Happiness was conceptualized as a relatively stable, positive, and affective trait that underlines subjective well-being and satisfaction with life in general (Holder and Klassen 2010; Kamp Dush et al. 2008). However, the term “happiness” has not been defined in a uniform way; in fact, it can also refer to positive emotions, pleasure, and feelings of contentment or fulfilment (Diener et al. 2009). The majority of the research about happiness has been conducted with adolescents (Al Nima et al. 2012; Buchanan 1991; Caprara et al. 2006; Natvig et al. 2003) and adults (Demir 2008; Demir et al. 2007; Demır and Weitekamp 2007; Glenn and Weaver 1981; Lyubomirsky et al. 2005; Requena 1995).

In the past two decades, research interest has increasingly focused on children’s happiness (Holder and Coleman 2008, 2009; Holder et al. 2016; Holder and Klassen 2010; Huebner 2004; Park and Peterson 2006; Uusitalo-Malmivaara and Lehto 2013) because the factors that contribute to subjective well-being in children may differ from those in adolescents and adults (Holder et al. 2010). For example, cognitive maturity, job satisfaction, spousal happiness, and marriage are associated with adults’ happiness (Headey et al. 1991; Stull 1988; Tait et al. 1989), but these variables are not evaluable in children (Holder et al. 2010).

In several studies, children’s happiness has been shown to be tightly related to positive social relationships (Chaplin 2009; Holder and Coleman 2009; Pallini et al. 2014) in terms of having a best friend and being that person’s best friend (Betts and Stiller 2014), having a high number of friends (Jover and Thoilliez 2010; Myers 2000), and having a greater perception of social self-efficacy (Bandura et al. 1999). Indeed, different studies have shown that friendship and positive social relationships represent an important aspect of psychological development and well-being (Smetana 2006). In the same way, social self-efficacy has been examined mainly in adolescents, and it represents the perceived efficacy in developing supportive social relationships that enhance life satisfaction and provide positive outcomes, including positive thinking and happiness (Caprara et al. 2010; Caprara et al. 2006)

Regarding other individual characteristics in children, some studies showed that, in both children (Phillips et al. 2008) and adolescents (Cheng and Furnham 2004; Huebner 1991), high levels of self-esteem, which is defined as a person’s confidence in his or her own worth or skills, and prosocial behavior, which is defined as voluntary behavior meant to benefit another (Cillessen and Rose 2005; Clark and Ladd 2000; Eisenberg et al. 2007; Hastings et al. 2007; Holder and Coleman 2008; Östberg 2003), are important predictors of happiness in children.

On the contrary, high levels of loneliness, which is defined as an undesirable feeling that may result from an unfulfilled desire to have friends; a lack of affective bonding (Asher et al. 1984; Asher and Paquette 2003; Csikszentmihalyi and Hunter 2003); behavioral problems—such as emotional symptoms, poor conduct, or peer problems (McAuley and Layte 2012; Moore and Keyes 2003; Vierhaus and Lohaus 2008; Vierhaus et al. 2013); and poor economic conditions (Breslin et al. 2017; McAuley and Layte 2012; Pugh 2010) have been identified as some of the main barriers to happiness in children.

1 Age and Gender Differences on Predictors of Happiness

Existing research indicated that younger children reported higher levels of life satisfaction and a more positive affect than older children did (Chang et al. 2003) whereas no significant differences were found in the subjective happiness levels between male and female children (Uusitalo-Malmivaara and Lehto 2013) or adolescents (Huebner et al. 2000; Park and Huebner 2005). However, a considerable amount of research documented gender differences in the majority of predictors of happiness in children above mentioned—such as self-esteem, social relationships, friendship, loneliness, and so on. In addition, several studies suggested that the predictors and the levels of subjective well-being and life satisfaction may differ for children of different ages (Chang et al. 2003; Holder et al. 2010). Thoilliez (2011) indicated that some components of friendship (such as reciprocity, positive affect, and trust) and the relationships with friends are more important for older children than for younger children. On the contrary, having friends has been found to be correlated from childhood through old age with well-being (Hartup and Stevens 1997), while children reporting more loneliness at younger ages than at older ages (McGuire and Clifford 2000). Moreover, research documented gender differences in friendship quality. Specifically, females reported higher levels of quality and lower levels of conflict in their friendships compared to the levels reported by males (Fehr 1996; Kretsch et al. 2016), but males and females do not differ in perceived social efficacy (Pastorelli et al. 2001) and loneliness both in children (Dunn et al. 2007) and adolescents (Kingery et al. 2011; Koening and Abrams 1999).

