1 Introduction

In recent years, subjective well-being (SWB) has become a new focus of inquiry in mainstream psychology. SWB is defined as a subjective appraisal of one’s life (Diener 2000). It is a tripartite entity that comprises an affective aspect that reflects the pleasant (positive affect, PA) and unpleasant (negative affect, NA) emotional experiences in one’s life, as well as a cognitive aspect that entails cognitive evaluation of one’s overall level of life satisfaction (LS). To our knowledge, there have been relatively few investigations of PA and NA with respect to LS. Although no consensus exists yet on a definitive structure of SWB, researchers (e.g., Busseri and Sadava 2011; Diener et al. 1999; Morgan et al. 2011) have emphasized that components of SWB are correlated, but have separable dimensions. Furthermore, it appears important to include PA, NA, and LS within the same study to provide a more complete depiction of SWB.

In the past, research on SWB has focused primarily on adult populations. Nonetheless, there has been increasing research evidence on the SWB of adolescents in recent years. SWB is as crucial in adolescence as it is in other stages of human development (Eryilmaz 2012). Low levels of SWB in adolescents have been associated with various problems such as suicide, sexual risk-taking behaviors, violent behaviors, aggression, depression, and problematic Internet use (Cao et al. 2011; Heisel and Flett 2004; MacDonald et al. 2005; Suldo and Huebner 2006; Valois et al. 2002). In contrast, a high level of SWB in adolescents has been associated with benefits such as school satisfaction, academic achievement and aspirations, self-esteem, life meaning, gratitude, and self-efficacy (Proctoret al. 2010; Suldo and Huebner 2006). Adolescence is considered an opportune time to establish a strong foundation for positive well-being to lead a satisfying life in adulthood (McCabe et al. 2011). Thus, researchers’ continuing efforts to examine adolescent SWB are crucial. Specifically, determining the underlying factors that link to adolescent SWB would be helpful in promoting positive well-being.

Thus far, SWB studies on Malaysian populations have comprised young adults, adults, older people, and indigenous people, in relation to variables such as loneliness and depression (Swami et al. 2007), materialism (Ang et al. 2014; Rakrachakarn et al. 2013), marital satisfaction (Ng et al. 2009), religiosity and spiritual engagement (Mellor et al. 2012), wealth (Howell et al. 2006), and living arrangements (Kooshiar et al. 2012). There is, however, relatively little SWB research on Malaysian adolescents, to our knowledge. A study conducted by Muhamad and Jaafar (2009) demonstrated the relationship between personality and SWB in Malaysian youths. In particular, conscientiousness, neuroticism, and extraversion appeared to be the best predictors for PA, NA, and LS. Furthermore, Yaacob, Tan, Tan, and Juhari (2012) discovered differences in LS among Malaysian adolescents, when grouped by sociodemographic factors such as sex, ethnicity, age group, and religion. However, SWB research in Malaysia seems to be in its preliminary stages. Little is known about the possible underlying variables that may account for Malaysian adolescent SWB.

The National Health Mobility Survey (III) that was conducted by the Ministry of Health in Malaysia highlighted an increase in mental health problems from 13 % (in 1996) to 20.32 % (in 2006) among Malaysian adolescents. The highest prevalence of acute suicidal ideation was also among youth. Moreover, teenage crime, such as snatch thefts, blackmail, gang activity, rape, murder, and drug abuse are all on the rise in Malaysia (Cheng 2014). From the perspective of positive psychology, SWB has been shown not only to be advantageous for adolescent adaptive functioning, but also, to serve as a protective factor against maladaptive functioning (Cao et al. 2011; Heisel and Flett 2004; Proctor et al. 2010; Suldo and Huebner 2004a, b, 2006; Valois et al. 2002). Therefore, there is a need to determine underlying factors that may contribute to SWB in Malaysian adolescents.

One of the crucial developmental milestones in adolescence is to gain autonomy, in which adolescents spend increasingly more time with peers (Rubin et al. 2006). Extant research, however, has identified a significant contribution of parenting behaviors to adolescent SWB (e.g., Cacioppo et al. 2013; Leung et al. 2010; Rasmi et al. 2012; Saha et al. 2010). Furthermore, internal resources such as self-efficacy, gratitude, optimism, self-esteem, hope, and resilience have been identified as psychological strengths that promote adolescent SWB (Khan 2013; Lightsey et al. 2013; Merkaš and Brajša-Žganec 2011; Sivis-Cetinkaya 2013; Suldo and Shaffer 2007; Sun and Shek 2012; Vecchio et al. 2007). Previous studies have suggested that parenting behaviors may affect adolescent developmental outcomes directly or indirectly via internal resources. For instance, parental control was found to influence adolescent developmental outcomes via attributional style (Schleider et al. 2014) and self-regulation (Lee et al. 2012).

