Introduction

The study of character has been prevalent among philosophers and educators for centuries (e.g., Aristotle, trans. 1985; Confucius, trans. 1999). Within developmental science, the empirical study of character has burgeoned in the last two decades (e.g., Berkowitz 2011; Lapsley and Narvaez 2006; Lerner and Callina 2014; Narvaez 2008; Nucci and Narvaez 2008). Within this literature, character has been conceptualized as reflecting virtuous or positive attributes of human functioning, for example, reflecting Aristotelian (1985) moral virtues or attributes reflecting or contributing to human thriving through productive performance of tasks, positive civic engagement, or intellectual functioning (Baehr 2013; Lickona and Davidson 2005; Seider 2012; Templeton 1995). That is, although not ignoring that people may show negative or no instantiations of the attributes associated with character, scholars contributing to this literature have focused on the positive instantiations of this construct and conceptualized character as involving interpersonal relations that reflect “a public system of universal concerns about human welfare, justice, and rights that all rational people would want others to adhere to” (Berkowitz 2012, p. 249). Similarly, Nucci (2001) conceives of character as a positive construct that contributes to human welfare, justice, and rights.

However, the research conducted in regard to character has been done mainly with adults and has focused on the content of character rather than on its development (Lerner and Callina 2014). That is, the process through which specific character attributes develop, especially as this development may occur in ecologically representative contexts, has not received enough research attention. Accordingly, in the present research we focus on the development of character in childhood and early adolescence, and focus on youth enrolled in a major out-of-school-time program, that is, the Cub Scouts program of the Boy Scouts of America. Out-of-school-time programs, such as those of the Boy Scouts of America, represent a major setting within the ecology of contemporary youth, and participation in such youth development programs is linked to the development of attributes of positive functioning, such as character (Bowers et al. 2015; Lerner et al. 2015; Vandell et al. 2015). Because of the focus on the Boy Scouts of America program participants, we have used a measure of the character attributes (Wang et al. 2015) that the leaders of the Boy Scouts of America seek to promote in youth involved in their programs, that is, cheerfulness, helpfulness, hopeful future expectations, kindness, obedience, religious reverence, thriftiness, and trustworthiness (Boy Scouts of America 2010). In short, and as we explain in more detail below, the present research focuses on the development of character, assesses such development across ages relatively unstudied in regard to this development, and appraises such development in relation to an ecologically prototypic context of contemporary young people that should be linked to character development.

Here, however, we should note that these research foci are consistent with discussions within developmental science about the nature of the character development process (e.g., Berkowitz 2011; Damon 1988, 2008; Lapsley and Narvaez 2006; Lerner and Callina 2014; Lickona and Davidson 2005; Sokol et al. 2010), which have involved increasing attention to the relational developmental systems metatheory as a framework potentially useful for conceptualizing the study of character development (e.g., Overton 2015). For instance, Lerner and Callina (2014) discuss character development as multidimensional and dynamic, and manifested in a specific set of mutually beneficial relations between individuals and their interpersonal and sociocultural contexts. They suggest that, as such relations vary across ontogenetic time and contextual location, there is a potential to promote positive character attributes among individuals and societies through theory-predicated program design and implementation.

Given that millions of American youth participate in out-of-school-time youth development programs each year, we have noted that such programs may provide a unique context in which to promote positive developmental outcomes related to character (Vandell et al. 2015). Indeed, the mission of major out-of-school-time youth development programs involves developing character virtues in young people (e.g., Eccles and Gootman 2002; Sherrod et al. 2010; Vandell et al. 2015). As described in relational developmental systems-based theories, the potential for youth thriving may be optimized when individual strengths are aligned with ecological developmental assets, such as those present in out-of-school-time youth development programs (e.g., positive adult mentors; Lerner et al. 2015). For instance, youth highly engaged in out-of-school-time youth development programs, such as Boy Scouts of America, who are nested within highly engaging program environments, experience the most positive developmental outcomes (e.g., Lynch et al. 2015). Thus, when characteristics of the out-of-school-time youth development program context align with the strengths of youth participating in the activity, positive developmental outcomes, including the development of character attributes, may be enhanced.

However, only a few studies have focused on capturing the developmental trajectories of specific character attributes or, consistent with a relational developmental systems-based focus on individual-context relations, on examining how ecological assets (e.g., represented by out-of-school-time youth development programs; Eccles and Gootman 2002; Vandell et al. 2015) in youth environments might contribute to the development of character attributes in specific groups of youth (Lapsley and Narvaez 2006; Lerner and Callina 2014). Some recent research on character development among youth from elementary and middle school has aligned character measures with key features of youth contexts (e.g., specific character themes of an out-of-school-time youth development program), but such work is still preliminary and little longitudinal data exist to allow for discussion of change in character developmental processes (e.g., Grier 2012; Wang et al. 2015).

