Assessing the Determinants and Stability of Self-Control Into Adulthood

Abstract

The stability hypothesis of Gottfredson and Hirschi’s general theory of crime has received limited support in the literature. However, recent research on the dimensions of self-control suggests varied developmental trends that may conflate the idea of stability. Using a sample of more than 300 individuals from the northeastern United States, this study examines the stability hypothesis across early adulthood and expands the current literature by considering the early childhood risk factors that may distinguish between longitudinal trajectories of self-control. Finally, a reduced, multidimensional version of the behavior problem index for measuring self-control is introduced. Results show evidence of relative stability in three dimensions of self-control, and that sex predicts trajectory group membership. Theoretical implications are discussed.

Keywords

self-control trajectory modeling stability life course longitudinal

Gottfredson and Hirschi (1990) presented their general theory of crime, purported to explain all crimes, at all times, among all persons, and across all cultures. Crime, to Gottfredson and Hirschi, constitutes “acts of force or fraud undertaken in pursuit of self-interest” (p. 14). They argue that most crime is impulsive, requiring little skill or forethought, and provides little monetary gain to the offender or harm to his victim. Therefore, people who engage in crime should exhibit the traits of impulsivity, lack of long-term focus, disregard for the well-being of others, and affinity for simple tasks. Gottfredson and Hirschi predict that criminal offenders will also engage in analogous behaviors that are maladaptive and antisocial, such as smoking and unprotected sex, and will be more likely to divorce or be unemployed.

To Gottfredson and Hirschi, self-control is primarily a function of parental socialization. Parents who can recognize, monitor, and appropriately discipline their child’s misconduct early in life will raise children with adequate self-control. Children whose parents fail to do so will not develop appropriate behavioral restraint, resulting in low self-control. While prominence is given to parents for instilling self-control, alternative sources such as school and churches—those institutions taking the place of parents in their absence—are viewed as capable of facilitating the development of self-control (see Turner, Piquero, & Pratt, 2005).

Borrowing from the earlier work of Patterson (1982), Gottfredson and Hirschi argue that the socialization process takes place in childhood and, therefore, by adolescence self-control is essentially stable. In large part, self-control will be mostly impervious to change beyond childhood. This stability hypothesis of Gottfredson and Hirschi’s theory can be misconceived and should be clarified. The theorists do not argue for absolute stability (individuals can gain or in some cases lose self-control over time), but they do contend for relative stability. Gottfredson and Hirschi (1990) asserted that “people differ in the likelihood that they will commit crimes and that these differences appear early and remain stable over much of the life course” (p. 108). Those with the highest self-control will at all times have higher self-control than their peers and those with the lowest will always have lower self-control than the rest. As discussed below, the research on the stability hypothesis provides considerable support; however, evidence suggests that a sizable group of individuals do in fact make significant gains and losses in self-control relative to their peers across adolescence (e.g., see Hay & Forrest, 2006; Turner & Piquero, 2002).

Research on the Stability Hypothesis

Our understanding of self-control stability has matured alongside advancements in statistical techniques. Arneklev, Cochran, and Gainey (1998) were the first to empirically assess the viability of the stability hypothesis with a sample of college students at three universities. Focusing on individual differences, Arneklev and colleagues showed that most individuals varied only slightly on their self-control rating across the semester. Furthermore, results of a hierarchical linear modeling analysis indicated significant mean differences in self-control but no evidence of within-individual changes in self-control across the semester; taken as support for the stability hypothesis.¹

Subsequent endeavors have attempted to improve upon many of the weaknesses of this initial effort by comparing researcher-induced groupings of delinquent and non-delinquent youth on self-control across a limited time frame. These studies show a modest correlation across waves, regardless of group, but find that delinquent youth exhibit lower self-control relative to non-delinquents at all time periods (Turner & Piquero, 2002; Winfree, Taylor, He, & Esbensen, 2006; Yun & Walsh, 2011). The conclusions drawn from these analyses are limited by the inability to follow youth from early childhood and beyond their teenage years as well as by looking at distinct groups rather than within-individual patterns of change in self-control. Current assessments of self-control stability rely upon group-based trajectory modeling, which permits the parceling of unique subgroups of persons who follow a similar pattern of self-control over time vis-à-vis other subgroups of persons who follow different patterns. These endeavors tend to show less evidence of stability when compared with earlier studies (Burt, Sweeten, & Simons, 2014; Hay & Forrest, 2006; Higgins, Jennings, Tewksbury, & Gibson, 2009; Jennings, Higgins, Akers, Khey, & Dobrow, 2013; Jo & Zhang, 2012; Na & Paternoster, 2012; Ray, Jones, Loughran, & Jennings, 2013). Although absolute stability appears to be the norm across these studies, sizable groups of individuals have been shown to gain or lose self-control over time, relative to their stable counterparts.

An important advancement in this line of research comes from the understanding that the overall self-control concept contains multiple underlying constructs that likely follow unique developmental trajectories. For example, while impulsivity drops off dramatically across childhood risk-seeking behavior follows an adolescence-limited trajectory for many individuals (Burt et al., 2014; Steinberg et al., 2008). Likewise, while stability is the norm in aggressive and hyperactive behavior the research characterizes their development as a sharp decrease across childhood with a sizable minority of individuals exhibiting high levels of aggression and hyperactivity across the life course (Nagin & Tremblay, 1999; Piquero, Carriaga, Diamond, Kazemian, & Farrington, 2012; Tremblay et al., 1999). Continuing to document these distinct trajectories and assessing their causes is of utmost importance to understanding the developmental landscape of a key correlate for criminal behavior.

Much of the research on the determinants and stability of self-control, however, has not considered a wider range of potential determinants (aside from parental socialization) that have been identified across disciplines to also influence human development and behavior. In the next section, specific attention is turned toward the influence of various early childhood risk factors as potential determinants of self-control.