The present study examined whether paternal involvement and maternal involvement are associated with PA, NA, and LS in Malaysian adolescents via the following self-efficacy beliefs: academic self-efficacy, social self-efficacy, and emotional self-efficacy. There is no single definition of parental involvement in the past literature. In the present study, “parental involvement” refers to parental interaction and engagement in various facets of their children’s lives (e.g., emotional development; intellectual development; developing independence and responsibility; leisure, fun, play; companionship), which may promote some aspects of development in children (Finley and Schwartz 2004). Rohner and his colleagues used a successful approach that showed a unique association between children’s retrospective perceptions of parental acceptance-rejection and various aspects of their development (e.g., Khaleque and Rohner 2002; Rohner and Veneziano 2001). By adapting this approach, other researchers have sought to confirm that the assessment of parental involvement, from the perspective of the adolescents themselves, can best capture the overall long-term impact of parental involvement on the adolescents’ development and well-being (Finley and Schwartz 2006). Aside from the work of Finley and colleagues (see Finley et al. 2008; Finley and Schwartz 2004, 2006, 2007, 2010), other studies (e.g., Allgood et al. 2012; Yeoh and Woo 2010) have used the same conceptualization of parental involvement and examined its associations with aspects of adolescent development.

Social cognitive theory (Bandura 1993, 1997, 1999, 2002, 2012) was used as the theoretical framework to guide the present study. Fundamentally, this theory asserts that human functioning has its foundation in the social environment and self-influences, in which self-efficacy is a constituent, concerning people’s beliefs about their capabilities to exercise control over their functioning. Adolescents exist within social environments and are continuously interacting with their parents. According to Schunk and Miller (2002), adolescents acquire much of their self-efficacy information from their families and home environment. Parental support and knowledge have been shown to correlate positively with adolescent efficacious belief (Frank et al. 2010), whereas the opposite was found for maternal rejection (Niditch and Varela 2012). Notably, self-efficacy is best understood as domain specific. The importance of specific efficacy domains varies by developmental phase (Berry and West 1993). Academic self-efficacy, social self-efficacy, and emotional self-efficacy are considered the most salient domains of efficacy belief in adolescents (Suldo and Shaffer 2007). Self-efficacy enhances the quality of human functioning via cognitive, affective, motivational, and decisional processes (Bandura 2012). Self-efficacy in different domains such as academic, social, emotional, regulating stress, and anger, has been shown to have significant relationships with components of SWB in adolescents (Lackaye et al. 2006; Lightsey et al. 2013; Meng et al. 2014; Putwain et al. 2013; Steca et al. 2011; Suldo and Shaffer 2007; Vecchio et al. 2007). To date, evidence has begun to accumulate on the links between parental involvement, self-efficacy beliefs, and the components of SWB in adolescents (Di Maggio and Zappulla 2013; Finley and Schwartz 2007, 2010; Weiser and Riggio 2010). Previous studies have identified self-efficacy as a mediator in promoting SWB among adolescents in both Western and non-Western settings (e.g., Fogle et al. 2002; Sun et al. 2014; Wang et al. 2014). Though efficacious beliefs have universal functional values, culture does play a role in shaping them (Bandura 2002).