Young children have higher self-esteem than older children and adolescents do, and this may be because young children do not yet possess the ability to engage in the process of self-evaluation, which protects them from thinking negatively about themselves (Harter 2012). On the contrary, prosocial behavior increases with age (Eisenberg and Fabes 1998; Tremblay 2003) whereas behavior problems are the same regardless of the children’s ages and genders (Schneiders et al. 2003). Other studies reported that females generally have lower levels of self-esteem (Gentile et al. 2009; Israel and Ivanova 2002; Mendelson and White 1985), fewer behavioral problems (Gissler et al. 1999), and higher levels of prosocial behavior (Eisenberg et al. 2007; Hay and Pawlby 2003) compared to males.

2 The Present Study

Most of the research reported in the previous section was conducted in the United States, Canada, and the United Kingdom. In Italy, to our knowledge, few studies on children’s happiness have been conducted (Businaro et al. 2015; Migliorini et al. 2018), and the research has produced more evidence about the subjective well-being of adolescents and adults (Capanna et al. 2005; Di Fabio and Ghizzani 2007; Monzani et al. 2011). Recently, Migliorini et al. (2018) conducted a study to investigate 8-year-old Italian children’s subjective well-being (n = 1145), exploring in more depth the role of the family and school contexts in their life satisfaction. According to previous research (Savahl et al. 2015), the authors found a high level of subjective well-being, and males and females were equally satisfied with their lives. Previously, in a sample of 132 Italian children between the ages of 8 and 11, Businaro et al. (2015) found that the family, friendships, sports, and school context were the most important sources of well-being, without significant differences between males and females.

Despite increased efforts in the research on children’s happiness, to our knowledge, no previous study has assessed happiness or children’s individual and social characteristics as perceived by both parents and children in a sample of child participants, especially in Italian context. Therefore, the primary objective of the present research was to evaluate the contribution of some background, relational, and individual characteristics of the children to their happiness. Specifically, in line with the literature (Breslin et al. 2017; McAuley and Layte 2012; Pugh 2010), we expected that (Hypothesis 1) high level of socioeconomic status of the family would be positively associated to children’s happiness: Children with higher socioeconomic status have access to more of the resources needed to support their well-being and happiness than do children with lower socioeconomic statuses (Bornstein and Bradley 2014). Consistent with other studies (Chaplin 2009; Holder and Coleman 2009; Pallini et al. 2014) that found a positive association between positive social relationships and happiness, we hypothesized that (Hypothesis 2) some relational variables of the children — such as having a best friend and friendship reciprocity, i.e. being the best friend’s best friend (Betts and Stiller 2014), having a high number of friends (Jover and Thoilliez 2010), and having a greater perception of social self-efficacy (Bandura et al. 1999) — would be positively associated with children’s happiness. Consequently, we suspected that (Hypothesis 3) the lack of friendships or affective bonding, which can be considered as high loneliness (Asher and Paquette 2003; Csikszentmihalyi and Hunter 2003), would represent the main barriers to happiness in children. According to the previous studies (Eisenberg et al. 2007; Hastings et al. 2007; Holder and Coleman 2008), we hypothesized that (Hypothesis 4) high levels of self-esteem of the children would be important predictors of their happiness. We also expected that (Hypothesis 5) high levels of prosocial behavior (Eisenberg et al. 2007; Hastings et al. 2007; Holder and Coleman 2008) and (Hypothesis 6) fewer behavioral problems (McAuley and Layte 2012; Moore and Keyes 2003; Vierhaus and Lohaus 2008; Vierhaus et al. 2013) a would be important predictors of children’s happiness . Finally, based on the aforementioned studies and given conflicting empirical findings, we had no specific hypotheses regarding the direction of gender’s and age’s effects on happiness.