Early Childhood Risk Factors for Low Self-Control

Prenatal and Perinatal Complications Leading to Neurological Deficits

Prenatal complications have been shown to predict chronic criminal behavior (McGloin, Pratt, & Piquero, 2006; Raine, Brennan, & Mednick, 1994; Tibbetts & Piquero, 1999). Beaver and Wright (2005) tested the influence of birth complications, such as premature birth, respiration problems, and meconium aspiration syndrome (the ingestion of feces in utero), on the General Theory’s explanation of self-control development. They found that premature birth and anoxia (restriction of oxygen supply to the fetus) were predictive of composite reports (teacher and parent combined) of low self-control in kindergarten. The effects of these complications were also found in the parental reports of low self-control, while the influence of premature birth was not replicated in the teacher reports. The remaining birth complications had varying effects based upon reporting source. In a recent analysis, Ratchford and Beaver (2009) assessed the influence that birth complications and low birth weight have on an individual’s self-control. They failed to find an effect of these factors on low self-control, although, these results are in need of replication. With regard to the dimensions of self-control, Huijbregts, Sequin, Zoccolillo, Bolvin, and Tremblay (2007) showed prenatal maternal smoking (which has been implicated in delinquent behavior: Gibson, Piquero, & Tibbetts, 2000; McGloin et al., 2006; Pratt, McGloin, & Fearn, 2006) to be predictive of high physical aggression and the dimensions of hyperactivity and impulsivity in childhood. Likewise, minor physical abnormalities at birth have been linked to attention problems, aggression, and impulsivity at age 3 and age 7 (Waldrop, Bell, McLaughlin, & Halverson, 1978).

As noted earlier, Gottfredson and Hirschi largely do not explicate a role for biological processes to influence the development of self-control. However, emerging evidence substantiates the role of neurological deficits in leading to an individual’s lower self-control. Research in cognitive neuroscience has shown that self-control processes are guided by the prefrontal cortex and constitute “executive functions” allowing for purposeful, planned, and goal-oriented behavior (Miller & Cohen, 2001; Posner & Rothbart, 2007). Damage to this area has been linked to disinhibitory behavior (Blumer & Benson, 1975; Damasio, Grabowski, Frank, Galaburda, & Damasio, 1994), as well as attention regulation and planning problems (Barceló & Knight, 2002; Shallice & Burgess, 1991). Furthermore, these deficits have been shown to stem from prenatal health problems, toxin exposure, and delivery complications (Arseneault, Tremblay, Boulerice, & Saucier, 2002; Barkley, 1997; Coles et al., 1997; Grandjean et al., 1997) and have been linked to antisocial behavior in adolescence and adulthood (Arseneault et al., 2002; Cauffman, Steinberg, & Piquero, 2005; Moffitt, Lynam, & Silva, 1994; Raine et al., 1994, 1997).

In regard to Gottfredson and Hirschi’s conceptualization of low self-control specifically, a handful of empirical tests have assessed the influence of neurological deficits on self-control. Beaver and colleagues (2007) used measures of fine and gross motor skills to indicate neurological deficits in the prefrontal cortex and showed that having more deficits in these tasks predicted low self-control in kindergarten independent of parental and neighborhood effects. Ratchford and Beaver (2009) replicated these findings (using verbal test scores as an indicator of neuropsychological deficits) in a sample of youth followed over adolescence and young adulthood. Furthermore, they indicated that birth complications and low birth weight might work indirectly through neuropsychological deficits to influence self-control development. Their findings provide evidence that what Gottfredson and Hirschi assume to be a strictly social process may be better explained in reality as impairments in higher order cognitive functioning.

Poor Early Childhood Social Conditions

The immediate social environment represents Gottfredson and Hirschi’s traditional view of self-control development. In support of their theory, evidence has shown that parenting practices have an influence on low self-control (Cecil, Barker, Jaffee, & Viding, 2012; Pratt, Turner, & Piquero, 2004). Parental punishment (such as spanking and being told they are not loved) has been shown to have a positive effect on low self-control (Ratchford & Beaver, 2009). Meanwhile, Huijbregts and colleagues (2007) also found that a hostile reactive parenting style predicted high levels of aggression, hyperactivity, and impulsivity in childhood. At the same time, evidence suggests that maternal self-control may predict both her child’s self-control and her choice of parenting practices (Nofziger, 2008), and that the parenting effect overall is weak when controlling for genetic effects (Harris, 1998; Wright & Beaver, 2005; Wright, Beaver, DeLisi, & Vaughn, 2008) and prenatal complications (Beaver & Wright, 2005).

Beyond the household, neighborhood disadvantage has been shown to predict levels of self-control independent of neuropsychological deficits and parental socialization (Beaver et al., 2007; Pratt et al., 2004; cf. Gibson, Sullivan, Jones, & Piquero, 2010). The findings on early childhood social and neurological risk factors have strongly supported their relationship to self-control (Turner, Livecchi, Beaver, & Booth, 2011); however, analyses have yet to explore their relationship to developmental trajectories of self-control. Importantly, these early biological and social risk factors seem to interact to produce negative outcomes. Pre- and perinatal risk factors that affect cognitive development coupled with poor social conditions have been found to increase the likelihood of antisocial behavior (Arseneault et al., 2002; Raine et al., 1994; Tibbetts & Piquero, 1999).

Current Study

In sum, growing evidence suggests that tracking stability in self-control masks the various developmental trends of the concept’s underlying constructs. Furthermore, previous presentations of these trajectories have rarely assessed the influence that various risk factors may have on these developmental paths. The present analysis represents an attempt to assess the underlying dimensions of self-control present in the behavior problem index (BPI), analyze stability in early adulthood self-control dimensions, and isolate the influence of childhood risk factors on early adulthood self-control trajectories.

The present study uses group-based trajectory analysis to address two specific research questions concerning the stability of self-control.

Research Question 1: Is relative stability present within the dimensions of self-control represented by the BPI in early adulthood?

Research Question 2: Do early life risk factors predict self-control trajectories across early adulthood?

In doing so, it presents a revised multidimensional measure of the BPI for measuring self-control.

