal study of a repre-
sentative sample of men. Journal of Personality 73:139–82.
Na, Chongmin, and Raymond Paternoster. 2012. Can self-control change substantially
over time? Rethinking the relationship between self- and social control. Criminology
Nagin, Daniel S. 2005. Group-Based Modeling of Development. Cambridge, MA: Harvard
University Press.
Nagin, Daniel S., and Candice L. Odgers. 2010. Group-based trajectory modeling in clin-
ical research. Annual Review of Clinical Psychology 6:109–38.
Nelson, Eric E., Ellen Leibenluft, Erin B. McClure, and Daniel S. Pine. 2005. The so-
cial re-orientation of adolescence: A neuroscience perspective on the process and its
relation to psychopathology. Psychological Medicine 35:163–74.
Patton, Jim H., Matthew S. Stanford, and Ernest S. Barratt. 1995. Factor structure of the
Barratt impulsiveness scale. Journal of Clinical Psychology 51:768–74.
Paus, Tomas. 2005. Mapping brain maturation and cognitive development during adoles-
cence. Trends in Cognitive Science 9:60–8.
Pfefferbaum, Betty, and Peter Wood. 1994. Self-report study of impulsive and delinquent
behavior in college students. Journal of Adolescent Health 15:295–302.
Pickering, Alan D., and Jeffrey A. Gray. 1999. The neuroscience of personality. In Hand-
book of Personality: Theory and Research, eds. Lawrence A. Pervin and Oliver P.
John. New York: Guilford Press.
Piquero, Alex R. 2008. Measuring self-control. In Out of Control: Assessing the General
Theory of Crime, ed. Erich Goode. Stanford, CA: Stanford University Press.
Piquero, Alex R., Randall MacIntosh, and Matthew Hickman. 2000. Does self-control
affect survey response? Applying exploratory, confirmatory, and item response theory
analysis to Grasmick et al.’s self-control scale. Criminology 38:897–930.
Piquero, Alex R., and Andre B. Rosay. 1998. The reliability and validity of Grasmick et
al.’s self-control scale: A comment on Longshore et al. Criminology 36:157–73.
Pratt, Travis C., and Francis T. Cullen. 2000. The empirical status of Gottfredson and
Hirschi’s general theory of crime: A meta-analysis. Criminology 38:931–64.
Quinn, Patrick D., and K. Paige Harden. 2012. Differential changes in impulsivity and
sensation seeking and the escalation of substance use from adolescence to early adult-
hood. Development and Psychopathology 25:223–39.
Quinn, Patrick D., Cynthia A. Stappenbeck, and Kim Fromme. 2011. Collegiate heavy
drinking prospectively predicts change in sensation seeking and impulsivity. Journal
of Abnormal Psychology 120:543–56.
Raffaeli, Marcela, Lisa J. Crockett, and Yuh-Ling Shen. 2005. Developmental stability
and change in self-regulation from childhood to adolescence. The Journal of Genetic
Psychology 166:54–75.
Ray, James V., Shayne Jones, Thomas A. Loughran, and Wesley G. Jennings. 2013. Test-
ing the stability of self-control: Identifying unique developmental patterns and asso-
ciated risk factors. Criminal Justice and Behavior. E-pub ahead of print.
Roberts, Brent W., and Wendy F. DelVecchio. 2000. The rank-order consistency of per-
sonality traits from childhood to old age: A quantitative review of longitudinal studies.
Psychological Bulletin 126:3–25.
Roberts, Brent W., and Daniel Mroczek. 2008. Personality trait change in adulthood. Cur-
rent Directions in Psychological Science 17:31–5.
Romer, Daniel. 2010. Adolescent risk taking, impulsivity, and brain development: Impli-
cations for prevention. Developmental Psychobiology 52:263–76.
Romer, Daniel, and Michael Hennessy. 2007. A biosocial-affect model of adolescent sen-
sation seeking: The role of affect evaluation and peer-group influence in adolescent
drug use. Prevention Science 8:89–101.
