Positive Childhood Experiences and Positive Adult Functioning: Prosocial Continuity and the Role of Adolescent Substance Use

Abstract

Purpose

To examine positive childhood experiences as predictors of positive adult functioning, including civic involvement, productivity and responsibility, interpersonal connection, and physical exercise; and to examine adolescent substance use as a mediator of prosocial continuity.

Methods

Four hundred and twenty-nine rural participants were interviewed across 7 waves from age 11 to 22. Structural equation models examined the relationship between positive childhood experiences and adult functioning, with adolescent substance use added to each model as a possible mediating mechanism.

Results

Positive childhood experiences predicted significantly better adult functioning for each model, even after accounting for adolescent substance use. Positive childhood experiences also consistently predicted significantly less adolescent substance use. In turn, adolescent substance use predicted significantly less civic involvement and less productivity and responsibility, but was not associated with interpersonal connection or physical exercise when accounting for childhood experiences. Results were largely consistent across gender and levels of family income.

Conclusions

Findings show the enduring importance of positive childhood experiences in predicting positive functioning in early adulthood. Although adolescent substance use increased risk for poorer functioning in important domains of adult life, results suggest that positive experiences in late childhood continued to have a significant prosocial effect into young adulthood. The study also highlights the late elementary grades as a time when parents, teachers, and others can potentially have a large influence in proactively providing prosocial opportunities for children.

The transition from adolescence to early adulthood is a challenging time for many young people. While avoiding problem behaviors remains important during this period [1, 2], a number of recent studies have stressed the importance of positive development and prosocial outcomes [3, 4]. Research has shown that positive functioning is conceptually and empirically distinct from problem behaviors and not simply the “opposite of” or absence of problems [3, 5, 6]. Most recently, an in-depth report from the National Research Council and Institute of Medicine [5] has called for a broad conceptual shift in the prevention of emotional and behavioral disorders to include the promotion of competencies and healthy functioning. The link between positive experiences in childhood and positive functioning in early adulthood, and the possible role of adolescent substance use as a mediator of this link, is the focus of the present investigation.

We examine four domains of positive functioning that have been identified as important indicators of healthy young adult development: civic involvement [3, 7], productivity and responsibility [3, 8, 9], interpersonal connection [3, 10, 11], and physical exercise [12]. In childhood, we identify experiences in family, school, and other social settings that correspond to these adult outcomes, such as doing volunteer work or sports with a parent, studying hard at school, and spending time with prosocial friends. In general, prior research suggests moderately positive associations between corresponding childhood and adult outcomes [13]. Few studies, however, have examined the continuity of positive functioning from childhood to adulthood. One exception is the work of Eisenberg and her colleagues, who found significant associations between a prosocial orientation in childhood and a similar orientation in early adulthood. Positive childhood experiences appeared to contribute to this consistency [14, 15].

An important consideration is the possible mediating influence of problem behaviors in adolescence. Social developmental theory [16, 17] proposes that problem behaviors such as substance use that are normative during the teen years [18, 19] may have enduring consequences if youth are diverted to an antisocial “pathway” that reinforces further health-risk behavior. Alternatively, positive childhood experiences may promote a prosocial developmental pathway that reinforces positive behavior and helps to keep adolescents from diverting to an antisocial pathway despite experimentation with risky behaviors [16, 17]. This study considers the role of adolescent substance use in prosocial continuity given its relatively high prevalence and potential for escalation to abuse and dependence. We examine the extent to which substance use during adolescence serves as an antisocial pathway which significantly mediates the link between positive childhood experiences and positive adult functioning within diverse domains of development.

