Introduction

Social experiences, positive social relationships, and social learning are essential for children and adolescents (Schonert-Reichl, 2017; Schunk & DiBenedetto, 2020; Van Lissa et al., 2017; Wolgast & Barnes-Holmes, 2018; Wolgast et al., 2022). Among other factors, social learning (Schonert-Reichl, 2017; Schunk & DiBenedetto, 2020; Voith et al., 2020) of school children is based on their perception of peer relationships (Furrer & Skinner, 2003; Gallardo & Barrasa, 2018; Gallardo et al., 2016).

There is however scant evidence on interrelations between phenomena in academic domain vs social domain in children. For example, evidence lacks whether and to what extent longitudinal effects exist between children’s school-related achievement goal orientations and their perception of peer relationships. Accounting for such effects would be of practical importance in peer learning during class and collaborative learning outside of class.

Students who focus on outstanding grades might feel distracted by peer interactions or even threatened through social comparisons with them and perceive their peer relationships in class accordingly rather negatively. In contrast, students who want to learn something might be also interested in learning something about their peers and perceive their peer relationships in class accordingly positively.

Prominent conceptualizations of achievement goal orientations are academic goal orientations including school-related mastery, performance-approach vs performance-avoidance goal orientations (Dweck, 1986; Dweck & Leggett, 1988). Students’ school-related goal orientations (assessed using the established inventory SELLMO, Dickhäuser et al., 2002), decreased in a longitudinal study (incl. grades 5–9, Fischer & Theis, 2014) and in a trend study over ten years in Germany (incl. grades 4–10 (Spinath et al., 2016). Few researchers examined, whether academic goal orientations (assessed using different measures) determine adolescents’ perceived social interactions (Gehlbach, 2006; Gong, 2003; Levy-Tossman et al., 2007). Positive perception of peer relationships in school emerge from positively experienced social interactions with classmates and the associated social-cognitive processes in those involved in the interactions (Furrer & Skinner, 2003). For example, Levy-Tossman et al. (2007) described small effects of school-related mastery and performance-approach goal orientations on seventh-grade (no age provided) students’ perceived social interactions except for the performance-avoidance goal orientation. The small effects regarded a significant positive effect of mastery goal orientation on students’ perceived social interactions and a significantly negative effect of performance-approach goal orientation on the students’ perceived social interactions (Levy-Tossman et al., 2007). Similar differential effects of school-related goal orientations in children on their later perception of peer relationships might thus exist. Accordingly, we initially asked whether and to what extend differential effects of children’s school-related mastery vs performance-approach goal orientation (described henceforth as ‘performance goal orientation’) on their later perception of peer relationships as criterion exist.

There is, however, evidence for the impact of children’s social interactions with peers on their achievement goal orientations as criterion each (Anderman, 2020; Anderman & Midgley, 1997; Anderman & Anderman, 1999), or academic achievement as criterion (Gallardo et al., 2016). Social comparisons (Anderman, 2020; Schunk & DiBenedetto, 2020; Steinmayr et al., 2019; Wigfield & Koenka, 2020). Cooperative learning activities in the class (Nichols, 1996; Nichols & Miller, 1994) often explain the impact on goal orientations or academic achievement. This evidence suggests effects of children’s perceived peer relationships (i.e., predictor) on school-related mastery vs performance goal orientation (i.e., criterion). It regards the converse direction to our initial research question mentioned above. We accordingly extended our research question to whether reciprocal effects exist between children’s mastery vs performance goal orientation and their perception of peer relationships. We expect negative reciprocal relations between students’ performance goal orientation and their perception of peer relationships in class. The theoretical foundation for these expectations is outlined next.

School-related goal orientations – theoretical focus

Goal orientations are conceptualized as motivational traits that are relatively stable over time within goal theory (Locke & Latham, 2002; Locke et al., 1981) and achievement goal theory (Dweck, 1986; Dweck & Leggett, 1988), which are classified as social-cognitive theories of achievement motivation (Urdan & Kaplan, 2020). Other examples of social-cognitive motivation theories include expectancies-by-values (Eccles et al., 1984), causal attributions (Weiner, 1980), reference norm orientation (Rheinberg, 1977), and social cognitive processes (Bandura, 1989). Within goal orientations, researchers particularly widely investigated mastery and performance goal orientations (Urdan & Kaplan, 2020).

