Development of adolescents’ self-esteem and general academic self-concept: perceived classroom climate makes a difference

Abstract

We investigated rank-order continuity and mean-level change in adolescents’ self-esteem, academic self-concept, and social self-concept and tested whether interindividual differences in intraindividual change could be explained by four dimensions of classroom climate (i.e., teachers’ focus on students, learning community, pressure related to social or achievement issues, and rivalry and disruptions in class). The effects of classroom climate dimensions were investigated at the individual and classroom levels. The study comprised 2722 students from 98 classes who participated in four waves from grades 5 to 8. Rank-order continuities of self-esteem, academic self-concept, and social self-concept were substantial. Multilevel regressions revealed a significant nonlinear mean-level change in all constructs under investigation, indicating an initial decrease that became smaller over time. In self-esteem and social self-concept, the trend shifted from an initial decrease to an increase. Multilevel regressions revealed significant interindividual differences in the changes over time in all constructs. Change in academic self-concept was significantly predicted by all classroom climate dimensions on the individual level. Change in self-esteem was predicted by teachers’ focus on students and pressure related to social or achievement issues on the individual level. Change in social self-concept was not predicted by any classroom climate dimension.

Keywords

Classroom climate self-esteem self-concept development of personality stability

Adolescence is described as a formative phase for the development of psychological variables, including traits such as self-esteem and self-concepts (Meeus, 2018). Stage-environment fit theory (Eccles et al., 1993) predicts a decline in adolescents’ school-related self-perceptions, especially during the middle school years (Eccles et al., 1993; Eccles & Roeser, 2009; Wigfield & Wagner, 2005). Recent meta-analyses of longitudinal studies have investigated how self-perceptions change throughout the school career (Orth et al., 2018; Scherrer & Preckel, 2019). Findings revealed that school-related self-perceptions show a substantial rank-order continuity over time (Orth et al., 2018; Scherrer & Preckel, 2019; Trzesniewski et al., 2003). In addition, mean levels of self-perceptions such as self-esteem and academic self-concept do not seem to change much during middle school (Orth et al., 2018; Scherrer & Preckel, 2019). At first glance, these meta-analytic findings seem to contradict the assumption of adolescence as a formative phase for students’ self-perceptions and of a decline in school-related self-perceptions as predicted by stage-environment fit theory (Eccles et al., 1993; Eccles & Roeser, 2009; Wigfield & Wagner, 2005). However, the meta-analytic findings did not take into account possible interindividual differences in the intraindividual development of students. Thus, they do not rule out the possibility that individual adolescents change systemically in these traits.

Stage-environment fit theory (Eccles et al., 1993) explains declines in adolescents’ self-perceptions and achievement motivation as resulting from a growing mismatch between students’ developing needs and the opportunities offered to them by their school environment, especially after the transition from elementary to middle school. Developmental changes in both the individual and the environment can lead to a poor stage-environment fit, which then results in deteriorating self-perceptions and achievement motivation. The theory further suggests that there are differences in the developmental changes in both the individual and the environment (Eccles et al., 1993; Eccles & Roeser, 2009; Gutman & Eccles, 2007). This means that for some students, the middle school environment promotes positive growth, whereas for other students, these environmental changes negatively affect their psychosocial development (Gutman & Eccles, 2007).

Regulatory processes within school environments are complex and can be described on several hierarchically structured levels (e.g., classrooms, schools, districts). The classroom comprises the most immediate educational environment for students and its general character, including teacher–student relationships, can be described as the classroom climate (Eccles & Roeser, 2009). A beneficial classroom climate is provided by teachers who care, trust, and respect their students and provide an emotionally and intellectually stimulating environment in which the students can engage, learn, and develop positive motivation, self-perceptions, and achievements (Eccles & Roeser, 2009). That is, according to stage-environment fit theory, differences in classroom climate affect interindividual differences in the intraindividual development of students’ self-perceptions.

In the present study, we tested some of these predictions of the stage-environment fit theory. To better understand the stability, variability, and change after the transition to middle school in adolescents’ self-perceptions, we investigated rank-order continuity and mean-level change in their self-esteem and self-concepts as well as the assumption that interindividual differences in intraindividual change can be explained by characteristics of the classroom environment. More specifically, we investigated four classroom climate dimensions (i.e., teachers’ focus on students, learning community, pressure related to social or achievement issues, and rivalry and disruptions in class) as moderators of students’ changes in self-esteem and the academic and social dimensions of their self-concept within a large longitudinal study covering the first 4 years of secondary school.

Self-esteem and its stability and change in adolescence

Self-esteem comprises individuals’ positive or negative attitude toward the self as a totality (Rosenberg et al., 1995). In their hierarchical model of self-concept, Shavelson et al. (1976) located self-esteem at the most general level of the self-concept hierarchy. High self-esteem reflects a favorable global evaluation of one’s own accomplishments and achievements in subjectively important areas, whereas low self-esteem reflects an unfavorable global evaluation of one’s previous accomplishments and achievements (Harter, 1990). Self-esteem is formed by different processes, such as reflected appraisals, social comparisons, and self-attributions (Rosenberg et al., 1995). It is positively related to social relationships (Harris & Orth, 2020; Hendrick et al., 1988; Neyer & Asendorpf, 2001) and subjective well-being (DeNeve & Cooper, 1998) and negatively related to depression (Sowislo & Orth, 2013) and health problems (O'Connor & Vallerand, 1998; Vingilis et al., 1998). Moreover, self-esteem shows small positive relations with school achievement (Freudenthaler et al., 2008; Hansford & Hattie, 1982; Körük, 2017).

Similar to most psychological variables (Geiser et al., 2017), self-esteem consists of both trait- and state-like components, implying that a certain stability and variability over time can be expected (Donnellan et al., 2012). Hank and Baltes-Götz (2019) investigated the stability and variability of self-esteem as distinct features of state self-esteem change and found that change in state self-esteem is a stable process largely determined by interindividual differences in trait self-esteem. To explain these interindividual differences, some researchers have argued that self-esteem is highly responsive to the immediate social context and related feedback, contradicting the concept of a stable personality trait (Conley, 1984; Crocker & Wolfe, 2001; Leary & Baumeister, 2000). According to this view, extrinsic forces cause the variability of self-esteem. In contrast, other researchers assumed intrinsic forces to initiate internally generated patterns of variability in self-esteem (Fortes et al., 2003; Leary & Baumeister, 2000; Wong et al., 2016). The dynamic system approach takes into account both intrinsic and extrinsic forces to explain intraindividual variability in self-esteem and underlines the importance of the interplay between these forces (De Ruiter et al., 2017; de Ruiter et al., 2018; van Geert & Steenbeek, 2005). According to this view, the extrinsic effects on self-esteem do not function in the sense of a simple stimulus-response paradigm but in connection to the intrinsic dynamics of self-esteem. For example, De Ruiter et al. (2018) found that state self-esteem variability was less dependent on the immediate context if strong intrinsic dynamics of self-esteem were present.

Regarding the stability of self-esteem, meta-analytic findings indicated a substantial rank-order continuity of (trait) self-esteem in the range of approximately r = .50–.70 (Trzesniewski et al., 2003), which is close to what has been reported for other psychological traits such as the Big Five personality traits (e.g., Roberts & DelVecchio, 2000: r = .31–.74). For adolescents, the continuity coefficients were typically slightly lower than those for adults (Trzesniewski et al., 2003). Regarding mean-level change, recent meta-analyses indicated no significant mean-level change during adolescence (Orth et al., 2018; Scherrer & Preckel, 2019).

General academic self-concept and its stability and change in adolescence

Academic self-concepts comprise the individual perceptions and beliefs of one’s academic abilities in general and in different academic domains (Marsh & Shavelson, 1985; Shavelson et al., 1976). In addition to being domain specific, academic self-concepts can be conceptualized as hierarchically organized, with the general academic self-concept at the apex of the hierarchy (Arens et al., 2021). Among the most important influences on students’ academic self-concepts are reference group effects, social comparisons, and performance-related feedback (Seaton et al., 2010; Trautwein & Möller, 2016). In educational research, the general academic self-concept is one of the best-studied predictors of achievement, showing positive reciprocal relations with achievement across longer time periods (Guay et al., 2003; Huang, 2011; Marsh & O'Mara, 2008) and very short time periods (Niepel et al., 2014). It has beneficial effects on a broad range of academic outcomes, such as academic interests and emotions and occupational aspirations (Trautwein & Möller, 2016).

Academic self-concept shows a crucial similarity to academic self-efficacy, defined as individuals’ specific beliefs about their efficacy to regulate their own learning activities and to master certain academic tasks (Bandura, 1993; Bandura et al., 1996; Wigfield & Eccles, 2002). As academic self-concept and academic self-efficacy are highly correlated, they are sometimes used interchangeably in the literature (Eccles & Wigfield, 2002; Trautwein & Möller, 2016). However, theoretical (Bong & Skaalvik, 2003) and empirical (Ferla et al., 2009; Peiffer et al., 2020; Scherer, 2013) research both indicate differences between these two constructs. In contrast to academic self-concept, which refers to comparative processes regarding one’s own competence in other domains and the competence of other individuals, self-efficacy usually refers to individuals’ own sense of whether they can or cannot accomplish a particular task (Wigfield & Eccles, 2002).

Regarding stability and change in individuals’ general academic self-concept, it is often assumed to be a malleable construct that can be fostered by educational interventions (O'Mara et al., 2006). A certain variability and change in the general academic self-concept is also in accordance with the classical theoretical assumption that self-concepts influence a person’s responses to situations and, in turn, are inferred from these responses (Shavelson et al., 1976). On the other hand, research has indicated relatively high rank-order continuity in academic self-concepts during adolescence that sometimes exceeds the rank-order continuity of the corresponding achievements (Marsh & Yeung, 1998; Niepel et al., 2014). In their meta-analysis, Scherrer and Preckel (2019) reported a rank-order continuity of r = .65 when the test-retest interval was shorter than 1 year, r = .55 when the test-retest interval comprised exactly 1 year, and r = .46 when the test-retest interval lasted longer than 1 year. Regarding mean-level change, Scherrer and Preckel (2019) obtained no significant change in general academic self-concept throughout students’ school careers, whereas they observed significant decreases in domain-specific academic self-concepts (i.e., math and language self-concepts) over time.

Social self-concept and its stability and change in adolescence

Social self-concepts comprise the perceptions and beliefs of one’s social acceptance by others and one’s competence to interact with others within a given social context (Byrne & Shavelson, 1996). The construct of social self-concepts can be distinguished from the related social self-efficacy construct that refers to specific behavior that underlies personal relationships (Connolly, 1989). Social self-concepts are made up of relationships with other people (Byrne & Shavelson, 1996) in feelings of being accepted by others (i.e., social self-concept of acceptance), or one’s ability to assert oneself (Preckel et al., 2013). Of note, in our study, we focus on the social self-concept of acceptance defined by one’s feeling of acceptance by peers in the school environment (Berndt & Burgy, 1996; Byrne & Shavelson, 1996).

There is much less research on social self-concept in the school context than on self-esteem or academic self-concept. Lindner-Müller et al. (2012) found that earlier negative peer status predicted later social self-concept, whereas the reversed relation was not significant. Spilt et al. (2014) found that peer rejection negatively predicted social self-concept over time, which—in turn—negatively predicted internalizing problems. Cross-sectionally, social self-concepts and academic self-concepts show small positive correlations (Preckel et al., 2013). However, in a longitudinal analysis, social self-concept of acceptance was found to negatively predict general academic self-concept. In the same study, social self-concept was not a significant longitudinal predictor of school achievement (Preckel et al., 2013).