3 Method

3.1 Participants

The non-representative sample consisted of 1363 Italian children (53% females, 47% males) enrolled in 16 primary and secondary schools in Central and Southern Italy. Participants’ ages ranged from 7 to 14 (female children: Mage = 11.12, SD = 1.53; male children: Mage = 11.13, SD = 1.44). No significant differences were found between the groups of females and males, (t[1361] = .13, p = .898) with respect to age. Each child was also given a parent questionnaire to take home, and only one parent of the child was asked to participate. The parents were asked to respond to sociodemographic questions as well as to complete the Strengths and Difficulties Questionnaire (SDQ; Goodman 1997). All parent questionnaires were returned. The parent sample consisted of more mothers (n = 1012, 74%) than fathers (n = 351, 26%). Almost 9% of parents (n = 71) reported their socioeconomic statuses to be above average, whereas 83% of them reported average socioeconomic statuses (n = 1135).

3.2 Procedures

All of the students completed the paper questionnaires during regular school hours. Written permission was obtained from the headmasters and headmistresses of the schools as well as from the institutional school committees, which included parents and teachers. Written informed consent was also obtained from the parents of the participating pupils, who brought the forms home for their parents to sign and subsequently returned them to their schools. The study was introduced to the students and the questionnaires were administered under the supervision of both the teacher and the researcher. We explained to them that the purpose of this research was to examine the happiness in children and preadolescents. Students were assured of anonymity and were provided with the option not to participate in the project. The study design was cross-sectional and the inclusion criteria were: (a) Italian nationality; (b) native speaker of Italian or spoke Italian as a second language; (c) age (≤14 years old). According to such criteria, 35 students were not included in the analyses: 20 were not Italian students; 10 were not native Italian speaker; 5 exceeded the age range. No compensation was provided for filling out the questionnaires. The respondents answered the same questionnaire packet individually, taking about 10 to 15 min to complete the questionnaires. A total of 97% of the distributed questionnaires were completely filled out. The study was conducted in accordance with ethical standards of the responsible committees on human experimentation and with the 1964 Helsinki Declaration. Data were collected from 2016 to 2017 and, before the data collection began, the protocol received the approval of the Ethics Commission of the Department of Developmental and Social Psychology of the Sapienza University of Rome.

3.3 Measures

Identifying Information

Children and parents completed an identifying form to collect data related to sociodemographic characteristics, such as children’s gender and age. Parents were asked to report their socioeconomic statuses by answering a single item (1 = very poor, 2 = poor, 3 = average; 4 = good, and 5 = very good).

Relational Variables

Each child was asked to indicate whether he or she had a best friend (“Do you have a best friend,” coded as “no” = 0, “yes” = 1). In the case of the “yes” alternative, the children could specify whether they felt that the individuals nominated would also consider them their best friends (friendship reciprocity), with the following question: “Are you his/her best friend, too,” coded as “no” = 0, “yes” = 1. Moreover, friendship quantity was measured by a single item: “Please indicate how many friends you have.”

Happiness and Satisfaction Subscale

The Happiness and Satisfaction Subscale (HAP) of the Piers-Harris Children’s Self-Concept Scale 2 (Piers-Harris 2; Piers & Herzberg 2002) was used as one estimate of children’s happiness and satisfaction with life in general. Children responded with “yes” or “no” to 10 items that expressed how they may feel happy and satisfied with life (e.g., “I’m a happy person” or “I like being the way I am”). Negative items were scored in reverse, whereby a higher score indicated greater happiness. The HAP was used as an estimate of children’s happiness in earlier studies (Holder and Coleman 2009; Wood et al. 1996; Young and Bradley 1998). In the present study, the Cronbach’s α was .75.

UCLA Loneliness Scale

The UCLA Loneliness Scale (UCLA-LS; Russell 1996) is a five-item questionnaire designed to assess individuals’ general levels of loneliness. Valkenburg and Peter (2007) adapted this short form. Children were asked to indicate how often they felt lonely (e.g., “There is no one I can turn to” or “I feel as if nobody really understands me”). A total score derived from the four-point Likert-type scale ranged from 1 (never) to 4 (always). A higher score indicated higher loneliness. In the present study, the Cronbach’s α was .73.