School-related mastery goal orientation refers to the intrinsic pursuit of cognitive stimulation, understanding, new learning content, and skill enhancement (Dickhäuser et al., 2002; Levy-Tossman et al., 2007; Spinath et al., 2016). Mastery goal-oriented students use understanding, problem solving, and intrapersonal criteria (e.g., the individual reference norm, Rheinberg, 1977) to assess their progress and success in completing school tasks. They predominantly apply the individual reference norm (Schöne et al., 2004) or criterial reference norm and attribute causes of difficulties and weaknesses predominantly to themselves (Dweck & Leggett, 1988). That is, they attribute the difficulties or weaknesses to themselves internally variable and are convinced of the changeability of their cognitive abilities. In particular understanding and problem solving might also help students in social interactions to understand their peers and, if needed, solve conflicts with them.

In contrast, school-related performance goal orientation involves the focus on reaching a goal, often extrinsically determined, in conjunction with social comparisons and the demonstration of high ability (Dickhäuser et al., 2002; Spinath et al., 2016); understanding learning contents and ability enhancement are subjectively of little relevance. Performance-goal-oriented students use social comparisons and demonstration of superior ability to evaluate their progress and success in completing school tasks (Levy-Tossman et al., 2007). Thus, they predominantly apply normative aspects such as outperforming others (Brophy, 2005) and often attribute difficulties and weaknesses externally to other children, adolescents, or adults. Performance-approach goal orientation is often accompanied by a higher competitive orientation than mastery goal orientation (Elliot & McGregor, 2001). In addition, high competitive orientation related to less positively perceived social relationships than low competitive orientation (Johnson & Johnson, 2009). Thus, performance-goal oriented students might perceive their peer relationships less positively than those who are mainly mastery goal oriented.

Elliot (1997) presented the subdimensions ‘approach learning goal orientation’ and ‘avoidance learning goal orientation’ vs ‘approach performance goal orientation’ and ‘avoidance performance goal orientation’. This model describes students who demonstrate own abilities and their proximity to goals on the ‘approach’ dimensions and hide subjectively perceived gaps in their abilities on the ‘avoidance’ dimensions. Elliot (1997) summarized these dimensions and subdimensions in a 2 × 2 model of performance goal theory. Elliot and McGregor (2001) already discussed problems with the avoidance learning goal orientation conceptualization because it seems counterintuitive that students hide gaps in own skills and concurrently strive for own skill enhancement (Elliot & McGregor, 2001). Accordingly, Pekrun et al. (2006) reported results from a model test without the avoidance learning goal orientation subdimension.

Further studies in different contexts suggest that children often cannot yet differentiate the subdimensions (i.e., approach and avoidance) of performance goal orientations because of their age. However, all children are already able to distinguish their learning goal orientation from their performance goal orientation in a questionnaire adapted to their age (Cumming et al., 2014; Smith et al., 2009). In addition, they are able to describe their perception of peer relationships in class (Furrer & Skinner, 2003). Moreover, the introduced previous research yielded only a positive effect of mastery goal orientation on students’ perceived social interactions (Levy-Tossman et al., 2007).

School-related goal orientations and students’ perceived peer-relations

School children already have extensive experiences from previous social interactions, especially from out-of-school situations with their caregivers. Theoretical explanations for more or less positive perception of social relationships in educational contexts include the development of social attachment (Furrer & Skinner, 2003), social cognitive processes (Bandura, 1989), and social comparisons (Festinger, 1954).

The present study is based on social cognitive approaches (Bandura, 1989; Dweck & Leggett, 1988; Schunk & DiBenedetto, 2020). These approaches consider both mastery/performance goal orientations in relation to tasks and social cognitive processes (e.g., perception of peer relationships). The approaches basically describe that processes in the person, environment, and observable behavior shape a dynamic system (Bandura, 1989; Schunk & DiBenedetto, 2020).