Method

Data

To answer these questions, data for the present study were extracted from the last three waves of the Simmons Longitudinal Study (SLS; Reinherz, 2009). The SLS is a community-based sample of youth entering kindergarten in a single school district in the northeastern United States. The communities served by this school district were predominantly Caucasian, working-class households. Wave 1 was collected in 1977 just prior to the youth entering their kindergarten year at 5 years of age and Wave 2 was completed at the end of the youth’s kindergarten year. In third grade, when the youth averaged 9 years of age, Wave 3 of data collection took place. Waves 4 and 5 were collected when the youth were in high school: at 15 and 18 years of age on average. Waves 6 and 7 followed youths into adulthood, measuring a variety of life events and aspirations, psychosocial functioning and problem behavior at ages 21 and 26.

The original SLS sample included 777 youth entering kindergarten in 1977. Most attrition occurred at Wave 3 when approximately one third of youth transferred to private schools. Of the remaining 519 youth in Wave 3, 403 were followed in subsequent waves. Analysis conducted by the SLS research team suggested that the post-attrition sample remained representative of the community from which it was drawn and did not vary in any substantive way from non-participants on factors predictive of life outcomes (the original purpose of the study; Reinherz, 2009). For inclusion in the present analyses, youth must have had complete self-control data on five of six waves (as per Hay & Forrest, 2006). In the end, this allowed for the analysis of 349 respondents from 18 to 26 years of age. Table 1 compares the analysis sample with the 54 excluded cases on key covariates and shows that there were no significant differences between retained and excluded observations.

Table 1:

Comparison of Final Sample to Excluded Cases

Variables	Retained^a	Excluded^b	t test	χ²
Variables	M (SD)	M (SD)	t test	χ²
Prenatal health index	0.43 (0.77)	0.30 (0.55)	1.14
Neonatal health index	0.20 (0.49)	0.29 (0.76)	−1.09
Low birth weight	0.06 (0.24)	0.12 (0.33)		2.27
Mother’s age	31.80 (5.52)	33.09 (4.86)	−1.62
Single-parent home	0.14 (0.35)	0.05 (0.23)		3.15
Family SES	2.57 (1.04)	2.42 (0.99)	1.02
School SES	1.94 (0.85)	1.87 (0.89)	0.53
Gender	0.50 (0.50)	0.52 (0.50)		.04
Race	0.98 (0.14)	0.97 (0.17)		.08
Self-control Wave 1	−0.02 (0.99)	0.24 (1.00)	−1.33
Self-control Wave 3	−0.01 (0.99)	0.09 (1.03)	−0.44
Self-control Wave 4	0.01 (1.02)	−0.12 (0.86)	0.82
Self-control Wave 5	0.01 (0.99)	0.05 (1.09)	−0.31
Self-control Wave 6	−0.01 (0.99)	0.14 (1.09)	−0.73
Self-control Wave 7	−0.00 (1.00)	0.00 (1.05)	−0.01

Note. SES = socioeconomic status.

n = 349.

n = 54.

Measures

Self-Control

In this study, self-control is measured through behavioral indicators as Gottfredson and Hirschi have expressed preference for behavioral over attitudinal measures of their concept (Hirschi & Gottfredson, 1993). In this study, youth reports of self-control are used in Waves 5, 6, and 7 (when respondents were 18, 21, and 26 years of age). The indicators of self-control are measured via the BPI, which were derived from the lengthy Child Behavior Checklist (Achenbach, 1978; Peterson & Zill, 1986) and has been used in previous literature to measure self-control (Hay & Forrest, 2006; Turner & Piquero, 2002).

In the present study, the BPI was subjected to an exploratory factor analysis to assess the dimensionality of the indicators present. Analysis of eigenvalues, residuals, and factor loadings suggested preference for five factors from the 19 indicators used in past research. Three items (“often confused or in a fog,” “nervous/tense,” and “overly fearful/anxious”) loaded heavily on a single factor. Although included in previous analyses as a measure of self-control, these three items were excluded from the present endeavor as they appeared to more accurately measure anxiety than self-control. Likewise, questions regarding the destruction of one’s own and others’ property and rule violations were omitted due to concerns that these measure outcomes of self-control rather than the phenomenon itself. Three final items measuring obsessive thoughts, lying/cheating, and not feeling guilty for misbehavior failed to strongly load on any given factor and bear little resemblance to Gottfredson and Hirschi’s concept of self-control (the latter perhaps more reflective of psychopathy than self-control).

Factor scores were estimated at each wave to allow individual indicators that more strongly influenced the measure of self-control to weigh more heavily on an individual’s value. Three factors remained with each item listed in Table 2 clearly loading on one factor score.² Furthermore, the same factors with largely similar factor loadings were realized across all three waves of data. These factors are named for the dimensions they appear to represent: impulsivity, aggression, and interpersonal difficulties. These measures and factor loadings by wave are presented in Table 2.

Table 2:

Eigenvalues for Indicators of Self-Control at Each Wave

	Wave 5	Wave 6	Wave 7
Impulsivity
Impulsive or act without thinking	0.52	0.49	0.75
Can’t concentrate, can’t pay attention for long	0.79	0.79	0.70
Cannot sit still, restless, or hyperactive	0.78	0.85	0.68
Interpersonal difficulties
Cruelty, bullying, or meanness to others	0.44	0.57	0.54
Don’t get along well with others	0.72	0.76	0.70
Not liked by others	0.65	0.91	0.83
Aggression
Have a hot temper	0.74	0.92	0.51
Argue a lot	0.62	0.73	0.78
Stubborn, sullen, or irritable	0.78	0.67	0.67

Note. Eigenvalues for each factor were above the standard 1.00 cutoff, and residuals analysis suggested the presence of multiple factors rather than one single factor.

Prenatal and Perinatal Problems

The secondary aim of this study was to assess the impact of early childhood risk factors on trajectory membership. The first of such risk factors is birth complications. Following Raine et al. (1994), the present analyses use indicators of pregnancy and delivery complications outlined by Baker and Mednick (1984) as important in long-term life outcomes. Specifically, the prenatal health index contained within the SLS will be used as a predictor of self-control trajectory membership. The specific items for the index can be found in Table 3, including items such as excessive bleeding during pregnancy, difficult delivery, and other pregnancy complications. The prenatal health index is a summated scale with a theoretical maximum value of 6.³ In addition, an indicator measuring low birth weight (weighing less than 5.5 pounds at birth—shown to have significant consequences for early onset offending; Tibbetts & Piquero, 1999) will be assessed for influence on trajectory membership.

