Introduction
The main interest of this article is the evaluative aspect of self-concept, i.e., self-esteem (Rosenberg
1965; Harter
2012a). Self-esteem can be regarded as both a stable trait and an unstable state, as several events (both successes and failures) can impact momentary feelings of self-worth, particularly in specific domains of self-esteem (e.g., scholastic self-esteem; see: James
1890/1952; Crocker and Wolfe
2001). The adolescent period may be specifically dynamic with regard to self-esteem. The period is a time of dramatic developmental transitions in many domains of development, which influence changes in global self-esteem and its domains (Harter
2012b), as well as a time of changing social and educational contexts during school transitions (Harter
1990,
2006,
2012b).
The main aim of the present study was to describe the change and consistency of self-esteem during school transition. Following Bornstein et al. (
2017), the term “consistency” is used to describe two main developmental qualities. The first is “continuity”, which may be defined as the lack of change at a group (i.e., mean level of the tested construct), which in this study is self-esteem. The second is “stability”, which may be defined as the lack of change at the individual (i.e., rank-order level), of the relevant construct, here self-esteem. Continuity and stability were tested with regard to global self-esteem and domain-specific self-evaluations. Also taken into account were the differences between early and middle adolescence with regard to the above-mentioned aspects of self-esteem. Therefore, from a more general point of view, the study aimed to compare early and middle adolescents’ self-esteem to discover differences between these two phases of development, and, at the same time, from a dynamic point of view, to explore change, continuity and stability in self-esteem during these two phases of adolescence.
Global self-esteem and its domains in adolescence
Both Shavelson (Shavelson et al.
1976) and Harter (
1987;
2012b) refer to “self-esteem” as a multidimensional, hierarchically-organized construct, with global self-esteem at the apex of the hierarchy. This multidimensionality is a core difference between these theories and Rosenberg’s (
1965) concept of self-esteem, which represents a unidimensional approach. According to Harter (
2012a), global self-esteem (she called it “self-worth”) refers to the general evaluation of how much one likes oneself and how much one is happy with oneself and the way one is as a human being. The specific domains of self-esteem in adolescence include scholastic competence, social acceptance, physical appearance, athletic competence, romantic appeal, close friendships, job competence and behavioral conduct (Harter
2012a,
2012b). It should be noted that, in this model, global self-esteem is not a simple resultant or mere sum of points in domain-specific dimensions of self-esteem, as specific dimensions can relate differently to global self-esteem (Harter
2012a).
Although early and middle adolescents’ self-esteem can be described using similar domains, several factors related to these two periods of development can influence the dynamics of self-esteem in different ways. First, early adolescence is a time of more rapid and dynamic changes associated with puberty that may influence one’s self-evaluation in domains related to physical appearance or athletic competence (Harter
2012b). The biological, pubertal changes influencing adolescents’ physical appearance and attractiveness may also influence their global self-esteem as physical appearance seems to be the most important domain for the formation of this general self-evaluation (Harter
2000). Second, social changes related to the period of early adolescence, including the change of social environment during the transition to junior high school (Wigfield et al.
1991) and the increasing role of social comparisons in the formation of self-esteem (Cole et al.
2001), may influence self-esteem in the domains related to social context. Third, the transition to junior high school and related changes in academic and behavioral requirements may influence these domains of self-esteem (Wigfield et al.
1991). Although the middle adolescence period is not free from changes in the physical, social and academic domains, the growing maturity of adolescents and decreasing maturity gap (see Moffitt
1993) may serve as protective factors that limit the fluctuation of self-esteem. In particular, neurodevelopmental processes during adolescence related to cognitive and affective development may influence self-esteem changes (Steinberg
2005; Dumontheil
2016).