Rosenberg Self-Esteem Scale

The Rosenberg Self-esteem Scale (RSE; Rosenberg 1965) is a 10-item questionnaire used to assess global self-esteem in children, defined as respect and acceptance toward oneself. The questions feature a four-point Likert-type scale, ranging from 1 (totally disagree) to 7 (totally agree). An example of an item is “On the whole, I am satisfied with myself” or “I take a positive attitude toward myself.” Negatively loaded items were scored in reverse. The total score of each participant was calculated as the sum of the 10 items, with higher values corresponding to higher degrees of self-esteem. The Italian version indicated good internal consistency (Antonelli et al. 2000). In the present study, the Cronbach’s α was .67.

Perceived Social Self-Efficacy Subscale

The Perceived Social Self-Efficacy Subscale (PSSE; Pastorelli et al. 2001) of the Children’s Perceived Self-Efficacy Scale (Bandura et al. 1996) included 13 items assessing the capacity for peer relationships, self-assertiveness, and leisure-time social activities as perceived by children. An example of an item is “How well can you learn regular physical education activities?” or “How well can you express your opinions when other classmates disagree with you?” A total score derived from the five-point Likert-type scale ranged from 1 (not at all capable) to 5 (fully capable), whereby a higher score indicated a greater ability to form and maintain social relationships, be assertive, and complete leisure activities. In the present study, the Cronbach’s α was .80.

Strengths & Difficulties Questionnaire

The SDQ is a 25-item scale of psychopathology and pro-social behavior for children ages 3 to 16 years that parents, teachers, or adolescents can complete. This scale is one of the most frequently used instrument to assess these constructs (Becker et al. 2004; McAuley and Layte 2012; Vierhaus et al. 2013). We administered the questionnaire only to parents. The questions feature a three-point Likert-type scale ranging from 0 (not true) to 2 (certainly true). The SDQ assesses four domains of behavioral problems as perceived by parents: emotional symptoms (e.g., “often unhappy, downhearted”), conduct problems (e.g., “often fights with other children”), hyperactivity problems (e.g., “easily distracted, concentration wanders”), and peer problems (e.g., “rather solitary, tends to play alone”). In the present study, the total difficulties score (total SDQ) was calculated as the sum of the four subscales, with higher values corresponding to higher degrees of behavioral problems and difficulties (range: 0–40). Negatively worded statements were scored in reverse. Moreover, the SDQ also assessed prosocial behavior (PB) using another five items. The prosocial subscale measured the child’s ability to act prosocially (e.g., “considerate of other people’s feelings”), independent of the difficulties that the other four subscales measure. A prosocial score was generated as the sum of the five items (range = 0–10). A higher score indicated greater prosocial behavior. The SDQ has good reliability and validity (Goodman 2001). In the present study, reliability analyses revealed a good level of internal consistency: total SDQ (α = .86) and PB (α = .77).

3.4 Statistical Analysis

We used the Statistical Package for the Social Sciences (SPSS 24.0) to conduct the analyses. Gender differences (male vs. female children, and father vs. mother) were analyzed using the chi-squared test and univariate analysis of variance (ANOVA), in which partials eta squared (ηp2) were calculated as measures of effect size. Bivariate correlations were performed to assess the relationships between variables. The internal consistency was measured by using Cronbach’s α. Moreover, hierarchical multiple regressions were conducted to test the interactive effects of gender, age, and the other variables inserted into this study on happiness as perceived by children. Before regression analysis were performed, linearity, homoscedasticity, normality of residuals, and multicollinearity assumptions were assessed. Additionally, our sample size seems adequate for the regression analysis, exceeding the most conservative recommendation of 30 participants per each independent variable (Porte 2010).

4 Results

4.1 Gender Differences between Variables

Descriptive statistics of the measures used in the study are displayed in Table 1. A chi-squared test and univariate analysis of variance was used to examine the differences between gender groups (male vs. female children, and father vs. mother) in the variables. No differences were found between fathers’ and mothers’ perceptions of their socioeconomic statuses, F(1, 1361) = .87, p = .349, ηp2 = .001, total SDQ, F(1, 1361) = .94, p = .334, ηp2 < .001, and PB scores, F(1, 1361) = 2.19, p = .139, ηp2 = .002. Conversely, male children presented higher numbers of friends, F(1, 1361) = 4.15, p = .042, ηp2 = .003, higher levels of happiness, F(1, 1361) = 12.81, p < .001, ηp2 = .009, higher perceived social self-efficacy, F(1, 1361) = 5.52, p = .019, ηp2 = .005, and lower levels of loneliness, F(1, 1361) = 45.67, p < .001, ηp2 = .032, in comparison with female children. For parsimonious reasons, the other results are not reported because no statistically significant differences were found between them.