Romero, Estrella, M. Angeles Luengo, and Jorge Sobral. 2001. Personality and antisocial
behavior: Study of temperamental dimensions. Personality and Individual Differences
Roth, Marcus, Philipp Hammelstein, and Elmar Br¨
ahler. 2007. Beyond a youthful be-
havioral style: Age and sex differences in sensation seeking based on need theory.
Personality and Individual Differences 43: 1839–1850.
Sampson, Robert J., and John H. Laub. 1993. Crime in the Making: Pathways and Turning
Points Through Life. Cambridge, MA: Harvard University Press.
Schulz, Stefan. 2006. Beyond Self-Control: Analysis and Critique of Gottfredson and
Hirschi’s General Theory of Crime (1990). Berlin, Germany: Duncker & Humblot.
Scollon, Christie Napa, and Ed Diener. 2006. Love, work, and changes in extraversion
and neuroticism over time. Journal of Personality and Social Psychology 91:1152–65.
Simons, Ronald L., and Callie H. Burt. 2011. Learning to be bad: Adverse social condi-
tions, social schemas, and crime. Criminology 49:553–98.
Simons, Ronald L., Velma Murry, Vonnie McLoyd, Kuei- Hsiu Lin, Carolyn Cutrona, and
Rand Conger. 2002. Discrimination, crime, ethnic identity and parenting as correlates
of depressive symptoms among African American children: A multilevel analysis. De-
velopment and Psychopathology 14:371–93.
Singer, Judith D., and John B. Willett. 2003. Applied Longitudinal Analysis: Modeling
Change and Event Occurrence. New York: Oxford University Press.
Smith, Gregory T., Sarah Fischer, Melissa A. Cyders, Agnews M. Annus, Nichea S.
Spillane, and Denis M. McCarthy. 2007. On the validity and utility of discriminating
among impulsivity-like traits. Assessment 14:155–70.
Somerville, Leah H., and B. J. Casey. 2010. Developmental neurobiology of cognitive
control and motivational systems. Current Opinion in Neurobiology 20:236–41.
Steinberg, Laurence. 2008. A social neuroscience perspective on adolescent risk-taking.
Developmental Review 28:78–106.
Steinberg, Laurence, Dustin Albert, Elizabeth Cauffman, Marie Banich, Sandra Graham,
and Jennifer Woolard. 2008. Age differences in sensation seeking and impulsivity as
indexed by behavior and self-report: Evidence for a dual systems model. Developmen-
tal Psychology 44:1764–78.
Steinberg, Laurence, Sandra Graham, Lia O’Brien, Jennifer Woolard, Elizabeth Cauff-
man, and Marie Banich. 2009. Age differences in future orientation and delay dis-
counting. Child Development 80:28–44.
Sweeten, Gary. 2012. Scaling criminal offending. Journal of Quantitative Criminology.
Tellegen, Auke. 1985. Structures of mood and personality and their relevance to assessing
anxiety with an emphasis on self-report. In Anxiety and the Anxiety Disorders, eds. A.
Husain Tuma and Jack D. Maser. Hillsdale, NJ: Lawrence Erlbaum.
Tittle, Charles R., David A. Ward, and Harold G. Grasmick. 2003. Self-control and
crime/deviance. Cognitive vs. behavioral measures. Journal of Quantitative Criminol-
ogy 19:333–65.
Turner, Michael G., and Alex R. Piquero. 2002. The stability of self-control. Journal of
Criminal Justice 30:457–71.
Vaidya, Jatin G., Elizabeth K. Gray, Jeffrey Haig, and David Watson. 2002. On the tem-
poral stability of personality: Evidence for differential stability and the role of life
experiences. Journal of Personality and Social Psychology 83:1469–84.
Vaidya, Jatin G., Robert D. Latzman, Kristian E. Markon, and David Watson. 2010. Age
differences on measures of disinhibition during young adulthood. Personality and In-
dividual Differences 48:8115–20.
Van Gelder, Jean-Louis, and Reinout E. de Vries. 2012. Traits and states: Integrating
personality and affect into a model of criminal decision making. Criminology 50:637–
Vazsonyi, Alexander T., Lloyd E. Pickering, Marianne Junger, and Dick Hessing. 2001.