For instance, the majority of character development studies among these age groups have involved adolescent, rather than child, samples (Bowers et al. 2011; Callina et al. 2014; Carlo et al. 2007; Hilliard et al. 2013; Nantel-Vivier et al. 2009; Stoddard et al. 2011). Of the smaller body of research focusing on children, the primary aims were to investigate the development of prosocial thinking and behaviors (e.g., helpfulness) (Cote et al. 2002; Jackson and Tisak 2001; Kokko et al. 2006) and not character development per se. In these studies, youth responded more generally to character-based scenarios (e.g., sharing, comforting a peer), rather than providing assessments of self-perceptions of the degree to which they possessed specific character attributes at different ages.

In turn, the Positive Action program, a school-based social-emotional and character development intervention program for low-income, ethnic minority youth from Grades 3 to 8 in urban Chicago public schools, underwent a longitudinal evaluation of character development among youth (Ji et al. 2005; Kaminski et al. 2009; Lewis et al. 2015). Character attributes reflected in the Positive Action curriculum, such as altruism, empathy, and honesty were measured. Over 6 years, a decline in self-reported positive attributes for all participating youth was observed; however, students in the Positive Action intervention group reported significantly less decline compared to youth in the control groups, which suggests a favorable developmental pattern for youth involved in character-development programs (Lewis et al. 2015). Declines in self-ratings of positive attributes with age has also been reported in various studies of youth prosocial behaviors, skills, and motivations (Carlo et al. 2007; Jozsa et al. 2014; Kokko et al. 2006; Nantel-Vivier et al. 2009).

Whereas these results provide knowledge about the character developmental process in middle childhood to adolescence, they do so only in regard to a structured school context. However, little to no such research has examined character attributes in the context of out-of-school-time youth development programs during childhood, an equally important context in which many youth are engaged (Vandell et al. 2015). Therefore, because empirical research examining the developmental trajectories of character attributes during early portions of life among youth involved in out-of-school-time youth development program is more limited, the longitudinal investigation of character attributes during childhood and early adolescence (e.g., ages 6–12 years) is needed to advance the understanding of character development across the ecology of youth during the first decade and early second decade of life.

Out-of-school-time youth development programs represent key contexts for development during childhood (for review, see Feldman and Matjasko 2007). Moreover, involvement in out-of-school-time youth development programs, such as the ones offered by the Boy Scouts of America, has also been associated with a host of positive academic, psychological, and social outcomes (American Institutes for Research 2013; Vandell et al. 2015; Vinson et al. 2013). Such involvement is also linked to the development of character attributes, such as helping behaviors, trustworthiness, kindness, and perseverance (Hilliard et al. 2014).

The latter, Hilliard et al. (2014) findings are derived from the analysis of a study of the Boy Scouts of America program participants, and a portion of this same data set is used in the present study. The explicit aim of the Boy Scouts of America programs is to imbue youth with the character attributes and life skills needed to thrive personally, and to develop into adults of character and responsible citizens who contribute positively to American democracy (Boy Scouts of America 2010). The Boy Scouts of America programs seek to adopt a developmental perspective by creating age-appropriate activities for boys across grade levels. From Grades 1 to 5, boys participate in Cub Scouting programs. Activities are mainly planned and led by adult volunteers, who are usually parents of the Scouts. Cub Scouts is considered to be a preparation stage for young boys to gradually develop character and skills valued by the organization (e.g., the character attributes noted earlier in this article), such that, Cub Scouts will have gained sufficient autonomy in goal-seeking abilities and leadership skills to allow them to succeed in the upper Boy Scout ranks (ideally, to attain the level of Eagle Scout) and, outside of Scouting, to live lives of honor and to make contributions to their local and global communities. The Cub Scout Promise and Law reflect the Boy Scouts of America’s emphasis on personal development and acquisition of character virtues, including helping behaviors, rule-following, and doing the right thing. These core values are incorporated into Cub Scouting activities and are taught through real life experiences, such as nature-based activities (e.g., camping), community service projects, and weekly meetings emphasizing different character virtues.

The Hilliard et al. (2014) study reported that, at initial assessment, the attributes of character reported by Scouts and non-Scout boys did not differ generally. As such, selection effects between Scouts and non-Scout youth did not appear salient in these data. This comparability set the stage for an assessment of character development among program participants and non-participants.