There has been extensive research on the effects of aversive childhood experiences (e.g., family conflict, social maladjustment) on adolescent substance use, as well as on the effects of adolescent substance use on adult problem outcomes [20, 21]. In contrast, relatively few longitudinal studies have focused specifically on the role of positive childhood experiences in protecting against adolescent substance use, or on possible substance use effects during adolescence on later positive outcomes. Notable exceptions include findings of negative consequences of adolescent substance use on young adult autonomy, perceived competence, and involvement in positive activities (sports, volunteer work, etc.) [22], as well as on college attendance and completion, job stability, and a general measure of young adult wellbeing [20, 23, 24]. This study extends these findings by looking at additional positive adult outcomes, and by examining the role of adolescent substance use in the context of positive childhood experiences.

Possible moderating effects on the role of adolescent substance use are also considered. Research suggests that the consequences of substance use may be less serious for males, or for middle or upper income individuals versus those from poverty, perhaps because they are less subject to social stigmas regarding substance use [25, 26]. We also have an opportunity in this study to examine the effects of a preventive drug-use intervention found previously to weaken the link between substance use in early adolescence and subsequent progression of use [27]. This study includes examination of gender, income, and intervention status as possible moderators.

The study was conducted with a rural sample well suited to this investigation. The rural settings provided prosocial community groups for children (e.g., 4-H, scouts), relatively small school and class sizes with opportunities for constructive class participation, a high proportion of intact dual-parent families that can increase opportunities for parent-child connection [28, 29], and, for some, open spaces and farming responsibilities that encourage physical activity. Yet, rates of tobacco and alcohol use in rural communities are often high [30] and in this study were similar to or somewhat higher than corresponding national rates [19].

This study examines three questions, applied to four different outcomes: civic involvement, productivity and responsibility, interpersonal connection, and physical exercise. First, what is the relationship between positive childhood experiences and positive adult functioning with respect to these outcomes? Based on related developmental research [13–15], we hypothesize moderate but significant positive associations. Second, to what extent does adolescent substance use mediate the link between positive childhood experiences and positive adult functioning? Previous studies suggest that family, school, and other social experiences in childhood are likely to be predictive of substance use, which in turn is expected to adversely affect positive outcomes, thus mediating prosocial continuity [20–24]. Third, potential moderating factors are considered. Do these relationships differ with respect to gender, family income, or intervention status? Research suggests less severe consequences of substance use for males, for those from middle or upper income families, and for those assigned to preventive intervention [25–27].

Method

Sample

In 1993, the study was initiated among families of sixth graders enrolled in rural schools in a Midwestern state. Schools were in districts which served small communities (population less than 8,500) and in which 15% or more of families were eligible for a school lunch program for families near or below the federal poverty level. Twenty-two schools were available for study.¹ Of all families with sixth-grade children in these schools (N = 883), 49% (429) completed the initial survey. This initial recruitment rate is comparable to rates reported for similar studies [31]. A prospective participation factor survey was conducted with a 90% response rate providing support for the representativeness of the sample. Examining a range of family characteristics, only parent education was found to be significantly associated with participation, with participating parents reporting 0.7 years more education on average than nonparticipating parents [32, 33].

When the study began, target children averaged 11.35 years of age; 52% were female; and 98% were White. Families had an average of 2.99 children, 85% of families were dual-parent, and mothers and fathers averaged 36.91 and 39.57 years of age, respectively. Median annual household income was $32,000 (compared with the national median for married-couple households in 1993 of $43,129), and 56% of mothers and 52% of fathers reported having some post-high school education.

Since the initial survey the sample has participated in six waves of follow-up, approximately at ages 12, 13, 14, 16, 18, and 22. The retention rate at age 22 was 73% (N = 313), and retention rates averaged 73% in the intervening waves. Extensive attrition analyses have been conducted comparing assessment dropouts versus completers across a range of sociodemographic, psychosocial, and outcome variables. Other than parent education, no differences in overall attrition have been observed [32, 34]. At each assessment, informed consent of participants and approval of the Human Subjects Review Committee at Iowa State University was obtained.