Change and consistency of self-esteem: a developmental perspective
The issue regarding self-esteem during adolescence that is still under debate is the change and consistency in the positivity/negativity of self-esteem. Empirical studies regarding this issue have produced contradictory results. Orth et al. (
2012) reported an increase, i.e., discontinuity in global self-esteem, from age 16 to middle adulthood and then a decrease into old age. On the other hand, a meta-analysis of 59 longitudinal studies on global self-esteem during the lifespan (Huang
2010) revealed only a small increase in global self-esteem during childhood, late adolescence and early adulthood and generally supported the idea of its continuity during the lifespan. The effect sizes for the change in global self-esteem were similar to those observed in personality traits (Huang
2010; see also Orth and Robins
2014). Similarly, the meta-analysis of individual stability, i.e., rank-order stability of global self-esteem during the lifespan (Trzesniewski et al.
2003), indicated an increase in stability of global self-esteem from childhood to the first decade of adulthood and then a decrease from the second decade of adulthood to old age. Therefore, one can conclude that global self-esteem obtains its consistent level in early adulthood, both at the mean and individual levels. Moreover, both meta-analyses revealed no effect of gender, suggesting a similar pattern of change for males and females (Huang
2010; Trzesniewski et al.
2003).
When concentrating on adolescence, the results relating to continuity and change in self-esteem are more contradictory. Theoretically, dynamic developmental changes during the adolescent period can result in a decrease in self-evaluations at a group level, and the tantrums of adolescence can be a specifically sensitive period in this area (Harter
2012b). Some cross-sectional (Robins et al.
2002) and longitudinal (Baldwin and Hoffman
2002) studies on self-esteem confirm this hypothesis, reporting a drop in global self-esteem between childhood and adolescence and between early and middle adolescence (Eccles et al.
1993). However, in other studies, this decrease was observed only in some domains of self-esteem and not for global self-esteem (Shapka and Keating
2005; Kuzucu et al.
2014). In Kuzucu et al. (
2014) longitudinal study of children between 9–16 years old, a significant—although only minor—decrease was found in the domains of physical appearance and behavioral conduct, accompanying an increase in the domains of athletic competence, social acceptance and academic competence. Global self-esteem remained continuous during this 7-year period of time. Some data also indicate continuity of self-esteem during the adolescence period, both regarding global self-esteem (Birkeland et al.
2012; Kuzucu et al.
2014) as well as specific dimensions of self-esteem (Young and Mroczek
2003). Finally, some studies have reported an increase in global self-esteem between early and middle adolescence (Moneta et al.
2001) and beyond (Erol and Orth
2011).
The discrepancies in these results regarding the continuity and change of self-esteem may stem from several reasons. One may be a differentiated conceptualization of self-esteem, with a focus on either global self-esteem, measured using different questionnaires, or on its specific dimensions (Kuzucu et al.
2014). The other may be tied to the different ages of participants as adolescents in the early and middle phase of this period face different developmental changes and challenges that may influence their self-esteem (Harter
2012b). Moreover, most studies included only a group level of analysis, studying changes in the mean level of self-esteem, whereas substantial individual differences regarding the trajectories of self-esteem have been observed (Kuzucu et al.
2014). For example, Harter and Whitesell (
2003) reported no change in the mean value of global self-esteem in adolescents transitioning from high school to college. However, at the individual level, the pattern of results was much more differentiated, with a decrease of global self-esteem in 17% of adolescents, an increase in 23% and stability in 40% of the group. Therefore, a strong need also exists to explore individual stability and change in self-esteem in early and middle adolescence. This claim is also supported by results of studies using latent growth modeling, which indicate that individual differences exist in global self-esteem, in specific domains of self-esteem and also in change in self-esteem during adolescence (Young and Mroczek
2003; Birkeland et al.
2012). Finally, the measures used to assess global self-esteem may also be an important factor influencing the results. The meta-analysis by Huang (
2010) revealed an important effect of using Harter’s Self-Perception Profile for Adolescence (SPPA; 2012a), which differs from both the Rosenberg Self-Esteem Scale (Rosenberg
1965) and the Coopersmith Self-Esteem Inventory (Coopersmith
1981). The differences may be due to a different conceptualization of the construct of global self-esteem (see Harter
2012a). It also should be noted that the observed patterns of change in self-esteem may depend on the time intervals taken into account, as some life events may create contexts for short-term changes in self-esteem (Chung et al.