Borrowing perspectives from social cognitive theory and extant empirical evidence, the present study proposes that adolescent functioning (SWB and self-efficacy) has its foundation in the social environment, in which parental involvement is an influential socio environmental factor in adolescent lives. Operating within a social environment, self-efficacy beliefs were envisioned as potential underlying self-influencing factors in the parental involvement–SWB link among adolescents. Importantly, self-efficacy beliefs vary across domains. The present study considered self-efficacy across three domains: academic, social, and emotional, leading to three proposed mediators. Instead of examining the mediation effects of academic self-efficacy, social self-efficacy, and emotional self-efficacy in three separate simple mediation models, the current study adopted the recommendations of Preacher and Hayes (2008) to propose a multiple mediator model (see Fig. 1). A multiple mediator model that allows mediators to be tested simultaneously can reduce the likelihood of parameter bias due to omitted variables. In addition, a multiple mediator model is more precise and parsimonious than separate simple mediation models. A proliferation of extant research has examined the separate influences of paternal parenting and maternal parenting. In reality, mothers and fathers do not parent in separate vacuums. Stolz, Barber, and Olsen (2005) emphasized the importance of considering both paternal and maternal parenting and suggested that they be included in the same model. In line with this suggestion, previous studies (e.g., Bean et al. 2006; Bean and Northrup 2009; Plunkett et al. 2009) assessed the unique influence of paternal and maternal parenting in the presence of each other using a single structural model. Hence, in the current study, paternal involvement and maternal involvement were included in the same model, as shown in Fig. 1. We hypothesized that self-efficacy beliefs would mediate the relationships of paternal involvement and maternal involvement to PA, NA, and LS. Given the limited amount of research that includes all of these self-efficacy beliefs, we did not formulate specific hypotheses about which self-efficacy beliefs would emerge as mediators.

Fig. 1
figure 1

Hypothesized multiple mediator model

Full size image

2 Method

2.1 Participants and Procedure

A total of 905 high-school adolescents participated in the current study. They were selected through a multi-stage cluster sampling approach. At the first stage, schools were selected randomly from all three zones in the Territory of Kuala Lumpur, Malaysia, according to the proportion of enrollment size in each zone. As a result, a total of 14 schools were selected (i.e., seven schools from Zone Bangsar Pudu, four schools from Zone Sentul, three schools from Zone Keramat). At the second stage, two classes of Grade 11 students in each selected school took part in the survey. The survey administration took place during regular school hours. To maintain anonymity, questionnaires did not contain a space for participant names, and participants were assured of the confidentiality of their responses.

Of the questionnaires collected, 103 data sets were excluded from the analyses because of problems such as major incompleteness, untrustworthy responses (i.e., placed the same value for all items under the same section), and absent data for gender and residency. As a result, a total of 802 participants were retained for further analyses. Of these participants, 55 % were female, 45 % were male, and the mean age was 16 years.

2.2 Instruments

2.2.1 Paternal Involvement and Maternal Involvement

Paternal involvement was measured using the Father Involvement Scale (Finley and Schwartz 2004). The scale measures a person’s perception of paternal involvement in 20 domains of life, such as emotional development; caregiving; leisure, play, fun; and mentoring/teaching. The same scale was used to assess maternal involvement by replacing the word “father” with “mother”. Items are rated on a five-point scale from 1 (never involved) to 5 (always involved). The Cronbach’s alpha coefficient based on the current sample was .94 for the father version and .92 for the mother version.

2.2.2 Self-Efficacy

Self-efficacy was assessed using the Self-Efficacy Questionnaire for Children (Muris 2001). The scale consists of three subscales: academic self-efficacy, social self-efficacy, and emotional self-efficacy. There are seven items in each subscale, rated on a five-point scale from 1 (not at all) to 5 (very well). Examples of items include: “How well do you succeed in understanding all subjects in school?” (academic self-efficacy); “How well can you work in harmony with your classmates?” (social self-efficacy); and “How well do you succeed in becoming calm again when you are very scared?” (emotional self-efficacy). In the present study, the Cronbach’s alpha coefficients for the overall scale and subscales of academic self-efficacy, social self-efficacy, and emotional self-efficacy were .83, .74, .67, and .72, respectively.

2.2.3 Positive Affect and Negative Affect

PA and NA were measured using the Positive and Negative Affect Schedule (PANAS; Watson et al. 1988). The scale consists of 10 items for PA (e.g., enthusiastic, determined, active) and 10 items for NA (e.g., upset, guilty, hostile). Participants indicated “in general” the degree to which they experienced each affect on a five-point scale from 1(never) to 5 (all the time). In this study, the Cronbach’s alpha coefficient was .80 for PA and .72 for NA.

2.2.4 Life Satisfaction

LS was assessed using the Students’ Life Satisfaction Scale (SLSS; Huebner 1991). The scale consists of seven items that are rated on a six-point scale from 1 (strongly disagree) to 6 (strongly agree). Examples of items include: “My life is going well” and “I have what I want in life”. In the present study, the Cronbach’s alpha coefficient for the SLSS was .75.