Following the initial assessment, half of the sample was included in a 5-week preventive intervention, Preparing for the Drug Free Years (PDFY; now called Guiding Good Choices). PDFY is a family competency training program to reduce children’s risk of early substance initiation by enhancing protective parent-child interactions and children’s peer resistance skills. The intervention has shown positive long-term effects on targeted outcomes, including reduced substance use (for further description, see [27, 32]).

Measures

With the exception of physical exercise, latent variables were modeled for each construct in the analyses, with three manifest indicators for each latent construct. The mean of standardized self-report items was calculated for most indicators, with exceptions noted below. Higher scores correspond to more of the behavior consistent with the construct label. Wording of items used in analyses is available from the first author.

Civic involvement

At age 22, the three indicators of the civic involvement latent construct consisted of (a) an item assessing involvement in community groups, (b) an item assessing time spent volunteering in the past year [3], and (c) a three-item Mokken scale [35] assessing political involvement (registered to vote, voting, attendance at meetings or rallies). Positive childhood experiences were assessed with a corresponding latent construct for civic involvement at age 11, indicated by (a) involvement with a parent in community groups (scouts, 4-H, etc.), (b) volunteering in the community with a parent, and (c) discussing news or current events with a parent (all single-item indicators).

Productivity and responsibility

Adult productivity and responsibility was indicated by (a) constructive engagement in school or work (summing the average time spent in class, homework, and paid employment in the past year) [3], (b) a summative index of success in school and/or work (self-reported grades, job performance and retention), and (c) responsibility (seven items assessing completion of assignments, effort, attendance, and financial prudence; α = .65). A corresponding latent construct at age 11 was indicated by (a) school engagement (four items assessing participation, ease of learning, studying, and class confidence; α = .73), (b) grades, and (c) an item about degree of planning for future goals.

Interpersonal connection

These constructs were intended to represent the closest relationships across different domains, rather than the quantity of relationships, by using the highest connection score within each respective domain of relationships. The first indicator at age 22 (a) used the greater of the mean prosocial connection with close friends or with romantic partner or spouse (e.g., happy, satisfied, depend on relationship) [3]. In order to code this measure for positive interpersonal connection it was adjusted for problem behavior attributed to friends or partner/spouse such that more problem behavior (including violence, crime, and substance abuse) attenuated the connection score; that is, the connection scores were multiplied by the inverse of the friends’ or partner’s problem behavior score. This method ensured that these connection scores were low to the extent friends or partners engaged in antisocial behavior [36]. Six items assessed connection with friends, and three items assessed connection with partner (mean α = .77). Other adult indicators included (b) connection with mother or father, taking the greater mean score (seven items each: e.g., feel close, share thoughts, loyal; mean α = .88), and (c) connection with other students or with prosocial coworkers, taking the greater score (three items each: like other students/coworkers, spend time together, talk often; mean α = .64).

Corresponding indicators at age 11 included (a) connection with friends (five items; α = .81), (b) connection with parent (28 items each for mother and father, mostly concerning involvement with the parent in specific activities; mean α = .94), and (c) connection with others (taking the greater of belonging at school or closeness to at least one teacher), each adjusted for friends’, parents’, or others’ problem behavior.

Physical exercise

Two items assessing frequency and intensity of exercise at age 22 were combined multiplicatively into a manifest outcome variable measuring physical exercise. One child-report item was available that assessed active involvement with a parent in sports or exercise and was included as a corresponding measure of physical activity at age 11.

Adolescent substance use

Our focus was on overall use across all substances assessed at different points in adolescence. We sought to capture this use in a parsimonious summary scale to provide a relatively straightforward test of mediation across multiple models of prosocial continuity. Indicators were created by taking the mean logged and standardized frequency of substance use reported at ages 13 through 18, including (a) tobacco use (α = .77), (b) alcohol use (α = .69), and (c) marijuana and other drug use (α = .85). Log transformation reduced skewness while preserving continuous frequency distributions, and standardization weighted items in the scales more equally without giving too much weight to certain ages or to more commonly used substances. During adolescence, 58% of the sample reported using tobacco, 79% reported using alcohol, and 21% reported using other drugs.