Within the framework of social cognitive theory (Bandura, 1989), self-efficacy is considered a particularly well-studied construct with regard to social perception and behavior (Bandura, 1989; Schunk & DiBenedetto, 2020). Social self-efficacy is the subjective expectation of successfully coping with demands or challenges in a social context.

Positive self-efficacy beliefs may aid positive thought patterns and anticipated success scenarios students mentalized. Such success scenarios guide for constructive and appropriate action in academic and social domains, although the domains are often full of impediments, failures, adversities, setbacks, frustrations, and inequities (Bandura, 1989). Researchers (Bandura, 1989; Furrer & Skinner, 2003; Schunk & DiBenedetto, 2020) suggested the consideration of self-efficacy when analyzing relationships between individual perception and action-related psychological constructs. Large evidence from academic and social domains supports their view (Huang, 2016; Huang et al., 2019; Unrau et al., 2018; Zell & Krizan, 2014).

Social self-efficacy and positively experienced social relationships with others are significant correlates of positive experiences in heterogeneous groups, cultural diversity, and inclusion in educational contexts (Abrams et al., 2017; Crisp & Turner, 2011). Motivation in this context refers to processes, including goal orientations, that initiate and sustain goal-directed activities and manifest in goal-directed actions.

There are few research findings on relations between achievement goal orientations and social-cognitive processes in students of grades 7–10 representing diverse ethnic groups with various socioeconomic and educational backgrounds (no age provided, Gehlbach, 2006; Levy-Tossman et al., 2007). These findings suggest effects from academic to social domain. Accordingly, mastery goal-oriented individuals primarily focused on their ability enhancement in both the academic domain and the social domain. Performance goal-oriented individuals particularly focused on their academic goal and positive self-presentation in the classroom. They further desired to belong to a "popular group", and to build a high social status (Levy et al., 2004). This desire is usually accompanied by social comparisons, in which performance goal-oriented students perceive themselves more positively than their peer relationships, preferring to conceal their own difficulties and weaknesses (Levy-Tossman et al., 2007).

Mastery goal-oriented individuals, on the other hand, indicated that their relationships with friends in the school class included trust, constructive social problem solving, and mutual sharing of difficulties and weaknesses (Levy-Tossman et al., 2007). Consequently, these two goal orientations in the academic domain may be associated with different consequences in the social domain.

Kindermann (1993) also showed that the nature of academic motivation can be systematically related to students' later social relationships. These students were in grades 4–5 (no age provided) in a suburban school in New York. For example, academic mastery goal-oriented school children were more likely to join one group and performance goal-oriented school children were more likely to join another group (Kindermann, 1993).

Further relevant characteristics of students

Gender repeatedly emerged as a significant predictor of both goal orientations in favor of girls (Elliot & McGregor, 2001) and social learning in favor of girls (Wolgast et al., 2018). Since social self-efficacy related to positive experiences in educational contexts (Abrams et al., 2017; Crisp & Turner, 2011), it might also be related to students ‘ perception of peer relationships.

In summary, it can be assumed that children’s mastery goal orientation determines later perception of peer relationships and vice versa, also when controlling for gender and self-efficacy. The corresponding theoretical model is depicted in Fig. 1.

Fig. 1
figure 1

Theoretical model for testing the hypothesis of reciprocal effects between students’ school-related goal orientations and their perception of peer relationships in class (adapted from Bandura, 1989; Dweck & Leggett, 1988; Schunk & DiBenedetto, 2020). Black arrows represent expected significantly positive relations, grey arrows represent expected negative relations

Full size image

Current research

The social cognitive theory framework (Bandura, 1989; Dweck & Leggett, 1988; Schunk & DiBenedetto, 2020) includes personal (e.g., goal orientations) goal-directed processes and actions in the academic and social domain. Especially in group learning, personal behavioral tendencies such as goal orientations and perception of peer relationships are closely related to each other. The extent to which mastery and performance goal orientations of school children co-determine their later perception of peer relationships has remained largely unexplained.

However, previous research let assume effects in both directions between academic and social constructs (e.g., Gehlbach, 2006; Wigfield & Koenka, 2020) suggesting reciprocal effects between school-related and social-cognitive constructs over time. Our research question was therefore: Are there significant reciprocal effects between students' school-related (mastery/performance) goal orientation and their perception of peer relationships over time?