The Current Study

As part of the Character and Merit Project (Hilliard et al. 2014)—a larger, longitudinal and mixed-methods investigation of character development among youth who were either associated or not associated with character-building programs of the Boy Scouts of America—the present research examined the developmental trajectories of several character attributes across two and a half years among Scouts and youth who were not involved in Scouting. Youth involved in the Boy Scouts of America beginning-level Cub Scouting programs (generally aged between 6 and 11 years old) were compared with youth who were not Cub Scouts but were from the same geographical region and in Grades 1–5. Such research may be useful for providing longitudinal evidence about the nature and course of character development. This research may also help clarify how out-of-school-time youth development programs aimed at character development, such as those of the Boy Scouts of America, might contribute to such development in childhood and early adolescent portions of ontogeny, wherein there may be the potential for considerable growth in positive features of development (Ferris et al. 2015; Vandell et al. 2015).

The present study examined if character attributes of focal concern to the Boy Scouts of America may develop differently among Scouts and non-Scout youth over the course of childhood and early adolescence. Based on prior research (e.g., Lewis et al. 2015), we expected younger youth to have higher self-ratings of character attributes than older youth. In addition, we predicted that youth participating in Cub Scouts would show more favorable developmental patterns of character attributes (e.g., less decline or greater increase) compared to non-Scouts.

Methods

Full details of the methodology of the Character and Merit Project have been presented in prior reports (e.g., Hilliard et al. 2014; Wang et al. 2015). Accordingly, we present only those features of methodology pertinent to the present article.

Participants

Participants included 1398 Scouts (Wave 1: M age  = 8.59, SD = 1.29, Range 6.17–11.92) and 325 non-Scout boys (Wave 1: M age  = 9.06, SD = 1.43, Range 6.20–11.81). All youth were recruited from the same region in the greater Philadelphia metropolitan area. Scouts were recruited from regularly-scheduled Scouting meetings held in local community locations, including churches, community centers, or schools. Non-Scout boys were recruited from Catholic schools, charter schools, and public schools in the same area to approximate the different ecological settings of our Scout sample. In the present study, 76.1 % of Scouts and 32.0 % of non-Scout boys held meetings, or attended school, respectively, at religious-based institutions. In addition, 82 % of Scouts were White or European American, 9.2 % were Black or African American, 2.9 % were Hispanic or Latino, 2.6 % were Asian or Pacific Islander, 2.2 % were other race/ethnicity, and 1.1 % identified as Multiethnic or Multiracial. Non-Scout boys were predominantly White or European American (63.8 %), and the remainder were 18.4 % Black or African American, 8.2 % Hispanic or Latino, 6.0 % Multiethnic or Multiracial, 2.5 % Asian or Pacific Islander, and 1.1 % American Indian.Footnote 1

Measures

The Assessment of Character in Children and Early Adolescents (ACCEA; Wang et al. 2015) was used to measure eight character attributes which are of focal concern to the Boy Scouts of America and are emphasized in the developmental literature (e.g., Hilliard et al. 2014). This assessment has good construct and concurrent validity, and previous research has identified ACCEA as having measurement invariance among Scouts, and non-Scout boys (and girls) (Wang et al. 2015). In addition, participants reported on their intentional self-regulation skills and school performance, both of which are important positive attributes for youth to develop during this period of life. An overview of the ACCEA’s eight subscales, and the measure of intentional self-regulation and school performance are presented below. All items were completed using a five-point Likert scale with response options ranging from 1 (Not at all like me) to 5 (Exactly like me). Mean-level composite scores were created by averaging participants’ responses across scale items for each character attribute examined; higher scores indicate greater embodiment of character attributes.

Obedience

Participants complete three items (α: .71–.74 across the five waves), including “I act the way I am supposed to”, “I do what my teachers say”, and “I do what my parents say”.

Religious Reverence

Four items (α: .76–.82), including “I like to read or listen to stories from my religion”, and “I pray” were completed by participants.

Cheerfulness

Three items (α: .78–.83) composed this subscale, including “I am happy”, “I smile a lot”, and “I am cheerful”.

Kindness

Participants completed four items (α: .80–.86), including “I’m kind to other kids” and “When my friends are upset, I try to make them feel better”.

Thriftiness

Four items (α: .60–.63), including “I save my money for something special” and “I find ways to reuse things” were completed by participants.

Hopeful Future Expectations

Participants were prompted, “Think about your future. What will your life be like when you grow up?”, and completed three items (α: .71–.73), including “I will be healthy”, “I will have a happy family”, and “People will think I am a good person”, which assessed hopeful future expectations.

Trustworthiness

Participants completed five items (α: .79–.82), including “I can be counted on to tell the truth” and “I take responsibility when I make a mistake”.

Helpfulness

Five items (α: .78–.82), including “I help people in my family”, and “I help my friends” were completed by participants.

Intentional Self-Regulation

Items were adapted from the selection, optimization, and compensation measure developed by Freund and Baltes (2002) to be more developmentally appropriate for young children. Six items, including “I am good at making plans” and “I am a hard worker” were completed by participants (α: .67–.72).