Regarding stability and change in social self-concepts, there are no meta-analyses investigating rank-order continuity or mean-level change in social self-concept. Rank-order continuity coefficients reported by single longitudinal studies (Lindner-Müller et al., 2012; Preckel et al., 2013; Spilt et al., 2014) range from r = .30 to r = .73 and are therefore comparable with those of self-esteem or general academic self-concept. In the same studies, the mean level of social self-concept remained stable (Spilt et al., 2014) or increased slightly over time (Preckel et al., 2013).

To conclude, empirical findings point to high mean-level stability and substantial rank-order continuity of self-esteem, general academic self-concept, and social self-concept during adolescence. These findings do not rule out the possibility that individual adolescents change systemically in these traits and that these changes could be predicted by context variables related to the classroom environment. Especially for self-concepts, there is strong evidence for changes due to contextual effects (i.e., reference group effects; Seaton et al., 2010).

Classroom climate

The classroom climate comprises the intellectual, social, emotional, and physical environment in which students learn (Ambrose et al., 2010, p. 170). Ever since the pioneering work of Walberg (Walberg & Anderson, 1968) and Moos (1987), theorists have distinguished between objective characteristics of the learning environment and students’ subjective perceptions of them. These individual perceptions lead to idiosyncratic cognitions regarding classroom climate (i.e., individual climate; Pekrun, 1985), which, being shared among the students within a class, make up the global climate for the class as a whole (i.e., classroom climate; Eder, 2010).

Which characteristics of the learning environment contribute to classroom climate has been a matter of debate since the earliest research in the field. For example, Avant et al. (2011) stress the meaning of the concept of emotional climate, whereas Allodi (2010) emphasize the social climate. According to this view, classroom climate is mainly shaped by the interactions between students or between students and teachers (Toren & Seginer, 2015). Using the Linzer Questionnaire of Classroom Climate for Grades 4 through 8 (LFSK 4-8; Eder & Mayr, 2000), Hank et al. (2022) assessed a total of 2084 students from 96 classrooms. The results of multilevel factor analyses revealed four major classroom climate dimensions: teachers’ focus on students, learning community, pressure related to social or achievement issues, and rivalry and disruptions in class, which can be measured at both the individual level and the classroom level. Teachers’ focus on students refers to nondirective teacher behavior and student participation in decision-making. Learning community refers to a positive classroom community and willingness to learn. Pressure related to social or achievement issues refers to authoritarian teacher behavior that is perceived as unfair, high teaching tempo, and individual overload. Rivalry and disruptions in class refers to detrimental competition and disruptive behavior. Like the different theoretical conceptualizations, there is a wide variety of methods and instruments for assessing classroom climate (for an overview, see, e.g., Eder, 2010; Rubie-Davies et al., 2016). Most often used questionnaires are specifically suited to assess students’ and/or teachers’ personal perceptions of the classroom climate (Babad, 2009).

Research has revealed statistically significant associations between classroom climate and students’ academic subject attitudes, motivation, school achievement, and satisfaction (Eder, 2010; Reyes et al., 2012; Rubie-Davies et al., 2016; Shewark et al., 2018). Classroom climate has also been linked to students’ well-being and sense of self-competence (Toren & Seginer, 2015; for a detailed discussion of crucial methodological issues in this research, see Lüdtke et al., 2009; Morin et al., 2014). Wang et al. (2020) conducted a meta-analysis on the relation between classroom climate and students’ academic and psychological wellbeing and found small-to-medium size positive associations between classroom climate and social competence, motivation, and engagement and small size negative associations with socioemotional distress and externalizing behaviors. Wang et al. (2020) noticed that there are very few longitudinal studies on classroom climate. Therefore, causal statements are difficult, and more research is needed to determine whether classroom climate predicts change in psychosocial constructs over time. One recent longitudinal study investigated the effects of students’ perceptions of school climate on mean-level changes in self-esteem, academic self-concept, and social self-concept in 25 classes from four schools before and after the transition to middle school (Coelho et al., 2020). The authors found that classroom climate positively predicted both the initial level of self-esteem, academic self-concept, and social self-concept and the mean-level change in these constructs. Although these results support the assumptions of the stage-environment fit theory that classroom climate moderates students’ changes in self-perceptions, it must be noted that Coelho et al. (2020) only analyzed between-students differences in classroom climate (i.e., the individual level) and did not take into account between-class differences (i.e., the classroom level). Furthermore, Coelho et al. (2020) only investigated the effects of a climate composite score, without investigating the unique contributions of different classroom climate dimensions.

Present research

Previous studies on the stability of self-esteem, general academic self-concept, and social self-concept revealed that their rank-order continuity coefficients were substantial. However, previous studies also indicated a certain variability over time. Findings for their mean-level change indicated no significant change in middle school for self-esteem and general academic self-concept, whereas the limited findings for social self-concept indicated no change or a slight increase. The combination of these (meta-analytic) findings of substantial rank-order continuity and no substantial mean-level change over time suggests that there are interindividual differences in the intraindividual mean-level change of self-esteem, general academic self-concept, and social self-concept over time. That is, the rank-order position of students changes over time because some students show mean-level increases while others show no change or decreases in their self-esteem and self-concepts.

Following the stage-environment fit theory, we focused on varying classroom environments and studied the relation of classroom climate to interindividual differences in mean-level changes. According to the dynamic systems approach, intraindividual variations in self-esteem can be explained by an active self-coordinating process “balancing between self-maintained stability and flexible adaptations to external influences” (De Ruiter et al., 2014, p. 14). Coelho et al. (2020) investigated classroom climate as an external influence and found that a positive classroom climate positively predicted the level and development of self-esteem, academic self-concept, and social self-concept. In the present study, we investigated four classroom climate dimensions: Teachers’ focus on students, learning community, pressure related to social or achievement issues, and rivalry and disruptions in class. According to the stage-environment fit theory and prior meta-analytic findings (Wang et al., 2020), teachers’ focus on students and a positive learning community can help students form positive self-perceptions (Eccles & Roeser, 2009). Feeling emotionally supported and belonging to the class leads students to view themselves as worthy individuals and as an important part of the class, which benefits the development of positive achievement-related self-perceptions and a sense of well-being in school (Eccles & Roeser, 2009; Wentzel, 2002). On the other hand, pressure related to social or achievement issues as well as rivalry and disruptions in class can negatively affect the development of self-perceptions, foster social comparison processes, and contribute to a shift in students’ view of ability from an incremental condition to an entity state (Eccles & Roeser, 2009).

In the present study, we investigated the rank-order continuity and mean-level change of self-esteem, general academic self-concept, and social self-concept of acceptance in a large longitudinal dataset covering the time span of 4 years. Previous research mainly applied linear methods (Coelho et al., 2020; Scherrer & Preckel, 2019), which can overlook nonlinear changes in self-esteem, general academic self-concept, and social self-concept. We therefore used multilevel growth curve modeling, which allows the investigation of linear and nonlinear change and the relations between classroom climate and self-esteem, general academic self-concept, and social self-concept on an individual level and on a classroom level. We tested whether there was significant interindividual variation in the intraindividual mean-level change. Finally, we tested whether differences in classroom climate predict interindividual differences in the intraindividual mean-level change. We preregistered the following hypotheses (H) before conducting the analyses:

https://osf.io/yp3b5/?view_only=365870a4954346ac974970c6862e12d1

• H1: We expect to find a substantial rank-order continuity (around r = .50) in self-esteem (H1a), general academic self-concept (H1b), and social self-concept of acceptance (H1c).

• H2: We do not expect to find significant mean-level change over time in self-esteem (H2a) or in general academic self-concept (H2b), but we expect a positive mean-level change in social self-concept of acceptance over time (H2c).

• H3: We expect to find significant variation in the mean-level change in self-esteem (i.e., interindividual differences in the intraindividual development of self-esteem) (H3a), general academic self-concept (H3b), and social self-concept of acceptance (H3c).

• H4: Interindividual differences in the intraindividual development of self-esteem (H4a), general academic self-concept (H4b), and social self-concept of acceptance (H4c) can be significantly predicted by the classroom climate variables: teachers’ focus on students (H4a-c I), learning community (H4a-c II), pressure related to social or achievement issues (H4a-c III), and rivalry and disruptions in class (H4a-c IV). For all three outcomes under study, we expect to find increases associated with teachers’ focus on students and learning community and decreases associated with pressure related to social or achievement issues and rivalry and disruptions in class.

• As an additional research question (RQ), we exploratively tested whether our findings for classroom climate dimensions regarding H4 remained robust after controlling for the effects of the remaining classroom climate dimensions (i.e., all classroom climate dimensions were investigated simultaneously in one model).

Of note, H1 to H4 were preregistered before conducting the analyses, whereas the RQ was added during the review process. Furthermore, H4 and the RQ were investigated at the individual and classroom levels (for more information, see the analyses section). For both levels of analysis, we expected to find comparable effects for the four classroom climate dimensions. That is, for instance, in H4a I, we assumed that the intraindividual development of self-esteem was positively predicted by teachers’ focus on students at both the individual level and the classroom level.

Methods

Participants and procedure

Data stemmed from two large longitudinal studies in Germany that assessed students’ development of motivation and achievement in the highest track of the German three-track secondary school system after the transition from elementary to middle school (PULSS Study and AVG Study). The PULSS Study was approved by the State Ministry for Education and Culture of Bavaria (“Bayerisches Staatsminesterium für Unterricht und Kultur”; protocol number: lll.4-5 LO355 lll 40 - 1.18 555). The AVG Study was approved by the Supervision and Services Directorate of Rhineland-Palatinate (“Aufsichts- und Dienstleistungsdirektion”; protocol number: 32-03 405/29/05).

The combined sample from both projects comprised 2770 students in 98 classes. Because of missing values on all investigated variables (i.e., self-esteem, academic self-concept, social self-concept, and classroom climate), 32 students were excluded from further analyses. Four further students were excluded because of missing information regarding the surveyed class. Finally, 12 students were excluded because they switched classes during the investigation.

Our final sample comprised 2722 students (mean age at T1 = 10.02; 44% girls; 89% German native speaker) from 12 schools and 98 classes¹. From this overall sample, 2058 students attended 69 regular classes and 664 students attended 29 classes for gifted students. To take into account the different class types, we used class type as a control variable in all subsequent analyses investigating the effects of classroom climate.

Students were tested by trained research assistants in groups in their classrooms at four measurement points. T1 occurred in 5th grade between 29 and 92 days after the transition to middle school. T2 occurred in 5th grade between 150 and 289 days after the transition to middle school. T3 occurred in 6th grade between 515 and 645 days after the transition to middle school. T4 occurred in either 7th or 8th grade between 911 and 1310 days after the transition to middle school. Data cannot be made openly available due to restrictions in the project approval of the school supervisory board of Rhineland-Palatinate (ADD). Covariation matrices of the investigated data can be found online. Subsets of the data were previously investigated in Preckel et al. (2013), Niepel et al. (2014), and Hank et al. (2022). None of these articles investigated the prediction of changes in self-esteem, general academic self-concept, and social self-concept by classroom climate.

Materials

Self-esteem, general academic self-concept, and social self-concept of acceptance were assessed via the same questionnaires at each time point. Classroom climate was assessed once in sixth grade at T3.

Self-esteem

Self-esteem (SE) was assessed with four positively worded items from the German version of the Rosenberg (1965) scale (e.g., “I feel that I have a number of good qualities”). Cronbach’s alpha was α = .758 at T1, α = .763 at T2, α = .780 at T2, and α = .786 at T4.