Table 1 Descriptive (means, standard deviations, percentages and gender differences) of the sample’s characteristics
Full size table

Table 2 presents the correlations between the key variables. We found that children’s happiness was positively related to socioeconomic status, r = .06, p < .05, having a best friend, r = .12, p < .01, friendship reciprocity, r = .14, p < .01, number of friends, r = .06, p < .05, global self-esteem, r = .35, p < .01, and level of perceived social self-efficacy, r = .38, p < .01. Conversely, children’s happiness was negatively correlated with children’s age, r = −.09, p < .01, loneliness as perceived by children, r = −.43, p < .01, and the total difficulties score as perceived by parents, r = −.09, p < .01. Prosocial behavior score was not associated with children’s happiness, r = .04, p > .05.

Table 2 Pearson’s r between happiness in children and other variables considered in the study
Full size table

4.2 Predicting Children’s Happiness

Using hierarchical multiple regression, we further examined how children’s gender and age, socioeconomic status, having a best friend and the friendship reciprocity, number of friends, loneliness, global self-esteem and perceived social self-efficacy as perceived by children, behavioral problems, and prosocial behavior as perceived by parents may relate to happiness in children (Table 3). A preliminary investigation showed that data met assumptions of linearity, homoscedasticity, normality of residuals, and multicollinearity. In the first step of the regression, we entered the children’s gender and age, as well as socioeconomic status (identifying variables). Having a best friend, the friendship reciprocity, and the number of friends (relational variables) were entered in the second step; the psychological adjustment variables that children reported (loneliness, self-esteem, and social self-efficacy) in the third step; behavior problems and prosocial behavior as evaluated by parents in the fourth step; and all possible interactions between these variables with gender and age in the last step. Prior to the analyses, all continuous measures were mean centered to limit multicollinearity (Aiken and West 1991).

Table 3 Hierarchical regression analyses for child and parent variables predicting children’s happiness
Full size table

The analysis of the sample revealed that a high score of happiness in children was associated with lower age, β = −.14, t = −2.95, p = .03, higher socioeconomic status, β = .25, t = 2.59, p = .01, and friendship reciprocity, β = .54, t = −2.06, p = .39. In addition, lower levels of UCLA-LS, β = −.41, t = −2.51, p = .01, and higher levels of RSE, β = .62, t = 3.81, p < .001, were significantly associated with happiness. Conversely, happiness was not associated with gender, β = −.78, t = − .85, p = .39, having a best friend, β = .13, t = .45, p = .65, number of friends, β = .02, t = .16, p = .87, PSSE, β = .12, t = .78, p = .43, total SDQ, β = .05, t = .03, p = .97, and PB score, β = .26, t = 1.74, p = .82.

Furthermore, the interactions in Step 5 of the regression analysis explain additional significant variance in happiness as perceived by children (3%). To further explore the nature of the interaction effects, simple slopes analyses were performed. In particular, a two-way interaction was found between age and UCLA-LS, β = − .48, t = −2.96, p = .003. The simple slopes analyses revealed that for high levels of UCLA-LS, higher happiness scores were associated with younger participants, β = −.31, t = − 4.79, p < .001, indicating that older participants with higher levels of loneliness were less happy than younger participants with high levels of loneliness were (Fig. 1). In contrast, the relation between happiness and age was nonsignificant for lower levels of loneliness, β = .05, t = .77, p = .44.

Fig. 1
figure 1

Children’s age as a moderator of happiness (HAP)* and loneliness (UCLA-LS)*

Full size image

The two-way interaction between gender and the PSSE scale was significantly related to happiness, β = .19, t = 1.97, p = .04. Simple slopes analyses indicated that for low levels of perceived social self-efficacy, higher happiness scores were associated with male participants, β = − .29, t = −2.15, p = .03, indicating that female participants with lower levels of perceived social self-efficacy were less happy than male participants with low levels of PSSE were (Fig. 2). In contrast, the relation between happiness and gender was nonsignificant for higher levels of perceived social self-efficacy, β = .12, t = 90, p = .37.