School Performance

Six items (α: .69–.71) were adapted from the academic competence subscale of the Self-perception Profile for Children (e.g., SPPC; Harter 1982) to assess youth school performance. Sample items include “I am very good at my schoolwork” and “I get good grades in school”.

Procedure

The present study used a cohort-sequential, longitudinal design whereby participants were recruited, consented, and surveyed in the first wave, and new participants were recruited at each subsequent wave of data collection (i.e., Waves 2–5). Attempts were made to follow all participants longitudinally across the five waves of data collected approximately every 6 months during the two and a half year study period. Survey data were collected from participating youth by trained Scouting leaders, school staff, or by members of our research team. A standard protocol was used to ensure that data collection was administered uniformly and that all study materials were returned to our research team (see Hilliard et al. 2014).

Attrition and Cohort Differences

The number of participants at each wave ranged from 726 (Wave 5) to 1062 (Wave 2). Of the 1723 participants in the present study, 69.76 % (n = 1202) of them had at least two waves of data and 12.77 % (n = 220) of them had all five waves of data. Figure 1 presents the frequencies of participants at each wave and by the number of waves they were surveyed.

Fig. 1
figure 1

Frequencies of participants by wave (a) and by the number of waves (b)

Full size image

Attrition in later waves was mainly due to an inability to reach participants because they and their parents did not report their contact information initially, their family had relocated, and/or because they lost interest in participating in the study and/or in Scouting. Pearson’s χ2 tests were conducted to determine whether youth in different waves, or with different numbers of waves of data, differed in race/ethnicity and religious versus non-religious institutional affiliation. Analyses were conducted independently for Scout and non-Scout samples. No systematic differences in race/ethnicity were found for youth in different waves or youth who had participated in different numbers of waves of the study. However, whereas a stable percentage of Scouts (about 75 %) were from religious sites across waves, the percentage of non-Scout boys affiliated with religion-based institutions (i.e., Catholic schools) decreased from Waves 1 to 5 (56.4, 32.6, 21.7, 16.9, and 9.7 %, respectively). In turn, among Scouts with one, two, three, four, or five waves of data, the percentages of Scouts from religious sites were very similar (77.2, 76.8, 76.6, 77.2, and 69.3 % respectively). However, the percentage of non-Scout boys from religious institutions decreased as the number of waves increased.

We also examined if religious versus non-religious institutional affiliation was correlated with youth ratings of their positive attributes at different waves. Only non-Scout boys’ ratings of religious reverence were consistently, positively, but weakly associated with their religious institutional affiliation (r = .14–.29 across waves). Non-Scout boys from religious institutions reported higher levels of religious reverence at every wave. Considering that the percentage of non-Scout boys from religious institutions systemically decreased across waves, and that their reverence self-rating was associated with their institutions’ religious affiliation, the religious versus non-religious institutional affiliation was taken into account in subsequent analyses examining religious reverence.

Missing Data

The Character and Merit Project contains both item and wave non-response data. Missing information was recovered using multiple imputation at the item level with the Amelia II package (Honaker et al. 2010) in R 3.1.2. Twenty imputed datasets were generated and then imported into Mplus 7.3 for analyses. The same analyses were performed on each completed dataset and effects were aggregated to create one estimate of parameters (Rubin 1987).

Analyses

A two-group latent growth curve model (LGM) was used to examine the growth trajectories of each positive attribute for Scouts and non-Scout boys. Participant age at each wave was included as a time-variant covariate to control for potential age effects on youth self-ratings of the attributes. Two latent growth factors were estimated in each model—the intercept and the slope (see Fig. 2). The intercept represents the estimated baseline value of each attribute at Wave 1. The five factor loadings for the latent intercept factor were fixed to 1. The slope represents the linear rate of change per wave in the growth trajectory of that attribute.Footnote 2 The factor loadings for the latent slope factor were fixed at values from 0 to 4 for Waves 1 (i.e., 0) to 5 (i.e., 4). The intercept and slope factors were freely correlated. The effects of age on youth self-ratings were fixed to be equal across waves and across Scouts and non-Scout boys. Effect sizes of longitudinal changes in the form of standardized mean differences were computed using estimated means at baseline (Wave 1) and study end-point (Wave 5) from estimated models and observed standard deviations (Card 2011; Lewis et al. 2015). After the LGMs were conducted, Wald’s tests were used to examine whether parameter magnitudes (e.g., the estimated means of the intercepts and slopes of the trajectories) differed significantly across Scout and non-Scout groups. For reasons noted in the previous section, follow-up analyses with the developmental trajectories of religious reverence were also conducted to account for the religious versus non-religious institutional affiliation from which youth were recruited.