Table 3:

Prenatal, Perinatal, and Early Childhood Risk Factors

Risk factors	M	SD
Prenatal health index (range = 0-4)	0.39	0.71
Bleeding (0 = no; 1 = yes)	0.09	0.28
Toxemia (0 = no; 1 = yes)	0.03	0.17
Difficult delivery (0 = no; 1 = yes)	0.10	0.31
Other pregnancy complications (0 = no; 1 = yes)	0.08	0.27
Other labor/delivery problems (0 = no; 1 = yes)	0.09	0.29
Low birth weight (0 = more than 5.5 lbs; 1 = less than 5.5 lbs)	0.07	0.25
Early childhood social conditions
Single-parent home (0 = no; 1 = yes)	0.15	0.33
Family SES (range = 1-5; 5 = low SES)	3.47	1.04
School SES (range = 1-3; 3 = low SES)	1.94	0.85
Sex (0 = female; 1 = male)	0.50	0.50

Note. SES = socioeconomic status.

Poor Early Childhood Social Conditions

This study will also assess the importance of early childhood social conditions in predicting trajectory group membership. Namely, family socioeconomic status (SES), school SES, and single-parent status will be included in these models. Family SES and school SES are measured on a 5-point scale where higher values indicate lower SES. SES is measured by the Hollingshead index of parental education and occupation. This two-factor index combines information on the highest level of education attained and the level of skill involved in an individual’s job to arrive at his or her SES (Hollingshead & Redlich, 1958). Single-parent status is a dichotomous variable taking a value of 1 if the mother or father is not present in the youth’s home. Individuals raised in a single-parent home have been shown to be at increased risk of maladjustment, including delinquency (Amato & Keith, 1991; McLanahan & Sandefur, 1994; Stouthamer-Loeber & Loeber, 1986). Finally, sex of the respondent will be included in the analyses (0 = female, 1 = male) along with parental report of self-control at age 15 (also using the BPI).

Analytical Plan

Question 1: Stability in Self-Control

Group-based trajectory modeling, specifically growth mixture modeling⁴ (which allows for both between- and within-group variation; see Muthén & Muthén, 2000), will be used to assess the number of distinct trajectories of the dimensions of self-control present in the data. Group-based trajectory modeling represents an improvement over the delineation of prospectively defined groups (as used in Arneklev et al., 1998; Turner & Piquero, 2002) as it uses statistical procedures to determine groups, thus permitting the identification of distinct groups following unique trajectories of a specific behavior. It also allows for the discovery of unpredicted groups and provides a method for calculating the probability of group membership (Nagin, 2005). While generally considered an improvement in modeling developmental pathways, decisions on the number of groups to include are still relatively subjective (although, fit statistics are available to guide this process). Furthermore, in many applications (including the current one) the trajectories are based on latent constructs and “groups should not be thought of as literally distinct entities” (Nagin, 2010, p. 59).

First, growth mixture models will be estimated with only one group, and fit statistics (Baysian Information Criterion [BIC], entropy, posterior probabilities, visualization plots, and class proportions) will be assessed. From there, a two-group model will be estimated and fit statistics compared with the one-group model for improvement. This process will continue until the addition of groups results in weaker models as indicated holistically by comparison of fit statistics across the models. The end result will be three final models (impulsivity, aggression, and interpersonal difficulties) with the number of varying trajectories best indicated by the data.

The trajectories modeled here occur during a period of the life course when Gottfredson and Hirschi argue that relative stability in self-control should be present. As such, one would expect rank ordering in these dimensions to be realized in the data. For example, as there should be “little to no movement from high self-control to low self-control” (Gottfredson & Hirschi, 1990, p. 107) the group highest in self-control at one wave should be the group highest in self-control at every other wave of data. If this is not the case, this is evidence against the assumption of relative stability in self-control.

Question 2: Influence of Early Childhood Risk Factors

From here, the prenatal and early childhood risk factors⁵ presented in Table 3 will be included in a logistic regression model predicting membership in one group over another. As the analyses will reveal, a large group exhibiting relatively few issues (i.e., high, stable self-control) across all three dimensions was present; therefore, a variable was created to capture those individuals who were classified into all three high, stable trajectory groups. This was a dichotomous variable where 1 = high, stable self-control and 0 = all others. Given prior evidence that these risk factors are associated with self-control deficits, one would expect these variables to be predictive of membership into the reference category (i.e., a significant odds ratio below 1) over the high self-control group.

Results

Question 1: Stability of Self-Control

Tables 4 through 6 report the statistics associated with models estimating trajectories of self-control.⁶ To determine the appropriate number of groups represented in the data for each dimension, BIC, entropy, class proportions, and posterior probabilities are taken into consideration. The best model shows a reduction in BIC without a sizable drop in entropy value, class proportions for all classes containing a meaningful number of individuals, and posterior probabilities remaining above the .70 cutoff as suggested by Nagin (2005).

Table 4:

Growth Mixture Model: Impulsivity

Classes	BIC	Entropy	Class proportions	Posterior probabilities
1	2,850.893		1. 100%
2	2,807.235	0.802	1. 0.24069	1. 0.917
			2. 0.75931	2. 0.949
3	2,806.494	0.796	1. 0.75358	1. 0.947
			2. 0.14040	2. 0.825
			3. 0.10602	3. 0.778
4	2,808.843	0.774	1. 0.69914	1. 0.908
			2. 0.11461	2. 0.781
			3. 0.13754	3. 0.834
			4. 0.04871	4. 0.802

Note: BIC = Baysian Information Criterion.