2014).
School transition as the context for self-esteem changes
The decrease in global self-esteem during childhood and adolescence may be related to the experience of stressful events (Baldwin and Hoffmann
2002), one of which may be school transitions (Wigfield et al.
1991; Eccles et al.
1993; Chung et al.
2014). School transitions can be a stressful event as it involves changes in academic demands and social contexts (Harter et al.
1992), involving a greater emphasis on evaluation and social comparison among students, stricter grading standards and a disruption of young adolescents’ social networks (Wigfield et al.
1991). Therefore, the domains of academic performance and social acceptance, as well as global self-esteem, may be mostly prone to change during this time. Seidman et al. (
1994) reported a decline in both self-esteem and perceived positivity of peer context (e.g., perceived social support) during school transition in early adolescence among poor youth. However, Cantin and Boivin (
2004) observed an increase in the self-evaluation of social acceptance during the transition from elementary school to junior high school, likely due to the intensification of supportive relationships with school friends. On the other hand, they observed a decrease in perceived scholastic competence, followed by a decrease in global self-esteem. As Cole et al. (
2001) indicated, school transitions for many adolescents may be related to the discontinuity of self-esteem; however, the direction of change may differ depending on the developmental period as they observed a drop in perceived academic competence during middle school transition but an increase in this domain during transition into high school (Cole et al.
2001). Moreover, in early adolescence, a significant drop in continuity of self-perceived competence in most domains was observed, whereas in middle adolescence, self-evaluations in most domains increased and reached their plateau (Cole et al.
2001), suggesting that school transition, combined with developmental changes of early adolescence, may create a context that is particularly challenging for young people and their self-evaluation (see also Simmons et al.
1987). On the other hand, studies also reported a lack of decrease (or even an increase) in self-esteem and perceived competence during transition from elementary to junior high school (Hirsch and Rapkin
1987; Proctor and Choi
1994), indicating a need for further research.
Current Study
The main aim of the present prospective cross-sectional study was to assess the continuity and stability of self-esteem in early and middle adolescence by focusing on global self-esteem and domain-specific self-evaluations. Moreover, it aimed to observe the short-term trajectory of self-esteem development just after the school transition for both age groups. The combined analyses of changes observed at the mean—or group—level and individual—or rank-order—level would fill an important gap in studies on self-esteem in adolescence. Moreover, analyzing the changes in both early and middle adolescence in the context of school transition may add an important contribution to understanding differences between younger and older adolescents, providing both a more detailed picture of the challenges they face, as well as implications for the support of their development.
Given the importance of school transitions, it should be specified that the Polish educational system has several transitions, and only one of them is obligatory: the transition from elementary school (which lasts for 6 years) to junior high school (3 years). After junior high school, several options are available: it is possible to end formal education or to transition to the next level, which may be high school (about 50% of adolescents choose this), technical college or vocational school. The current study included adolescents after the transition to junior high school and high school.
Taking into account that the adolescent period and its challenges (among them, school transition) may create fluctuations in self-esteem, a decrease in self-esteem over the tested period, i.e., in the context of school transition, was hypothesized (Hypothesis 1). As several studies have shown that after the initial decrease in self-esteem in early adolescence, a growth of self-esteem in later adolescence and adulthood is observed, it was expected that self-esteem in middle adolescence would be higher than in early adolescence, both regarding global self-esteem and specific domains (Hypothesis 2a) and that the decrease observed in self-esteem in early adolescence would be steeper than in middle adolescence (Hypothesis 2b). Taking into account the individual level of analyses, it was hypothesized (Hypothesis 3) that there is a stability of individual differences in self-esteem throughout adolescence over a year and a half, and this stability may be higher in middle than in early adolescence, as previous studies have revealed considerably high (increasing with age) rank-order stability of self-esteem.