2.2.5 Control Variables

The present study considered a number of sociodemographic variables as potential control variables: gender (1 = male; 2 = female), ethnicity (1 = Malay; 2 = non-Malay), religion (1 = Muslim; 2 = non-Muslim), family form (1 = intact family; 2 = non-intact family), fathers’ and mothers’ highest level of education (1 = no formal education to 7 = graduate degree), number of children at home (ranging from 1 to 12), and family monthly income (ranging from RM 0 to RM 107,000, approximately US$ 0 to US$ 33,170). These variables have been shown to influence parental involvement and/or adolescent SWB in past studies (Finley and Schwartz 2010; Gray et al. 2013; Gugl and Welling 2012; Juhari et al. 2013; Schwartz and Finley 2009; Updegraff et al. 2009; Wilcox 2002; Yaacob et al. 2012). Furthermore, these variables have also been used frequently as control variables in research that investigated relationships between parenting behaviors and adolescent developmental outcomes (Baker 2014; Gault-Sherman 2012; Harper 2010; Karre and Mounts 2012; Moon et al. 2014; Padilla-Walker et al. 2011).

2.3 Analysis Strategy

Following the two-step procedure introduced by Anderson and Gerbing (1988), the measurement model was first evaluated using confirmatory factor analysis. Specifically, convergent and discriminant validity were examined following Fornell and Larcker’s (1981) technique. Factor loadings and average variance extracted (AVE) values of ≥ 0.5 and composite reliabilities of ≥ 0.6 suggest that a construct fulfills convergent validity. In addition, a squared correlation between two constructs that is less than the constructs’ AVE scores suggests that discriminant validity can be assumed between those constructs. To control for inflated measurement errors as a result of having multiple items for the latent variables, item parcels were created. Specifically, according to Little, Cunningham, Shahar, and Widaman (2002), the item-to-construct balance approach was applied to create parcels for the latent variables PA, NA, LS, academic self-efficacy, social self-efficacy, and emotional self-efficacy. The domain-representative approach was used to create parcels for the latent variables of paternal involvement and maternal involvement. According to Little and his collaborators (2002), item parceling is appropriate when researchers are interested in relationships between the latent constructs and not among the items. Two to four parcels per latent construct can be created.

The structural model was evaluated after the measurement model was accepted. To test mediation effects in the context of a multiple mediator model, the approach introduced by Preacher and Hayes (2008) was followed. Specifically, total indirect effects, specific indirect effects, and pairwise contrasts were estimated. Total indirect effect is the mediation effect of the set of mediators. Specific indirect effect is the unique mediator effect of a mediator above and beyond other mediators in the model. Pairwise contrasts compare the strengths of each specific indirect effect in relation to all other specific indirect effects. Bias-corrected bootstrapped 95 % confidence intervals of the indirect effects were derived from 10,000 resamples. If the interval does not include zero, a mediated effect is considered significant. The bootstrap method is preferred over other methods as it does not impose the assumption of normality of the sampling distribution of indirect effects, has a lower type I error rate, and has greater power to detect mediation (MacKinnon et al. 2002, 2004). According to Hayes (2009), unstandardized metrics are preferred in the mediation model. Hence, unstandardized regression coefficients are reported in this article.

All the analyses were conducted in Mplus 7 (Muthén and Muthén 1998–2012). As recommended by Hair, Black, Babin, and Anderson (2010), the extent of missing data can be determined for individual cases, individual variables, and the overall data set. In the present study, 54.5 % of the cases had complete data. The remaining cases were missing up to 8.33 % of their data. According to Hair et al. (2010), missing data under 10 % for an individual case can generally be ignored. Each key variable had between 0.00 and 3.24 % of missing values. Only 0.97 % of the entire data set contained missing values. No consistent relationships were found for the missing data among the variables. The full information maximum likelihood (FIML) estimate procedure was used to address missing values as it produces less biased estimates than list-wise deletion or mean substitution (Acock 2005) under the assumption of missing at random (MAR). It was not possible to confirm that the missing data in the present study were, in fact, missing at random. Hence, the MAR assumption is possibly violated. However, FIML is an appropriate method as it is robust even when the assumption of MAR is not fully met (Schafer and Graham 2002). The fit of the model was evaluated using the comparative fit index (CFI), Tucker-Lewis index (TLI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). The criteria of an acceptable model fit were CFI and TLI > .90; RMSEA < .06, and SRMR < .08 (Hu and Bentler 1999).