An empirical test of a general theory of crime: A four-nation comparative study of
self-control and the prediction of deviance. Journal of Research in Crime and Delin-
quency 38:91–131.
Whiteside, Stephen P., and Donald R. Lynam. 2001. The five factor model and impulsiv-
ity: Using a structural model of personality to understand impulsivity. Personality and
Individual Differences 30:669–89.
Winfree, L. Thomas, Terrance J. Taylor, Ni He, and Finn-Aage Esbensen. 2006. Self-
control and variability over time: Multivariate results using a 5-year, multisite panel
of youth. Crime & Delinquency 52:253–86.
Wood, Peter B., Betty Pfefferbaum, and Bruce J. Arneklev. 1993. Risk-taking and self-
control: Social psychological correlates of delinquency. Journal of Crime and Justice
Zuckerman, Marvin. 1979. Sensation Seeking: Beyond the Optimal Level of Arousal. Hills-
dale, NJ: Lawrence Erlbaum.
Callie H. Burt is an assistant professor in the School of Criminology and Criminal
Justice at Arizona State University. Her research takes a biopsychosocial life-course ap-
proach, and current research foci include interpersonal racial discrimination, racial social-
ization, and offending and self-control and crime.
Gary Sweeten is an associate professor in the School of Criminology and Criminal
Justice at Arizona State University. His research interests include criminological theory,
transitions to adulthood, and quantitative methods.
Ronald L. Simons is a Foundation Professor in the School of Criminology and Crimi-
nal Justice at Arizona State University. His current research interests include theories of
crime and biosocial perspectives on deviant behavior.
Additional Supporting Information may be found in the online version of this article at
the publisher’s web site:
Appendix A. Low Self-Control Items
Appendix B. Measures of Offending Across Waves
Appendix C. One-Group Trajectory Models
Appendix D. Comments on Hirschi’s (2004) Reconceptualization
... We believe there are advantages to studying the specific dimension of impulsivity rather than the composite concept of self-control. Recent research has demonstrated subtle but important differences in empirical findings among the various dimensions of low self-control (Burt and Simons 2013;Burt et al. 2014). Given measurement inconsistency across self-control research, these differences imply substantial problems for replication, interpretation, and generalizability inherent in composite concepts. ...
... First, our study focuses on impulsivity, a single dimension within Gottfredson and Hirschi's broader concept. This allows us to capture the dimension most central to low self-control while still recognizing that self-control's separate elements may not develop uniformly (Burt and Simons 2013;Burt et al. 2014). Second, and perhaps more importantly, we go beyond prior work's focus on testing Gottfredson and Hirschi's claim that a single characteristic is sufficient for explaining crime and criminal peers. ...
Full-text available
Objectives Drawing on criminological research about peer delinquency and self-control, we employ a network perspective to identify the potential paths linking impulsivity, peers, and delinquency. We systematically integrate relevant processes into a set of dynamic network models that evaluate these interconnected pathways. Methods Our analyses use data from more than 14,000 students in Pennsylvania and Iowa collected from the evaluation of the PROSPER partnership model. We estimate longitudinal social network models to disentangle the paths through which impulsivity and delinquency are linked in adolescent friendship networks. Results We find evidence of both peer influence and homophilic selection for both impulsivity and delinquency. Further, results indicate that peer impulsivity is linked to individual delinquent behavior through peer influence on delinquency, but not on impulsivity. Finally, the results suggest that impulsivity moderates both influence and selection processes, as adolescents with higher levels of impulsivity are more likely to select delinquent peers but less likely to change their behavior due to peers. Conclusions In sum, this study offers a more holistic framework and stronger theoretical tests than similar studies of the past. Our results illustrate the need to consider the simultaneous network processes related to peers, impulsivity, and delinquency. Further, our findings reveal that a large dataset with ample statistical power is a valuable advantage for detecting the selection processes that shape friendship networks.