Table 5:

Growth Mixture Model: Aggression

Classes	BIC	Entropy	Class proportions	Posterior probabilities
1	2,889.046		1. 100%
2	2,763.266	0.934	1. 0.89929	1. 0.986
			2. 0.10071	2. 0.957
3	2,692.354	0.925	1. 0.06268	1. 0.979
			2. 0.76004	2. 0.971
			3. 0.17727	3. 0.934
4	2,668.246	0.931	1. 0.03152	1. 0.961
			2. 0.05731	2. 0.879
			3. 0.17765	3. 0.929
			4. 0.73352	4. 0.974
5	2,645.896	0.934	1. 0.68768	1. 0.971
			2. 0.02579	2. 0.987
			3. 0.12321	3. 0.930
			4. 0.11748	4. 0.927
			5. 0.04585	5. 0.967

Note: BIC = Baysian Information Criterion.

Table 6:

Growth Mixture Model: Interpersonal Issues

Classes	BIC	Entropy	Class proportions	Posterior probabilities
1	2,891.729		1. 100%
2	2,726.241	0.963	1. 0.93696	1. 0.993
			2. 0.06304	2. 0.942
3	2,657.904	0.937	1. 0.03438	1. 0.979
			2. 0.14327	2. 0.901
			3. 0.82235	3. 0.984
4	2,613.914	0.937	1. 0.09169	1. 0.909
			2. 0.02292	2. 0.996
			3. 0.81375	3. 0.977
			4. 0.07163	4. 0.915
5	2,558.604	0.946	1. 0.01433	1. 0.993
			2. 0.10315	2. 0.928
			3. 0.07163	3. 0.953
			4. 0.02292	4. 0.995
			5. 0.78797	5. 0.973

Note: BIC = Baysian Information Criterion.

In these data, a two-group impulsivity model and three-group aggression and interpersonal problems models were deemed the best fit. It is important to note that individuals are placed into the group that best reflects their observed trajectories (i.e., for which they have the highest posterior probability of membership). Therefore, the estimated trajectory line will not constitute a perfect fit for all or even most individuals but constitutes the average trajectory for a group of individuals exhibiting a similar trend over time.

The relative (rank order) stability espoused by Gottfredson and Hirschi was largely present in these early adulthood data.⁷ Impulsivity (see Figure 1) was represented by two groups: one with low, stable impulsivity from the age of 18 to 26 and another group with higher levels of impulsivity. The high impulsivity group, 24% of the sample, was characterized by a roughly 1 standard deviation rise and fall in impulsivity from age 18 to 26. Overall, it appeared that one group, representing a quarter of youth in these data, exhibited relatively high impulsivity in early adulthood than another, larger group who at each time point reported little, if any impulsivity.

Figure 1:

Trajectories of Impulsivity

A three-group model best characterized the aggression dimension (see Figure 2). One group (76% of the sample) was reflective of the low impulsivity group, in that their aggression was, overall, lower compared with their counterparts across waves. Their trajectory was characterized by stable, low aggression that reduced slightly to the age of 26. This trajectory profile aligns well with the predictions of Gottfredson and Hirschi who expect that “socialization continues to occur throughout life,” thus resulting in incremental gains in self-control over time (p. 107). The other two groups were characterized by increases in aggression across early adulthood at varying rates. Six percent of the sample was characterized by a trajectory of sharply increasing aggression, while around 18% of the sample was estimated to follow a steadier, low grade increase in aggression over time.

Figure 2:

Trajectories of Aggression

Finally, the three-group model best fit the data on interpersonal difficulties (see Figure 3). As was the case for impulsivity and aggression, a large proportion (82%) of individuals were characterized to have little, if any interpersonal difficulties—at all times estimated to have a low, stable trajectory on this dimension. Two other groups fell in rank order, reporting an increase in interpersonal difficulties at age 21 that diminishes by age 26. Three percent of the sample showed relatively high interpersonal difficulties (at age 21, beyond 3 standard deviations above the mean) than the rest of the sample, while about 14% of individuals exhibited slightly elevated difficulties with interpersonal relations across time.

Figure 3:

Trajectories of Interpersonal Difficulties

Question 2: Influence of Early Childhood Risk Factors

For these analyses, a binary variable was created to capture those individuals who exhibited high self-control across all three dimensions (i.e., were characterized by low, stable levels of impulsivity, aggression, and interpersonal difficulties; 1 = high, stable self-control, 0 = all others). Fifty-three percent of the sample fell into this high, stable self-control category. Logistic regression was used to predict membership into the high self-control group. As seen in Table 7, only one risk factor predicted high self-control group membership. Prenatal health problems, low birth weight, single-parent household, family and neighborhood SES failed to distinguish the high, stable self-control group. Sex, however, significantly predicted high self-control membership. Females were 2 times as likely to be characterized by high self-control across all dimensions than were males (p < .01). A prior measure of high self-control taken from parents at Wave 4 was not significantly related to high self-control here, indicating either instability or (more likely) a lack of congruence between parental and respondent reports of the construct.

Table 7:

Logistic Regression of High Self-Control on Early Childhood Risk Factors

Variables^a	Odds ratio	95% CI
Prenatal health problems	1.08	[0.87, 1.33]
Low birth weight	0.69	[0.28, 1.70]
Sex	0.49**	[0.32, 0.77]
Single-parent home	0.85	[0.43, 1.67]
Low parent SES	1.16	[0.91, 1.47]
Low school SES	1.16	[0.92, 1.46]
Self-control age 15^b	1.26	[0.77, 2.06]
R ²	.04
χ²	15.46*
N	332

Note. CI = confidence interval; SES = socioeconomic status.

All variables were standardized prior to inclusion in the model.

Self-control at age 15 measured by parent report.

p < .05. **p < .01.

Discussion

This study investigated Gottfredson and Hirschi’s (1990) stability hypothesis across early adulthood using a reduced, multidimensional version of the BPI culled from longitudinal data on a sizable sample of individuals in the Northeast. Furthermore, given the presence of varying trajectory profiles for three dimensions of self-control, analyses of the predictors of trajectory group membership were conducted. In all, the stability hypothesis received considerable support. More than half of the sample exhibited absolute stability in each of the dimensions of self-control—at all times reporting relatively less issues than their counterparts in the areas of impulsivity, aggression, and interpersonal relations. Relative stability was also largely present in the remaining trajectories; although marked gains and/or losses in all three dimensions were modeled in these data. It appears that while a large number of individuals may establish strong self-control early in life and maintain it into adulthood, for many, self-control is a dynamic characteristic susceptible to influence (both positively and negatively) across early adulthood.