Fig. 2
figure 2

Children’s gender as a moderator of happiness (HAP)* and social self-efficacy (PSSE)*

Full size image

Moreover, the three-way interactions among age, gender, and RSE, β = .21, t = 2.04, p = .04, and between age, gender, and PB, β = .24, t = 2.77, p = .006, proved to be significantly and positively related to happiness. To advance interpretations of the significant three-way interactions further, we plotted the interaction effects for two levels of gender (males vs. females). Thus, for each level of gender, we plotted the relationship between age and happiness for low and high levels of global self-esteem (Fig. 3), and for low and high levels of prosocial behavior as perceived by parents (Fig. 4).

Fig. 3
figure 3

Children’s age as a moderator of happiness (HAP)* and self-esteem (RSE)* only in female children

Full size image
Fig. 4
figure 4

Children’s age as a moderator of happiness (HAP)* and prosocial behavior (PB)* only in female participants

Full size image

Regarding the interaction effect of global self-esteem scores, Fig. 3 indicates that age had the strongest relationship with happiness when the levels of self-esteem in females were low. On the contrary, in male participants or in all children with high levels of self-esteem, age was not associated with happiness. According to a simple slope, only when the levels of RSE in female participants were low was age significantly and negatively related to high levels of happiness, β = −.47, t = −5.07, p < .001, indicating that older females with lower levels of global self-esteem were less happy than younger females with low levels of RSE were. When the levels of global self-esteem in female participants, β = .01, t = − .01, p = .99, and in male participants, β = − .01, t = − .03, p = .97, were high, or when these scores were low in male participants, β = − .03, t = − .37, p = .71, the relation between age and happiness was not significant.

Figure 4 reveals the interaction effect of prosocial scores. Simple slopes analyses indicated that only when the levels of PB as perceived by parents in female participants were low was age significantly and negatively related to high levels of happiness, β = −.37, t = − 4.09, p < .001. This indicated that older females with lower levels of prosocial behavior were less happy than younger participants with low levels of PB were. When the levels of prosocial behavior in female participants, β = −.11, t = − 1.15, p = .25, and in male participants, β = −.12, t = − 1.29, p = .19, were high, or when these scores were low in male participants, β = .77, t = .91, p = .36, no significant relation was found between age and happiness. The adjusted R2 for the entire model was .40.

5 General Discussion

This study was intended primarily to contribute to the extension of knowledge about children’s subjective well-being and to analyze the factors that could improve their sense of happiness. Moreover, the effect of children’s gender and age on their happiness was studied along with all possible interactions with the potential factors underlying their subjective well-being, such as the role of social relationships, loneliness, self-esteem, social self-efficacy, behavior problems, and prosocial behavior. In addition, an element of originality of this research involved taking into account the perspective of both children and parents when assessing children’s subjective well-being.

First and foremost, to explore the predictors of children’s happiness, we hypothesized that some background, relational, and individual characteristics of children may be associated with high levels of happiness. The findings of hierarchical multiple regression confirmed the hypotheses partially. Indeed, the results revealed that younger children with high socioeconomic conditions (Hypothesis 1) were happier. These results are consistent with previous research indicating that younger children (Chang et al. 2003) and those with high socioeconomic statuses (Breslin et al. 2017; McAuley and Layte 2012; Pugh 2010) reported higher subjective well-being. Previous studies (Marsh 1989; Wigfield et al. 1991) revealed declines in self-concepts and life satisfaction from early childhood to preadolescence (age 11–14). Indeed, childhood is characterized by positive self-perceptions (Chang et al. 2003), whereas preadolescence and adolescence are more emotionally stressful periods. Moreover, a near-worldwide agreement exists that children with high socioeconomic statuses have access to more of the resources needed to support their positive development than do children with lower socioeconomic statuses (Bornstein and Bradley 2014).

Happiness has been positively associated with having a mutual friend (friendship reciprocity) but not with number of friends or having a best friend (Hypothesis 2). A possible explanation for this result exists: Having a best friend or a high number of friends is less likely to provide experiences for subjective well-being (Bukowski et al. 1993). For example, previous studies found that spending time with many friends, without friendship reciprocity, probably influences children to use strategies that may be negative to their subjective well-being because they are non-congruent with and contradictory to their own self-concepts (Al Nima et al. 2012). Consequently, the development of false self-behavior may lead to negative outcomes and feelings of inadequacy (Tkach and Lyubomirsky 2006). Conversely, children who do not have mutual friends are at risk for victimization and behavior problems (Hodges et al. 1999), psychopathology, and psychological maladjustment (Fink et al. 2015).

