- 1School Pedagogy, Institute of Educational Science, University of Greifswald, Greifswald, Germany
- 2Institute of Educational Science, University of Potsdam, Potsdam, Germany
The social context plays a decisive role in the formation of the academic self-concept (ASC) and has been widely studied as the big-fish-little-pond-effect (BFLPE). This effect describes that comparable talented students in high-achieving school settings have a lower ASC compared to equally talented students attending low-achieving settings. Past research has focused on students’ domain-specific ASC, while little is known about the relation of achievement-related classroom compositions and the various facets of ASC. Additionally, BFLPE-research has been critiqued to build its theoretical frame on social comparison theory, without providing sufficient empirical support. To address this gap, we analyzed how the single student’s social, criterial, absolute, and individual ASC relate to class-level achievement of 8th graders. Applying Multilevel Structural Equation Modeling (MLSEM) we found that all facets of ASC were significantly related to average-class achievement, while student’s social ASC revealed the strongest associated. The results reveal explicitly that average-class achievement is strongly related to social comparison processes.
Introduction
The academic self-concept (ASC) has been established as one of the key constructs that determines academic success (Valentine et al., 2004) and shares vital structural characteristics with academic interest (Gogol et al., 2016). Following a rather broad definition, ASC is formed through academic-related experience and interpretations, combining attitudes, beliefs, and perceptions a student has about his or her own abilities and skills within the school context (Marsh and Hattie, 1996; Lent et al., 1997). Stressing the cognitive part of ASC, it is also defined as the entity of cognitive representations of one’s own abilities in academic achievement situations (Schöne et al., 2002).
Current research has widely shown that the domain-specific ASC of high-achieving students (e.g., in subjects such as math, languages, biology, and physics) is impacted by the composition of the class (Köller and Baumert, 2001; Dumont et al., 2017; Stäbler et al., 2017) as well as by the composition of the school (Harker and Tymms, 2004; Köller et al., 2006; for an overview see Marsh et al., 2008). However, little is known about the interrelations of achievement-related classroom compositions and the various facets of ASC, which are differentiated into social, criterial, absolute, and individual ASCs (Schöne et al., 2002). The present study addresses this gap by examining how these four facets of ASC are related to students’ overall grades at the class level.
Academic Self-Concept and the Big-Fish-Little-Pond-Effect
As the ASC reflects a student’s beliefs about his or her scholastic abilities, it is closely and positively related to academic achievement, which was confirmed by numerous studies and described in previous reviews (West et al., 1980; Hansford and Hattie, 1982; Hattie, 1992; Zimmerman, 1995; Valentine et al., 2004). This solid finding is based on the rationale that individuals strive to maintain consistent self-views (Swann, 1997), according to self-affirmation and self-regulation theory (Steele, 1988; Bandura, 1989). Hence, students with high academic self-beliefs are more likely to show performance behavior or adjust their behavior accordingly (self-regulation) to increase their chances for academic success, while subsequent academic success confirms their high self-beliefs (self-affirmation). In this sense, the reciprocal effects model (Marsh, 1990a; Marsh and Martin, 2011) postulates that ASC and achievement mutually reinforce each other, which has been validated in numerous studies (Valentine et al., 2004; Arens et al., 2016).
Besides the impact of individual academic achievement on ASC, there is consent among researchers that social comparison impacts students’ ASC decisively (Markus and Wurf, 1987; Marsh and Craven, 2002; Zeinz and Köller, 2006). How social comparison mechanisms relate to ASC in academic settings is described by the big-fish-little-pond-effect (BFLPE), according to which comparable talented students who attend high-ability schools have a lower ASC compared to equally talented students who attend comprehensive schools (Marsh, 1990a; Seaton et al., 2009). Köller et al. (2006) explain this effect as follows: A student (little fish) who attends a high-ability class or school (big pond) has many opportunities for upward-comparisons, which in turn are related to a low self-concept. However, if a student (big fish) would attend a rather low-ability class or school (small pond), the opportunity for downward-comparisons increase, leading to an increase of the student’s self-concept. In fact, research has shown that the ASC of high-tracked students suffers a substantial decline, compared to students from lower-tracked schools (Becker et al., 2014; Arens and Watermann, 2015). This finding calls for further research of high-tracked students, as low self-concept impairs for example achievement, educational attainment, effort, and career aspirations (Valentine et al., 2004; Marsh and Martin, 2011; Nagengast and Marsh, 2012). Following this rationale, the current study focuses on students from high-tracked schools.