These results fall in line with past stability research in uncovering varying trajectory profiles based upon the dimensions of self-control. Subjecting the BPI to factor analysis uncovered a number of underlying dimensions that both align with the self-control concept (impulsivity, aggression, interpersonal difficulties) and others that should possibly be excluded when using the BPI as a measure of self-control in future endeavors (such as anxiety and non-criminal deviance). These findings will need to be replicated with a larger dataset to further explore the multidimensionality of the BPI. Validating evidence for the use of the reduced BPI as a measure of self-control was found in a post hoc analysis of maladaptive behavior at age 26 (see Table 8). Consistent with theoretical expectations, those individuals who were characterized as having high, stable self-control in early adulthood across all three dimensions of self-control reported significantly less lifetime drug use, alcohol problems, and arrest than did the rest of the sample.

Table 8:

Comparison of Groups on Maladaptive Behaviors (Wave 7)

	High, stable self-control^a (%)	Reference group^b (%)	χ²
Drug use	41	57	8.78**
Drinking problem	17	25	9.93**
Arrest	40	58	8.39**

n = 186.

n = 163.

**p < .01.

Sex was shown to be a robust predictor of self-control stability with males being significantly less likely to be characterized by high, stable self-control in early adulthood. This finding falls in line with the theoretical understanding of gender and self-control supported by numerous cross-sectional/short-term studies (e.g., Hay & Forrest, 2006; Pratt et al., 2004; Wright & Beaver, 2005). They may run counter to recent longitudinal findings, however, that do not show marked gender differences in developmental trajectories (Jo & Bouffard, 2014; Ray et al., 2013). Many of the early childhood risk factors uncovered in the literature were not significant predictors of self-control stability in these data. It is important to note, however, that many of the other risk factors (viz., prenatal issues and low birth weight) were quite rare occurrences in this dataset, which may account for their lack of influence here. This limitation, along with others discussed next, should be addressed in future research.

Limitations

These conclusions must be tempered due to the presence of some data constraints. First, these data come from a few hundred (mostly Caucasian) youth in a single school district in the Northeast United States. This data collection was not intended to achieve national representativeness nor should the findings presented here be necessarily considered as such. The generalizability of these results beyond this sample is therefore limited and in need of replication. Furthermore, the attrition at Wave 3 may have influenced the representativeness of this sample, as a sizable proportion of youth were transferred to private school at this time. Arguably, these more affluent youth may have had higher, more stable self-control that could have influenced the trajectories presented here. These findings must be considered in light of this possibility.

In addition, these data are limited by the inability to account for heritability. Emerging evidence on the importance of genetic factors in the expression of self-control discussed above suggests that a substantial portion of the development of self-control may be due to heritable traits. Including such measures in the models predicting group membership would greatly improve the explanatory power of these analyses. Finally, in a similar vein, direct measures of parental socialization were unfortunately not available in the data. Given the theoretical importance of this concept, it would be ideal for parental socialization to be included in predicting trajectory group membership with other datasets containing these measures.

Conclusion

Gottfredson and Hirschi’s concept of self-control has been shown to largely overlap with personality traits such as conscientiousness and agreeableness (Ahadi & Rothbart, 1994; O’Gorman & Baxter, 2002; Tobin, Graziano, & Vanman, 2000), which (along with numerous personality traits) can exhibit changes well into adulthood (although consistency in personality strengthens with age; Roberts & DelVecchio, 2000).⁸ It follows that this phenomenon likely exhibits a fair amount of elasticity for some individuals well into the life course. An important caveat for future research will be to investigate which events across the life course influence these changes in self-control.

The finding that a sizable proportion of individuals are characterized by instability in early adulthood could be explained by expectations in personality psychology that time periods of novel experiences illicit the most change (Glenn, 1980; Tyler & Schuller, 1991). Clearly, the age range included here—from the end of high school, across college, and the first few years in the workforce—is ripe with fresh experiences that may lead to changes in self-control however permanent or temporary. Changes in marital, employment, and substance use status have been suggested as potential catalysts to the sorts of developmental pathways exhibited here (Hay & Forrest, 2006). Early life factors did not seem to shape the stability of self-control in adulthood, yet it remains possible that the significance of sex on self-control might manifest in these later life events. Future analyses should use growth curve modeling, for example, to consult the host of life events that may serve to bolster or diminish self-control across the life course.

Finally, the multidimensional self-control measure presented here may serve as a guide for future stability inquiries as it used objective criteria to parcel out those indicators that are strongly related to—or outcomes of—Gottfredson and Hirschi’s self-control. The reduced number of indicators provided in the present measures may better reflect the concept than the entirety of the BPI used in previous analyses—an assertion worthy of future inquiry.

Footnotes

Appendix

Notes

Brie Diamond is an assistant professor of criminal justice at Texas Christian University. Her research focuses on criminological theory and corrections and can be found in outlets such as Crime and Delinquency, Journal of Criminal Justice, Intelligence, and Justice Quarterly.

References

Achenbach

T. M.

(1978). The child behavior profile: I. boys aged 6-11. Journal of Consulting and Clinical Psychology, 46, 478-488.

Ahadi

Rothbart

M. K.

(1994). Temperament, development and the big five. In Halverson

C. F.

Kohnstamm

Martin

(Eds.), Development of the structure of temperament and personality from infancy to adulthood (pp. 189-208). Hillsdale, NJ: Lawrence Erlbaum.

Amato

P. R.

Keith

(1991). Parental divorce and the well-being of children: A meta-analysis. Psychological Bulletin, 110, 26-46.

Arneklev

B. J.

Cochran

J. K.

Gainey

R. R.

(1998). Testing Gottfredson and Hirschi’s “low self-control” stability hypothesis: An exploratory study. American Journal of Criminal Justice, 23, 107-127.

Arseneault

Tremblay

R. E.

Boulerice

Saucier

J. F.

(2002). Obstetrical complications and violent delinquency: Testing two developmental pathways. Child Development, 73, 496-508.