3 Results

3.1 Preliminary Analyses

Means, standard deviations, skewness, kurtosis, and correlations among the variables are presented in Table 1. The skewness and kurtosis values of all the variables, except for family income, were acceptably normal with no item possessing absolute skewness and kurtosis higher than 3 and 5, respectively (Kline 2011). Family income was transformed using natural log transformation and the resulting variable was well within acceptable skewness and kurtosis ranges (Kline 2011). Examination of the correlation matrix indicated that correlations between the key variables were of low to moderate degree (r = −.03 to .53). A series of multiple regressions were conducted with parental involvement and self-efficacy beliefs as the predictors of PA, NA, and LS. The collinearity statistics were well within the acceptable ranges (Hair et al. 2010) with no tolerance values less than .60 and no VIF values greater than 1.6, suggesting that the data of the present study did not exhibit a statistically significant multicollinearity problem. Adolescent gender, race, religion, family form, fathers’ and mothers’ education level, and number of children at home were significantly related to at least one of the independent and dependent variables. Therefore, these variables were controlled for when testing the mediation model to reduce confounding effects. Family income was not included as a control variable because it was not significantly related to any key variables. According to Becker (2005), control variables are recommended for inclusion only in cases of significant correlation with focal variables, to avoid spurious suppression.

Table 1 Means, standard deviations, skewness, kurtosis, and correlations among variables
Full size table

3.2 Measurement Model

Confirmatory factor analysis showed that the measurement model fit the data well (CFI = 0.974, TLI = 0.966, RMSEA = 0.039 and SRMR = 0.029). Factor loadings for each of the indicators, AVE estimates, and construct reliabilities are shown in Table 2. All the factor loadings for the indicators on the latent variables were greater than .50 and significant (p < .001). Although the AVE estimate of NA was lower than .50, it is considered acceptable as its composite reliability was adequate (.711). None of the squared correlations (see Table 3) were higher than the AVE estimates. Overall, convergent and discriminant validity were supported.

Table 2 Results of the convergent and discriminant validity analyses
Full size table
Table 3 Squared correlations between the constructs
Full size table

3.3 Structural Model

The test of the hypothesized multiple mediator model as depicted in Fig. 1 resulted in a good fit to the data (CFI = 0.939, TLI = 0.925, RMSEA = 0.050 and SRMR = 0.045). Table 4 shows the bootstrap results for indirect effects and pairwise contrasts. Inspection of the total indirect effects indicated that the set of the proposed mediators significantly mediated the relationships of paternal and maternal involvement to PA (paternal involvement: CI.95 .020, .064; maternal involvement: CI.95 .034, .072), NA (paternal involvement: CI.95 −.022, −.004; maternal involvement: CI.95 −.029, −.006), and LS (paternal involvement: CI.95 .006, .054; maternal involvement: CI.95 .018, .067).

Table 4 Bootstrap results for indirect effects and pairwise contrasts
Full size table

Examination of specific indirect effects revealed that academic self-efficacy and social self-efficacy reached statistical significance as unique mediators of the paternal involvement–PA relationship (academic self-efficacy: CI.95 .004, .024; social self-efficacy: CI.95 .011, .048). Comparable results were found in the maternal involvement–PA relationship (academic self-efficacy: CI.95 .006, .031; academic self-efficacy: CI.95 .019, .053). There were no significant specific indirect effects detected in the relationships of paternal involvement and maternal involvement to NA. Academic self-efficacy was found to be a unique mediator in the relationships of paternal involvement and maternal involvement to LS (paternal involvement: CI.95 .015, .068; maternal involvement: CI.95 .031, .086).

Inspection of pairwise contrasts showed a few statistically significant results. However, these results deserved minimal attention because only one mediator was found to be significant in all pairwise comparisons. For example, the pairwise comparison between academic self-efficacy and social self-efficacy was significant in the relationship of paternal involvement to LS. However, in the specific indirect results, academic self-efficacy was identified as a unique mediator, whereas social self-efficacy was not.

Effects of control variables on the mediators and dependent variables in the structural model are presented in Table 5. A few significant effects were detected. Specifically, adolescent gender was found to be significantly related to academic self-efficacy (βunstandardized = .339, p < .05), PA (βunstandardized = −.291, p < .05), and NA (βunstandardized = .343, p < .05). These results suggest that female adolescents had higher academic self-efficacy and NA but lower PA than male adolescents. In addition, family form was found to be significantly related to LS (βunstandardized = −1.172, p < .05), suggesting that adolescents from intact families had higher LS than those from non-intact families.