Based on the social cognitive theoretical framework (Bandura, 1989; Dweck & Leggett, 1988; Schunk & DiBenedetto, 2020), we assumed that a more pronounced goal orientation in terms of school competence enhancement has stronger positive effects on later perception of peer relationships than performance goal orientation. Since performance goal orientation is close to competitive behavior, it is probably negatively related to the perception of peer relationships.

The perception of social relationships, especially prosocial relationships, with classmates is characterized by the relationship of the own person to the others or ´me and the classmates’ (Furrer & Skinner, 2003). Students’ perception of their peer relationships probably corresponds with their traits. Of importance to prosocial behavior is social self-efficacy as an individual trait (Bandura, 2001). We consequently included social self-efficacy as control variable in the current study. As introduced, gender predicted individual goal orientations in favor of girls (Elliot & McGregor, 2001) and social learning again in favor of girls (Wolgast et al., 2018). Gender is therefore also included as control variable in our study. Figure 1 provides an overview of the theoretically grounded and expected significant relations between students’ mastery goal orientation, performance goal orientation, and their perception of peer relationships at two time points (T1–T2).

Method

Results from an a priori power analysis for a structural equation model, an anticipated small effect size of Cohen’s d = 0.30 based on a moderate effect in previous research (r = 0.42, Furrer & Skinner, 2003, Tables 1, 2), a power of 80% (Lakens & Evers, 2014; Soper, 2020), the significance level α = 0.05, and six latent factors measured by 28 indicators yielded N = 89 participants as minimum sample size to identify the model structure and N = 161 participants as recommended sample size (Cohen, 1988; Soper, 2020; Westland, 2010). Furthermore, simulation studies for latent cross-lagged SEM suggested a power of 80% (Falkenström et al., 2020; Lakens & Evers, 2014) with a sample size of at least n = 50 individuals (Falkenström et al., 2020). Other simulation studies comparing latent cross-lagged and latent change score models and including samples of n = 200 yielded different estimates only for estimated slopes of latent change score models (Usami et al., 2016).

Table 1 Tests of differences in students’ school-related mastery goal orientation (MGO), school-related performance goal orientation (PGO), perception of peer relationships (SOR) in class (Keller-Schneider & Albisser, 2014) between subsamples
Full size table
Table 2 Measures used, sample items, and internal consistency of the constructs assessed in the RUMBA-S study
Full size table

Data are available from the student survey of the study RUMBA-S, a sub-study of the school development study RUMBA (Keller-Schneider & Albisser, 2013, 2015). RUMBA has not been preregistered. The study aimed to gain insights into the interaction of individual and collective resources in schools or in school classes as social system. Students of entire classes (Keller-Schneider, 2019), or teachers of entire schools, in the canton of Zurich participated in the questionnaire-based school development study with the first measurement time (T1) in November (Keller-Schneider & Albisser, 2013, 2015). One month later, in December, the teacher responsible for the participating school class received the mean with variance for that class (i.e., no individual student results) and the average mean with variance across participating schools (i.e., no individual school was mentioned or named) as a reflection of results for the corresponding construct (e.g., class climate from the student perspective).

Data source

The sub-study RUMBA-S provides data from 341 primary students. Of the 341 students n = 204 students (60% of N = 341) participated at T1 and T2, n = 135 students (40% of N = 341) at T2 and T3, and n = 64 students (19% of N = 341) at T1–T3. The multivariate analysis of variance yielded similar goal orientations and perceived peer-relationships in the class between the students who participated at T1 and T2, and the students who did not participate at T1 and T2 (see Table 1 for the detailed results).

Furthermore, the Χ2-Test indicated no differences in the variable gender between the students who participated at T1 and T2 and those who did not participate at T1 and T2 (see Table 1). Since we aimed to test a latent cross-lag SEM from T1 to a later time point, we included this largest longitudinal sample in the current analyses.