Analyses

A schematic of the analysis model is shown in Figure 1. First, prosocial continuity was examined in four separate models predicting each domain of positive adult functioning from the corresponding measure of positive childhood experiences. Next, adolescent substance use was added to each model as a possible mediating mechanism. Analyses included all participants with baseline data (N = 429) using Mplus [37] with maximum likelihood estimation to compute full-information maximum likelihood (FIML) estimates with incomplete data. Prior analyses have found very few between-school differences, indicating the appropriateness of family-level analysis [32].

Figure 1.

Analysis model. First, prosocial continuity was examined in four separate models predicting each domain of positive adult functioning from the corresponding measure of positive childhood experiences. Next, adolescent substance use was added to each model as a possible mediating mechanism.

Results

As described above, we sought to create latent constructs reflecting a priori domains of positive functioning identified in the literature from indicator variables assessing specific facets of life (school, work, parents, peers, etc.). This approach resulted in consistently significant factor loadings (all p < .01) for each of the four models described above, and significant correlations among the indicators of each construct (see Table 1). Model fit ranged from a CFI of .900 to .987, and an RMSEA of .067 to .029. Though adequate, intra-factor correlations and model fit were of moderate magnitude in some cases due to the need for consistency in approach across models and with prior studies.

Table 1.

Correlations among Indicator Variables Used in Analyses

Construct (age)	Civ. inv. (11)		Civ. inv. (22)			Sub. use (13–18)			Construct (age)	Pro./res. (11)		Pro./res. (22)			Sub. use (13–18)
	(b)	(c)	(a)	(b)	(c)	(a)	(b)	(c)		(b)	(c)	(a)	(b)	(c)	(a)	(b)	(c)
Civ. inv. (11)									Pro./res. (11)
(a)a	.40*	.33*	.16*	.14*	.21*	−.11*	−.12*	−.04	(a)	.46*	.27*	.14*	.15*	.23*	−.16*	−.07	−.24*
(b)		.36*	.09	.11*	.12*	−.09	−.14*	−.06	(b)		.14*	.24*	.13*	.07	−.21*	−.01	−.12*
(c)			.25*	.19*	.24*	−.07	−.04	−.01	(c)			.04	.05	.05	−.12*	−.07	−.06
Civ. inv. (22)									Pro./res. (22)
(a)				.42*	.24*	−.11	−.13*	−.07	(a)				.18*	.13*	−.15*	.07	.01
(b)					.30*	−.21*	−.15*	−.06	(b)					.41*	−.17*	−.14*	−.27*
(c)						−.12*	−.13*	−.15*	(c)						−.10	−.19*	−.29*
	Int. con. (11)		Int. con. (22)			Sub. use (13–18)						Phys. exe. (22)			Sub. use (13–18)
Construct (age)	(b)	(c)	(a)	(b)	(c)	(a)	(b)	(c)	Construct (age)						(a)	(b)	(c)
Int. con. (11)									Phys. exe. (11)				.21*		−.12*	−.09	−.11*
(a)	.28*	.31*	.17*	.18*	.09	−.22*	−.08	−.28*	Phys. exe. (22)						−.11	−.04	−.18*
(b)		.16*	.10	.34*	.24*	−.17*	−.15*	−.12*
(c)			−.01	.13*	.02	−.19*	−.11*	−.25*
Int. con. (22)									Sub. use (13–18)
(a)				.14*	.13*	−.09	−.12*	−.08	(a)							.53*	.42*
(b)					.27*	−.10	−.03	−.12*	(b)								.39*
(c)						−.12	−.11	−.06

Note. civ. inv. = civic involvement; pro./res. = productivity and responsibility; int. con. = interpersonal connection; phys. exe. = physical exercise; sub. = substance.

^*p < .05.