In the frame of social comparison, negative contrast as well as assimilation (or reflected glory) effects have been investigated as counterbalancing mechanisms in BFLP research. While school-average ability is related to a negative self-concept, as students compare themselves to brighter classmates with higher achievement (negative contrast effects), students of high-achieving schools feel pride in belonging to this type of school, which has a positive effect on their self-concept (assimilation effect or reflected glory effects). Research has shown that the BFLPE is the net effect of these counteracting mechanisms, while the negative contrast effect was shown to have a stronger impact, leading to an overall low self-concept of students (Marsh et al., 2000; Trautwein et al., 2009).
Since the beginning of BFLP research about 20 years ago, it has been speculated that the BFLPE would be a consequence of social comparison in high ability schools (Marsh, 1987; for review see Marsh et al., 2008). However, BFLP research has been critiqued for interpreting results in light of social comparison mechanisms, but lacking empirical proof for such underlying mechanisms (Dai and Rinn, 2008). As a response to this critique, Huguet et al. (2009) provide first empirical support of the role social comparison plays for the BFLPE: The study’s findings show that BFLPE is eliminated after controlling for students’ perceived relative standing in class and coexists with contrastive and assimilative effects (Huguet et al., 2009).
Different Facets of Self-Concept and the Big-Fish-Little-Pond Effect
Current research on the BFLPE focuses on domain-specific mechanisms and often uses the Academic Self-Description Questionnaire (Marsh, 1990b) to measure ASC in 12–15 school subjects (Rindermann and Heller, 2005). Hence, most studies have disregarded measuring ASC in a format that represents general cognitive skills and academic achievement of students, independent of specific school subjects and grades. Most studies have also disregarded that students’ ASCs can be differentiated into different dimensions.
Schöne et al. (2002) developed a scale to assess ASC (SESSKO) that differentiates four non-domain-specific facets of ASC, namely social ASC (individual performance compared to performance of peers), criterial ASC (individual performance evaluated according to an objective criterion), absolute ASC (individual performance unrelated to any internal or external frames of reference), and individual ASC (individual performance compared to past performance). SESSKO items are related to aspects of giftedness, intelligence, learning ability, and the management of requirements (Schöne et al., 2002). It might be assumed that, at an individual level, all facets of ASC are positively related to previous achievement (Valentine et al., 2004). At the classroom level, it can be assumed that social ASC is mostly affected by average class achievement, as the BFLPE stresses the importance of social comparison. In this sense, social comparison theory postulates that there “exist[s], in the human organism, a drive to evaluate his opinions and abilities” (Festinger, 1954, p. 117). This pervasive social phenomenon is particularly powerful when individuals compare themselves to those similar to them (Suls et al., 2002; Huguet et al., 2009). Following this, it might be assumed that high-achieving students compare themselves to same-aged high-achieving students from their class by asking the question, consciously or unconsciously, “Am I as good as I ought to be?” (Suls et al., 2002, p. 159).
In contrast, the absolute ASC is not related to any frame of reference, which is explicitly mentioned by the authors (Schöne et al., 2002, p. 45). Hence, it can be assumed that the absolute ASC is not significantly related to the achievement of the class. Similarly to the general ASC, the absolute ASC captures general beliefs of students’ skills and abilities in school. In this sense, it is somewhat speculative whether the BFLPE holds for absolute ASC, as previous studies have yielded inconsistent results when examining general ASC. Becker and Neumann (2016) as well as Marsh and Craven (1997) did find evidence for the BFLPE, relating general ASC to class- and school-average ability respectively, but Jackman et al. (2011) did not find the BFLPE for university students attending a high-ability course. In fact, Marsh et al. (2008) argue that BFLPEs are stronger when using domain-specific effects, as they are more proximal compared to non-domain effects, which is the case for absolute ASC.
The assumptions about the individual and criterial ASC are similarly speculative: The criterial ASC represents students’ abilities and performance compared to school standards, so the frame of reference is rather abstract, lacking proximity to the social comparison process which is the essence of the BFLPE (Marsh et al., 2008). We therefore expect a small effect of average class achievement on criterial ASC.
