A validation study of the Multidimensional Life Satisfaction Scale for Children

Studies in the area of psychology have turned to examining the role of life satisfaction in the adjustment of people in life stages other than adulthood, as a reflection of society and the changes it has experienced. Research that initially focused on adults (e.g., Solomon & Winfield, 1945) was extended to adolescents and children in the 1970s and 1980s (see Daly & Carpenter, 1985; Lessing, 1972). A search made by the authors in the PsycNET database indicates that more than 6000 articles on life satisfaction have been published up to 2016, but only a quarter of them focus on adolescents (13–17 years) and schoolchildren (6–12 years).

Satisfaction with life can be defined as the overall assessment of quality of life made by people according to their own criteria (Diener, Suh, Lucas, & Smith, 1999; Huebner, 1994) or as a cognitive response referring to evaluative judgments of satisfaction with life as a whole or in its multiple dimensions (Lucas, Diener, & Suh, 1996). This definition somehow implies the cognitive and emotional ability to consider and weigh several different sources of information.

For Lucas, Diener, and Suh (1996), satisfaction with life is (along with positive and negative affect) one of the dimensions of subjective well-being that refers to the frequency of positive or negative feelings. In other studies, however, the concepts of satisfaction with life and subjective or general well-being are usually superimposed or interchangeable (Cheung & Lucas, 2015; Kuppens, Realo, & Diener, 2008; Lyubomirsky, King, & Diener, 2005; Pavot & Diener, 2008; Poletto & Koller, 2011).

Studies on school-age children and adolescents have found associations between satisfaction with life and vulnerability and discrimination (Gilligan & Huebner, 2002; Poletto & Koller, 2011), school environment (Gilligan & Huebner, 2002; Poletto & Koller, 2011), work (Arteche & Bandeira, 2003), and an individual’s network of relationships (Serafini & Bandeira, 2011). Satisfaction with life has also been associated with perceptions of support from teachers (Guess & McCane-Bowling, 2016) and parents (Jiang, Huebner, & Hills, 2013), as well as satisfaction with the neighborhood (Shin, Morgan, Buhin, Truitt, & Vera, 2010). Satisfaction with life also predicts externalizing and internalizing problems (Gao et al., 2015; Haranin, Huebner, & Suldo, 2007).

Given the multiplicity and diversity of these associations, researchers have proposed models that cover the unidimensionality and multidimensionality of life satisfaction. The unidimensional, general, or global models (Diener, Emmons, Larsen, & Griffin, 1985; Pavot & Diener, 1993; Huebner, 2004) use a single scale to indicate levels of satisfaction with life in general. The multidimensional models (Giacomoni & Hutz, 2008; Huebner, 1994) provide a profile of satisfaction with life in various specific domains.

Proctor, Linley, and Maltby (2009) have reviewed the instruments developed from the unidimensional and multidimensional models to assess life satisfaction. Among the models of the second group, the authors present the Multidimensional Students’ Life Satisfaction Scale (MSLSS) (Huebner, 1994), a 40-item self-report scale designed to provide a profile of life satisfaction within five specific domains (family, friends, school, context, and self) and an overall assessment of general life satisfaction. The MSLSS is applicable for use with students aged 8–18. High indices of internal consistency were reported for each dimension (.82 for family, .85 for school, .85 for friends, .83 for context, and .82 for self) and for the full scale (.92).

Today, the MSLSS is used in different cultures, having been adapted and validated for young people in France (Fenouillet, Heutte, Martin-Krumm, & Boniwell, 2015), Iran (Hatami, Motamed, & Ashrafzadeh, 2010), Turkey (Irmak & Kuruüzüm, 2009), Serbia (Jovanovic & Zuljevic, 2013), and Italy (Zappulla, Pace, Cascio, Guzzo, & Huebner, 2014). The studies of adaptation for these countries clearly presented the stages of translation and semantic adaptation, exploratory and confirmatory factor analysis (CFA), and reliability indices of the instrument, maintaining the five-dimension structure, with good indices of fit and satisfactory internal consistency values (Cronbach’s alpha greater than .70).

The studies also explored the construct validity of the scale. For convergent validity, instruments were used to measure self-esteem (Jovanovic & Zuljevic, 2013; Zappulla et al., 2014), self-concept (Zappulla et al., 2014), positive relationships with others (Zappulla et al., 2014), positive affect (Jovanovic & Zuljevic, 2013), and other instruments of life satisfaction (Fenouillet et al., 2015; Irmak & Kuruüzüm, 2009). For divergent validity, instruments were used to measure depression (Fenouillet et al., 2015; Jovanovic & Zuljevic, 2013; Zappulla et al., 2014), mental health (Zappulla et al., 2014), and anxiety and stress (Jovanovic & Zuljevic, 2013).