^aLetters correspond to indicators described in the text.

To address our first research question, the first step in analyses of structural relationships was to examine the direct path from positive childhood experiences to positive adult functioning (before adding adolescent substance use to the model), shown in the first column of coefficients in Table 2. This path was positive and significant for each model.

Table 2.

Effects of Adolescent Substance Use on the Relationship between Positive Childhood Experiences and Positive Adult Functioning within Each Domain

Model: Domain of positive functioning	Model path coefficients a (95% confidence interval)				Model fit b
	Step 1	Step 2			χ2(df)	CFI	RMSEA
	Childhood experiences → Adult functioning (w/o adol. sub. use)	Childhood experiences → Adult functioning (w/adol. sub. use)	Childhood experiences → Adolescent sub. use.	Adolescent sub. use → Adult functioning
Civic involvement	.47*** (.30 to .63)	.41*** (.23 to .58)	−.20* (−.36 to −.05)	−.23* (−.41 to −.06)	32.59 (24)	.982	.029
Productivity and responsibility	.35* (.18 to .51)	.26* (.09 to .44)	−.26** (−.42 to −.11)	−.35* (−.53 to −.17)	70.50***(24)	.900	.067
Interpersonal connection	.55* (.31 to .78)	.55* (.24 to .85)	−.49*** (−.65 to −.33)	−.03 (−.32 to .26)	45.54** (23)	.940	.048
Physical exercise	.22*** (.11 to .32)	.20*** (.09 to .30)	−.16* (−.28 to −.03)	−.12 (−.26 to .02)	6.75 (4)	.987	.040

Note. adol. = adolescent; sub. = substance.

^*p < .05;

^**p < .01;

^***p < .001.

^aEntries are standardized path coefficients with 95% confidence intervals in parentheses.

^bModel fit indices are based on the final models including adolescent substance use. N = 429 for all models.

The next column of Table 2 shows that, in Step 2, these direct, longitudinal paths from positive childhood experiences to positive adult functioning remained significant for each model even after accounting for the possible effects of adolescent substance use. In most cases, the magnitude of the path coefficient was diminished only minimally.

Analyses for each model also found a significant path from positive childhood experiences predicting less adolescent substance use (third column in Table 2). In turn, the path from adolescent substance use to adult functioning varied in significance (fourth column). Civic involvement and productivity and responsibility in adulthood were significantly related to less substance use in adolescence. However, neither interpersonal connection nor physical exercise in adulthood were significantly associated with adolescent substance use after taking into account related childhood experiences. In line with these results, the indirect effect of childhood experiences on adult functioning was significant for civic involvement and productivity and responsibility (β = .05 and .09, respectively, both p < .05), but not for interpersonal connection or physical exercise (β = .02 and .02, respectively, both p > .15). Notably, all adult outcomes were significantly correlated at the zero-order level with adolescent substance use (r’s were −.32, −.42, −.30, and −.15, for the correlation of substance use with civic involvement, productivity and responsibility, interpersonal connection, and physical exercise, respectively, all p < .05).

Gender, Family Income, and Intervention

Possible differences in the relationships shown in Table 2 by gender, family income, and intervention status were examined for each of the four models by constraining parameter estimates to be equal across groups and comparing fit to parallel unconstrained models [37, 38]. These analyses revealed no significant differences by gender, and only one significant difference in one model when comparing below-median-income families (< $32,000 in 1993) with median- or above-median-income families: Childhood interpersonal connection was related to significantly less adolescent substance use for those from relatively low-income families (B = −.75, SE = .20, p < .001), but not for those from median or higher income families (B = −.18, SE = .13, p = .17) (the decrement in fit for the constrained model was Δχ²(4) = 13.50, p < .01). In comparing intervention and control groups, two differences were found: Adolescent substance use was associated with significantly less adult productivity and responsibility among controls (B = −.34, SE = .13, p < .01), but not among intervention participants (B = −.10, SE = .06, p < .08) (Δχ²(3) = 9.22, p < .05), whereas childhood interpersonal connection was significantly related to less adolescent substance use among intervention participants (B = −.95, SE = .23, p < .001), but not among controls (B = −.24, SE = .14, p < .08) (Δχ²(4) = 12.99, p < .05). Overall, only 3 of 36 structural paths examined were significantly different across groups. Additionally, when household income or intervention status was added to the models to control for their effects, the pattern of results shown in Table 2 was unchanged.