Concerning individual ASC, students compare their actual performance within the high-ability class to their own previous performance. In our study, students were assessed 1.5 years after their transition from elementary to secondary school. Consequently, it might be assumed that students already knew their teachers and classmates and thus, were familiar with the average class achievement. Marsh et al. (2001) have shown that about 1 year after the school transition, students had adjusted to the new learning context and the BFLPE was present. Drawing on this finding, we expect to find the BFLPE by relating average class achievement to individual ASC.
The Present Study
The present study investigates the BFLPE among students from high-ability-tracking secondary schools by relating class-level grades (averaged over Math, German, English as a Foreign Language, and Biology) to social, criterial, absolute, and individual ASC. In Germany students are tracked according to their achievement after 4th or 6th grade, depending on the federal state. Students with relatively high achievement are tracked into high-ability-tracking schools (Gymnasium), with the aim to obtain a diploma (Abitur) which will allow them to study at a University or College. All other students are tracked to secondary schools, in which they can obtain various kinds of certificates. Based on the reviewed empirical and theoretical rationale, we formulated and tested the following hypotheses:
(1) Overall achievement at the end of 7th grade are positively related to social, criterial, absolute, and individual ASC in 8th grade at the student level.
(2) The social ASC is strongly related to the average class achievement, supporting the assumption that social comparison mechanisms drive the BFLPE.
(3) The absolute ASC is weakly related to the average class achievement.
(4) It is speculative how criterial ASC is affected by average class achievement. However, based on the theoretical outline above, we expect a BFLPE to be present, relating average class achievement to criterial ASC.
(5) We expect the individual ASC to be impacted negatively by average class achievement.
Materials and Methods
Sample and Procedure
The analyses are based on a cross-sectional early adolescent sample (N = 779; Mage = 13.09; SD = 0.50) in the federal state of Brandenburg, Germany. The participating students are from high-track schools (57% girls) in 47 classes, which were collected in 2011 at the beginning of 8th grade. In this study, entire classes were assessed, while the average number of students examined was 16.57 students per class.
After obtaining the formal approval from the government’s Department of Education, Youth and Sport of the federal state of Brandenburg to conduct the study, parents and/or legal representatives were asked for written informed consent so that students would be able to participate in the study. Afterward, the students themselves were asked for their consent to participate in the study. It should be noted that therefore self-selected biases could occur. However, the response rate was relatively high, namely 85%. The data was collected via anonymous, written, class-based questionnaires. At least two trained research assistants were present during each session to introduce the paper-pencil questionnaire format and to clarify any questions related to the items or the use of a Likert scale. To minimize potential difficulties with self-report data, operationalization and data-handling was executed following Chan’s (2009) suggestion that only well-established scales, for which construct validity was supported, should be used.
Measures
Dependent Variables
In order to evaluate adolescents’ perceived ASC, we used SESSKO, developed by Schöne et al. (2002). SESSKO differentiates four facets of ASC measured on a five-point Likert scale: absolute ASC, individual ASC, criterial ASC, and social ASC. These scales were conceptualized for measurement of ASC in German speaking countries and validated on 3326 students (Grades 5–9): The absolute ASC showed the internal consistency of 0.83, and split-half reliability was 0.84, the individual ASC showed the internal consistency of 0.87, and split-half reliability was 0.88, the criterial ASC showed the internal consistency of 0.80, and split-half reliability was 0.81 and the social ASC showed the internal consistency of 0.88, and split-half reliability was 0.89.
(1) The social academic self-concept consists of six items and showed a satisfactory reliability in the present sample (αα = 0.92). This subscale focuses on the comparison of a student’s own abilities with that of other classmates, for example “I think I am more talented at school than my classmates” or “Learning new things is easier for me compared to my classmates.”
(2) The absolute academic self-concept consists of five items and showed a good reliability in the present sample (α = 0.88). This subscale is not related to any frame of reference model as it focuses on a rather general ASC including items such as “I am gifted at school” or “In school, many tasks are easy for me.”