Discussion
Previous studies on the development of self-esteem have concentrated mainly on the description of the change observed at mean (or group) level in global self-esteem in the context of school transitions. In this study, Harter’s Self-Perception Profile for Adolescents was used and thus both global self-esteem and self-evaluations in eight domains were assessed. Moreover, these were assessed three times over the period of a year and a half in two age-groups: early and middle adolescents (i.e., 13- and 16-year-olds). Moreover, both the change and the consistency of global self-esteem and domains’ self-evaluations were analyzed, taking into account both group (mean) and individual level. This seems particularly important, as the pattern of change or consistency may differ depending on the level of analysis. Additionally, a comparison was conducted between patterns of development observed during early and middle adolescence.
On the whole group level, a significant change following school transition was observed in three domains of self-esteem: increase in behavioral conduct and job competence and decrease in social acceptance. The remaining five domains of self-esteem, as well as global self-esteem, remained stable during the 18 months following the school transition. The decrease in social acceptance was congruent with expectations, and this may result from changes in adolescents’ social networks connected to school transition (Wigfield et al.
1991), as many adolescents face challenges finding a place in a new group of peers and teachers. An increase in behavioral conduct and job competence may stem from an increasing insight into self-responsibility regarding correctness of behavior and social expectancies with regard to “adult-like” behavior. In line with this interpretation, Crocetti et al. (
2019), in their longitudinal study with middle adolescents, used items from the behavioral conduct subscale of the SPPA to measure self-perceived morality and also found that females’ morality increased over the 2-year period. However, one should note that, generally, at a group level, self-esteem seems rather continuous for most domains of self-esteem, as well as for global self-esteem, and no changes were observed following the school transition, indicating that this context may be not as challenging for the evaluative aspect of self-concept as one may have supposed based on earlier research (Cole et al.
2001; Harter et al.
1992; Simmons et al.
1987). Note that most of this research was conducted in the XX century, and thus the results of this study are important as the new insight in self-esteem development in XXI-century adolescents was provided.
It should be noted that the observed pattern of change in self-esteem was inter-individually differentiated, but, in general, the age group (early versus middle adolescence) was not responsible for this differentiation. In other words, although there generally were no differences between the age groups in self-esteem development based on the results of the conditional LGC, when age was taken as a covariate, likely in middle adolescence, changes in physical appearance and global self-esteem were more differentiated than in early adolescence. One may suggest that in comparison with previous research (Cole et al.
2001), the development of self-esteem as identity development (Klimstra et al.
2010) might be prolonged or delayed, and thus changes in the most subjectively important aspect of self-esteem in adolescence (Harter
2012b); i.e., global self-worth and physical appearance, are only observed in middle adolescents. However, this effect should also be further analyzed (probably with a larger sample) as at the whole group level, these two aspects of self-esteem did not change. Therefore, the hypothesis that a change or even a decrease in self-esteem in early adolescence may be more pronounced (Cole et al.
2001; Harter
2012b) was not confirmed. It may be supposed that, generally, school transition in adolescence may influence self-esteem in a similar way, independently of the level of education. On the other hand, taking into account that there are inter-individual differences in self-esteem development, the lack of age differences when the whole group was taken into account could suggest additional factors not taken into consideration in the present study that exist and influence different courses of self-esteem development. These additional factors may be related to peers’ social support (Hirsch and DuBois
1991), body image and relations with parents (Birkeland et al.
2012) or shifts in life events and family cohesion (Baldwin and Hoffmann
2002), as well as individual factors related to socio-economic status (SES), race or social class (Rhodes et al.
2004). Analyzing these factors definitely deserves further studies.