Fig. 2
figure 2

The latent growth curve model estimated for each positive attribute

Full size image

Results

Descriptive statistics for the ten positive attributes distinguished by Scouts versus non-Scout group status, and data collection wave, are presented in Table 1. In general, youth ratings on all of the positive attributes across all five waves clustered between 3.59 and 4.61, which is indicative of a moderate to high range given the five-point rating scale used for all items. Thus, all youth reported having moderate to high levels of each positive attribute assessed in the present study.

Table 1 Descriptive statistics for positive attributes of Scouts and non-Scout boys
Full size table

In order to investigate developmental trajectories of the 10 positive attributes assessed among Scouts and non-Scout boys, two-group LGMs were run for each of the ten attributes. The overall goodness of fit of each model was evaluated using the χ2 statistic, as well as four widely used model fit indices (Kline 2011): (a) comparative fit index (CFI), (b) Tucker–Lewis index (TLI), (c) root-mean-square error of approximation (RMSEA), and (d) standardized root-mean-square residual (SRMR). The recommended cut-off points are .90 for CFI and TLI (Bentler 1992; Kenny and McCoach 2003), and .08 for RMSEA and SRMR (Browne and Cudeck 1993; Hu and Bentler 1999). As shown in Table 2, all LGMs we conducted showed good or acceptable model fit.

Table 2 Model Fit indices for the two-group latent growth curve model of each positive attribute
Full size table

As noted previously, participant age at each wave was included in all LGMs as a time-variant covariate to control for variance in youth self-evaluation processes attributable to age differences. As presented in Table 3, age showed significant negative effects on religious reverence (B = −.035, SE = .014, p < .05), thriftiness (B = −.023, SE = .009, p < .05), helpfulness (B = −.017, SE = .008, p < .05), and school performance (B = −.035, SE = .009, p < .001). At each wave of this study, older youth reported lower religious reverence, thriftiness, helpfulness, and school performance compared to younger youth.

Table 3 Unstandardized regression coefficients of age on positive attributes in the two-group latent growth curve models
Full size table

Table 4 presents the intercept and slope estimates of the developmental trajectories of Scouts and non-Scout boys’ 10 positive attributes. All intercept means were significantly different from zero, ranging from 3.84 for Scouts and non-Scout boys’ intentional self-regulation to 4.52 for non-Scout boys’ hopeful future expectations. The slope means were positively and significantly different from zero for six of the 10 positive attributes among Scouts, including obedience (M = .040, SE = .008, p < .0001), cheerfulness (M = .021, SE = .009, p < .05), kindness (M = .025, SE = .009, p < .01), hopeful future expectations (M = .029, SE = .007, p < .0001), trustworthiness (M = .032, SE = .008, p < .0001), and helpfulness (M = .019, SE = .009, p < .05). The slope means were not significantly different from zero for Scouts’ religious reverence, thriftiness, intentional self-regulation, or school performance. For non-Scout boys, none of the 10 slope means were significantly different from zero. However, the slope mean was negatively and significantly different from zero for non-Scout boys’ religious reverence (M = −.051, SE = .024, p < .05).

Table 4 Estimated intercepts, slopes, and effect sizes of the positive attribute trajectories
Full size table

Thus, Scouts’ self-ratings across five waves showed a significant increase in obedience, cheerfulness, kindness, hopeful future expectations, trustworthiness, and helpfulness. Scouts’ ratings showed no significant change in religious reverence, thriftiness, intentional self-regulation, and school performance. Non-Scout boys’ self-ratings showed a significant decrease in religious reverence across the five waves, and no significant change in any other attribute.

The variances of the intercepts and the slopes were not significantly different from zero for all of the measured attributes of both the Scouts and non-Scout boys. Thus, within both groups of participants, there was no statistically significant variability in youth initial levels of the attributes and in youth rates of change over time. Similarly, the covariances of the intercepts and the slopes were not significantly different from zero for all of the measured attributes of both the Scouts and non-Scout boys. Therefore, there were no statistically significant associations between the initial level and the rate of change of the measured attributes.

To examine whether the estimated means of the intercepts and slopes differed significantly in magnitude between Scout and non-Scout groups, we conducted Wald’s tests in each of the 10 LGMs. Results indicated that Scouts and non-Scout boys differed significantly in the intercept means of obedience (Wald χ 2 = 12.17, df = 1, p < .001) and religious reverence (Wald χ 2 = 19.21, df = 1, p < .0001). As depicted in Table 4, non-Scout boys reported significantly higher levels of obedience and religious reverence compared to Scouts at Wave 1. Wald’s tests of the slope means showed that Scouts and non-Scout boys differed significantly in the rates of change of obedience (Wald χ 2 = 6.58, df = 1, p < .05), religious reverence (Wald χ 2 = 6.56, df = 1, p < .05), cheerfulness (Wald χ 2 = 5.56, df = 1, p < .05), and helpfulness (Wald χ 2 = 4.31, df = 1, p < .05). Table 4 illustrates that the changes in self-ratings of obedience, cheerfulness, and helpfulness were significantly more positive for Scouts than non-Scout boys. However, the change in religious reverence across five waves was significantly more negative for non-Scout boys compared to Scouts.