(3) The criterial academic self-concept consists of five items and showed a good reliability in the present sample (α = 0.86). This subscale focuses on the ranking of an individual’s school achievement based on objective criteria, for example “When I look at what we have to do for school, I think I am talented” or “When I look at what we have to know in school, I think I am able to handle the tasks quite well.”
(4) The individual academic self-concept scale consists of six items and showed a good reliability in the present sample (α = 0.81). Questions focus on the comparison of a student’s own abilities in the past vs. the present and include items such as “Learning new things at school is easier for me than before” or “I am more intelligent than before.”
Independent Variables
To measure students’ achievement, students were asked to report their grades in four major subjects on their last report card (from the end of 7th grade): German, Math, English, and Biology. The German grading system ranges from 1 (very good) to 6 (insufficient). As grades were recoded, a high score indicates high achievement.
Statistical Analyses
The following analyses were estimated using Mplus 7.2 (Muthén and Muthén, 1998–2012). Due to the nested structure of the data (i.e., students being clustered in classrooms), we used multilevel structural equation models (MLSEM) (Bouvaird, 2007). This method allows analyzing hierarchic, structured data-sets – in our case, students (individual-level) nested in classes (context-level) (Richter and Naumann, 2002). To analyze the impact of the educational context, we considered the classroom level and not the school level, as classrooms present the immediate context in which students learn and are therefore better suited to investigate the BFLPE (van Ewijk and Sleegers, 2010). To test the hypotheses, four MLSEMs were built to examine the associations between end-of-year grades and the four facets of ASC. Separate models were assessed to avoid possible suppression effects (Paulhus et al., 2004) as well as vertical multicollinearity, which may occur when analyzing correlated subscales that measure the same underlying attribute (Kock and Lynn, 2012). In each MLSEM, the independent variables (end-of-year grades) are manifest, whereas the dependent variables (four facets of ASC) are latent. Hence, the classroom-level construct (grades) is based on a formative aggregation of the student-level construct (see Marsh et al., 2009), which means that the aggregation is based on a group average of individual-grade characteristics. This is a common technique in various disciplines, such as educational science, sociology, and psychology (Marsh et al., 2009). Moreover, based on their simulation study with formative constructs, Lüdtke et al. (2008) have shown that when the sampling ratio was low, the models based on the latent-aggregation approach were suitable for formative variables as there was considerable sampling error. We chose to apply a latent-measurement/manifest-aggregation approach. The L2 ASC is latent as it is based on multiple indicators, but manifest regarding its aggregation from L1 to L2. In their simulation study, Lüdtke et al. (2011) have shown that this approach is very promising as it corrects error in multilevel data when the data provide only limited information in terms of the L2 construct, i.e., low ICC.
As recommended in most cases, if group differences are of interest, all parameters are group-mean centered, whereby only the in-group variance of the predictor variable is included in the prediction (Kreft et al., 1995; Richter and Naumann, 2002; Enders and Tofighi, 2007). Hence, group-mean centering allows interpreting the intercept as the expected outcome for a specific student in a specific classroom, whose covariate values are equal to the values of the specific classroom mean.
In order to investigate how classroom composition affects students’ achievement, separating the between-group and the within-group components from the total variation is a common procedure (Cronbach and Webb, 1975; Paccagnella, 2006). As the aim of this study is to examine the BFLPE, which examines frame-of-reference effects, separating the between-group and the within-group components is necessary, and therefore group-mean centering is appropriate. Missing data were completely at random, as Little’s MCAR test confirmed [χ2(12) = 15.81; p > 0.05], and therefore handled – by default – with full information maximum likelihood in Mplus.
Results
Table 1 shows the means, standard deviations (SD), minimum, and maximum of the variables of interest.
Unconditional Models
Before we designed the final MLSEMs, we tested a serious of four separate unconditioned models for each facet of ASC, which included only the outcome variables. These models were built to provide information on the amount of variance at both levels (within and between), which are necessary to estimate the intraclass correlations (ICC). Whereas ICC1 estimates the total variance of individual students’ ratings, the ICC2 gives information about the reliability of classroom mean ratings (Lüdtke et al., 2009). Table 2 shows the results of the ICCs. The ICCs are relatively low, however, it has to be mentioned that there are no standard values for acceptable reliability using ICC (Koo and Li, 2016). A low ICC-value can not only reflect the low degree of rater- or measurement-agreement, but also relate to the lack of variability among the sampled subjects, the small number of cases, or the small amount of raters being asked (Portney and Watkins, 1993; Lee et al., 2012). In case data provide only limited information (e.g., low ICCs and small number of groups) it is adequate to use the latent-measurement/manifest-aggregation approach, as it partially corrects for biased group effects (Lüdtke et al., 2011).