Giacomoni and Hutz (2008) developed the Multidimensional Life Satisfaction Scale for Children (MLSS-C), a Brazilian instrument, based on the scale originally developed by Huebner (1994), seeking to adapt Huebner multidimensional model to the characteristics of the Brazilian culture. To ensure the ecological validity of the scale, the authors derived the MLSS-C items mainly from interviews conducted with Brazilian children aged 5–12.

The MLSS-C contains 50 items, distributed among six dimensions, including domains of life satisfaction not covered in Huebner’s original model, such as compared-self and non-violence. The self dimension consists of items that describe the self with positive characteristics, such as self-esteem, good mood, the ability to relate, and the ability to show affection. The compared-self dimension groups items characterized by evaluations comparative with peers. The non-violence dimension includes items with content associated with aggressive behaviors. The family dimension includes items that describe a healthy, harmonious, and affectionate family environment and satisfying relationships. Items in the friendship dimension refer to relationships with peers, the level of satisfaction in these relationships, and some indications of leisure, fun, and support. Finally, the school dimension includes items that describe the importance of school, the school environment, interpersonal relationships in this space, and levels of satisfaction with this environment.

Giacomoni and Hutz (2008) investigated the psychometric properties of the MLSS-C. They first conducted an exploratory factor analysis (EFA) with a sample of 661 students aged 7–12 (mean 10.6) from public and private schools in Porto Alegre in Rio Grande do Sul. The EFA provided a six-factor structure as the best solution, with factors corresponding to the proposed model. Five dimensions presented internal consistency rates ranging from .82 and .86; only one dimension, non-violence, obtained an index of less than .70 (.66). The total Cronbach’s alpha was .93. In a further study with 230 children, the authors found moderate positive correlations between all dimensions and a measure of self-esteem, as well as moderate negative correlations with measures of depression and anxiety (Giacomoni & Hutz, 2008).

The current evidence of adequate psychometric properties of the MLSS-C indicates that it is worth pursuing the validation process. As part of this process, there is a need to confirm the factorial structure derived from the EFA. Will the six-factor model maintain in a CFA? Given Brazil’s cultural diversity, will data from children in another region fit the same model? These questions guided the present study, which aims to validate the structure of the scale using CFA in a sample of children living in the state of São Paulo, Brazil. The sample participated in a broader prospective study from the fifth to the sixth year of elementary school.

In addition to the factorial validity, other contributions to the construct validity are presented through the study of some indicators of discriminant, convergent, and divergent validity. For the discriminant validity, each dimension’s average variance extracted (AVE) must be compared to its squared correlations with other dimensions in the model (Fornell & Larcker, 1981).

Convergent validity is demonstrated when there are correlations between instruments that measure similar traits or characteristics (Pasquali, 2007). Social skills and self-concept constructs were therefore chosen which (according to the literature) are moderately and positively associated with self, family, school and friendship dimensions of life satisfaction as measured by the MSLSS, an instrument similar to the MLSS-C (Jovanovic & Zuljevic, 2013; Zappulla et al., 2014). The hypotheses for the convergent validity analysis are that the dimensions of life satisfaction will positively correlate with positive self-concept and social skills.

Divergent validity is achieved when no correlations are obtained between the measures (Pasquali, 2007). The hypothesis is that life satisfaction will not correlate with stress symptoms, since it is described as a different construct not associated with life satisfaction, as indicated by Jovanovic and Zuljevic (2013).

Test-retest consistency, or stability, was assessed to verify the reliability of the instrument, as well as the internal consistency of the various dimensions of the MLSS-C using Cronbach’s Alpha (α) and composite reliability (CR).

This study aimed to contribute to the progressive ecological validation process of the MLSS-C, initially constructed for Brazil by Giacomoni and Hutz (2008). This instrument was based on the scale developed by Huebner (1994), and the aim of the current study was to improve its accuracy and appropriateness for use with Brazilian children. CFA was used to test the theoretical structure of the instrument from the empirical data (Marôco, 2010). Through the CFA and other construct validity indicators, this study confirmed the feasibility of the model with six dimensions proposed by the authors.