Despite non-Scout boys’ higher obedience ratings at Wave 1, Scouts’ obedience increased significantly over time, whereas non-Scout boys’ self-rated obedience remained stable, resulting in equivalent levels of obedience between the two groups at Wave 5 (Fig. 3). In addition, although non-Scout boys had higher religious reverence scores at beginning of the study, their ratings decreased significantly over time, whereas Scouts’ religious reverence showed no significant change. As a result, Scouts and non-Scout had equivalent levels of religious reverence by Wave 5. For cheerfulness and helpfulness, Scouts and non-Scout boys had equivalent initial levels of these two attributes at Wave 1, but changes in Scouts’ ratings were significantly more positive over time compared to non-Scout boys.

Fig. 3
figure 3

Significantly different developmental trajectories of obedience, religious reverence, cheerfulness, and helpfulness between Scouts and non-Scout boys based on parameters estimated with data across the five waves of the study

Full size image

Considering the association between the religious versus non-religious institutional affiliations and non-Scout boys’ ratings on religious reverence, as well as the changing percentage of non-Scout boys from religious institutions over the five waves of this study, additional multi-group LGMs were conducted to examine the developmental trajectories of four groups of youth: (1) Scouts affiliated with religious institutions, (2) Scouts affiliated with non-religious institutions, (3) non-Scout boys attending religious schools, and (4) non-Scout boys attending secular schools. The model fit the data well (χ 2 = 197.99, df = 139; CFI = .924; TLI = .924; RMSEA = .031; SRMR = .044). The negative age effect on religious reverence was similar to what was found in the two-group LGM (B = −.035, SE = .014, p < .05). The intercept and slope estimates of the trajectories of the four groups are presented in Table 5. The mean of the intercept was the highest for non-Scout boys from religious affiliated institutions, followed by non-Scout boys from non-religious affiliated institutions, Scouts from religious affiliated institutions, and Scouts from non-religious affiliated institutions. The slope mean was negative and significantly different from zero for non-Scout boys who were from religious affiliated institutions (M = −.102, SE = .033, p < .01). The slope means for the other three groups of youth were not significantly different from zero.

Table 5 Estimated intercepts, slopes, and effect sizes of the religious reverence trajectories of scouts from religious sites, Scouts from non-religious sites, non-Scouts from religious institutions, and non-Scouts from non-religious institutions
Full size table

Follow-up Wald’s tests suggest that the intercept mean of religious reverence among non-Scout boys from religious affiliated institutions was significantly higher than each of the other three groups (e.g., Wald χ 2 = 24.87, df = 1, p < .0001, for the comparison between non-Scout boys from religious and non-religious affiliated institutions). The intercept mean of Scouts from non-religious affiliated institutions was significantly lower than each of the other three groups (e.g., Wald χ 2 = 10.58, df = 1, p < .01, for the comparison between Scouts from religious and non-religious affiliated institutions). But Scouts from religious affiliated institutions and non-Scout boys from non-religious affiliated institutions were not significantly different in religious reverence at Wave 1.

When the rate of change in the trajectory of religious reverence was considered, Wald tests suggest that the decreasing trend of non-Scout boys from religious affiliated institutions was significantly different from the two Scout groups, whose trends were generally stable across data collection waves (e.g., Wald χ 2 = 9.61, df = 1, p < .01, for the comparison with Scouts from religious affiliated sites). However, no significant difference in the rate of change was found between non-Scout boys from religious and non-religious affiliated institutions. Together, findings from the two- and four-group LGMs of religious reverence, indicate that, of all youth in the present study, non-Scout boys from religious affiliated institutions started with the highest level of reverence and Scouts from non-religious affiliated institutions started with the lowest level of reverence. Over time, only non-Scout boys from religious affiliated institutions significantly decreased in the level of their religious reverence, whereas other youth maintained stable levels of religious reverence throughout the study.

Discussion

Research on character development has received increasing attention from developmental scientists across the past 20 years (Lerner and Callina 2014). Reflecting a relational developmental systems-based emphasis on the importance of youth-context relations for youth thriving, applied work has pointed to the usefulness of aligning individual strengths with ecological developmental assets to promote positive developmental outcomes. This work is represented by out-of-school-time youth development programs aimed at promoting character virtues, prosocial behaviors, and positive civic actions (Eccles and Gootman 2002; Sherrod et al. 2010; Vandell et al. 2015).