General academic self-concept

General academic self-concept (ASC) was assessed with three positively worded items from the German version of the Self-Description Questionnaire (Marsh, 1990; e.g., “In most subjects, I learn quickly”). Cronbach’s alpha was α = .777 at T1, α = .806 at T2, α = .819 at T2, and α = .836 at T4.

Social self-concept of acceptance

Social self-concept of acceptance (SSC) was assessed with three reverse-coded items provided by Fend, 1986, e.g., “In class, I sometimes feel like an outsider”). Cronbach’s alpha was α = .790 at T1, α = .831 at T2, α = .842 at T2, and α = .861 at T4.

Classroom climate

Classroom climate was assessed with 42 items from the Linzer Questionnaire of Classroom Climate for Grades 4 through 8 (LFSK 4-8 questionnaire; Eder & Mayr, 2000, see also Hank et al., 2022). These items can be organized into 14 clusters, each representing a narrow facet of classroom climate (i.e., three items form one cluster). In turn, these 14 clusters can be organized into four higher-order classroom climate dimensions, teachers’ focus on students (TF), learning community (LC), pressure related to social or achievement issues (Pres), and rivalry and disruptions in class (Riv), which were investigated in the present study. The factor structure of the LFSK was recently demonstrated to be acceptable by multilevel confirmatory factor analyses (Hank et al., 2022). Figure 1 shows the assignment of the items to clusters and of the clusters to the higher-order classroom climate dimensions. Table 1 presents the content of clusters and higher-order classroom climate dimensions.

Figure 1.

Five steps to calculate the four classroom climate dimensions. The 42 classroom climate items were assessed by students’ self-reports on the LFSK questionnaire. TF = Teachers’ focus on students; LC = Learning community; Pres = Pressure related to social or achievement issues; Riv = Rivalry and disruptions in class.

Table 1.

Content of the 14 clusters and four higher-order classroom climate dimensions.

Cluster	Classroom climate dimensions with descriptions and items examples
	Teachers’ focus on students (TF)
	Facilitated, nondirective teacher behavior, student participation in decision-making, and performance monitoring
Educational engagement	“I think it is very important to the teachers that we understand the material.”
Teaching quality	“Our teachers can explain even complicated things so well that we understand them.”
Autonomy	“Our teachers decide many things all together with us.”
Student involvement	“Our teachers are teaching in such a way that the students always have the chance to add their own thoughts.”
Monitoring	“In our class, the teachers pay a lot attention to how we work and what we can achieve.”
	Learning community (LC)
	A positive classroom community and willingness to learn
Community	“In our class, no one leaves a student hanging from the class.”
Willingness to learn	“Most students in our class enjoy learning.”
	Pressure related to social or achievement issues (pres)
	Authoritarian teacher behavior that is perceived as unfair, high teaching tempo, and individual overload
Fairness (−)	“Teachers do their best to treat all students fairly.”
Restrictiveness	“Students are immediately threatened with a bad grade, if they do not participate effectively.”
Comparison	“If we get back test results, the best and worst students are discussed in front of the class.”
Performance pressure	“If you missed a few days, you have to catch up a lot.”
Teaching pressure	“The teachers often explain so quickly that one can barely keep up.”
	Rivalry and disruptions in class (Riv)
	Competitive drive to the detriment of classmates and disruptive behavior
Rivalry	“If someone says something wrong, the others are secretly happy about it.”
Disruption	“Some students are always disrupting the class.”

Each cluster was calculated based on three items. Item examples were translated into English and paraphrased for presentation purposes.

We calculated TF, LC, Pres, and Riv scores on the individual level and on the classroom level in five steps (see Figure 1). Following the guidelines of Eder and Mayr (2000), (1) 14 classroom climate clusters were calculated based on the available 42 items (i.e., a mean value was calculated from each set of three items), and (2) four classroom climate dimensions were calculated based on the 14 clusters (i.e., a mean value was calculated from 2 or 5 clusters). (3) The four classroom climate dimensions were z-transformed based on the complete sample (i.e., grand mean z-transformation). (4) For each of the 98 classes, a class average value was calculated based on the z scores (i.e., grand mean centering). These class average scores represent the between-class differences on the four classroom climate dimensions, whereas students within classes do not differ on these scores (i.e., the scores represent classroom climate on the classroom level). These scores were labeled TF_L3, LC_L3, Pres_L3, and Riv_L3. (5) For each individual student, the class average scores TF_L3, LC_L3, Pres_L3, and Riv_L3 were subtracted from the individual TF, LC, Pres, and Riv z scores (i.e., group-mean centering). The resulting group-mean centered scores represent individual differences within the classes (i.e., the scores represent classroom climate on the individual level). These scores were labeled TF_L2, LC_L2, Pres_L2, and Riv_L2.

At the individual level, Cronbach’s alpha for TF_L2, LC_L2, Pres_L2, and Riv_L2 was α = .813, α = .515, α = .742, and α = .506, respectively. At the classroom level, Cronbach’s alpha for TF_L3, LC_L3, Pres_L3, and Riv_L3 was α = .909, α = .700, α = .716, and α = .648, respectively.

Analyses

The Mplus and R Code and covariation matrices of the investigated data are accessible at https://osf.io/3wqgp/?view_only=0a45a1756a6742e3b3ad980c9e5aad35

Measurement invariance over time

Before testing our hypotheses, we tested whether SE, ASC, and SSC were measurement invariant over the four time points with Mplus (Muthén & Muthén, 2017). We applied the type is complex option to account for the nested data structure (i.e., students within classes). We conducted several autocorrelative structural equation models (SEMs) and successively tested increasing measurement invariance levels against each other (i.e., configural, metric, scalar, and strict invariances). We used effect coding to identify the conducted models (Little et al., 2006). Following Chen (2007), we used comparative fit indices (ΔCFI) to compare two models of different measurement invariance levels. ΔCFI values of .01 or less were interpreted as a tolerable deterioration in model fit.

Hypothesis 1

Latent rank-order continuity coefficients in SE (H1a), ASC (H1b), and SSC (H1c) were estimated based on the respective scalar measurement invariant autocorrelative SEMs.

Hypotheses 2 to 4

Hypotheses 2 to 4 were tested separately for SE (a), ASC (b), and SSC (c) based on multilevel regressions that were successively expanded. All multilevel regressions were conducted with the lme4 package in R (Bates et al., 2018) by using the log-likelihood estimator. We conducted the same analyses for SE, ASC, and SSC. For simplicity, we detail the analyses of SE only.

Prior to testing our hypotheses, we calculated a three-level intercept-only model (Model 1) to estimate the SE variance within students over time (Level 1), between students in one class (Level 2), and between classes (Level 3). In this model, the repeated measurement points within the persons represent Level 1, differences between the persons represent Level 2, and differences between the classes represent Level 3. Next, we added linear and quadratic time effects as Level 1 predictors into this model to test H2a (Model 2). Note that in this and the following models, we shifted the intercept for all variables to the third measurement point, as classroom climate was only assessed at the third measurement point. H3a was tested by allowing the linear time effects to vary between Level 2 and Level 3 (Model 3; i.e., random slope model). We used the likelihood ratio test to test whether the global model fit decreased significantly if the random slope between students and the random slope between classes was omitted from the model. In Model 4, we added class type as a Level 3 predictor to control for differences between regular and gifted classes.

To test whether interindividual differences in the intraindividual change in SE can be explained by each of the four classroom climate dimensions (H4a), we conducted four models (one for each classroom climate dimension: Model 5I, Model 5II, Model 5III, and Model 5IV). These models extended Model 4 by the respective classroom climate variable that was added as a Level 2 and Level 3 predictor. Classroom climate on Level 2 represents the differences in the evaluation of the climate within the classes (i.e., differences between individual ratings and the class average). Classroom climate on Level 3 represents the classroom climate differences between classes (i.e., class average scores). The cross-level interaction terms between the linear time effects and classroom climate at Level 2 and Level 3 were also added into Model 5I to Model 5IV. These cross-level interaction terms test whether interindividual differences in the linear time slope can be explained by classroom climate at Level 2 and Level 3.

Research question

Finally, in Model 6, all classroom climate variables were added as simultaneous predictors of SE, ASC, or SSC on Level 2 and Level 3.

Results

The correlations and descriptive statistics for the study variables are reported in Table 2.

Table 2.

Means, standard deviations, and correlations of study variables.

	M	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19
1. SE_T1	4.14	0.66
2. SE_T2	4.04	0.71	.48**
3. SE_T3	3.97	0.73	.39**	.50**
4. SE_T3	3.96	0.73	.32**	.37**	.49**
5. ASC_T1	4.10	0.67	.58**	.41**	.30**	.23**
6. ASC_T2	3.96	0.75	.34**	.61**	.39**	.28**	.57**
7. ASC_T3	3.83	0.79	.28**	.38**	.53**	.32**	.42**	.56**
8. ASC_T4	3.70	0.84	.26**	.33**	.34**	.44**	.38**	.48**	.58**
9. SSC_T1	4.34	0.85	.27**	.20**	.17**	.15**	.19**	.17**	.13**	.12**
10. SSC_T2	4.19	0.95	.16**	.31**	.21**	.18**	.15**	.23**	.17**	.13**	.52**
11. SSC_T3	4.25	0.94	.11**	.15**	.34**	.18**	.10**	.12**	.19**	.08**	.33**	.43**
12. SSC_T4	4.30	0.89	.07**	.13**	.18**	.33**	.08**	.10**	.12**	.13**	.30**	.33**	.44**
13. TF_L2	−0.00	0.67	.11**	.17**	.23**	.14**	.13**	.18**	.23**	.20**	.11**	.14**	.14**	.10**
14. LC_L2	−0.00	0.76	.12**	.12**	.18**	.09**	.10**	.12**	.16**	.13**	.13**	.09**	.14**	.08**	.52**
15. Pres_L2	0.00	0.62	−.01	−.03	−.09**	−.06*	−.01	−.05*	−.14**	−.10**	−.10**	−.13**	−.17**	−.09**	−.41**	−.14**
16. Riv_L2	0.00	0.74	−.02	−.01	−.07**	−.03	.01	.01	−.07**	−.06*	−.14**	−.16**	−.24**	−.14**	−.28**	−.31**	.50**
17. TF_L3	−0.01	0.30	.03	.06**	.04	−.02	.04	.07**	.08**	.07**	−.01	.06**	.01	−.03	−.00	−.00	.00	.00
18. LC_L3	−0.01	0.32	−.00	.04	.04*	−.01	.02	.06**	.08**	.03	−.01	.05*	.04	−.01	.00	−.00	.00	−.00	.72**
19. Pres_L3	0.01	0.22	.00	−.05*	−.04*	.01	−.03	−.07**	−.07**	−.04	−.01	−.05*	−.07**	−.02	−.00	.00	−.00	−.00	−.60**	−.49**
20. Riv_L3	−0.01	0.36	.04	−.02	−.06**	−.01	−.01	−.05*	−.08**	−.02	−.01	−.05*	−.10**	−.06**	.00	.00	−.00	−.00	−.48**	−.67**	.59**

*p < .05. **p < .01. SE = Self-esteem; ASC = General academic self-concept; SSC = Social self-concept of acceptance; TF = Teachers’ focus on students; LC = Learning community; Pres = Pressure related to social or achievement issues; Riv = Rivalry and disruptions in class; T1-T4 = First to fourth point of measurement; L2 = Differences within the classes; L3 = Differences between the classes.