... While the two constructs are conceptually different in their origin in that one is theorized to largely inform interpersonal norms around confrontation and violence and the other is more general in its influence on criminal behavior they are far from separate, they are indelibly linked, both theoretically and empirically. Much has been made of the theorized stability and empirical reality of the temporal durability of low self-control (Burt et al., 2014;Hay & Forrest, 2006;Hay et al., 2018). Anderson (1999) has suggested that while most "codeswitch" situationally there are temporal developmental patterns that apply generally to all, and for those most likely to commit crime the street code may be a more significant and consistent part of their identity and decision-making processes (Anderson, 1999). ...
Full-text available
The code of the street and low self-control are prominent theories of crime. However, there is no research that examines if these criminogenic dispositions inform each other over time. We utilize the G.R.E.A.T. data to analyze the development of street code adherence and low selfcontrol longitudinally. We find a portion of the stability associated with street code adherence and low self-control to be a product of measurement, as evidenced by correlating error terms across waves. Additionally, we find low self-control to be related to increases in street code adherence especially in later waves and, to a lesser extent, we show effects of street code adherence on subsequent levels of low selfcontrol. We also discuss results from analyses split by race, gender, and neighborhood type. We discuss the theoretical implications of our findings and argue for the development and utility of a broad measure of criminal propensity.
... Again, few studies have tackled the full complement and complexity of these theoretical predictions, for the previously noted reasons, but have instead focused mostly or exclusively on the rank-order stability (between individual differences) issue of self-control, for instance, (Beaver & Wright, 2007;Coyne & Wright, 2014;Diamond, 2016;Jo, 2015;Yun & Walsh, 2011) or have conflated the between and within individual difference changes over time, or have erroneously interpreted theory as predicting no additional changes at all in self-control during the second decade of life and beyond. Some of this work has provided evidence supporting rank-order stability (Arneklev et al., 1998;Coyne et al., 2015;Diamond et al., 2015;Hay & Forrest, 2006;Ray et al., 2013;Turner & Piquero, 2002), while work has not (Burt et al., 2006(Burt et al., , 2014Meinert & Reinecke, 2018;Mitchell & MacKenzie, 2006;Na & Paternoster, 2012;Winfree, et al., 2006). ...
Full-text available
The current investigation tested changes in low self-control and the developmental links between parenting and the developmental course of self-control. It was hypothesized that (1) low self-control would change over time (within individual changes); (2) parenting would negatively predict both the intercept and slope of low self-control during childhood; (3) parenting would negatively predict only the intercept during early and late adolescence. Self-report data from the Korean Children and Youth Panel Survey (KCYPS) were used, from the (1) first-grade elementary school panel (childhood; N = 2342), (2) the fourth-grade elementary school panel (early adolescence; N = 2378), and the (3) first-grade junior high school panel (late adolescence; N = 2351). Second-order latent growth curve models provided support that low self-control decreased over time. Findings also partially supported hypothesis 2, as parenting negatively predicted the intercept of low self-control, not the slope. Finally, they supported hypothesis 3, as a significant negative parenting effect predicted the low self-control intercept during both early and late adolescence. The current study contributes to research on the link between positive parenting and low self-control development, tested across three distinct developmental periods or age groups and by studying these questions among Korean youth.
... However, within-person changes in impulsive symptoms are heterogeneous. Some highly impulsive adolescents show relative stability in impulsivity or decrease only slightly during this period, potentially leaving these individuals vulnerable to persistent negative outcomes into their 20's (Burt, Sweeten, & Simons, 2014;Quinn & Harden, 2013). ...