Baker

R. L.

Mednick

B. R.

(1984). Influences on development. Boston, MA: Kluwer.

Barceló

Knight

R. T.

(2002). Both random and preservative errors underlie WCST deficits in prefrontal patients. Neuropsychologia, 40, 349-356.

Barkley

R. A.

(1997). Behavioral inhibition, sustained attention, and executive functions: Constructing a unifying theory of ADHD. Psychological Bulletin, 121, 65-94.

Beaver

K. M.

Wright

J. P.

(2005). Evaluating the effects of birth complications on low self-control in a sample of twins. International Journal of Offender Therapy and Comparative Criminology, 49, 450-471.

10.

Beaver

K. M.

Wright

J. P.

DeLisi

(2007). Self-control as an executive function: Reformulating Gottfredson and Hirschi’s parental socialization thesis. Criminal Justice and Behavior, 34, 1345-1361.

11.

Blumer

Benson

D. F.

(1975). Personality changes with frontal and temporal lesions. In Benson

D. F.

Blumer

(Eds.), Psychiatric aspects of neurologic disease (pp. 151-169). New York, NY: Grune & Stratton.

12.

Burt

C. H.

Sweeten

Simons

R. L.

(2014). Self-control through emerging adulthood: Instability, multidimensionality, and criminological significance. Criminology, 52, 450-487.

13.

Cauffman

Steinberg

Piquero

A. R.

(2005). Psychological, neuropsychological and physiological correlates of serious antisocial behavior in adolescence: The role of self-control. Criminology, 43, 133-176.

14.

Cecil

C. A. M.

Barker

E. D.

Jaffee

S. R.

Viding

(2012). Association between maladaptive parenting and child self-control over time: Cross-lagged study using a monozygotic twin difference design. The British Journal of Psychiatry, 201, 291-297.

15.

Coles

C. D.

Platzman

K. A.

Raskind-Hood

C. L.

Brown

R. T.

Falek

Smith

I. E.

(1997). Comparison of children affected by prenatal alcohol exposure and attention deficit, hyperactivity disorder. Alcoholism: Clinical & Experimental Research, 21, 150-161.

16.

Damasio

Grabowski

Frank

Galaburda

A. M.

Damasio

A. R.

(1994). The return of Phineas Gage: Clues about the brain from the skull of a famous patient. Science, 264, 1102-1105.

17.

Gibson

C. L.

Piquero

A. R.

Tibbetts

S. G.

(2000). Assessing the relationship between maternal cigarette smoking during pregnancy and age at first police contact. Justice Quarterly, 17, 519-542.

18.

Gibson

C. L.

Sullivan

C. J.

Jones

Piquero

A. R.

(2010). “Does it take a village?” Assessing neighborhood influences on children’s self-control. Journal of Research in Crime & Delinquency, 47, 31-62.

19.

Glenn

(1980). Values, attitudes, and beliefs. In Brim

O. G.

Jr. Kagan

(Eds.), Constancy and change in human development (pp. 596-640). Cambridge, MA: Harvard University Press.

20.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

21.

Grandjean

Weihe

White

R. F.

Debes

Araki

Yokoyama

. . . Jorgensen

P. J.

(1997). Cognitive deficit in 7-year-old children with prenatal exposure to methylmercury. Neurotoxicology and Teratology, 19, 417-428.

22.

Harris

J. R.

(1998). The nurture assumption: Why children turn out the way they do. New York, NY: Free Press.

23.

Hay

Forrest

(2006). The development of self-control: Examining self-control theory’s stability thesis. Criminology, 44, 739-774.

24.

Higgins

G. E.

Jennings

W. G.

Tewksbury

Gibson

C. L.

(2009). Exploring the link between low self-control and violent victimization trajectories in adolescents. Criminal Justice and Behavior, 36, 1070-1084.

25.

Hirschi

Gottfredson

M. R.

(1993). Commentary: Testing the general theory of crime. Journal of Research in Crime & Delinquency, 30, 47-54.

26.

Hollingshead

A. G.

Redlich

F. C.

(1958). Social class and mental illness (Vol. 49). New York, NY: Wiley.

27.

Huijbregts

S. C.

Sequin

J. R.

Zoccolillo

Bolvin

Tremblay

R. E.

(2007). Associations of maternal prenatal smoking with early childhood physical aggression, hyperactivity-impulsivity, and their co-occurrence. Journal of Abnormal Child Psychology, 35, 203-215.

28.

Jennings

W. G.

Higgins

G. E.

Akers

R. L.

Khey

D. N.

Dobrow

(2013). Examining the influence of delinquent peer association on the stability of self-control in late childhood and early adolescence: Toward an integrated theoretical model. Deviant Behavior, 34, 407-422.

29.

Bouffard

(2014). Stability of self-control and gender. Journal of Criminal Justice, 42, 356-365.

30.

Zhang

(2012). The stability of self-control: A group-based approach. Asian Journal of Criminology, 7, 173-191.

31.

McGloin

J. M.

Pratt

T. C.

Piquero

A. R.

(2006). A life-course analysis of the criminogenic effects of maternal cigarette smoking during pregnancy: A research note on the mediating impact of neuropsychological deficit. Journal of Research in Crime & Delinquency, 43, 412-426.

32.

McLanahan

Sandefur

G. D.

(1994). Growing up with a single parent: What hurts, what helps. Cambridge, MA: Harvard University Press.

33.

Miller

E. K.

Cohen

J. D.

(2001). An integrative theory of prefrontal cortex function. Annual Review of Neuroscience, 24, 167-202.

34.

Moffitt

T. E.

Lynam

D. R.

Silva

P. A.

(1994). Neuropsychological tests predicting persistent male delinquency. Criminology, 32, 277-300.

35.

Muthén

L. K.

(2000). Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism: Clinical & Experimental Research, 24, 882-891.

36.

Paternoster

(2012). Can self-control change substantially over time? Rethinking the relationship between self- and social control. Criminology, 50, 427-462.

37.

Nagin

D. S.

(2005). Group-based modeling of development. Cambridge, MA: Harvard University Press.