Measurement invariance over time

Detailed results of the measurement invariance tests for SE, ASC, and SSC are presented in Table 3. In all scales, we found strict measurement invariance over measurement points (i.e., ΔCFI <.01 between models of different measurement invariance levels), indicating that the manifest mean values of SE, ASC, and SSC were comparable over measurement points. Therefore, subsequent multilevel analyses were conducted based on manifest means of SE, ASC, and SSC instead of latent factor scores.

Table 3.

Results for the tests of measurement invariance over time for the scales assessing self-esteem, general academic self-concept, and social self-concept of acceptance.

Model	χ ²	df	SCF	Δ χ²	Δdf	Δ p	CFI	ΔCFI	RMSEA
SE_configural	187.149	74	1.285				.985		.024
SE_metric	202.959	83	1.294	16.202	9	.063	.984	.001	.023
SE_scalar	249.763	92	1.269	52.364	9	.000	.979	.005	.025
SE_strict	248.482	104	1.378	11.515	12	.485	.981	.002	.023
ASC_configural	31.570	30	1.362				1		.004
ASC_metric	50.514	36	1.398	17.496	6	.008	.998	.002	.012
ASC_scalar	104.384	42	1.361	62.845	6	.000	.990	.008	.023
ASC_strict	163.930	51	1.632	43.310	9	.000	.983	.007	.029
SSC_configural	54.342	30	1.305				.997		.017
SSC_metric	64.950	36	1.267	10.562	6	.103	.996	.001	.017
SSC_scalar	88.759	42	1.260	24.277	6	.000	.994	.002	.020
SSC_strict	101.705	51	1.404	14.914	9	.093	.993	.001	.019

SE = Self-esteem; ASC = General academic self-concept; SSC = Social self-concept of acceptance.

Hypothesis 1

Latent rank-order continuity coefficients in SE (H1a), ASC (H1b), and SSC (H1c) are reported in Table 4. As expected, rank-order continuity in all scales was substantial (continuity range: r = .387 to r = .701). Descriptively, ASC was more stable than SE or SSC.

Table 4.

Latent rank-order continuity coefficients in self-esteem (H1a), general academic self-concept (H1b), and social self-concept of acceptance (H1c).

	T1 - T2	T1 - T3	T1 - T4	T2 - T3	T2 - T4	T3 - T4
SE	r = .586	r = .475	r = .378	r = .594	r = .470	r = .592
ASC	r = .669	r = .498	r = .461	r = .670	r = .591	r = .701
SSC	r = .613	r = .396	r = .352	r = .520	r = .404	r = .515

Note. All coefficients were significant (p < .001). Coefficients were estimated based on scalar measurement invariant autocorrelative SEMs. ASC = General academic self-concept; SSC = Social self-concept of acceptance.

Hypotheses 2 to 4

The results of multilevel regressions testing Hypotheses 2 to 4 for SE (Hypotheses a), ASC (Hypotheses b), and SSC (Hypotheses c) are presented in Tables 5–7, respectively.

Table 5.

Multilevel analyses investigating the effects of class climate on change in self-esteem.

	Model 1				Model 3	Model 4	Model 5I	Model 5II	Model 5III	Model 5IV	Model 6
Fixed effects
Intercept_L1	4.022***			3.935***	3.942***	3.914***	3.937***	3.938***	3.936***	3.937***	3.939**
Time_l_L1				−.071***	−.075***	−.075***	−.076***	−.076***	−.076***	−.073***	−.078***
Time_q_L1				.048***	.040***	.040***	.035***	.036***	.035***	.044***	.034***
TF_L2							.181***				.141***
LC_L2								.131***			.070***
Pres_L2									−.066**		−.001
Riv_L2										−.039*	.016
TF_L3							.060				.082
LC_L3								.044			−.076
Pres_L3									−.069		.061
Riv_L3										−.064	−.011
Time_l_L1: TF_L2							.022*				.021
Time_l_L1: LC_L2								.005			−.008
Time_l_L1: Pres_L2									−.023*		−.009
Time_l_L1: Riv_L2										−.014	−.011
Time_l_L1:TF_L3							−.019				−.012
Time_l_L1: LC_L3								−.003			−.049
Time_l_L1: Pres_L3									.004		.042
Time_l_L1: Riv_L3										−.025	−.087*
Class_type_L3						.104**	.082*	.086*	.089*	.068*	.082*
Random effects variances
Intercept_L2		.202	.211		.225	.225	.224	.230	.238	.220	.224
Intercept_L3		.008	.009		.011	.010	.008	.008	.008	.004	.007
Residual		.295	.288		.255	.255	.252	.252	.252	.288	.251
Time_l_Slope_L2					.017	.017	.021	.021	.021		.021
Time_l_Slope_L3					.002	.002	.002	.002	.002		.002

Note. Self-esteem was used as a criterion variable in all reported models. *p < .05. **p < .01. ***p < .001. TF = Teachers’ focus on students; LC = Learning community; Pres = Pressure related to social or achievement issues; Riv = Rivalry and disruptions in class; L2 = Differences within the classes; L3 = Differences between the classes. Model 5I, Model 5II, and Model 6 failed to converge with the default optimizer; therefore, we applied the optimizer = optimix and the algorithm = L-BFGS-B to estimate these models. Model 5III was estimated without random slopes because of convergence issues.

Table 6.

Multilevel analyses investigating the effects of class climate on change in general academic self-concept.

	Model 1			Model 3	Model 4	Model 5I	Model 5II	Model 5III	Model 5IV	Model 6
Fixed effects
Intercept_L1	3.886***		3.818***	3.817***	3.767***	3.805***	3.808***	3.805***	3.809***	3.808***
Time_l_L1			−.143***	−.149***	−.148***	−.144***	−.143***	−.143***	−.139***	−.145***
Time_q_L1			.027***	.024***	.024 ***	.022***	.023***	.021***	.024***	.022***
TF_L2						.235***				.200***
LC_L2							.142***			.058**
Pres_L2								−.107***		−.034
Riv_L2									−.032	.044*
TF_L3						.127*				.114
LC_L3							.082			−.056
Pres_L3								−.166*		−.054
Riv_L3									−.081	−.052
Time_l_L1: TF_L2						.050***				.035*
Time_l_L1: LC_L2							.025**			.003
Time_l_L1: Pres_L2								−.049***		−.025
Time_l_L1: Riv_L2									−.028***	−.013
Time_l_L1:TF_L3						.025				.026
Time_l_L1: LC_L3							.026			−.020
Time_l_L1: Pres_L3								−.028		.036
Time_l_L1: Riv_L3									−.031	−.050
Class_type_L3					.184***	.145***	.150***	.156***	.155***	.147***
Random effects variances
Intercept_L2		.272	.280	.315	.315	.289	.301	.310	.271	.286
Intercept_L3		.018	.016	.015	.009	.008	.008	.007	.008	.008
Residual		.322	.291	.243	.243	.238	.238	.238	.287	.238
Time_l_Slope_L2				.028	.028	.030	.030	.030		.030
Time_l_Slope_L3				.001	.001	.001	.001	.001		.001

General academic self-concept was used as a criterion variable in all reported models. *p < .05. **p < .01. ***p < .001. TF = Teachers’ focus on students; LC = Learning community; Pres = Pressure related to social or achievement issues; Riv = Rivalry and disruptions in class; L2 = Differences within the classes; L3 = Differences between the classes. Models 4, 5I, and 6 failed to converge with the default optimizer; therefore, we applied the optimizer = optimix and the algorithm = L-BFGS-B to estimate this model. Model 5 IV was estimated without random slopes because of convergence issues.

Table 7.

Multilevel analyses investigating the effects of class climate on change in social self-concept of acceptance.

	Model 1		Model 3	Model 4	Model 5I	Model 5II	Model 5III	Model 5IV	Model 6
Fixed effects
Intercept_L1	4.251***	4.171***	4.179***	4.166***	4.179***	4.181***	4.179***	4.178***	4.181***
Time_l_L1		−.012	−.016	−.017	−.022*	−.022	−.023*	−.023*	−.023*
Time_q_L1		.048***	.039***	.039***	.037***	.038***	.038***	.038***	.036***
TF_L2					.178***				.080**
LC_L2						.142***			.056*
Pres_L2							−.169***		−.028
Riv_L2								−.216***	−.164***
TF_L3					−.005				−.101
LC_L3						.038			−.105
Pres_L3							.169		−.103
Riv_L3								−.153**	−.217*
Time_l_L1: TF_L2					.004				.010
Time_l_L1: LC_L2						−.004			−.012
Time_l_L1: Pres_L2							.000		.012
Time_l_L1: Riv_L2								−.008	−.012
Time_l_L1:TF_L3					−.028				−.027
Time_l_L1: LC_L3						−.017			−.060
Time_l_L1: Pres_L3							−.015		–.004
Time_l_L1: Riv_L3								−.028	−.074
Class_type_L3				.049	.069	.066	.065	.066	.077
Random effects variances
Intercept_L2	.328	.338	.339	.339	.313	.310	.315	.296	.289
Intercept_L3	.015	.013	.017	.018	.017	.018	.015	.015	.013
Residual	.494	.498	.461	.461	.453	.454	.453	.453	.453
Time_l_Slope_L2			.020	.020	.022	.022	.022	.022	.022
Time_l_Slope_L3			.004	.004	.004	.004	.004	.004	.004

The social self-concept of acceptance was used as a criterion variable in all reported models. *p < .05. **p < .01. ***p < .001. TF = Teachers’ focus on students; LC = Learning community; Pres = Pressure related to social or achievement issues; Riv = Rivalry and disruptions in class; L2 = Differences within the classes; L3 = Differences between the classes. Models 5I, 5III, 5IV, and 6 failed to converge with the default optimizer; therefore, we applied the optimizer = optimix and the algorithm = L-BFGS-B to estimate this model.

Self-esteem

Model 1 revealed that 58% of the overall variance in SE could be allocated to differences between Level 1 units, 40% to differences between Level 2 units, and 2% to differences between Level 3 units. SE showed a nonlinear mean-level change over time (Model 2; linear effect: b = −.071, p < .001; quadratic effect: b = .048, p < .001; H2a). That is, initial declines in SE decreased over time. Moreover, SE stabilized and then even slightly increased. The exact SE mean-level change trajectory over time calculated based on Model 2 is depicted in Figure 2. Model 3 revealed that the time effect on SE significantly varied between Level 2 units (slope variance = .017, p < .001) and Level 3 units (slope variance = .002, p < .001). That is, we found significant interindividual differences in the development of SE (H3a). Class type was a significant Level 3 predictor of SE (Model 4) with students from gifted classes showing higher SE (b = .104, p = .002).

Figure 2.

Mean-level change over time. The thick line represents the average development trajectories based on the respective Model 2 results in SE, ASC, and SSC. The gray lines represent the individual trajectories for a random subsample of students. The model intercept was set to the third measurement point. SE = self-esteem; ASC = General academic self-concept; SSC = Social self-concept of acceptance.

The results of Model 5I to Model 6 are presented in Figure 3 (for exact numbers, see Table 5). SE was positively predicted by TF_L2 and LC_L2 and negatively predicted by Pres_L2 and Riv_L2. Interindividual differences in the development of SE were positively predicted by TF_L2 and negatively predicted by Pres_L2. Finally, Model 6 revealed that TF_L2 and LC_L2 remained positive predictors of SE even after controlling for the other classroom climate dimensions. Surprisingly, intraindividual development of SE was only significantly and negatively predicted by Riv_L3 in this model, although Riv_L3 did not significantly predict SE in Model 5IV.

Figure 3.