Although it cannot be completely confirmed that clear differences exist in the rate of change in self-esteem between early and middle adolescence, differences were found in the initial levels between age groups. Contrary to expectations, self-esteem was higher in the younger group (in the cases of scholastic competence, social acceptance, athletic competence, physical appearance and global self-esteem). No differences between age groups were found in close friendship, job competence, behavioral conduct or romantic appeal. The results highlight the complexity of self-esteem in the period of adolescence. In general, self-esteem was lower among middle adolescents in comparison to early adolescents; however, with regard to some domains, this rule was not observed. It may be speculated that the drop in self-esteem observed between childhood and adolescence (Baldwin and Hoffman
2002; Robins et al.
2002) is more prolonged, and the later observed increase in self-esteem (Orth et al.
2012) is more delayed. However, this is only an hypothesis, and the cross-sectional data do not allow for verification of this speculation. On the one hand, it may be concluded that the period of early adolescence does not seem as challenging for self-esteem as supposed (Harter
2012b), and young adolescents generally have a more positive view of their competences than their older counterparts. On the other hand, it must be emphasized that this conclusion should be considered cautiously as one of the studied age samples was gender-biased: in the middle adolescence group there were more girls than boys (107 versus 34); and thus the lower self-esteem in this middle adolescent group may be related to the gender or gender x age interaction, even though the meta-analyses on self-esteem trajectories indicated the same pattern for both males and females (Huang
2010; Trzesniewski et al.
2003). This suggests that gender is not a significant factor influencing the continuity or stability of self-esteem. It has been shown, however, that self-esteem tends to be lower in females than in males (Bleidorn et al.
2016; Kling et al.
1999; Moksnes and Espnes
2013; Quatman and Watson
2001), and this pattern may be more pronounced in middle adolescence (Kling et al.
1999). Therefore, the observed pattern of age differences in self-esteem may be intertwined with gender differences during this age period.
The pattern of results slightly changed at the individual level, when the rank-order stability of self-esteem was analyzed. The results of autoregressive analyses revealed that, for the younger group, global self-esteem just after the school transition explained only 14% of its variance 9 months later, even though at the group level, global self-esteem seemed stable. This is also true for behavioral conduct in this group (13% of variance at T2) and for romantic appeal (15% of variance at T3). For most domains of self-esteem, particularly in early adolescence, the level of self-esteem at T1 explained only a small part of its variance 9 months later. These results suggest that school transitions create a context for some kind of instability in self-esteem at the individual level. Afterward, during the more extended period of time, this stability seems to be restored, which seems congruent with the growing stability of individual differences reported in the literature (Trzesniewski et al.
2003).
Limitations
This study’s conclusions have limitations worth noting. They are limited due to the specific context of school transitions in Poland, which may interact with developmental changes typical for the period of early and middle adolescence. Specifically, in early adolescence, there is a transition to junior high school, which comprises the last level of obligatory education in Poland. On the other hand, the transition to high school (in middle adolescence) is not obligatory, and it refers to about 50% of teenagers graduating from junior high schools. Additionally, the attrition rate was higher among middle adolescents, which may be related to a higher school absence rate in this age group (Motyka
2018).
One also can argue that short periods of time between measurement points may make observing developmental change in self-esteem difficult or may influence the observed developmental trends as usually the studied time periods have been much longer (e.g., Kuzucu et al.
2014). However, the short periods of time between measurement points enabled us to reveal short-term changes, observable particularly at the individual level (instability in self-esteem just after school transition), which may not be visible in a broader time perspective.
Moreover, the study was limited in terms of the participants and measures. The study mainly included participants from an urban area, so the sample was rather homogeneous. The sample was not equal regarding gender distribution as in the middle adolescent group, there were many more girls than boys (107 versus 34), and this factor could have influenced the obtained results, particularly regarding the level of self-esteem (Kling et al.
1999; Quatman and Watson
2001). It should also be highlighted that some of the presented results should be considered with caution as some of the analyzed models (for example, regarding behavioral conduct) did not reach acceptable fit indices, and the reliability coefficients of some of the SPPA scales were rather low. However, the reliability of the behavioral conduct subscale was also below 0.7 for the original version of the SPPA (Harter
2012a). Low alphas (below 0.6) for the romantic appeal scale in early adolescents might have been affected by the fact that they are not yet motivated to be interested in one romantic partner. On the other hand, middle adolescents might be not so focused on school achievements, and thus their answers on the scholastic competence subscale were not so consistent.