TABLE 2. Number of items, reliability (Cronbach’s α), ICC1 and ICC2 for dependent variables and example item.
In sum, the unconditional models for social ASC [χ2(23) = 193.06, p < 0.001; RMSEA = 0.09; CFI = 0.94; SRMRwithin = 0.02, SRMRbetween = 0.14], absolute ASC [χ2(14) = 57.12, p < 0.001; RMSEA = 0.06; CFI = 0.97; SRMRwithin = 0.03, SRMRbetween = 0.08], criterial ASC [χ2(14) = 76.74, p < 0.001; RMSEA = 0.07; CFI = 0.97; SRMRwithin = 0.03, SRMRbetween = 0.07], and individual ASC [χ2(7) = 83.63, p < 0.001; RMSEA = 0.12; CFI = 0.95; SRMRwithin = 0.04, SRMRbetween = 0.08] showed satisfactory model fits.
After estimating the unconditional models, the four MLSEMs were computed. The model fit for the social ASC model [χ2(33) = 149.80, p < 0.001; RMSEA = 0.07; CFI = 0.96; SRMRwithin = 0.03, SRMRbetween = 0.16], absolute ASC model [χ2(22) = 69.51, p < 0.001; RMSEA = 0.05; CFI = 0.97; SRMRwithin = 0.03, SRMRbetween = 0.07], criterial ASC model [χ2(22) 70.93, p < 0.001; RMSEA = 0.05; CFI = 0.98; SRMRwithin = 0.02, SRMRbetween = 0.06], and individual ASC model [χ2(13) = 83.42, p < 0.001; RMSEA = 0.08; CFI = 0.95; SRMRwithin = 0.04, SRMRbetween = 0.10] each showed an acceptable fit to the empirical data.
Multilevel Structural Equation Models (MLSEMs)
Table 3 shows the associations between grades and the four facets of ASC.
TABLE 3. Results of the Multilevel Structural Equation Models with grades as independent variable and school self-concept components as dependent variables.
The Relationship Between Grades and Social Academic Self-Concept
Figure 1 shows the results of the relationship between grades and social ASC. This association is positive and highly significant on the within-level (β = 0.54, SE = 0.06, p < 0.001). Students with higher grades at the end of 7th grade reported higher social ASC at the beginning of 8th grade. This association is not significant on the between-level, which means that classes with higher mean grades do not show higher levels of social ASC. However, the context effect (β = -0.47, SE = 0.09, p < 0.001), as well as the centered context effect (β = -0.32, SE = 0.06, p < 0.001), are negative and highly significant (see Table 3, Model 4), supporting the BFLPE (Marsh and Craven, 2002). The results of the centered context effect indicate that, with every increase of grades at the end of 7th grade by one SD, the social ASC decreases by almost one SD.
FIGURE 1. The relationship between averaged grades in Math, German, English as a foreign language. Biology and social academic self-concept on the within and between level. Level 1, student level; Level 2, classroom level; regression coefficients were standardized by group-mean-centering; grades, averaged grades at the end of the school year 7 (Math, German, English as a foreign language, Biology); ASC_soc_w, social academic self-concept on LI (within); M_grades, aggregated mean-grades on L2; ASC_soc_b, social academic self-concept on L2 (between); bold pathways are significant at ∗p < 0.05, ∗∗p < 0.001; dotted pathways are not significant.
The Relationship Between Grades and Absolute Academic Self-Concept
Figure 2 shows the MLSEM examining the relationship between grades and absolute ASC. The association is positive and highly significant on the within-level (β = 0.45, SE = 0.05, p < 0.001), which means that students with better grades at the end of 7th grade reported a higher absolute ASC at the beginning of 8th grade. In addition, results show a positive and significant association on the between-level (β = 0.18, SE = 0.07, p < 0.05), which means that classrooms with higher average achievement levels also show higher mean levels for absolute ASC. Moreover, the context effect (see in Figure 3, Model 1) is negative and significant (β = -0.27, SE = 0.09, p < 0.05). When comparing two students with the same grades, but from different classes, the students that were part of classes with higher grades showed a lower absolute ASC, supporting the BFLPE (Marsh and Craven, 2002). Moreover, the centered context effect was negative and significant (β = -0.20, SE = 0.06, p < 0.05), which means that with every increase of grades (by 1 SD) the absolute ASC of students decreases by 0.20 SDs.