Table 5 Effects of control variables on the mediators and dependent variables in the structural model
Full size table

4 Discussion

The primary purpose of the current study was to examine the extent to which academic self-efficacy, social self-efficacy, and emotional self-efficacy explain the associations between parental involvement and SWB in adolescents. The overall picture illustrated by the current study suggests that PA, NA, and LS are influenced differently. This provides support for the recommendations of other researchers (e.g., Busseri and Sadava 2011; Diener et al. 1999; Morgan et al. 2011) that investigating all three components of SWB within the same study and treating them as separate dimensions is crucial. In addition, the present study adds to a growing body of literature suggesting that paternal involvement is just as important to adolescent positive development as maternal involvement. In line with the ultimate goal of positive psychology to discover factors that promote human well-being, the results of the current study suggest that self-efficacy could be an internal resource that promotes adolescent well-being.

A multiple mediator model provided information on the mediation effect of the set of proposed mediators, the mediation effect of each mediator above and beyond other proposed mediators, and the relative magnitude of each mediator (Preacher and Hayes 2008). In the present study, the significant total indirect effects suggest that parental involvement may have beneficial effects on adolescent academic self-efficacy, social self-efficacy, and emotional self-efficacy. This effect suggests also, to some extent, these self-efficacy beliefs may come into play, influence each other, and together contribute to adolescent SWB. The results for specific indirect effects revealed that paternal and maternal involvement may increase adolescent efficacious beliefs in the academic field and social relationships and that, therefore, each of these beliefs may play a unique role in promoting PA and/or LS.

According to Bandura (2005), individual functioning has its foundation in the social environment. Self-efficacy is developed through mastery experiences, social modeling, and persuasions (Bandura 2012). Despite the occurrence of the individuation process in the adolescence period, parents continue to be salient within the adolescent social environment (Schwarz et al. 2012). Past studies have demonstrated that parental advice and encouragement have positive impacts on adolescent academic self-efficacy (Fan and Williams 2010; Mena 2011). In addition, parental guidance for social problem-solving and the provision of social opportunities predicted children’s social competence (McDowell and Parke 2009). Moreover, parents may contribute to adolescent emotional competence by being role model, demonstrating social referencing, providing optimal encouragement, and coaching adolescents in expressing and regulating emotions (Morris et al. 2007). Hence, it may be via positive involvement in various aspects of adolescent children’s lives that a parent communicates efficacious beliefs through being a role model, providing guidance, and being encouraging in difficult encounters and accomplishments. Thus, this behavior creates opportunities for efficacious actions and support of mastery experiences, which are then internalized by adolescent children to develop their own efficacy level (Weiser and Riggio 2010; Whitbeck 1987). In turn, these efficacious beliefs serve as an internal resource that enhances adolescent SWB.

Self-efficacy has been posited to enhance human functioning via affective, cognitive, motivational, and decisional processes (Bandura 2012). Higher self-efficacy establishes a sense of control through beliefs that positive outcomes are achievable, resulting in positive guides, support for performance, less vulnerability to stress and disturbing thought patterns, and perseverance in the face of difficulty, thus promoting accomplishments. With accomplishments, one experiences PA (Snyder 2002; Snyder et al. 1996) and derives satisfaction from life (Sahan et al. 2012). Adolescence has been recognized as a particularly stressful period of life (Hostinar and Gunnar 2013). Increased academic demands are one of the achievement-related stressors reported at this life stage (Hankin et al. 2007; Mezulis et al. 2010). Additionally, time spent in academic learning accounts for a large portion of adolescent life (Bassi et al. 2011). Empirical findings corroborate the positive contribution of adolescent academic self-efficacy to PA (e.g., Lackaye et al. 2006; Putwain et al. 2013; Steca et al. 2011) and LS (e.g., Suldo and Shaffer 2007; Vecchio et al. 2007). It is possible that adolescents with a strong sense of academic self-efficacy possess greater academic interest and motivation, manage academic stressors more successfully, expend greater efforts, and remain persistent in the face of difficulty and failure (Bandura 1997; Bandura et al. 1996; Bassi et al. 2007; Zimmerman 1995; Zimmerman and Bandura 1994). These qualities likely lead to accomplishment, which in turn leads to PA and satisfaction. These arguments support the unique mediation role of academic self-efficacy in the relationship between parental involvement and adolescent PA and LS.