38.

Nagin

D. S.

(2010). Group-based trajectory modeling: An overview. In Piquero

A. R.

Weisburd

(Eds.), Handbook of quantitative criminology (pp. 53-68). New York, NY: Springer.

39.

Nagin

Tremblay

R. E.

(1999). Trajectories of boys’ physical aggression, opposition, and hyperactivity on the path to physically violent and nonviolent juvenile delinquency. Child Development, 70, 1181-1196.

40.

Nofziger

(2008). The “cause” of low self-control. Journal of Research in Crime & Delinquency, 45, 191-224.

41.

O’Gorman

J. G.

Baxter

(2002). Self-control as a personality measure. Personality and Individual Differences, 32, 533-539.

42.

Patterson

G. R.

(1982). Coercive family process. Eugene, OR: Castalia.

43.

Peterson

J. L.

Zill

(1986). Marital disruption, parent-child relationships, and behavior problems in children. Journal of Marriage and the Family, 48, 295-307.

44.

Piquero

A. R.

Carriaga

M. L.

Diamond

Kazemian

Farrington

D. P.

(2012). Stability in aggression revisited. Aggression and Violent Behavior, 17, 365-372.

45.

Posner

M. I.

Rothbart

M. K.

(2007). Research on attention networks as a model for the integration of psychological science. Annual Review of Psychology, 58, 1-23.

46.

Pratt

T. C.

McGloin

J. M.

Fearn

N. E.

(2006). Maternal cigarette smoking during pregnancy and criminal/deviant behavior: A meta-analysis. International Journal of Offender Therapy and Comparative Criminology, 50, 672-690.

47.

Pratt

T. C.

Turner

M. G.

Piquero

A. R.

(2004). Parental socialization and community context: A longitudinal analysis of the structural sources of low self-control. Journal of Research in Crime & Delinquency, 41, 219-243.

48.

Raine

Brennan

Mednick

S. A.

(1994). Birth complications combined with early maternal rejection at age 1 year predispose to violent crime at age 18 years. Archives of General Psychiatry, 51, 984-988.

49.

Raine

Brennan

Mednick

S. A.

(1997). Interaction between birth complications and early maternal rejection in predisposing individuals to adult violence: Specificity to serious, early-onset violence. American Journal of Psychiatry, 154, 1265-1271.

50.

Ratchford

Beaver

K. M.

(2009). Neuropsychological deficits, low self-control, and delinquent involvement: Toward a biosocial explanation of delinquency. Criminal Justice and Behavior, 36, 147-162.

51.

Ray

J. V.

Jones

Loughran

T. A.

Jennings

W. G.

(2013). Testing the stability of self-control: Identifying unique developmental patterns and associated risk factors. Criminal Justice and Behavior, 40, 588-607.

52.

Reinherz

H. Z.

(2009). Simmons Longitudinal Study: Adaptation and development across the lifespan [Data Set]. doi:10.3886/ICPSR22800.v1

53.

Roberts

B. W.

DelVecchio

W. F.

(2000). Rank order consistency of personality traits from childhood to old age: A quantitative review of longitudinal studies. Psychological Bulletin, 126, 3-25.

54.

Shallice

Burgess

P. W.

(1991). Deficits in strategy application following frontal lobe damage in man. Brain, 114, 727-741.

55.

Steinberg

Albert

Cauffman

Banich

Graham

Woolard

(2008). Age differences in sensation seeking and impulsivity as indexed by behavior and self-report: Evidence for a dual systems model. Developmental Psychology, 44, 1764-1778.

56.

Stouthamer-Loeber

Loeber

(1986). Boys who lie. Journal of Abnormal Child Psychology, 14, 551-564.

57.

Tibbetts

S. G.

Piquero

A. R.

(1999). The influence of gender, low birth weight, and disadvantaged environment in predicting early onset of offending: A test of Moffitt’s interactional hypothesis. Criminology, 37, 843-878.

58.

Tobin

R. M.

Graziano

W. G.

Vanman

E. J.

(2000). Personality, emotional experience, and efforts to control emotions. Journal of Personality and Social Psychology, 79, 656-669.

59.

Tremblay

R. E.

Japel

Perusse

McDuff

Boivin

Zoccolillo

Montplaisir

(1999). The search for the age of “onset” of physical aggression: Rousseau and Bandura revisited. Criminal Behavior and Mental Health, 9, 8-23.

60.

Turner

M. G.

Livecchi

C. M.

Beaver

K. M.

Booth

(2011). Moving beyond the socialization hypothesis: An empirical investigation of the sources of self-control. Journal of Criminal Justice, 39, 120-127.

61.

Turner

M. G.

Piquero

A. R.

(2002). The stability of self-control. Journal of Criminal Justice, 30, 457-471.

62.

Turner

M. G.

Piquero

A. R.

Pratt

T. C.

(2005). The school context as a source of self-control. Journal of Criminal Justice, 33, 327-339.

63.

Tyler

T. R.

Schuller

R. A.

(1991). Aging and attitude change. Journal of Personality and Social Psychology, 61, 689-697.

64.

Waldrop

M. F.

Bell

R. Q.

McLaughlin

Halverson

C. F.

(1978). Newborn minor physical anomalies predict short attention span, peer aggression, and impulsivity at age 3. Science, 199, 563-565.

65.

Winfree

L. T.

Taylor

T. J.

Esbensen

(2006). Self-control and variability over time: Multivariate results using a 5-year, multisite panel of youths. Crime & Delinquency, 52, 253-286.

66.

Wright

J. P.

Beaver

K. M.

(2005). Do parents matter in creating self-control in their children? A genetically informed test of Gottfredson and Hirschi’s theory of low self-control. Criminology, 43, 1169-1202.

67.

Wright

J. P.

Beaver

K. M.

DeLisi

Vaughn

(2008). Evidence of negligible parenting influences on self-control, delinquent peers, and delinquency in a sample of twins. Justice Quarterly, 25, 544-569.

68.

Yun

Walsh

(2011). The stability of self-control among South Korean adolescents. International Journal of Offender Therapy and Comparative Criminology, 55, 445-459.