However, empirical research regarding how character develops in varied ecological contexts, in particular out-of-school-time youth development programs, is very limited, especially during the relatively early portions of the life span. To address this gap in research, the present study aimed at extending the developmental studies of character into childhood and early adolescence and contextualizing the study in a representative high-quality youth development setting focused on character development—the Cub Scout program. Based on the Boy Scouts of America focus on character (Boy Scouts of America 2010), we expected that youth involved in Cub Scouting would show more favorable developmental patterns of character than youth not involved in Cub Scouting. In addition, based on past research about age-related changes in youth and in self-evaluations (Lewis et al. 2015), we expected younger youth to rate themselves more positively than older youth.

The findings were largely consistent with our expectations. Participating youth in our study generally had moderate to high levels of self-ratings on all of the measured attributes. However, younger youth rated themselves even higher than older youth in several of the positive attributes we assessed, including religious reverence, thriftiness, helpfulness, and school performance, supporting the heightened positivity bias discovered in younger children’s self-evaluations in prior studies (e.g., Boseovski 2010). After age effects in self-evaluative processes were accounted for, Scouts’ self-ratings across the study period of two and a half years showed small but significant increases in obedience, cheerfulness, kindness, hopeful future expectations, trustworthiness, and helpfulness. Thus, Scouts showed positive developmental changes in several of the core character attributes emphasized by the Boy Scouts of America and in the developmental literature. Non-Scout boys’ self-ratings generally showed no significant changes during the study period, except that non-Scout boys from religious affiliated institutions reported a medium to large degree of decrease in their religious reverence. Thus, there is support for a more favorable developmental pattern of character among Scouts than non-Scout boys, for at least some of the attributes assessed in the study.

The findings of unique character developmental trajectories of Scouts and non-Scout boys are in line with prior research on adolescents indicating that different activities expose youth to distinct institutional structures, goals, skills, and relationships, so that specific developmental outcomes derive from different types of involvement in youth development programs (Hansen et al. 2003; Larson et al. 2006). Based on findings of children and early adolescents, the present study contributed to the literature that early out-of-school-time youth development programs, such as those of the Boy Scouts of America beginning-level Cub Scouting program, are already playing a role in channeling youth character developmental trajectories. It may be that, by engaging Scouts in themed meetings and activities around a variety of character virtues, forming sustained positive relationship between Scouts and adults in the program, and empowering Scouts to practice and apply their learned skills in different ecologically-valid settings (steps reflecting the “Big 3” attributes of effective youth development programs; Hershberg et al. in press; Lerner 2004), the Boy Scouts of America programs capitalize on youth strengths and provide youth with resources and opportunities to develop character attributes that are particularly valued by the Boy Scouts of America.

Although the sizes of positive changes found for Cub Scouts in the present study are still small, they were obtained in a relatively short period of time (2.5 years), and involved changes that began at quite high starting points (youth self-ratings at the beginning of the study were already at moderate to high levels). In addition, boys in the comparison group either remained stable in their self-rated attributes or showed substantial decrease (as in the case of Catholic school students’ religious reverence). The potential promotive and protective effects of the Boy Scouts of America program on youth positive development are worthy of future research and practical attention, as the cascading effects of earlier strengths on future thriving are important developmental phenomena and have long-term impact on both individuals and society (Masten et al. 2005).

Declines in religiosity from adolescence to early adulthood have been widely documented in earlier studies (e.g., Hayward and Krause 2013; Starbuck 1901; Van Tuyl 1938). Globally, researchers have identified rationalization and individualization as the major factors leading to declined religiosity at the societal level (van Ingen and Moor 2015). By examining the developmental trajectories of religious reverence among young boys from different ecological contexts, the present study contributed to the literature by showing that declined religiosity might occur as early as childhood and early adolescence, but only for youth embedded in a highly religious context (e.g., Catholic school). The importance of considering person-context synergy in understanding the similarity and variations of developmental patterns may once again be evident.

However, we do not contend that character attributes associated with the Boy Scouts of America programs constitute the features of character thoroughly or finally defining the character construct in this developmental period. There may be both commonality and variability in views of character development between the Boy Scouts of America and other out-of-school-time youth development programs (and across the Boy Scouts of America programs in different countries). For instance, whereas some character attributes (e.g., obedience, religious reverence) are explicitly inscribed in the Scout Law, they are not unequivocally emphasized in all youth developmental contexts. The relational developmental systems-based emphasis on bidirectionally influential person-context relations (Lerner and Callina 2014), which means that the content of character may vary across time and place (Elder et al. 2015), and empirical evidence that specific developmental outcomes derive from specific youth development programs (Hansen et al. 2003; Larson et al. 2006), should be considered in interpreting the current findings.