Accordingly, the longitudinal sample for the current analyses includes 204 primary school children with a mean age of 10.6 years (SD = 0.56) at T1 (54% female, 46% male). The students’ demographic background was controlled by the study design (e.g., families with middle to high socioeconomic status and vocational or higher education in the Canton of Zurich). All students attended one type of public school since there is just one type of public school and no private school, special school, or other types of school in Switzerland. Assignment to a school is based on the residential district in which the student lives.

The first measurement took place in November and the second measurement in November about 12 months later. The students were in the same peer group at both measurement times.

In Switzerland, the students do not change the school between 4 and 6th grade. The primary school students who participated in the study were in the 4th or 5th grade at T1 and in the 5th or 6th grade at T2. The students responded to the questionnaire presented in German within one lesson (Keller-Schneider, 2019).

The participation of the schools, teachers and students was voluntary, the teachers agreed via the school management, the parents were asked by the school management. The data collection was anonymized. There was no deception. All steps of the study followed international ethical standards (AERA et al., 2014). Responsible teachers and parents gave their consent for the participation of their child in the current study.

Materials

Since we follow Open Science approaches (Obels et al., 2020; Simmons et al., 2011), we provide a complete list of constructs captured in the current large scale assessment in Supplement 1. The measures used in the current study are established instruments published in German language. Table 2 summarizes these measures in English for the presentation purpose here, two sample items each, and internal consistency using McDonalds ω (McDonald, 1999; Revelle, 2019) in the RUMBA-S sample.

McDonald's coefficient ω is a measure to estimate the degree of measurement accuracy (AERA et al., 2014) and indicates the extent to which a latent variable (construct) reflects the common variance of all items (Dunn et al., 2014) in contrast to Cronbach’s α, that just measures the inner consistence of each scale. Accordingly, ω captures the construct as a whole. McDonald's ω (Dunn et al., 2014) can range from 0 to 1. The interpretation of the coefficient ω is equivalent to Cronbach's α coefficient (Schweizer, 2011). Thus, an ω > 0.60 allows the interpretation that there is an acceptable internal consistency of the items capturing a construct with respect to the data used for the statistical analyses. The measures presented in the next paragraphs were assessed using a 5-point rating scale (1 = not true at all to 5 = absolutely true).

Mastery goal orientation (1) was assessed with items whose wording was adapted to children aged approximately 10–12 years in Switzerland (adapted to the current study’s target group, SELLMO, Dickhäuser et al., 2002). The items assessing mastery goal orientation include aspired thinking or understanding of complex relationships (see Table 2). The assessed mastery goal orientation at T1 served as predictor variable and the assessed mastery goal orientation at T2 as criterion variable in the cross-lag structural equation model (SEM).

Performance goal orientation (2) was assessed with items whose wording was also adapted to children aged approximately 10–12 years in Switzerland (adapted to the current study’s target group, SELLMO, Dickhäuser et al., 2002). Performance goal orientation was included as predictor variable at T1 and as criterion variable at T2 in the SEM.

Perception of peer relationships (3) was assessed with the focus on ‘me and the others’ in students’ social relationships to their classmates (modified from SEMOS, Nakamura 2008). This individual evaluation represents internal representations of experiences that have arisen in previous interaction and communication contexts. The method is aimed at primary school students. The factor 'me and the others' was included as predictor variable at T1 and as criterion variable at T2 in the SEM.

Social self-efficacy (4) (Schwarzer & Jerusalem, 1999) was introduced using "How accurately do the following statements apply to you?" Four items assessed the subjective likelihood of dealing appropriately with others in social situations. The manifest variable (i.e., the mean of) social self-efficacy at T1 served as control variable in the SEM. The questionnaire included other instruments that are not relevant to the investigation of the present research question (see Keller-Schneider & Albisser, 2014, for an overview of RUMBA).