Background Borderline personality disorder (BPD) is associated with altered activity in the prefrontal cortex (PFC) and amygdala, yet no studies have examined fronto-limbic circuitry in borderline adolescents and emerging adults. Here, we examined the contribution of fronto-limbic effective connectivity (EC) to the longitudinal stability of emotion-related impulsivity, a key feature of BPD, in symptomatic adolescents and young adults. Methods We compared resting-state EC in 82 adolescents and emerging adults with and without clinically significant borderline symptoms ( n BPD = 40, ages 13–30). Group-specific directed networks were estimated amongst fronto-limbic nodes including PFC, ventral striatum (VS), central amygdala (CeN), and basolateral amygdala (BLA). We examined the association of directed centrality metrics with initial levels and rates of change in emotion-related impulsivity symptoms over a one-year follow-up using latent growth curve models (LGCMs). Results In controls, ventromedial prefrontal cortex (vmPFC) and dorsal ACC had a directed influence on CeN and VS, respectively. In the BPD group, bilateral BLA had a directed influence on CeN, whereas in the healthy group CeN influenced BLA. LGCMs indicated that emotion-related impulsivity was stable across a one-year follow-up in the BPD group. Further, higher EC of R CeN to other regions in controls was associated with stronger within-person decreases in emotion-related impulsivity. Conclusions Functional inputs from BLA and vmPFC appear to play competing roles in influencing CeN activity. In borderline adolescents and young adults, BLA may predominate over CeN activity, while in controls the ability of CeN to influence BLA activity predicted more rapid reductions in emotion-related impulsivity.
... However, many studies also find changes in self-regulation over time (e.g. Burt et al., 2014;Hay et al., 2018;Höing et al., 2015). Especially in individuals with low self-regulation, more change and less stability of self-regulation is found, while those who start with high levels of self-regulation show a stable pattern, which may indicate a ceiling effect as high self-regulation offers little room for further improvement (Ray et al., 2013). ...
... While self-control increases most rapidly during the first decade of life, it continues to develop-albeit at a slower pace-during subsequent phases of life (Vazsonyi & Jiskrova, 2018). On average, self-control matures or increases across adolescence and young adulthood (Forrest et al., 2019;Shulman et al., 2015;Winfree et al., 2006;Zondervan-Zwijnenburg et al., 2020), but there is substantial heterogeneity in the development of self-control, reflected in diverse trajectories of self-control development in adolescents and young adults (Burt et al., 2014;Forrest et al., 2019). This suggests that self-control is not stable during adolescence and young adulthood and that its development is probably not predetermined. ...
Full-text available
Adolescents’ self-control develops in the context of mental health and family functioning, but it is unclear how the interplay of self-control, mental health, and family functioning unfolds across time within individuals. Separating within-person from between-person effects, random-intercept cross-lagged panel models were applied to adolescents (from ages 11 to 26) from a Dutch cohort ( n = 2228, 51% female). Adolescents with low self-control were likely to have mental health problems and poorly functioning families. Although within-person changes in the study variables were not meaningfully associated in a reciprocal manner, changes in self-control and mental health were concurrently associated. This suggests that besides stable connections between self-control, mental health, and family functioning in adolescence and young adulthood, changes in self-control and mental health are developmentally linked as well.
... The dual systems model of adolescent risk-taking has risen to prominence as a strong psychological framework for understanding engagement in a range of antisocial behaviors (Burt et al., 2014;Ellingson et al., 2019;Forrest et al., 2019;Jonas & Kochanska, 2018;Wojciechowski, 2020a). This framework posits that differential developmental patterns of impulse control and sensation-seeking are predictive of why engagement in risky behaviors tends to peak in adolescence. ...
Full-text available
Stimulant/amphetamine use presents a major public health problem. There is a dearth of research which has studied this behavior from a dual systems model perspective. This study examined the relevance of sensation-seeking and impulse control for predicting stimulant/amphetamine use and tested whether these relationships varied as a function of time. The Pathways to Desistance data were used in analyses, comprising the responses of 1,354 justice-involved youth across 84 months with 11 data points each. Mixed effects modeling was used to examine these relationships. Results indicated that greater sensation-seeking was associated with greater odds of stimulant/amphetamine use. This relationship varied as a linear function of time, with the salience of sensation-seeking for predicting stimulant/amphetamine use declining as participants got older. However, this interaction only reached marginal significance at the p < .09 level. Interventions focused on sensation-seeking may help reduce stimulant/amphetamine use, but effects may be greater for adolescents relative to young adults.