Future directions and implications
This study’s results supported the validity of the adopted approach. The combination of cross-sectional and longitudinal designs and providing both group and individual analyses in one study allowed to adequately describe the complexity of the development of self-esteem in adolescence. Despite finding a general continuity in the development of self-esteem in the periods of early and middle adolescence, individual differences in the trajectories of development of self-esteem were also discovered. This likely means that, in some domains, the change is more complex. Thus, it may be suggested that future research is needed in the area of self-esteem in adolescence because the courses and the consequences of individual differences in the development of self-esteem should be analyzed in depth. One may speculate, for example, that the direction of change can vary: it can decrease in one subgroup, increase in another, and in the next maintain an equilibrium. This interpretation is also supported by the analyses of the changes at the individual level, suggesting some weak instability in self-esteem, particularly just after the school transition. Therefore, one should note that the general, group-pattern changes in self-esteem may not automatically refer to all individuals, as individual patterns of change may be much more differentiated. The present study only revealed the inter-individual differences in the course of the development of self-esteem; thus, additional analyses aimed at drawing a distinction between the groups with distinct courses of change in self-esteem are needed.
The need for more research seems particularly important in the context of practical implications. Different trajectories of self-esteem may lead to different developmental outcomes not only in adolescence (Zimmerman et al.
1997) but also throughout the course of one’s entire life (Steiger et al.
2014). The implications of such an observation for educational policy are very important, as people may require different supports at different life periods.
These implications are even more significant if one considers that self-esteem is a key determinant of well-being and success in life (Deci and Ryan
1995; James
1890/1952; Leary
1999). This research indicates that the development of self-esteem does not have a uniform trajectory. Firstly, self-esteem develops differently in particular domains and, secondly, it develops differently for different age groups during adolescence; both of these observations are important, as the precise description of these changes can provide guidance for those working with young people. For example, when assuming the key importance of self-esteem in the lives of young people, one should be aware that supporting people in early and late adolescence should be approached in different ways. According to Topolewska-Siedzik and Cieciuch (
2018), these two periods in research and development are often treated together, without proper distinctions made between them. The specificity of each of these periods should be further tested, as results obtained confirm that younger and older adolescents develop differently and probably require different kinds of support during this process. This applies also to other developmental periods, as there is also a need to describe change and consistency in self-esteem in late adolescence and in emerging adulthood.
Last but not least, based on the present results, the theoretical implications of understanding the period of adolescence should be underscored, as adolescence may be viewed as a formative period of life (Blakemore
2018) and understanding how young people develop a sense of self (particularly, self-concept and self-esteem) may be seen as especially important. This is one of the most crucial challenges of this age group (James
1890/1952; Harter
2012b). Additionally, as the individual differences in the continuity and stability of different domains of self-esteem are observed, they may be related to changes in the importance of different domains of self-esteem in the course of adolescence generally (Harter
2012b). Definitely, the issue of the importance of different domains of self-esteem and their significance for global self-esteem also deserves future studies.
Conclusion
Although the development of self-esteem in adolescence have been widely studied, previous results have been mixed and questions about patterns of change and consistency at both the mean (group) and the individual level have not been asked. In the present study, using multidimensional and multilevel designs, the complexity of the patterns of development of self-esteem during early and middle adolescence was described. First, in most domains of self-esteem and in global self-esteem over the period of a year-and-a-half, the continuity was observed, not change. Second, in most domains, a weak but quickly restored stability at the individual level was found. Third, also observed was higher self-esteem in early (not middle) adolescence and significant individual differences in the levels of global and domain-specific self-evaluations not explained by age. In general, at the beginning of the 21st century, school transitions turned out to be not as significant turning point as it had been previously thought.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.