Of the individual child variables entered into the regression analyses in Step 3, only low level of loneliness (Hypothesis 3) and high level of self-esteem (Hypothesis 4) strongly predicted children’s happiness. The findings are also consistent with previous studies revealing that prolonged loneliness has the potential to undermine an individual’s psychological and emotional well-being (Dunn et al. 2007). Research has strongly supported a relationship among routine rejection by a peer group, low self-esteem, and loneliness (Buhs and Ladd, 2001), reporting that children whom their peers rejected showed higher levels of loneliness, and consequently lower self-esteem and subjective well-being, compared with children whom their peers did not reject.

The further purpose of this study was to verify if some variables that parents reported, such as their perceptions of children’s prosocial behavior (Hypothesis 5) or behavioral problems (Hypothesis 6), represented potential predictors of children’s happiness. Contrary to our expectations, the results revealed that low levels of behavior problems and high levels of prosocial behavior as perceived by parents were not significant predictors of higher levels of children’s subjective well-being.

Different meta-analysis reported the relevance of the “source of information” as a moderator of the relation between variables: Studies in which the same person reported two variables exhibited a higher correlation or effect size compared with those studies in which different people were used to measure the same variables (Pallini et al. 2018). In the present study, children evaluated happiness, and the psychological variables that the children reported were more highly correlated (from .38 to .43) compared with the variables that parents evaluated (from .04 to .09).

Gender was not found to be related to happiness. Although the literature found no gender differences on children’s happiness (Huebner et al. 2000; Park and Huebner 2005; Uusitalo-Malmivaara and Lehto 2013), we expected a significant association between these two variables since that the previous Anova (Table 1) reported that males were happier than females. On the other hand, several studies found that the predictors of happiness may differ for children’s age (Chang et al. 2003; Holder et al. 2010). For these reasons, the third purpose of the research was to explore interaction effects among gender, age, and each of the variables considered in this paper (Hypothesis 3). The first significant two-way interaction was found between age and loneliness, indicating that older children with high levels of loneliness were less happy than younger participants with high levels of loneliness (Fig. 1). This result is in line with the literature, which found an association between loneliness and younger age (McGuire and Clifford 2000). Consequently, when older children feel lonely, their levels of happiness decrease probably because they are in an emotionally stressful period (Marsh 1989; Wigfield et al. 1991) or because their peer groups have rejected them as discussed previously (Buhs and Ladd, 2001; Dunn et al. 2007).

The only other significant two-way interaction, between gender and PSSE, indicated that females with low levels of perceived social self-efficacy were less happy than males with low PSSE were (Fig. 2). This result may be explained through the greater importance that females attribute to friendships, in terms of quality, satisfaction, attachment, trust, and loyalty (Fehr 1996). In addition, females tend to have smaller and more intimate friendships, whereas males generally prefer larger and less-intimate friendship networks, emphasizing shared interests (Kretsch et al. 2016). When female children are not satisfied with their social relationships, their levels of happiness decrease. This is probably due to the fact that the relationships among them are important regardless of their age.

The first three-way interaction revealed that gender and age moderate the effects of self-esteem on children’s happiness (Fig. 3). In particular, older female children with low self-esteem reported lower happiness than younger females with low levels of RSE did. This result can be interpreted in light of previous studies on self-esteem, which found that self-esteem was lower in older children (Harter 2012) and in females (Israel and Ivanova 2002; Mendelson & White, 1985) compared with younger children and males respectively. This is in line with the traditional theory, which suggested that lower levels of self-esteem in female children are associated with the physical changes of puberty (Gentile et al. 2009; Rosenberg 1986). Indeed, females begin puberty earlier compared with male children, whose development is more delayed.