Data analyses

Testing the hypothesis first requires fitting the theoretical model including the six factors (see Supplement 1) to the data using confirmatory factor analysis (CFA). The CFA model is constructed with the following six latent factors (measured by the respective items each presented in Table 2): the mastery goal orientation at (1) T1 and (2) T2, performance goal orientation at (3) T1 and (4) T2, ‘me and the others' classmates in social relationships at (5) T1 and (6) T2. All variables included in the CFA model were z-standardized. The WLSMV estimator (weighted least square mean of variance adjusted estimation; Rosseel, 2010) was used as well as adjustments for complex data. The fit between the CFA model including the six factors and the data was tested. In addition to that, measurement invariance across gender and over time was tested in multigroup analyses (Oberski, 2015; Rosseel, 2010). According to recommendations from simulation studies (Rutkowski & Svetina, 2014), it was determined prior to analysis that the hypothesis of scalar measurement invariance for comparisons of means over time is supported at values ΔCFI < 0.020 and ΔRMSEA < 0.010.

Finally, the CFA model including the six latent factors allows an extension to the intended cross-lag SEM by adding the control variables (i.e., gender and self-efficacy) and cross-lag paths as depicted in Fig. 1. The SEM was tested using the R package lavaan (Rosseel, 2010). The SEM included the predictor variables and latent factors ‘mastery goal orientation’, ‘performance goal orientation’, and 'me and the others’ classmates in social relationships at T1 as well as the control variables gender and social self-efficacy (see Fig. 1 and Supplement 1) specified on the latent factors ‘mastery goal orientation’, ‘performance goal orientation’, and 'me and the others' classmates in social relationships at T2 that were included as criterion variables in the SEM. The covariates were considered random, and the means, variances and covariances were free parameters. The advantage of the latent measurement is that a latent factor can be measured by means of the covariance between observable indicators (if the indicators form one factor). Thereby, the measurement error can be determined.

Results

The CFA model suggested a good fit between its structure and the structure found in the data: χ2 (145) = 173.643, p = 0.053, Comparative Fit Index (CFI) = 0.973, Standardized Root Mean Square Residual (SRMR) = 0.081, Root Mean Square Error of Approximation (RMSEA) = 0.038; 95% CI [< 0.001, 0.057], see Supplement 1). The multi-group analysis and measurement invariance test yielded the scalar measurement invariance across gender (Delta Comparative Fit Index ΔCFI = 0.013; Delta Root Mean Square Error of Approximation ΔRMSEA = 0.001), and over time (ΔCFI < 0.014; ΔRMSEA < 0.009, see Supplements 4 and 5 for details).

The chi-square test result and fit indices of the SEM including cross-lags between the three latent factors indicated an acceptable fit between the assumed and found structure of the data: χ2 (207) = 287.177, p < 0.001, CFI = 0.917, SRMR = 0.091, RMSEA = 0.053; 95% CI [0.037, 0.067], see Table 3 and Supplement 2). A post hoc power analysis to detect the degree of misspecification of the SEM corresponding to the RMSEA (RMSEA = 0.053, df = 207, α = 0.05) using the R package semPower (Jobst et al., 2021; Moshagen, 2021; Moshagen & Erdfelder, 2016) gave the probability to falsify our SEM when it is actually wrong is 96% (see Supplement 5 for details). Results from the SEM are shown in Table 3.

Table 3 Results from the latent SEM: Standardized beta-coefficients, standard errors, z-values, significance levels, and 95% confidence intervals of the cross-lags between students’ mastery goal orientation, performance goal orientation and their perception of social relationships to their classmates
Full size table

Figure 2 presents an overview of the significant beta coefficients of the SEM with cross-lags. Significant effects existed of the latent factor ‘mastery goal orientation’ at T1 on the latent factor 'me and the others' at T2 (β = 0.413, p = 0.002). These effects persist despite competing high positive autoregressive effects of mastery goal orientation at T1 on itself at T2 and of performance goal orientation at T1 on itself at T2 (see Fig. 2 and Table 3).

Fig. 2
figure 2

Theoretical model including the standardized beta coefficients from the SEM with cross-lags between students’ school-related goal orientations and their perception of peer relationships (see Supplement 2 for a statistical model presentation). Black arrows represent significant positive relations, dashed arrows represent insignificant relations

Full size image

However, the latent factor ‘me and the others’ at T1 did not predict the mastery goal orientation at T2 (β = 0.238, p = 0.317). Furthermore, neither performance goal orientation at T1 significantly predicted 'me and the others' classmates in social relationships at T2 (β = 0.142, p = 0.222), nor vice versa: There was also no significant effect of 'me and the others' at T1 on students’ performance goal orientation at T2 (β = -0.241, p = 0.085).