Discussion

Results stress the enduring importance of positive experiences in childhood through adolescence and into early adulthood. The findings are consistent with the research of Eisenberg and her colleagues [14, 15] on the childhood origins of stable prosocial dispositions. Eisenberg et al. [15] found greater stability for early prosocial behaviors that were more other-oriented (e.g., spontaneous sharing), compared with those that appeared less altruistic (e.g., compliant helping). Similarly, although we examined substantially different constructs from Eisenberg, our social measures of positive functioning (civic involvement and interpersonal connection) demonstrated greater bivariate stability than less explicitly social measures (productivity and responsibility and physical exercise). The magnitude of the prosocial continuity coefficients also were similar to those found by Eisenberg and colleagues [14, 15], with correlations mostly in the .35 to .55 range.

Notably, findings suggest that childhood experiences may have an enduring influence into early adulthood across multiple domains of positive functioning even in the presence of potentially problematic substance use in adolescence. Adolescent substance use mediated some of the continuity in civic involvement and productivity and responsibility, but not in interpersonal connection and physical exercise, and all direct paths from childhood experiences to adult functioning remained significant with adolescent substance use in the models.

The differential effects of adolescent substance use across models suggest that there may be important qualitative differences in the outcomes examined. Behaviors such as community involvement or productive engagement in school or work may be more easily derailed by substance use in adolescence, while more intimate interpersonal connections or individual exercise habits may be less vulnerable to adolescent substance use if childhood experiences in these domains generally promote a prosocial developmental pathway [16, 17]. One possible explanation for this is that opportunities for involvement in conventional community or school activities may be diminished by adolescent substance use, reducing later positive opportunities in these domains, while other interpersonal or individual opportunities are less affected.

Although research suggests female gender and poverty may increase risks associated with the use of substances in adolescence [25, 26], there was little evidence here for consistent gender or income differences in the models. Two moderating effects of the preventive intervention in childhood were found, suggesting that the intervention increased the protective effects of positive interpersonal connection on adolescent substance use, and decreased the negative effects of substance use on adult productivity and responsibility.

Limitations of the study include an ethnically homogeneous, rural sample, possibly restricting the generalizability of the findings if positive experiences or adolescent substance use have different effects in different settings. The study also was limited to self-reported data, with the possibility of differential underreporting or overreporting of behaviors. However, research has indicated that self-report surveys administered privately and confidentially, as in this study, provide reliable and valid data [39]. Finally, due to the need for consistency in approach across models and with prior studies, some fit indices and intra-factor correlations were adequate but moderate in magnitude. Further development of measures of positive development and prosocial outcomes may be important foci for future studies.

The findings of this study are timely given the latest recommendations of the National Research Council and Institute of Medicine [5] stressing the promotion of positive development as a key component of public health programs. While much research has focused on reducing developmental risk factors, this study bolsters the promise of programs and policies that target the promotion of prosocial involvement of children in families, school, and communities. Findings suggest that such positive childhood experiences are not only likely to be protective against adolescent substance use, but may also promote healthy, engaged, and productive functioning into young adulthood. Results also highlight the late elementary grades as a time when parents, teachers, and others can potentially have a large influence in proactively providing prosocial opportunities for children. Compared with earlier or later ages, late childhood is likely to be an especially opportune time developmentally to implement universal family interventions through the schools because nearly all children in the U.S. are attending elementary school and parents remain relatively involved in their socialization [40].