Self-esteem, academic self-concept, and social self-concept predicted by classroom climate. Regressions stemmed from the respective Model 5. Transparent lines represent nonsignificant regressions. *p < .05. **p < .01. ***p < .001. SE = self-esteem; ASC = General academic self-concept; SSC = Social self-concept of acceptance; TF = Teachers’ focus on students; LC = Learning community; Pres = Pressure related to social or achievement issues; Riv = Rivalry and disruptions in class.

General academic self-concept

For ASC, 53% of the overall variance could be allocated to differences between Level 1 units, 44% to differences between Level 2 units, and 3% to differences between Level 3 units (Model 1). ASC showed a nonlinear mean-level decline over time (Model 2, linear effect: b = −.143, p < .001; quadratic effect: b = .027, p < .001; H2b). That is, the decline in ASC decreased over time. The exact ASC mean-level change trajectory over time is depicted in Figure 2 (calculated based on Model 2). The time effect on ASC (Model 3) significantly varied between Level 2 units (slope variance = .028, p < .001) and Level 3 units (slope variance = .001, p = .066). This indicates significant interindividual differences in the intraindividual development of ASC (H3b). Class type was a significant Level 3 predictor, with students from gifted classes showing higher ASC (b = .184, p < .001; Model 4).

The results of Model 5I to Model 6 are presented in Figure 3 (for exact numbers, see Table 6). ASC was positively predicted by TF_L2, TF_L3, and LC_L2 and negatively predicted by Pres_L2 and Pres_L3. Interindividual differences in the development of ASC were positively predicted by TF_L2 and LC_L2 and negatively predicted by Pres_L2 and Riv_L2. Finally, Model 6 revealed that TF_L2 and LC_L2 remained positive predictors of ASC even after controlling for the other classroom climate dimensions. Surprisingly, Riv_L2 positively predicted ASC in this model, although Riv_L2 did not significantly predict ASC in Model 5IV. Finally, only LC_L2 remained a significant predictor of the development of ASC after controlling for the other classroom climate dimensions in Model 6.

Social self-concept of acceptance

In SSC, 59% of the overall variance could be explained by differences between Level 1 units, 39% by differences between Level 2 units, and 2% by differences between Level 3 units (Model 1). SSE showed nonlinear mean-level change over time (Model 2; linear effect: b = −.012, p = .113; quadratic effect: b = .048, p < .001; H2c). Mean SSC initially declined but then reversed and increased over time. The exact SSC mean-level change trajectory calculated based on Model 2 is depicted in Figure 2. The time effect on SSC (Model 3) significantly varied between Level 2 (slope variance = .020, p < .001) and Level 3 units (slope variance = .004, p < .001). This indicates significant interindividual differences in the development of SSC (H3c). Class type did not significantly predict SSC (b = .049, p = .250; Model 4).

The results of Model 5I to Model 6 are presented in Figure 3 (for exact numbers, see Table 7). SSC was positively predicted by TF_L2 and LC_L2 and negatively predicted by Pres_L2, Riv_L2, and Riv_L3. The interindividual development of SSC was not predicted by classroom climate. Model 6 revealed that TF_L2 and LC_L2 remained positive predictors of SSC and Riv_L2 and Riv_L3 remained negative predictors of SSC even after controlling for the other classroom climate dimensions.

Discussion

In our longitudinal study, we investigated rank-order continuity and the mean-level change in adolescents’ SE, ASC, and SSC in their first 4 years in secondary school. Based on the assumptions of stage-environment-fit theory, we tested whether there are interindividual differences in intraindividual change and whether these differences can be explained by characteristics of the classroom environment (i.e., classroom climate). All relevant results are summarized in Table 8. As expected, rank-order continuities of SE, ASC, and SSC were substantial. Multilevel regressions revealed that approximately half of the overall variability in SE, ASC, and SSC occurred due to individual changes over time. We found a significant nonlinear mean-level change in all constructs under investigation, indicating an initial decrease that decreased over time. In SE and SSC, the declines even reversed, so that there were no substantial mean differences between the first and last measurement points. Furthermore, for all constructs under investigation, we found significant interindividual differences in development. In SE, these differences were related to students’ individual experiences of teacher-student-related climate dimensions only and not related to differences between classrooms in these dimensions or to individual or classroom differences in student-student-related climate dimensions. In ASC, interindividual differences in development were related to students’ individual experiences in all climate dimensions but unrelated to differences between classrooms. Finally, interindividual differences in the development of SSC were not related to differences in students’ individual experiences or to classroom differences in the climate dimensions.

Table 8.

Summary of the results.

	Variance proportions on levels of investigation	H1: Continuity coefficients	H2: Mean-level change	H3: Interindividual differences in the intraindividual development	H4: Classroom climate as predictor of the intraindividual development
Self-esteem	Within students = 58%	r = .378–.592	Non-linear change	Yes	Individual TF	Positive
	Within classes = 40%				Individual LC	n.s.
	Between classes = 2%				Individual Pres	Negative
					Individual Riv	n.s.
					Classroom TF	n.s.
					Classroom LC	n.s.
					Classroom Pres	n.s.
					Classroom Riv	n.s.
General academic self-concept	Within students = 53%	r = .461–.701	Non-linear change	Yes	Individual TF	Positive
	Within classes = 44%				Individual LC	Positive
	Between classes = 4%				Individual Pres	Negative
					Individual Riv	Negative
					Classroom TF	n.s.
					Classroom LC	n.s.
					Classroom Pres	n.s.
					Classroom Riv	n.s.
Social self-concept of acceptance	Within students = 59%	r = .352–.613	Non-linear change	Yes	Individual TF	n.s.
	Within classes = 39%				Individual LC	n.s.
	Between classes = 2%				Individual Pres	n.s.
					Individual Riv	n.s.
					Classroom TF	n.s.
					Classroom LC	n.s
					Classroom Pres	n.s
					Classroom Riv	n.s.

Note. TF = Teachers’ focus on students; LC = Learning community; Pres = Pressure related to social or achievement issues; Riv = Rivalry and disruptions in class; n.s. = not significant.

Rank-order continuity

In accordance with our expectations and previous research (Lindner-Müller et al., 2012; Preckel et al., 2013; Scherrer & Preckel, 2019; Spilt et al., 2014), we observed substantial rank-order continuity in SE, ASC, and SSC (range: r = .387 to r = .701). That is, many students kept their rank position compared to peers within the first 4 years of middle school. Stated differently, students with a high/low SE, ASC, or SSC at the beginning of middle school compared to their peers tended to show a similarly high/low SE, ASC, or SSC at the subsequent measurement points. However, autocorrelations also demonstrated a certain variability over time (i.e., autocorrelations were less than r = 1), which indicates that there were students who showed relatively strong decreases or increases in their self-perceptions. Furthermore, multilevel analyses revealed considerable variability in SE, ASC, and SSC within individuals over time. Of note, most variability was present due to changes within persons: The variance proportion within persons was higher than the proportions of the variance between individuals and between classes altogether. Finally, ASC showed a higher rank-order continuity than SE and SSC. This could be because ASC is positively related to school grades (Marsh & O'Mara, 2008), which show high stability over time (e.g., Gallardo & Barrasa, 2018; Tang et al., 2021; Weidinger et al., 2020).

Mean-level change

In line with our expectations and with prior findings for SE (Lindner-Müller et al., 2012; Orth et al., 2018; Scherrer & Preckel, 2019; Spilt et al., 2014), we did not observe substantial mean-level change over time when comparing mean levels at the first and the last point of measurement. The same finding held for SSC, where we had expected to find slight increases. However, the change was nonlinear. SE and SSC initially decreased, stabilized over time, and increased again at the end of the study. The first measurement point occurred shortly after the students moved from elementary to middle school. According to stage-environment fit theory, initial decreases could be expected after this transition (Eccles et al., 1993). Our findings of an increase in the end indicate, however, that for SE and SSC, this negative development in the new school environment was not permanent.

For ASC, we expected no mean-level change. However, we found a significant and nonlinear decrease in ASC over time. Decreases decreased over time but lasted until the end of the investigation. The discrepancies between our findings and previous meta-analytic findings (Scherrer & Preckel, 2019) might be explained by the fact that we observed a relatively long time interval of almost 4 years, whereas the longitudinal studies reviewed by Scherrer and Preckel (2019) on average referred to interval durations of 1.83 years for repeated measures of ASC. Moreover, in our analyses, we took into account nonlinear time effects and the varying distances between measurement points between the students. ASC, SE, and SSC all showed nonlinear changes over time. We therefore propose that future research assess multiple measurement points and take into account possible nonlinear development trends.

Interindividual variation in the mean-level change and the role of classroom climate

In accordance with our expectations, we found significant variation in the mean-level change of all constructs under study. Some students reported stronger declines in their self-perceptions than other students, whereas some students even reported increased self-perceptions over time. These findings support the assumption of interindividual differences in developmental change (De Ruiter et al., 2017; Eccles et al., 1993; Eccles & Roeser, 2009; Gutman & Eccles, 2007). We investigated whether these differences can be explained by differences in classroom climate. In line with prior findings from cross-sectional studies (Wang et al., 2020), we found that the level of all self-perceptions was related to classroom climate. We observed these relations with individual climate (i.e., students’ individual perceptions of classroom climate dimensions) and with classroom climate (i.e., climate perceptions aggregated for all students within a classroom). Note that the relationships between climate and the level of self-perceptions are purely correlational and allow no causal interpretation. That is, it remains unclear whether students report low (or high) self-perceptions because of a bad (or good) classroom climate or whether there is a bad (or good) classroom climate because of students’ self-perceptions. Therefore, we limit the following interpretation of our findings to longitudinal relations.

Regarding classroom climate and its relation to interindividual differences in the development in SE, ASC, and SSC, we found a rather differentiated results pattern that varied by construct under study. The only general tendency we found was that classroom climate did not significantly moderate the development of any self-perception. Individual climate, on the other hand, moderated the development of SE and ASC (but not SSC). Students who experienced a better teachers’ focus on students and lower pressure related to social or achievement issues developed better in their self-esteem over time than students who experienced less supportive and more directive, authoritarian teacher behavior. When including all classroom climate dimensions in one model, only rivalry and disruption in class at the classroom level negatively predicted the development in SE, while the other dimensions showed no incremental effects. Since rivalry and disruption in class were not significant single moderator variables, this finding is most likely due to suppression effects. Note that the four classroom climate dimensions were relatively highly correlated with each other on the individual level (positive and negative correlations ranged from .14 to .52, see Table 2) as well as on the classroom level (positive and negative correlations ranged from .49 to .72, see Table 2). That is, most bivariate relationships between each classroom climate dimension and SE, ASC, and SSC can be explained by the shared variance of all dimensions. On the other hand, it seems that the shared variance of all dimensions covered up the relationship between rivalry and disruption and development in SE as this relationship was only observed after controlling for the effect of other classroom climate dimensions.

Supportive and nondirective teacher behavior also supported the development of students’ academic self-perceptions (ASC). Moreover, experiencing a positive classroom community and a willingness to learn among students, as well as less competition and disruptions in class, accompanied a more positive development of students’ ASC. When analyzing all classroom climate dimensions simultaneously, teachers’ focus on students on the individual level remained a positive predictor of ASC development, which supports its incremental validity for the development of students’ ASC.