FIGURE 2. The relationship between averaged grades in Math, German, English as a foreign language. Biology and absolute academic self-concept on the within and between level. Level 1, student level; Level 2, classroom level; regression coefficients were standardized by group-mean-centering; grades, averaged grades at the end of the school year 7 (Math, German, English as a foreign language, Biology); ASC_abs_w, absolute academic self-concept on LI (within); M_grades, aggregated mean-grades on L2; ASC_abs_b, absolute academic self-concept on L2 (between); bold pathways are significant at ∗p < 0.05, ∗∗p < 0.001; dotted pathways are not significant.
FIGURE 3. The relationship between averaged grades in Math, German, English as a foreign language. Biology and criterial academic self-concept on the within and between level. Level 1, student level; Level 2, classroom level; regression coefficients were standardized by group-mean-centering; grades, averaged grades at the end of the school year 7 (Math, German, English as a foreign language, Biology): ASC_crit_w, criterial academic self-concept on LI (within); M_grades, aggregated mean-grades on L2; ASC_crit_b, criterial academic self-concept on L2 (between); bold pathways are significant at ∗p < 0.05, ∗∗p < 0.001; dotted pathways are not significant.
The Relationship Between Grades and Criterial Academic Self-Concept
Figure 3 shows the results of the relationship between grades at the end of 7th grade and criterial ASC at beginning of 8th grade. The association is positive and highly significant on the within-level (β = 0.51, SE = 0.05, p < 0.001), which means that students with higher grades also reported a higher criterial ASC. On the between level, the association is significant and positive (β = 0.25, SE = 0.09, p < 0.05), which means that classes with higher mean grades show higher mean levels for criterial ASC. Moreover, the context effect (β = -0.26, SE = 0.10, p < 0.05) and the centered context effect (β = -0.18, SE = 0.07, p < 0.05) (see Table 3, Model 3) are negative and significant. When comparing two students with the same grades, but from different classrooms, students from classes with a higher mean grade reported significantly lower levels of criterial ASC, supporting the BFLPE (Marsh and Craven, 2002).
The Relationship Between Grades and Individual Academic Self-Concept
Figure 4 shows the results of the MLSEM examining the relationship of grades and individual ASC. This association is positive and highly significant on the within-level (β = 0.30, SE = 0.08, p < 0.001), which means that students with better grades at the end of 7th grade reported a higher individual ASC at the beginning of 8th grade. This association is not significant on the between-level. This means that classes with higher mean grades do not significantly differ from classes with lower mean grades in their mean individual ASC. The context effect for this association (see Table 3, Model 2) is negative and significant (β = -0.25, SE = 0.12, p < 0.05). When comparing two students with the same grades, but from different classes, students from classes with a higher mean grade reported lower individual ASC, supporting the BFLPE (Marsh and Craven, 2002). This is also the case for the centered context effect (β = -0.13, SE = 0.06, p < 0.05) (see Table 3, Model 2).
FIGURE 4. The relationship between averaged grades in Math, German, English as a foreign language. Biology and individual academic self-concept on the within and between level. Level 1, student level; Level 2, classroom level; regression coefficients were standardized by group-mean-centering; grades, averaged grades at the end of the school year 7 (Math, German, English as a foreign language, Biology); ASC_ind_w, individual academic self-concept on LI (within); M_grades, aggregated mean-grades on L2; ASC_ind_b, individual academic self-concept on L2 (between); bold pathways are significant at ∗p < 0.05, ∗∗p < 0.001; dotted pathways are not significant.
Overall, the results show the BFLPE was present in our sample, as the context effects were significant for all facets of the ASC.