During adolescence, relationships with nonparental figures such as peers take on increased meaning (Roeser et al. 1998). In this study, social self-efficacy refers to perceived capabilities for peer relationships (Muris 2001). Fundamentally, social connectedness and positive interactions contribute to one’s PA (Hawkley et al. 2003, 2007; Ingersoll-Dayton et al. 1997; Newsom et al. 2003; Rook 2001). Extant literature corroborates the positive relationship between social self-efficacy and PA in adolescents (e.g., Caprara et al. 2006; Meng et al. 2014). Further, adolescent peer relations have been shown to be associated with higher PA (e.g., Proctor et al. 2010). It is plausible that adolescents with a strong sense of social self-efficacy are higher in sociability and able to make more friends. They may interact more frequently and effectively with their peers and have a greater sense of control over the outcome of a conflict situation (Connolly 1989; Gresham 1984; Proctor et al. 2009; Torquati and Vazsonyi 1999). This might lead them to be more socially connected and increase their chances of having more positive interactions with peers, thereby increasing their PA. These arguments shed light on the unique mediation role of social self-efficacy in the relationship between parental involvement and adolescent PA.

Despite some past studies (e.g., Meng et al. 2014; Suldo and Shaffer 2007; Vecchio et al. 2007; Wright and Perrone 2010) indicating positive contributions of social self-efficacy to adolescent LS, this study did not find support for social self-efficacy as a unique mediator in the relationship between parental involvement and adolescent LS. Joint consideration of this result and the significant mediation role of social self-efficacy in the relationship between parental involvement and adolescent PA suggests that perceived competency in peer relations may contribute to adolescents’ pleasant emotions but may not necessarily affect overall LS like perceived competency in the academic domain. These findings are consistent with a study conducted by Morgan et al. (2011) which found that distal variables such as peers only influence adolescent PA but not LS. As adolescents spend a lot of time in school, it is understandable that their perceived abilities in dealing with academics and peers may influence their day-to-day affective feelings (e.g., PA). According to Haybron (2008), LS is closely linked to one’s priorities. Correspondingly, Diener, Napa Scollon, and Lucas (2009) argued that domains perceived as being very important to an individual are likely to serve as a source of information used to evaluate LS. As in many Asian countries, particular emphasis is placed on academic achievement in Malaysia (Chew et al. 2012; Chong 2007; Dzulkifli and Alias 2012; Remali et al. 2013), as it is considered a ladder to future success, e.g., gaining admission to a university, receiving a scholarship, securing a good job, having a high standard of living (Ismail 1998; Lau et al. 2000; Yuen and Fong 2013). Hence, it is plausible that Malaysian adolescents perceive academic achievement as important and of higher priority than other aspects of life (e.g., peer relations), and therefore, the adolescents’ evaluation of overall LS (e.g., “My life is going well”) is judged primarily by how well they can perform academically at school.

Notably, in the presence of academic self-efficacy and social self-efficacy, emotional self-efficacy did not mediate the relationship of paternal involvement and maternal involvement to PA and LS. In this study, emotional self-efficacy refers to the perceived capability of regulating negative emotions (Muris 2001). Existing literature (e.g., Diener and Emmons 1984; Lucas et al. 1996) suggests that there is some degree of independence between positive and negative emotions (e.g., people with low negative emotions are not necessarily full of positive emotions) and affect and LS (e.g., people may report low LS in conjunction with frequent PA). Besides, the experience of positive emotions was more strongly related to LS than the absence of negative emotions (Kuppens et al. 2008). Hence, regulation of negative emotions may not result in a corresponding increase in PA and LS. Further, Suh, Diener, Oishi, and Triandis (1998) found that emotions played less of a role in LS judgments among collectivists. Together, these offer a plausible explanation of why emotional self-efficacy played a less potent mediation role when tested simultaneously with academic self-efficacy and social self-efficacy in the relationship of paternal involvement and maternal involvement to adolescent PA and LS.