The significant relations between all classroom climate dimensions and the development of ASC are in accordance with the assumption of the stage-environment fit theory that teacher–student relationships are a key aspect of classroom climate and that peer relations play an increasingly important role in adolescence (Eccles et al., 1993; Eccles & Roeser, 2009). Our findings support the meaning of both teacher-student-related climate dimensions and student-student-related climate dimensions for the development of ASC, with teachers’ focus on students and a good learning community in class promoting growth in students’ academic self-perceptions, and pressure and rivalry disrupting this developmental process. However, it seems that the development of SE is mainly related to teacher-student-related climate dimensions. This finding is in accordance with the literature assuming that teacher–student relationships are an important factor for positive development in middle school students (Myers & Pianta, 2008). Positive teacher–student relationships evolve through warm, sensitive, and responsive interactions with the teacher within the classroom, whereas negative teacher–students relationships often come along with a high level of teacher conflict (Engels et al., 2016). Positive teacher–student relationships seems to contribute to positive self-perceptions and consequently to a growing SE in students. The finding that classroom climate dimensions referring to student-student relationships were not associated with development in SE can possibly be explained by the assumption that students’ interactions with teachers are more relevant to satisfy their desire for security, harmony, and comfort than students’ interactions with peers.

The differences between SE and ASC could further be explained by the fact that the formation of students’ ASC is highly dependent on social comparison processes (Marsh, 1986; Möller et al., 2011). Therefore, one could speculate that a positive learning community reduces social comparisons between students and thus enables them to develop a positive ASC regarding themselves based on internal comparison processes. On the other hand, rivalry and disruptions in class could increase the salience of extrinsic motivation, performance-avoidance goals, and social comparison processes, which, in turn, could have had a negative impact on the development of achievement-relevant self-perceptions (Eccles & Roeser, 2009).

Other than expected, the development of students’ self-concept of social acceptance was unrelated to the class climate dimensions under study. Cross-sectionally, we found negative relations with rivalry and disruptions in class at both the individual and classroom levels. It is very plausible that students who experience more competition in class to the detriment of others in classes that are characterized by such a climate feel less socially accepted. But why is the development of their SSC unrelated to this class climate dimension? Compared to the other self-perceptions, SSC overall showed higher means, and ceiling effects of the scale might have led to range restrictions (see Figure 2) that impeded us from finding relations of change with classroom climate. Therefore, further research with better discriminating scales is needed before interpreting this finding with regard to content.

Interestingly, our findings indicate that individual climate (i.e., classroom climate on the individual level) seems to be more important than classroom climate (i.e., classroom climate on the classroom level) for the development of students’ self-perceptions. A possible explanation for our findings is that individual climate may reflect the subjective person-environment fit. That is, students may have reported a positive classroom climate if their individual needs matched the opportunities afforded to them by the school environment, regardless of whether other students in their class evaluated the classroom climate differently based on their needs. However, this explanation is quite speculative, as we did not observe students’ individual needs and thus could not estimate the actual fit between students’ needs and opportunities afforded to them by the school environment. Nevertheless, our findings and explanations are in line with the stage environment fit theory. According to the stage environment fit theory, the individual fit between students’ needs and opportunities afforded to them by the environment is a key factor for students’ motivational development. Accordingly, to the extent that individual climate reflects subjective person-environment fit, the finding that individual climate is even more important for personal development than classroom climate is theoretically plausible. Although the classroom climate indicated by the aggregated ratings of all class members may be an indicator of environmental influences, it does not reflect whether these environmental influences fit to the individual needs. According to the dynamic systems approach, intraindividual variations in SE can be explained by the interplay between extrinsic and intrinsic forces. Our finding that individual climate is more important for the development of students’ self-perceptions than classroom climate supports the view that extrinsic forces (indicated by classroom climate) are not sufficient to explain the variability in students’ self-perceptions. Note that we did not focus our investigation on intrinsic forces and the interplay of intrinsic and extrinsic forces. Therefore, we cannot say much about their importance for the development of students’ self-perceptions. In addition, our explanations are quite speculative, and the nonsignificant findings regarding the aggregated classroom climate ratings could be alternatively explained by the fact that this variable needs to be measured directly at the classroom level (e.g., by external observations) instead of aggregating ratings at the individual level. Further research should focus on using multiple methods to assess classroom climate (external observations, teacher ratings in addition to student ratings, etc.).

Our findings add to the ongoing debate about whether classroom climate at the individual level is a reasonable estimate of students’ classroom climate or simply reflects measurement error that should not be interpreted systematically (Marsh et al., 2012). Hank et al. (2022) found that the classroom climate dimensions that we used in our study can be validly and reliably measured at both the individual and classroom levels. Our study adds evidence that individual climate is meaningfully related to the development of students’ self-perceptions. To conclude, our findings reveal that classroom climate needs to be investigated and analyzed on both levels in subsequent research on the effects of classroom climate on SE, ASC, and SSC and their development.

Limitations and recommendations for further research

We conducted a large-scale longitudinal study over almost 4 years based on 2722 students in 98 classes, which is a rarity in the literature (see Wang et al., 2020). However, one limitation of this study is the fact that it only focused on the highest school track of the German secondary school system. Therefore, the scope of our work is limited to a Western European educational context, and it is therefore crucial to replicate our results in other school systems and countries before concluding general implications based on our findings. A further limitation is that the exact timing of the measurement points varied considerably among the students. For example, the last measurement point took place in the eighth grade for some students, while for others, it occurred in the seventh grade, which could have affected the stability of the investigated constructs. However, it must be highlighted that we took into account the exact timing of the measurement points (i.e., the time effect was calculated based on varying dates between students) in all calculated multilevel regression models. Furthermore, the varied timing of the measurement points could be seen as a methodological strength of our investigation, as this allowed us to obtain more systematic variance of the time effects, which allowed us to appropriately model nonlinear developmental processes.

Another limitation is that classroom climate dimensions were only assessed by student self-report. Therefore, ratings of classroom climate could be biased by how well the students performed in the class and how likeable the teacher they found the teacher (Wang et al., 2020). In addition, it must be noted that all variables under investigation were assessed by self-report questionnaires. Although self-report questionnaires are the most frequently used method for assessing students’ self-perceptions, self-report questionnaires have been criticized for numerous issues, including common method bias, socially desirable responding (Podsakoff & Organ, 1986), and the inability to asses person-environment interactions (Ortner & van Vijver, 2015). To rule out common method bias as a possible explanation for the observed relations between classroom climate and students’ SE and ASC, different assessment tools, such as teacher ratings, should be used in future research to assess some of the variables under investigation (Podsakoff et al., 2003; Podsakoff et al., 2012). However, alternative assessment strategies, such as teacher reports or external observations, also have limitations. For example, teacher reports are often biased in that teachers may be motivated to report a more favorable classroom climate than it truly is (Wang et al., 2020). Students’ reports can be used to calculate both aggregated class averages (i.e., classroom climate level), which average out individual biases, and individual deviation from the aggregated class averages (i.e., individual climate level), which specifically reflect students’ individual evaluation of the classroom climate. Finally, since we did not implement any experimental manipulation, our analyses were purely correlative. Thus, common-method variance of the self-report measures used is a possible alternative explanation for the found relationships (Antonakis et al., 2010). In addition, further variables such as students’ gender, cognitive ability, socioeconomic status, or achievement motivation might have affected our findings. Furthermore, it must be noted that we only investigated the effects of class climate on the change in students’ self-perceptions, whereas reversed effects (i.e., effects of self-perceptions on the change in classroom climate) were not investigated due to the design of the present study. While for SE, ASC, and SSC, there were four common measurement points, for classroom climate, there was only one common measurement point for all students. It would certainly be interesting to investigate reciprocal effects between students’ self-perceptions and class climate in further research.

Conclusions

In summary, we found substantial rank-order continuities and small nonlinear mean-level changes in SE, ASC, and SSC, indicating an initial decrease that diminished over the first years in secondary school. Furthermore, we observed a large intraindividual variability in SE, ASC, and SSC, along with significant interindividual differences of the intraindividual development over time. These findings map nicely onto the perspective that students’ self-perceptions reflect both dispositional and situationally responsive components (De Ruiter et al., 2018; Jayawickreme et al., 2019). According to the stage-environment-fit theory, it is plausible that declines at the beginning of middle school are partly due to a mismatch between students’ needs and the opportunities afforded by the school. Eccles and Roeser (2009) suggested several ecological levels where this mismatch can be mitigated (i.e., classrooms, schools, school districts, and communities). Smaller class sizes that make teachers more accessible, extracurricular activities that support students' well-being (Eccles & Templeton, 2002; Lee & Smith, 1997), and later start times for school that minimize the discrepancy between students' circadian preferences and school attendance times (Goldin et al., 2020; Scherrer & Preckel, 2021) are just a few examples of general recommendations to make learning more enjoyable and to ensure a positive development in students’ self-perceptions.

Interestingly, we found that intraindividual development in SE and ASC was predicted by students’ individual perceptions of classroom climate, whereas classroom climate perceptions aggregated for all students within a classroom did not predict the intraindividual development in any construct under investigation. Accordingly, it seems that students perceive their school environment differently and that these differences are important for the development of their self-perceptions. We conclude that in addition to general measures to improve the school environment, it is also important to focus on the individual students by assessing their individual needs and their perception of the current school environment to ensure that everyone feels comfortable and develops positive self-perceptions.

Footnotes

Acknowledgments

Data of Study 1 stem from the PULSS project conducted by Wolfgang Schneider, Eva Stumpf, Franzis Preckel, and Albert Ziegler. The PULSS project is a two-cohort longitudinal study on school achievement and academic development at the beginning of secondary education. The project was supported by the Ministry of Cultural Affairs and Education in Bavaria, by the Ministry of Education, Youth and Sports in Baden-Wuerttemberg, and by the Karg Foundation. The project was carried out in cooperation of the University of Wuerzburg, the University of Trier, and the University Erlangen-Nuremberg. Data from Study 2 stem from the AVG project conducted by Franzis Preckel. The AVG project focuses on motivational and self-concept development in secondary school students. This longitudinal study is supported by the Ministry of Education, Science, Adolescence, and Culture of Rhineland-Palatinate. The funding sources were not involved in decisions referring to study design, analysis, or interpretation of the data, respectively.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Data accessibility statement

The hypotheses and analyses of this article were preregistered on OSF before conducting the analyses:

The study analyses scripts and the covariation matrices of the investigated data can be accessed at:

Data cannot be made openly available due to restrictions in the project approval of the school supervisory board of Rhineland-Palatinate (ADD). Example items for each measurement are given in the methods section. The complete material cannot be made publicly available due to third-party rights on the questionnaires.

Notes

References

Allodi

M. W.

(2010). The meaning of social climate of learning environments: Some reasons why we do not care enough about it. Learning Environments Research, 13(2), 89–104. https://doi.org/10.1007/s10984-010-9072-9

Ambrose

S. A.

Bridges

M. W.

DiPietro

Lovett

M. C.

(2010). How learning works: Seven research-based principles for smart teaching. Jossey Bass.

Antonakis

Bendahan

Jacquart

Lalive

(2010). On making causal claims: A review and recommendations. The Leadership Quarterly, 21(6), 1086–1120. https://doi.org/10.1016/j.leaqua.2010.10.010

Arens

A. K.

Jansen

Preckel

Schmidt

Brunner

(2021). The structure of academic self-concept: A methodological review and empirical illustration of central models. Review of Educational Research, 91(1), 34–72. https://doi.org/10.3102/0034654320972186

Avant

T. S.

Gazelle

Faldowski

(2011). Classroom emotional climate as a moderator of anxious solitary children's longitudinal risk for peer exclusion: A child × environment model. Developmental Psychology, 47(6), 1711–1727. https://doi.org/10.1037/a0024021

Babad

(2009). The social psychology of the classroom. Routledge. https://doi.org/10.4324/9780203872475

Bandura

(1993). Perceived self-efficacy in cognitive development and functioning. Educational Psychologist, 28(2), 117–148. https://doi.org/10.1207/s15326985ep2802_3

Bandura

Barbaranelli

Caprara

G. V.

Pastorelli

(1996). Multifaceted impact of self-efficacy beliefs on academic functioning. Child Development, 67(3), 1206–1222. https://doi.org/10.1111/j.1467-8624.1996.tb01791.x

Bates

Maechler

Bolker

Walker

Christensen

R. H. B.

Singmann

Green

(2018). Linear mixed-effects models using 'eigen' and S4. https://github.com/lme4/lme4/

10.