There was no difference between girls and boys in their mastery goal orientation at T1 (β = -0.035, p = 0.731) or performance goal orientation at T1 (β = 0.175, p = 0.065, see Table 3). Those who indicated relatively low social self-efficacy, however, reported also relatively low positive perception of their peer relationships (β = 0.687, p < 0.001, see Table 3). The SEM explained 77.7% in the variance of the latent measured factor ‘mastery goal orientations’ at T2, 60.1% of the latent factor ‘performance goal orientation’ at T2, and 36.7% of the variance in the latent measured factor ‘me and others’ classmates in social relationships.

Discussion

The social cognitive theory framework (Bandura, 1989; Dweck & Leggett, 1988; Schunk & DiBenedetto, 2020) involves personal (e.g., goal orientations) goal-directed processes and actions which correspond to the academic domain and the social domain respectively. Personal behavioral tendencies such as goal orientations and perception of peer relationships are closely related to each other, for example, in peer learning in school contexts. The extent to which mastery and performance goal orientation of school children co-determine their later perception of peer relationships has remained largely unexplained (as worked out in the second part of this article).

The present longitudinal study over two measurement times examined the extent to which reciprocal effects exist between school children’s mastery vs performance goal orientation and their perception of peer relationships. Mastery goal orientation involves the pursuit of cognitive challenge and skill enhancement (e.g. through peer learning) that let us assume a positive effect on the school children's later perception of peer relationships, and vice versa, a positive effect of this perception on their mastery goal orientation (Bandura, 1971, 1989, 2001; Locke & Latham, 2002; Locke et al., 1981). The results from cross-lag SEM support this theoretically grounded hypothesis in part: Students' mastery goal orientation is significantly related to their later perception of social relationships but not vice versa. Thus, no significant reciprocal effects between mastery goal orientation and the perception of peer relationships existed in the included sample.

Perception of peer relationships did not significantly explain the students’ later mastery goal orientation. The presented theoretical model (see Fig. 1) provides an explanatory approach for these results. Performance goal orientation focuses on an external performance goal, with social relationships being less relevant to goal achievement. Performance goal orientation as a focus on a desired goal is more strongly associated with subsequent competitive behavior and social comparisons with others than with social interactions (Dweck & Leggett, 1988). Performance goal orientation may thus be associated only with less positively perceived peer relationships. In contrast, the characteristics of mastery goal orientation (reflecting, thinking, understanding complex contexts) may not only be applied to the completion of school tasks, but also significant for the positive perception of social relationships. Examples are understanding the complex behaviors of classmates or behave appropriately in and outside class. In this case, understanding the complex behaviors of peers would be a mediating variable between learning goal orientation and subsequent perception of peer relationships. This assumption could be tested in future studies.

Moreover, the results presented in Table 3, supplemental Table S2a and b suggest interindividual differences between students’ intraindividual changes in their ‘perception of peer relationships’ and social self-efficacy as a relevant third variable. Social self-efficacy seems to be a relevant third variable because including it in the model changed the respective coefficients of the autoregressive path ‘perception of peer relationships’ and the relationship between goal orientations and ‘perception of peer relationships’ from T1 to T2. However, these changes in the coefficients do not change the interpretations with regard to the current hypotheses. Longitudinal research including three measurement time points would allow to test the hypothesis that social self-efficacy mediates the relationship between students’ mastery goal orientation and their perception of peer relationships.

The present study results support previous research findings that also showed the empirical link between mastery goal orientation and social processes in school children, but rarely between performance goal orientation and social processes in school children (Levy-Tossman et al., 2007). Mastery goal orientation involves a tendency to think about, reason about, and understand complex relationships (Dickhäuser et al., 2002). Attempting to understand other people requires thinking about and empathizing with the complexities of their behaviors. Prosocial relationships with others are accompanied by a willingness to think and empathize with them (Wolgast & Barnes-Holmes, 2018; Wolgast et al., 2019).