As a set, academic self-efficacy, social self-efficacy, and emotional self-efficacy mediated the relationships of paternal involvement and maternal involvement to NA. However, none of the proposed mediators uniquely mediated the relationships of paternal involvement and maternal involvement to NA. Negative emotions are highly differentiated in nature (Ben-Ze’ev 2000; Fredrickson 1998; Grandey 2008). Furthermore, when the specificity of self-efficacy has low correspondence to the criterion variable, the predictive power of self-efficacy is weakened (Bandura 1986). Researchers (e.g., Lightsey et al. 2013) suggest that self-efficacy beliefs targeted to more specific negative emotions could be more accountable. Correspondingly, the null result could be attributed to a lack of specificity of the self-efficacy measures in relation to NA. Alternatively, the present study viewed the development of adolescent SWB as rooted in the social environment, and the null findings could be due to this perspective. Past studies (e.g., Goldsmith et al. 1999; Novgorodovaa et al. 2013) demonstrated a strong genetic influence on negative emotions, whereas there was a strong environmental influence on PA.

The results of this study need to be considered in light of a number of limitations. First, this was a cross-sectional study, which limits any causal inferences. Parenting, self-efficacy beliefs, and SWB are dynamic constructs that could change over time (Bandura and Wood 1989; Bastaits and Mortelmans 2014; Busseri and Sadava 2013), and their interplay is also dynamic (Bandura 2012). The data of the present study were interpreted in fairly unidirectional terms, with parenting and internal resources affecting adolescent well-being. However, the reverse could also be true. For instance, using longitudinal data, Saha et al. (2010) showed that after controlling for baseline levels of LS, none of the parenting behaviors at Time 1 predicted changes in adolescent LS. Nevertheless, the reverse direction was supported, with baseline LS predicting positive changes in parental support. Hence, it remains unclear whether the mediation effects found in the present study would be supported in a longitudinal study. Future studies could consider testing reciprocal models using longitudinal design to provide a better understanding of the relationships between parental involvement, self-efficacy beliefs, and SWB, as well as the function of self-efficacy beliefs serving as mediators over time.

Second, because all information was derived from adolescent self-reports, future studies that simultaneously employ a multiple-informants approach are recommended. Third, as all the students in the selected classes were involved in the current study, students from the same class may share similar characteristics. Hence, the demographic backgrounds of the sample in the present study may not be as diverse as they could have been, had a simple random sampling approach been used. Fourth, there might be other psychological constructs connecting parental involvement and adolescent SWB. Some possible mediators may include (but are not limited to) optimism, self-esteem, and resilience, which have been shown to be related to parenting behaviors and SWB (e.g., Ben-Zur 2003; Kong et al. 2013; Palomar-Lever and Victorio-Estrada 2014). Fifth, although the present study included various sociodemographic variables as control variables, the obtained results may still be subject to omitted variable bias, because of the inability to control for some factors (e.g., personality, genetics, and academic performance) that have been identified in previous literature (e.g., Baker et al. 1992; Cheng and Furnham 2002; Goldsmith et al. 1999; Goswami 2014; Lykken and Tellegen 1996; Novgorodovaa et al. 2013; Schimmack et al. 2008; Weiss et al. 2008) as important correlates of and/or contributors to SWB.

Finally, our sample presented several challenges in terms of generalizability. The samples were obtained from high school students who resided in the Territory of Kuala Lumpur (an urban area in Malaysia), limiting the generalizability of the results to Malaysian adolescents who are not enrolled in schools, individuals of other age groups, and/or people from rural areas. Furthermore, although the characteristics of collectivistic culture offer plausible explanations for the findings of the present study, cautious interpretation is needed, as the survey was administered in a school setting where much social interaction occurs and students are expected to achieve. Therefore, participants could place extra weight on the importance of academic self-efficacy and social self-efficacy when reporting PA and/or LS. This could cause an overestimation of these particular efficacious beliefs in relation to the components of SWB. Moreover, as the plausible explanations were interpreted using rather general collectivistic cultural values shared by Malaysians as a whole (ethnicity and religion were controlled for), valuable information concerning the variations in ethnicity and religion, as they relate to Malaysian SWB, was lost. Thus, these specific cultural aspects could be further examined in future studies. As cultural differences influence parenting, self-efficacy beliefs, and SWB, the results of this study should also be interpreted carefully in populations with different cultural origins.

Despite these limitations, the present study extends our insight into the roles of academic, social, and emotional self-efficacy in the relationships between parental involvement and the components of SWB among adolescents. Future psychological interventions aiming to promote adolescent SWB could consider the specific roles of these efficacy beliefs in promoting specific aspects of SWB.