Discussion
The current study investigates the BFLPE among high-ability-tracked students in 8th grade, relating average class-level grades to a student’s social, absolute, criterial, and individual ASC measured by SESSKO (Schöne et al., 2002). We thereby expand knowledge on the BFLPE, as so far no study has investigated the BFLPE on the various facets of cognitive ASC independent of school domains. The focus of the present study on different facets of students’ ASC is motivated by the fact that SESSKO measures cognitive representations of individual skills by using three facets that refer to frames of reference (social, criterial, and individual ASC) as well as absolute ASC without any frame of reference.
The results indicate that overall grades (the average grades in Math, German, English as a Foreign Language, and Biology) at the end of the previous school year were positively related to all facets of ASC at the student level, indicating that students with high achievement tend to have a high social, absolute, criterial, and individual ASC. Consequently, the first hypothesis was confirmed. These results are in line with previous studies, which found that academic achievement is positively related to self-concept (Becker and Neumann, 2016; Stäbler et al., 2017; for a review see Valentine et al., 2004).
Furthermore, the social ASC was shown to be strongly related to average class achievement, confirming the second hypothesis. When comparing two students with equal achievement, but from different classrooms, students from classrooms with higher overall achievement reported significantly lower levels of social ASC. This effect is particularly striking, because with every improvement of the overall grades on the class level, a student’s social ASC decreases. This result not only confirms the BFLPE in homogeneous high-achievement settings, but also supports the idea that the BFLPE is driven by social comparison mechanisms based on social comparison theory (Festinger, 1954; Suls et al., 2002). Hence, this study adds to BFLP research investigating social mechanisms (e.g., Huguet et al., 2009) and provides a response to the criticism of Dai and Rinn (2008, p. 286) that there is a “bulk of social comparison research that does not provide direct evidence regarding the BFLPE per se.”
The results further show that the absolute ASC was significantly affected by average class achievement, supporting the BFLPE and the third hypothesis (Marsh et al., 2001, 2007, 2008; Trautwein et al., 2006a,b). However, compared to social ASC, the absolute ASC was not as strong associated with the overall class achievement. This is in line with past research arguing that ASC that does not relate to a domain-specific frame of reference is not affected as strongly by the BFLPE (Marsh and Craven, 1997; Becker and Neumann, 2016).
Similarly, criterial ASC was significantly related to overall class achievement, though not as strongly as social ASC, supporting the fourth hypothesis, which was rather speculative. This may be explained by the lack of proximity of criterial ASC, asking students to compare their achievement to school standards. Hence, this small effect (compared to social ASC) may be explained by the lack of proximity to frames of reference, while the proximal nature of comparison processes is one postulate of the BFLPE. Concerning the individual ASC, the results indicate a significant relationship between class level achievement and individual ASC, supporting hypothesis five. Hence, as expected and in line with Marsh et al. (2001), the students in grade eight have adjusted to the new classroom setting that took place about 1.5 years ago.
In sum, if we assume that ASC is associated with social comparison mechanisms, which is the theoretical underpinning of the BFLPE (Marsh, 1990a; Marsh et al., 1995; Marsh and Rowe, 1996; Marsh and Craven, 2002; Zeinz and Köller, 2006), criterial as well as individual ASC may implicitly ask for social comparison mechanisms. At this point, the question arises whether a student’s evaluation of their own ability within the school context may ever occur without relating to the performance of his or her classmates, as learning and growing up takes place primarily in a social school context. In this sense, according to Rogers (1947), self-concept is a social product, formed and developed through social, interpersonal relationships.
Strengths, Limitations, and Further Directions
This study is apparently unique as it is the first investigation of how class-average achievement is related to social, absolute, criterial, and individual ASC in the frame of BFLP research. The results reveal explicitly that class-average achievement are strongly related to social comparison processes. The findings complement research providing empirical support for social comparison mechanisms related to the BFLPE. Furthermore, in the current study, we examined whole classes. We were therefore able to tackle the BFLPE more precisely (van Ewijk and Sleegers, 2010) compared to studies that base their findings on only few students per class at the school level, using large, nationally representative samples.
Each study has its limitations, which in our case is the use of teacher-assigned grades and not standardized test results. As the current study was not part of a large, national representative study, no standardized tests were conducted. However, within our means we tried to handle grades in a way that they would represent students’ overall achievement by summing up end of year grades in different subjects assigned by different teachers, like Math, German, English as a Foreign Language, and Biology.