Prior research with adolescents has also suggested that parental encouragement and modeling might influence youth participation in youth development programs (e.g., McGinley et al. 2010). In order to garner more information about how parents may play a part in youth program participation and engagement, in our larger Character and Merit Project we asked parents about reasons for participating in the program and for leaving the program (for those who left). From the parents who provided information about why their children no longer participated in the program, we learned that reasons varied and included reasons such as their children had no time for the program, did not enjoy the program, and/or that sports took precedence. However, parents did not say that they no longer had an interest in the program (as a reason for their children’s lack of participation). Of course, more information about parent influence on youth program participation would be very relevant in future studies to help interpret our current findings further, especially with youth in this age group.

More research like the present study should also be conducted with ethnic-minority youth. Specifically, the Boy Scouts of America programs have an initiative, ScoutReach, that delivers the program to youth in urban city centers and does so for free. ScoutReach youth may not have access to the program without such support. Youth in ScoutReach are predominately ethnic-minority youth from low-income backgrounds. Other studies from the Character and Merit Project have found, for example, that leaders in this instance of the Boy Scouts of America programs believe the program influences youth character, specifically through influencing youth self-confidence and hope for the future (e.g., Hershberg et al. in press). The Assessment of Character in Children and Early Adolescents used in the current study has also been found to have good psychometric validity with this population, and that character attributes of kindness and helpfulness appear to be modeled for these youth as part of their participation in ScoutReach (e.g., Ferris et al. 2015). Moreover, ScoutReach youth reported that they engaged in meaningful conversations with their ScoutReach leaders about issues of acceptance and tolerance (of girls and all youth, regardless of the community they are from or their ethnicity). However, those findings were exploratory, and no longitudinal examination of character has been conducted among ScoutReach youth (due to the still relatively low participation rates in ScoutReach among ethnic-minority youth). More research is needed to clarify what kind of programs works for youth from what kind of socioeconomic and cultural backgrounds and in which ways.

In addition, findings from the present study should be interpreted in light of several limitations. First, data analyzed in the present study involved youth self-reports on a rating scale with a quite narrow range. Although the Assessment of Character in Children and Early Adolescents has been found to have good psychometric properties (Wang et al. 2015), self-ratings provide only one means to assess youth character development. The narrow range of the rating scale may have constrained the identification of meaningful intraindividual changes and interindividual differences. Thus, future studies should consider adopting a multitrait-multimethod approach to obtain information regarding youth character development from different respondents, such as parents and teachers. Behavioral measures can also be incorporated to complement findings from survey data. Qualitative data should also be collected to help triangulate the findings from quantitative methods. These additional data may help elucidate how the identified positive changes among Scouts and the differences between different groups of youth may translate into youth performance and contribution in ecologically-valid settings.

Second, the sample size of non-Scouts was relatively small for such complex longitudinal data collection and analyses (especially in comparison to the Scouts’ sample), leading to greater standard errors of the estimated coefficients for non-Scouts. Therefore, few significant developmental changes have been identified for the non-Scout boys and only a few between-group differences remained significantly different when Scouts and non-Scout boys’ trajectories were compared directly. Future research should keep improving measurement accuracy by refining measures and data collection procedures, and by increasing the sample size of the comparison groups.

Third, the heterogeneity within the Scout and non-Scout groups can be further examined in future studies. For example, Scouts with different length, participation frequency, and engagement levels in Scouting were found to differ in their levels of character attributes (Lynch et al. 2015). Future study can delve deeper into within-group variations in character development of both the Scouts and non-Scout youth.

Conclusion

The present study extended the research on character development into childhood and early adolescence. All character attributes examined in the study reflected current literature on character development (Lerner and Callina 2014), and were contextualized in ecologically-valid experiences of youth in the Boy Scouts of America programs. To our knowledge, no prior research has empirically examined character development in this age group in the out-of-school-time youth development program context. In addition, previous studies of youth self-reported positive attributes often failed to distinguish between the development of self-evaluation processes and the development of specific attributes (e.g., Jozsa et al. 2014; Lewis et al. 2015). The present study examined the two distinct processes separately and provided evidence for both a declined positivity bias involved in youth self-evaluation and the development of specific character attributes. The developmental trajectories of several character attributes of different groups of youth spanning two and a half years supported the goal of the Boy Scouts of America of promoting youth character development. The findings also indicated variations in the development of different character attributes and across youth from different ecological contexts. Most important, the longitudinal findings from this study provided valuable baseline data for charting subsequent changes in character development across childhood and into the breadth of the adolescent years.