The only other significant three-way interaction, among gender, age, and PB (Fig. 4), indicated that older female children reported lower levels of happiness when their parents perceived them to have low levels of prosocial behavior. This pattern was not significant when the levels of PB were low in young female children or when they were high in males of all ages. In line with the literature, prosocial behavior is higher in females compared with males (Hay and Pawlby 2003), and they increase with age (Eisenberg and Fabes 1998; Tremblay 2003). According to Eisenberg and Fabes’s (1998) meta-analysis, children ages 13–15 years tend to be higher in prosocial and altruistic behaviors compared with those ages 7–12 years. Children over time develop an increasingly refined understanding of others’ emotional states and can better decode their emotional states. Moreover, older children are more likely to have the social experience necessary to perceive another’s need (Eisenberg 1986; Fabes et al. 1999).

Another possible explanation for this result may be the fact that females are expected to be more prosocial than males are among teachers, peers, and parents and other family members. This expectation is stronger for older but not younger female children. As reported previously, several studies revealed the tendency of females to be more prosocial than males are (Eisenberg et al. 2007; Hay and Pawlby 2003). These gender differences, based on stereotypic gender roles, may increase when children become more aware of and internalize these gender role stereotypes (Karniol et al. 1998). Consequently, older female children who violate gender role expectations regarding prosocial behavior may feel disapproved and unaccepted by others, and this may decrease their self-reported happiness. Our result seems to suggest that these traditional gender norms and expectations based on gender-role beliefs are still rife in Italian culture, according to previous research in the Italian context (Baiocco et al. 2013; Pistella et al. 2018; Salvati et al. 2018). However, we did not consider the adherence to gender norms in our study. Thus, this explanation is only speculative and is not supported by data from empirical study.

Finally, contrary to previous Italian studies on subjective well-being in children (Businaro et al. 2015; Migliorini et al. 2018), a significant difference was found between males and females regarding their happiness in this study (Table 1). Gender differences may vary in strength by the type of subjective well-being studied or by the happiness measured with self-reports, observational methods, or reports from others. Thus, this difference may stem from the various instruments used to assess children’s happiness. For example, the Piers-Harris 2 used in this study assesses both happiness and satisfaction with life (Piers and Hezberg, 2002), whereas the instruments used in Italian studies were different (Businaro et al. 2015; Migliorini et al. 2018). Indeed, in the literature are many tools that assess happiness and that measure different types of subjective well-being indices, and no uniform way in which to define the nature of happiness exists. Diener et al. (2009) suggested that happiness is composed of various levels of specificity and should be studied as a multi-faceted phenomenon.

6 Limitations of the Study and Future Research

This research featured several limitations. First, the generalizability of our findings is limited due to the use of a convenience sample and due to the use of self-report instruments that social desirability may have influenced. Moreover, all of the participants were Italian, and thus, these findings may not apply to children living in other national contexts. In addition, the age range was limited to 7–14 years, so the results may not be applicable to adolescents and young adults. This study was conducted in Italy, a country where traditional gender norms and gender inequalities still persist, so the results may not be generalizable to less sexist cultural contexts as well (Pistella et al. 2018). Furthermore, the present research was limited because not all variables related to friendships relations were included. For example, number of friends, having a best friend, friendship reciprocity, perceived social efficacy, and prosocial behavior were included, whereas the quality of children’s friendships, types of social interactions, and family relations were not.

Future research should consider individual, relational, and parental variables when investigating factors contributing to children’s happiness. It would be of interest to use qualitative methods to further analyze the predictors of children’s happiness. Furthermore, although no generally accepted measure of well-being or happiness exists for children (Diener et al. 2009), future research should use multiple measures of happiness and not be based only on self-report instruments. Future research could include measures designed to evaluate adherence to the gender norms of children or their internalization of gender roles (Salvati et al. 2018). Longitudinal studies would be particularly useful because the predictors of happiness may differ with age, and it is possible to monitor the course and development of happiness determinants.

Scientific literature should continue to study children’s happiness. Indeed, research may identify and promote interventions and school programs for enhancing happiness in children, including the knowledge of those predictors that may contribute to increasing their happiness. Finally, research into the assessment and enhancement of positive mental health is determinants, since that several studies reported that happiness and satisfaction with life represent more than the absence of psychopathological symptoms disorders (Diener et al. 2009). Hence, evaluating the statuses and determinants of children’s happiness and quality of life may help with understanding and preventing the onset of various illnesses in later childhood/adolescence, such as depression or other disorders.