Performance goal orientation, on the other hand, involves focusing on a goal combined with competitive behavior and demonstrating superior skills to others, so that fewer cognitive resources are available for empathizing with and maintaining peer relationships. The direct and indirect relationships between goal and competitive orientations, understanding complex (academic/social) contexts, subsequent perception of peer relationships, and prosocial behavioral tendencies could be explored in future studies.

It is noteworthy at this point that school-related performance goal orientation and its importance for belonging to a high social-status group in school has dominated school related mastery goal orientation in previous studies (Dalbert & Stöber, 2008). Urdan and Kaplan (2020) already discussed that changes in the goal structures of students in complex classroom processes are often difficult. Classroom processes are embedded in larger systems, the school, the living environment, and the social system. School children are encouraged to engage in social comparison and competition in classroom processes and outside of school (e.g., at home) in their leisure time (Urdan & Kaplan, 2020).

However, other studies showed how mastery goal orientation can be promoted in the classroom, for example, via autonomy support, positive error culture, and individual reference norms (Ames, 1992; Schöne et al., 2004; Theis et al., 2020). These studies suggest that fostering a positive group climate increases students’ mastery goal orientation. The results of the present study show that mastery goal orientation by school children is not only associated with positive perception of new or challenging learning content, but also outside the academic domain, with positive perception of peer relationships.

Limitations.

The sample for longitudinal analyses with cross-lag SEM is relatively small. The anticipated small effect size of Cohen’s d = 0.30 (based on r = 0.42, Furrer & Skinner, 2003, Table 2) might be optimistic according to a reviewer and Klein et al. (2018). The researchers (Klein et al., 2018) found in replications of 28 classic and comparative findings median comparable medium effect sizes (Cohen’s d) for the original findings and small effect sizes for the corresponding replications. For considering these research findings, the reviewer recommended using the pwrSEM package (Wang & Rhemtulla, 2021) for a priori power analyses in further research.

To avoid over-specification of the model (more parameters than cases), only a relatively small number of parameters could be included in the model. In this regard, latent modeling was preferred over including only manifest variables because the latent measured factors in SEM account for measurement error. We have only used self-report measures. Due to the difficult comprehensibility of the original measures for approximately ten-year-old children in Switzerland, an adaptation of the item formulations was necessary. The extent to which the linguistic adaptations affect the reliability and validity of the self-report measures can hardly be estimated. The adapted versions suggested acceptable to good internal consistency each. Peer relationships could be validated by external assessment or even by objective observers. It would also be interesting to relate the data to learning outcomes. Further research on the presented statistical relationships based on a larger longitudinal sample would strengthen the robustness of the results and allow for integrations of additional variables into the model, such as mastery and achievement emotions that significantly predicted learning in school in previous research (Pekrun et al., 2006, 2017; Sommet et al., 2021) and might be also relevant for students’ perception of social relationships, their social interactions and social learning (Voith et al., 2020).

Implications for research and practice

The results presented here might be tested in samples with adolescents and young adults, especially after an educational transition as relationships with other people are important in the transition to vocational training or university studies, and candidates who are motivated to learn and able to work in a team are sought, especially in the labor market (European Council, 2005). Social relationships that are perceived as less positive, supportive, and empowering may limit cooperation and constructive teamwork. A growing number of experience sampling studies suggest daily fluctuations in different forms of motivation (Bellhäuser et al., 2019; Ketonen et al., 2018; Liborius et al., 2019; Patall et al., 2018). These findings may be extended by a cross-lag hierarchical Bayesian continuous time dynamic model (Driver & Voelkle, 2018) including goal orientations and perceived peer-relationships. Moreover, the question arises whether this finding would change over longer periods of time and in non-school social environments. This line of research underlines the immediate and prospective effects of students’ mastery goal orientation on their academic (peer-)learning and perceived peer relationships in the class. The current findings provide helpful insights into unobservable associations between phenomena of the academic domain and social domain for teachers and students. Such associations might be highly important after transitions (e.g., from primary to secondary school, from school to vocational or higher education) when students or young adults interact and collaborate with people in new academic, vocational, and social environments.