The dimensions of the MLSS-C cover different positive characteristics such as self-esteem, a good mood and the ability to relate, and perceived relationships within different contexts (family, school, and peers). The information assessed through the MLSS-C also comprises information regarding perceived satisfaction with fun and support, the ability to demonstrate affection, and interpersonal relationships in different contexts. It is a self-report measure that elicits self-assessment and social comparisons (with peers), leading to the production of both global and dimensional scores.

In addition to the CFA, the convergent, divergent, and discriminant validity of this scale was investigated. Our results seem to support convergent validity, since all MLSS-C dimensions show a moderate positive correlation with self-concept and social skills. The results concerning social skills are in line with those reported for the MSLSS by Jovanovic and Zuljevic (2013) and Zappulla et al. (2014) with adolescents. Both studies showed associations between dimensions of life satisfaction and positive relationships. Concerning self-concept, there is also some previous evidence of association between life satisfaction and positive self-evaluations in children and adolescents. Giacomoni and Hutz (2008) found a positive correlation between a measure of self-esteem and the six dimensions of the MLSS-C. More recently, Zappulla et al. (2014) reported positive correlations between self-acceptance and all dimensions of the MSLSS.

The results clearly support divergent validity for the school dimension of life satisfaction in relation to stress. For self, friendship, and non-violence, the correlations with stress were very low and negative. For compared self, family, and the total MSLSS score, low negative correlations with stress were observed. These last results are consistent with the findings reported by Jovanovic and Zuljevic (2013) in their study with adolescents aged 16–19. Since we worked with younger subjects in a narrow grade school range, the pattern of results we found concerning stress deserves further attention, especially with regard to its longitudinal evolution.

The discriminant validity was mostly satisfactory (below .50 for all factors). Results concerning AVE and the correlation matrix between each factor and total score suggest that the appraisals of life satisfaction are quite differentiated by domain among children of this age. The self and family dimensions were more associated with the global score, probably because they both include items that describe affection and positive characteristics of self and because they are still less influenced by meso- and macro systems’ feedback.

The internal consistency estimates were acceptable for five of the six dimensions. The non-violence dimension, however, presented a low Cronbach’s alpha value and low CR. Possibly, the small number of items (there were only two: “I fight a lot with my friends” and “Fighting solves problems”) influenced this result. It is worth noticing that a lower Cronbach’s alpha value was also observed in the original version, which had four items, including “I like fights” and “I’m angry” (Giacomoni & Hutz, 2008). This dimension thus seems less reliable for drawing conclusions about children’s life satisfaction, probably because of social desirability and/or the lack of cognitive maturity for self-assessment.

Given the low consistence of this dimension, it can be argued that it should be dropped out of the scale. Nevertheless, the authors consider it untimely to do so at this stage of MLSS-C validation, since later studies may develop or improve it. As Giacomoni and Hutz (2008) suggested, one possible direction for this improvement would be to increase the number of the items of the subscale. The non-violence dimension needs further exploration because of its striking importance for primary and secondary prevention of social victimization.

A possible limitation of this study lies in the timing of the retesting, which occurred after an important contextual change: the transition from elementary school I to elementary school II, between the fifth and sixth year. This transition requires adaptation in the children, as it involves changes that have effects ranging from the daily routine to the composition of the peer group, when students from the same class are allocated to different schools. This situation probably increased the variation of the results in the sixth year, as the transition is associated with positive consequences for some adolescents in terms of quality of life, while for others, it can lead to negative results (Bélanger & Marcotte, 2013). The bias thereby introduced increases the probability that the stability of MLSS-C has been underestimated, since the children were experiencing changes due to the transition when responding to the scale for the second time.

Another limitation of this study concerns the composition of the sample. While representativeness was ensured in relation to the pool of the city’s public schools, the exclusive focus on public school students in their fifth year may have restricted the variability of the data, affecting the results of the correlations.

The growing interest in the area of child and adolescent life satisfaction demands instruments with good psychometric properties. This study aimed to contribute to the progressive validation process of the MLSS-C. Together, the reported results broaden the evidence basis for the reliability and validity of this scale, suggesting that it is suitable for use with Brazilian schoolchildren in terms of both research and practice. Its multidimensional nature allows for obtaining global and specific scores, providing detailed information regarding the life satisfaction profile of the children. It would be important to add more waves to the research to investigate its longitudinal predictive power, as well as the relationship and/or differentiation between dimensions as children grow older and progress through school.

Considering the considerable cultural and socioeconomic diversity of Brazil, future studies should include samples from different states and cities to consolidate the ecological validity of the instrument. For further evidence of the scale’s external validity, it would also be interesting to triangulate information obtained through self-report with quantitative and qualitative information collected from parents, teachers, and other significant people in the life of the children.