Berndt

T. J.

Burgy

(1996). Social self-concept. In Bracken

B. A.

(Ed.), A Wiley-Interscience publication. Handbook of self-concept: Developmental, social, and clinical considerations. Wiley.

11.

Bong

Skaalvik

E. M.

(2003). Academic self-concept and self-efficacy: How different are they really? Educational Psychology Review, 15(1), 1–40. https://doi.org/10.1023/a:1021302408382

12.

Byrne

B. M.

Shavelson

R. J (

1996). On the structure of social self-concept for pre-, early, and late adolescents: A test of the Shavelson, Hubner, and Stanton (1976) model. Journal of Personality and Social Psychology, 70(3), 599–613. https://doi.org/10.1037//0022-3514.70.3.599

13.

Chen

F. F.

(2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14(3), 464–504. https://doi.org/10.1080/10705510701301834

14.

Coelho

V. A.

Bear

G. G.

Brás

(2020). A multilevel analysis of the importance of school climate for the trajectories of students' self-concept and self-esteem throughout the middle school transition. Journal of Youth and Adolescence, 49(9), 1793–1804. https://doi.org/10.1007/s10964-020-01245-7

15.

Conley

J. J.

(1984). The hierarchy of consistency: A review and model of longitudinal findings on adult individual differences in intelligence, personality and self-opinion. Personality and Individual Differences, 5(1), 11–25. https://doi.org/10.1016/0191-8869(84)90133-8

16.

Connolly

(1989). Social self-efficacy in adolescence: Relations with self-concept, social adjustment, and mental health. Canadian Journal of Behavioural Science/Revue Canadienne Des Sciences Du Comportement, 21(3), 258–269. https://doi.org/10.1037/h0079809

17.

Crocker

Wolfe

C. T.

(2001). Contingencies of self-worth. Psychological Review, 108(3), 593–623. https://doi.org/10.1037/0033-295X.108.3.593

18.

DeNeve

K. M.

Cooper

(1998). The happy personality: A meta-analysis of 137 personality traits and subjective well-being. Psychological Bulletin, 124(2), 197–229. https://doi.org/10.1037/0033-2909.124.2.197

19.

De Ruiter

N. M. P.

Den Hartigh

R. J. R.

Cox

R. F. A.

Van Geert

P. L. C.

Kunnen

E. S.

(2014). The temporal structure of state self-esteem variability during Parent–adolescent interactions: More than random fluctuations. Self and Identity, 14(3), 314–333. https://doi.org/10.1080/15298868.2014.994026

20.

De Ruiter

N. M. P.

Hollenstein

van Geert

P. L. C.

Saskia Kunnen

(2018). Self-esteem as a complex dynamic system: Intrinsic and extrinsic microlevel dynamics. Complexity, 2018, 4781563. https://doi.org/10.1155/2018/4781563

21.

De Ruiter

N. M. P.

Van Geert

P. L. C.

Kunnen

E. S.

(2017). Explaining the “how” of self-esteem development: The self-organizing self-esteem model. Review of General Psychology, 21(1), 49–68. https://doi.org/10.1037/gpr0000099

22.

Donnellan

M. B.

Kenny

D. A.

Trzesniewski

K. H.

Lucas

R. E.

Conger

R. D.

(2012). Using trait-state models to evaluate the longitudinal consistency of global self-esteem from adolescence to adulthood. Journal of Research in Personality, 46(6), 634–645. https://doi.org/10.1016/j.jrp.2012.07.005

23.

Eccles

J. S.

Midgley

Wigfield

Buchanan

C. M.

Reuman

Flanagan

Mac Iver

(1993). Development during adolescence: The impact of stage-environment fit on young adolescents' experiences in schools and in families. American Psychologist, 48(2), 90–101. https://doi.org/10.1037/0003-066X.48.2.90

24.

Eccles

J. S.

Roeser

R. W.

(2009). Schools, academic motivation, and stage-environment fit. In Lerner

R. M.

Steinberg

(Eds.), Handbook of adolescent psychology. John Wiley & Sons, Inc. https://doi.org/10.1002/9780470479193.adlpsy001013

25.

Eccles

J. S.

Templeton

(2002). Chapter 4: Extracurricular and other after-school activities for youth. Review of Research in Education, 26(1), 113–180. https://doi.org/10.3102/0091732X026001113

26.

Eccles

J. S.

Wigfield

(2002). Motivational beliefs, values, and goals. Annual Review of Psychology, 53, 109–132. https://doi.org/10.1146/annurev.psych.53.100901.135153

27.

Eder

(2010). Schul- und Klassenklima. In Rost

D. H.

(Ed.), Handwörterbuch pädagogische psychologie (pp. 694–703). Beltz PVU.

28.

Eder

Mayr

(2000). Linzer Fragebogen zum Schul-und Klassenklima für die 4.-8. Klassenstufe [Linzer Questionnaire of Classroom Climate for Grade Fourth through Eight]. Hogrefe.

29.

Engels

M. C.

Colpin

van Leeuwen

Bijttebier

van den Noortgate

Claes

Goossens

Verschueren

(2016). Behavioral engagement, peer status, and teacher-student relationships in adolescence: A longitudinal study on reciprocal influences. Journal of Youth and Adolescence, 45(6), 1192–1207. https://doi.org/10.1007/s10964-016-0414-5

30.

Fend

(1986). Dokumentation der Skalen des Projekts“Entwicklung im Jugendalter”[Documentation of scales of the project“development inadolescence”]. University of Konstanz.

31.

Ferla

Valcke

Cai

(2009). Academic self-efficacy and academic self-concept: Reconsidering structural relationships. Learning and Individual Differences, 19(4), 499–505. https://doi.org/10.1016/j.lindif.2009.05.004

32.

Fortes

Delignieres

Ninot

(2003). PsycEXTRA dataset. https://doi.org/10.1037/e547922012-114

33.

Freudenthaler

H. H.

Spinath

Neubauer

A. C.

(2008). Predicting school achievement in boys and girls. European Journal of Personality, 22(3), 231–245. https://doi.org/10.1002/per.678

34.

Gallardo

L. O.

Barrasa

(2018). Self-other agreement measures of acceptance in predicting academic achievement: A longitudinal analysis. Current Psychology, 37(1), 107–115. https://doi.org/10.1007/s12144-016-9494-3

35.

Geiser

Götz

Preckel

Freund

P. A.

(2017). States and traits. European Journal of Psychological Assessment, 33(4), 219–223. https://doi.org/10.1027/1015-5759/a000413

36.

Goldin

A. P.

Sigman

Braier

Golombek

D. A.

Leone

M. J.

(2020). Interplay of chronotype and school timing predicts school performance. Nature Human Behaviour, 4(4), 387–396. https://doi.org/10.1038/s41562-020-0820-2

37.

Guay

Marsh

H. W

Boivin

(2003). Academic self-concept and academic achievement: Developmental perspectives on their causal ordering. Journal of Educational Psychology, 95(1), 124–136. https://doi.org/10.1037/0022-0663.95.1.124

38.

Gutman

L. M.

Eccles

J. S.

(2007). Stage-environment fit during adolescence: Trajectories of family relations and adolescent outcomes. Developmental Psychology, 43(2), 522–537. https://doi.org/10.1037/0012-1649.43.2.522

39.

Hank

Baltes-Götz

(2019). The stability of self-esteem variability: A real-time assessment. Journal of Research in Personality, 79, 143–150. https://doi.org/10.1016/j.jrp.2019.03.004

40.

Hank

Baltes-Götz

Preckel

(2022). Welche Klimadimensionen misst der Linzer Fragebogen zum Schul- und Klassenklima für die 4. – 8. Klasse (LFSK 4 – 8; Eder, 2000)? Diagnostica, 68(1), 28–38. https://doi.org/10.1026/0012-1924/a000280

41.

Hansford

B. C.

Hattie

J. A.

(1982). The relationship between self and achievement/performance measures. Review of Educational Research, 52(1), 123. https://doi.org/10.2307/1170275

42.

Harris

M. A.

Orth

(2020). The link between self-esteem and social relationships: A meta-analysis of longitudinal studies. Journal of Personality and Social Psychology, 119(6), 1459–1477. https://doi.org/10.1037/pspp0000265

43.

Harter

(1990). Processes underlying adolescent self-concept formation. In Montemayor

Adams

G. R.

Gullotta

T. P.

(Eds.), From childhood to adolescence: A transitional period? (pp. 205–239). Sage Publications.

44.

Hendrick

S. S.

Hendrick

Adler

N. L.

(1988). Romantic relationships: Love, satisfaction, and staying together. Journal of Personality and Social Psychology, 54(6), 980–988. https://doi.org/10.1037/0022-3514.54.6.980

45.

Huang

(2011). Self-concept and academic achievement: A meta-analysis of longitudinal relations. Journal of School Psychology, 49(5), 505–528. https://doi.org/10.1016/j.jsp.2011.07.001

46.

Jayawickreme

Zachry

C. E.

Fleeson

(2019). Whole trait theory: An integrative approach to examining personality structure and process. Personality and Individual Differences, 136, 2–11. https://doi.org/10.1016/j.paid.2018.06.045

47.

Körük

(2017). The effect of self-esteem on student achievement. In Karadag

(Ed.), The factors effecting student achievement (pp. 247–257). Springer International Publishing. https://doi.org/10.1007/978-3-319-56083-0_15

48.

Leary

M. R.

Baumeister

R. F.

(2000). The nature and function of self-esteem: Sociometer theory. Advances in Experimental Social Psychology, 32, 1–62. https://doi.org/10.1016/S0065-2601(00)80003-9

49.

Lee

V. E.

Smith

J. B.

(1997). High school size: Which works best and for whom? Educational Evaluation and Policy Analysis, 19(3), 205–227. https://doi.org/10.3102/01623737019003205

50.

Lindner-Müller

John

Arnold

K.-H.

(2012). Longitudinal assessment of elementary school students’ social self-concept in relation to social preference. Journal for Educational Research, 4(1), 47–72. https://doi.org/10.25656/01:7051

51.

Little

T. D.

Slegers

D. W.

Card

N. A.

(2006). A non-arbitrary method of identifying and scaling latent variables in sEM and MACS models. Structural Equation Modeling: A Multidisciplinary Journal, 13(1), 59–72. https://doi.org/10.1207/s15328007sem1301_3

52.

Lüdtke

Robitzsch

Trautwein

Kunter

(2009). Assessing the impact of learning environments: How to use student ratings of classroom or school characteristics in multilevel modeling. Contemporary Educational Psychology, 34(2), 120–131. https://doi.org/10.1016/j.cedpsych.2008.12.001

53.

Marsh

H. W

(1986). Verbal and math self-concepts: An internal/external frame of reference model. American Educational Research Journal, 23(1), 129–149. https://doi.org/10.3102/00028312023001129

54.

Marsh

H. W

(1990). The structure of academic self-concept: The Marsh/Shavelson model. Journal of Educational Psychology, 82(4), 623–636. https://doi.org/10.1037/0022-0663.82.4.623

55.