Based on in-depth interviews with 29 active drug robbers (25 male, 4 female) from St. Louis, MO (USA), we explore restraint among people and in circumstances where there should be none. Focusing on greed restraint at the crime’s payoff point (i.e., not taking everything one could when rewards are seized), we identify the decision-making constructs and conceptual pathways by which this happens and discuss their implications for improved specification of the relationship between criminal propensity, self-regulation, and risk sensitivity. We contend that self-centeredness is the one dimension of criminal propensity that is sufficiently receptive to risk sensitivity to make self-regulation possible, and that individuals with low trait self-control can show state self-control when ambiguity aversion and reference point expectations align to sate anomic greed. This refinement offers novel pathways for future study of dual-influence models of crime, and suggests that offender decision-making is best conceptualized as a process that unfolds during crimes rather than a discrete event that precedes them.
Full-text available
Purpose: Research surrounding the intergenerational transmission of self-control has expanded recently. Yet, findings are mixed, and key limitations regarding the inclusion of distinct measures of parental attachment toward children and parenting practices within a longitudinal framework remain. We seek to address these limitations by providing a longitudinal test of serial mediation linking maternal low self-control, maternal attachment toward children, maternal parenting practices, and adolescent low self-control. Methods: We employed structural equation modeling to examine direct and indirect effects between maternal low self-control measured when children were six months-old, maternal attachment toward children measured when target children were seven years old, maternal parenting practices assessed when children were twelve years-old, and adolescent low self-control when children were fifteen years old. Results: Findings partially support assertions from self-control theory in that maternal low self-control was positively related to later adolescent low self-control indirectly via maternal attachment toward children and, in turn, maternal parenting practices. However, direct associations between maternal low self-control and maternal parenting practices and, even more, between maternal low self-control and adolescent low self-control, were also observed. Limitations and implications of our findings are discussed.
Full-text available
The purpose of the present study was to revise the Barratt Impulsiveness Scale Version 10 (BIS-10), identify the factor structure of the items among normals, and compare their scores on the revised form (BIS-11) with psychiatric inpatients and prison inmates. The scale was administered to 412 college undergraduates, 248 psychiatric inpatients, and 73 male prison inmates. Exploratory principal components analysis of the items identified six primary factors and three second-order factors. The three second-order factors were labeled Attentional Impulsiveness, Motor Impulsiveness, and Nonplanning Impulsiveness. Two of the three second-order factors identified in the BIS-11 were consistent with those proposed by Barratt (1985), but no cognitive impulsiveness component was identified per se. The results of the present study suggest that the total score of the BIS-11 is an internally consistent measure of impulsiveness and has potential clinical utility for measuring impulsiveness among selected patient and inmate populations.
1 Introduction.- 2 The Demographic Distribution of Delinquency and ADM Problems.- 3 Prevalence and General Offending/Use Patterns: The Joint Occurrence of Delinquent Behavior and ADM Problems.- 4 Age, Period, and Cohort Effects.- 5 Developmental Patterns.- 6 The Etiology of Delinquency and ADM Problems.- 7 Prediction of Delinquent and ADM Behavior from Other Delinquent and ADM Behavior.- 8 Summary and Implications.- References.- Appendix A Frequency of Alcohol Use.- Appendix B Mental Health Measures.- Appendix C Prevalence and Offending/Use Rates for Multiple Problem Types.- Appendix D Annual Transition Matrices for Problem Behavior Types.- Author Index.
Normative personality change over 40 years was shown in 2 longitudinal cohorts with hierarchical linear modeling of California Psychological Inventory data obtained at multiple times between ages 21-75. Although themes of change and the paucity of differences attributable to gender and cohort largely supported findings of multiethnic cross-sectional samples, the authors also found much quadratic change and much individual variability. The form of quadratic change supported predictions about the influence of period of life and social climate as factors in change over the adult years: Scores on Dominance and Independence peaked in the middle age of both cohorts, and scores on Responsibility were lowest during peak years of the culture of individualism. The idea that personality change is most pronounced before age 30 and then reaches a plateau received no support.