Marsh

H. W

Lüdtke

Nagengast

Trautwein

Morin

A. J. S.

Abduljabbar

A. S.

Köller

(2012). Classroom climate and contextual effects: Conceptual and methodological issues in the evaluation of group-level effects. Educational Psychologist, 47(2), 106–124. https://doi.org/10.1080/00461520.2012.670488

56.

Marsh

H. W

O'Mara

(2008). Reciprocal effects between academic self-concept, self-esteem, achievement, and attainment over seven adolescent years: Unidimensional and multidimensional perspectives of self-concept. Personality & Social Psychology Bulletin, 34(4), 542–552. https://doi.org/10.1177/0146167207312313

57.

Marsh

H. W

Shavelson

(1985). Self-concept: Its multifaceted, hierarchical structure. Educational Psychologist, 20(3), 107–123. https://doi.org/10.1207/s15326985ep2003_1

58.

Marsh

H. W

Yeung

A. S.

(1998). Top-down, bottom-up, and horizontal models: The direction of causality in multidimensional, hierarchical self-concept models. Journal of Personality and Social Psychology, 75(2), 509–527. https://doi.org/10.1037/0022-3514.75.2.509

59.

Meeus

(2018). Adolescent development. Routledge. https://doi.org/10.4324/9780429465338

60.

Möller

Retelsdorf

Köller

Marsh

H. W.

(2011). The reciprocal internal/external frame of reference model. American Educational Research Journal, 48(6), 1315–1346. https://doi.org/10.3102/0002831211419649

61.

Moos

R. H.

(1987). Evaluating educational environments (2. print). A joint publication in the Jossey-Bass higher education & social and behavioral science series. Jossey-Bass Publ.

62.

Morin

A. J. S.

Marsh

H. W

Nagengast

Scalas

L. F.

(2014). Doubly latent multilevel analyses of classroom climate: An illustration. The Journal of Experimental Education, 82(2), 143–167. https://doi.org/10.1080/00220973.2013.769412

63.

Muthén

L. K.

Muthén

B. O.

(2017). Mplus user’s guide. Eighth edition. Muthén & Muthén.

64.

Myers

S. S.

Pianta

R. C.

(2008). Developmental commentary: Individual and contextual influences on student-teacher relationships and children's early problem behaviors. Journal of Clinical Child and Adolescent Psychology, 37(3), 600–608. https://doi.org/10.1080/15374410802148160

65.

Neyer

F. J.

Asendorpf

J. B.

(2001). Personality–relationship transaction in young adulthood. Journal of Personality and Social Psychology, 81(6), 1190–1204. https://doi.org/10.1037/0022-3514.81.6.1190

66.

Niepel

Brunner

Preckel

(2014). The longitudinal interplay of students’ academic self-concepts and achievements within and across domains: Replicating and extending the reciprocal internal/external frame of reference model. Journal of Educational Psychology, 106(4), 1170–1191. https://doi.org/10.1037/a0036307

67.

O'Connor

B. P.

Vallerand

R. J.

(1998). Psychological adjustment variables as predictors of mortality among nursing home residents. Psychology and Aging, 13(3), 368–374. https://doi.org/10.1037/0882-7974.13.3.368

68.

O'Mara

A. J.

Marsh

H. W

Craven

R. G.

Debus

R. L.

(2006). Do self-concept interventions make a difference? A synergistic blend of construct validation and meta-analysis. Educational Psychologist, 41(3), 181–206. https://doi.org/10.1207/s15326985ep4103_4

69.

Orth

Erol

R. Y.

Luciano

E. C.

(2018). Development of self-esteem from age 4 to 94 years: A meta-analysis of longitudinal studies. Psychological Bulletin, 144(10), 1045–1080. https://doi.org/10.1037/bul0000161

70.

Ortner

T. M.

van Vijver

F. J. R. de.

(2015). Behavior-based assessment in psychology: Going beyond self-report in the personality, affective, motivation, and social domains. Psychological assessment--science and practice (Vol. 1). Hogrefe.

71.

Peiffer

Ellwart

Preckel

(2020). Ability self-concept and self-efficacy in higher education: An empirical differentiation based on their factorial structure. Plos One, 15(7), e0234604. https://doi.org/10.1371/journal.pone.0234604

72.

Pekrun

(1985). Schulklima. In Twelmann

(Ed.), Handbuch Schule und Unterricht. Schwann.

73.

Podsakoff

P. M.

MacKenzie

S. B.

Lee

J.-Y.

Podsakoff

N. P.

(2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. The Journal of Applied Psychology, 88(5), 879–903. https://doi.org/10.1037/0021-9010.88.5.879

74.

Podsakoff

P. M.

MacKenzie

S. B.

Podsakoff

N. P.

(2012). Sources of method bias in social science research and recommendations on how to control it. Annual Review of Psychology, 63, 539–569. https://doi.org/10.1146/annurev-psych-120710-100452

75.

Podsakoff

P. M.

Organ

D. W.

(1986). Self-reports in organizational research: Problems and prospects. Journal of Management, 12(4), 531–544. https://doi.org/10.1177/014920638601200408

76.

Preckel

Niepel

Schneider

Brunner

(2013). Self-concept in adolescence: A longitudinal study on reciprocal effects of self-perceptions in academic and social domains. Journal of Adolescence, 36(6), 1165–1175. https://doi.org/10.1016/j.adolescence.2013.09.001

77.

Reyes

M. R.

Brackett

M. A.

Rivers

S. E.

White

Salovey

(2012). Classroom emotional climate, student engagement, and academic achievement. Journal of Educational Psychology, 104(3), 700–712. https://doi.org/10.1037/a0027268

78.

Roberts

B. W.

DelVecchio

W. F.

(2000). The rank-order consistency of personality traits from childhood to old age: A quantitative review of longitudinal studies. Psychological Bulletin, 126(1), 3–25. https://doi.org/10.1037/0033-2909.126.1.3

79.

Rosenberg

(1965). Society and the adolescent self-image. Princeton university press.

80.

Rosenberg

Schooler

Schoenbach

Rosenberg

(1995). Global self-esteem and specific self-esteem: Different concepts, different outcomes. American Sociological Review, 60(1), 141. https://doi.org/10.2307/2096350

81.

Rubie-Davies

Asil

Teo

(2016). Assessing measurement invariance of the student personal perception of classroom climate across different ethnic groups. Journal of Psychoeducational Assessment, 34(5), 442–460. https://doi.org/10.1177/0734282915612689

82.

Scherer

(2013). Further evidence on the structural relationship between academic self-concept and self-efficacy: On the effects of domain specificity. Learning and Individual Differences, 28, 9–19. https://doi.org/10.1016/j.lindif.2013.09.008

83.

Scherrer

Preckel

(2019). Development of motivational variables and self-esteem during the school career: A meta-analysis of longitudinal studies. Review of Educational Research, 89(2), 211–258. https://doi.org/10.3102/0034654318819127

84.

Scherrer

Preckel

(2021). Circadian preference and academic achievement in school-aged students: A systematic review and a longitudinal investigation of reciprocal relations. Chronobiology International, 38(8), 1195–1214. https://doi.org/10.1080/07420528.2021.1921788

85.

Seaton

Marsh

H. W

Craven

R. G.

(2010). Big-fish-little-pond effect. American Educational Research Journal, 47(2), 390–433. https://doi.org/10.3102/0002831209350493

86.

Shavelson

R. J.

Hubner

J. J.

Stanton

G. C.

(1976). Self-concept: Validation of construct interpretations. Review of Educational Research, 46(3), 407. https://doi.org/10.2307/1170010

87.

Shewark

E. A.

Zinsser

K. M.

Denham

S. A.

(2018). Teachers’ perspectives on the consequences of managing classroom climate. Child & Youth Care Forum, 47(6), 787–802. https://doi.org/10.1007/s10566-018-9461-2

88.

Sowislo

J. F.

Orth

(2013). Does low self-esteem predict depression and anxiety? A meta-analysis of longitudinal studies. Psychological Bulletin, 139(1), 213–240. https://doi.org/10.1037/a0028931

89.

Spilt

J. L.

van Lier

P. A. C.

Leflot

Onghena

Colpin

(2014). Children’s social self-concept and internalizing problems: The influence of peers and teachers. Child Development, 85(3), 1248–1256. https://doi.org/10.1111/cdev.12181

90.

Tang

Wang

M.-T.

Parada

Salmela-Aro

(2021). Putting the goal back into grit: Academic goal commitment, grit, and academic achievement. Journal of Youth and Adolescence, 50(3), 470–484. https://doi.org/10.1007/s10964-020-01348-1

91.

Toren

N. K.

Seginer

(2015). Classroom climate, parental educational involvement, and student school functioning in early adolescence: A longitudinal study. Social Psychology of Education, 18(4), 811–827. https://doi.org/10.1007/s11218-015-9316-8

92.

Trautwein

Möller

(2016). Self-concept: Determinants and consequences of academic self-concept in school contexts. In Lipnevich

A. A.

Preckel

Roberts

R. D.

(Eds.), The Springer Series on Human Exceptionality. Psychosocial skills and school systems in the 21st century (pp. 187–214). Springer International Publishing. https://doi.org/10.1007/978-3-319-28606-8_8

93.

Trzesniewski

K. H.

Donnellan

M. B.

Robins

R. W.

(2003). Stability of self-esteem across the life span. Journal of Personality and Social Psychology, 84(1), 205–220. https://doi.org/10.1037/0022-3514.84.1.205

94.

van Geert

Steenbeek

(2005). Explaining after by before: Basic aspects of a dynamic systems approach to the study of development. Developmental Review, 25(3-4), 408–442. https://doi.org/10.1016/j.dr.2005.10.003

95.

Vingilis

Wade

T. J.

Adlaf

(1998). What factors predict student self-rated physical health? Journal of Adolescence, 21(1), 83–97. https://doi.org/10.1006/jado.1997.0131

96.

Walberg

H. J.

Anderson

G. J.

(1968). Classroom climate and individual learning. Journal of Educational Psychology, 59, 414–419. https://doi.org/10.1037/h0026490

97.

Wang

M.-T.

Degol

Amemiya

Parr

Guo

(2020). Classroom climate and children’s academic and psychological wellbeing: A systematic review and meta-analysis. Developmental Review, 57, 100912. https://doi.org/10.1016/j.dr.2020.100912

98.

Weidinger

A. F.

Spinath

Steinmayr

(2020). The value of valuing math: Longitudinal links between students’ intrinsic, attainment, and utility values and grades in math. Motivation Science, 6(4), 413–422. https://doi.org/10.1037/mot0000179

99.

Wentzel

K. R.

(2002). Are effective teachers like good parents? Teaching styles and student adjustment in early adolescence. Child Development, 73(1), 287–301. https://doi.org/10.1111/1467-8624.00406

100.

Wigfield

Eccles

J. S.

(2002). The development of competence beliefs, expectancies for success, and achievement values from childhood through adolescence. In Development of achievement motivation (pp. 91–120). Elsevier. https://doi.org/10.1016/b978-012750053-9/50006-1

101.

Wigfield

Wagner

A. L.

(2005). Competence, motivation, and identity development during adolescence. In Elliot

A. J.

Dweck

C. S.

(Eds.), Handbook of competence and motivation (222–239). The Guilford Press.

102.

Wong

A. E.

Vallacher

R. R.

Nowak

(2016). Intrinsic dynamics of state self-esteem: The role of self-concept clarity. Personality and Individual Differences, 100, 167–172. https://doi.org/10.1016/j.paid.2016.05.024