Furthermore, the cross-sectional nature of the data can be rated as a limitation as well as the sensitivity to self-selected biases due to the voluntariness to participate. However, as the current study aims to examine potential differences between the four facets of ASC, this design is adequate. Future longitudinal studies are warranted to confirm and expand these findings by identifying underlying mechanisms and processes. Additional research should also investigate how social-motivational mechanisms impact the BFLPE by investigating questions such as how the BFLPE is distinct among (a) students oriented toward mastery vs. performance goals (Butler, 1992), (b) students who belong to different socio-motivational types (Raufelder et al., 2013), or (c) students socialized in collectivistic and individualistic societies (McFarland and Buehler, 1995; Hoferichter et al., 2018).
Author Contributions
FH wrote the main part of the paper. DR and AL did the statistical analyses. RL was mainly involved in the conceptualization, statistical advisement and helped reviewing the manuscript.
Funding
This work was supported by a grant (Schumpeter Fellowship, II/84 452) from the Volkswagen Foundation.
Conflict of Interest Statement
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
References
Arens, A. K., Marsh, H. W., Pekrun, R., Lichtenfeld, S., Murayama, K., and vom Hofe, R. (2016). Math self-concept, grades, and achievement test scores: long-term reciprocal effects across five waves and three achievement tracks. Perspect. Psychol. Sci. 1, 133–163. doi: 10.1111/j.1745-6916.2006.00010.x
Arens, A. K., and Watermann, R. (2015). How an early transition to high-ability secondary schools affects students’ academic self-concept: contrast effects, assimilation effects, and differential stability. Learn. Individ. Differ. 37, 64–71. doi: 10.1016/j.lindif.2014.11.007
Bandura, A. (1989). Social Foundations of Thought and Action: A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall.
Becker, M., and Neumann, M. (2016). Context-related changes in academic self-concept development: on the long-term persistence of big-fish-little-pond effects. Learn. Instr. 45, 31–39. doi: 10.1016/j.learninstruc.2016.06.003
Becker, M., Neumann, M., Tetzner, J., Böse, S., Knoppick, H., Maaz, K., et al. (2014). Is early ability grouping good for high-achieving students’ psychosocial development? Effects of the transition into academically selective schools. J. Educ. Psychol. 106, 555–568. doi: 10.1037/a0035425
Bouvaird, J. A. (2007). “Multilevel structural equation models for contextual factors,” in Modeling Contextual Effects in Longitudinal Studies, eds T. D. Little, J. A. Bovaird, and N. A. Card (Mahwah, NJ: Erlbaum), 149–182.
Butler, R. (1992). What young people want to know when: effects of mastery and ability goals on interest in different kinds of social comparisons. J. Educ. Psychol. 82, 934–943. doi: 10.1037/0022-3514.62.6.934
Chan, D. (2009). “So why ask me? Are self-report data really that bad?” in Statistical and Methodological Myths and Urban Legends: Doctrine, Verity and Fable in Organizational and Social Sciences, eds C. E. Lance and R. J. Vandenberg (New York, NY: Routledge), 309–336.
Cronbach, L. J., and Webb, N. (1975). Between-class and within-class effects in a reported aptitude treatment interaction: reanalysis of a study by G. L. Anderson. J. Educ. Psychol. 67, 717–724. doi: 10.1037/0022-0663.67.6.717
Dai, D. Y., and Rinn, A. N. (2008). The big-fish-little-pond effect: What do we know and where do we go from here? Educ. Psychol. Rev. 20, 283–317. doi: 10.1007/s10648-008-9071-x
Dumont, H., Protsch, P., Jansen, M., and Becker, M. (2017). Fish swimming into the ocean: how tracking relates to students’ self-beliefs and school disengagement at the end of schooling. J. Educ. Psychol. 109, 855–870. doi: 10.1037/edu0000175
Enders, C. K., and Tofighi, D. (2007). Centering predictor variables in cross-sectional multilevel models: a new look at an old issue. Psychol. Methods 12, 121–l38. doi: 10.1037/1082-989X.12.2.121
Festinger, L. (1954). A theory of social comparison processes. Hum. Relat. 7, 117–140. doi: 10.1177/001872675400700202
Gogol, K., Brunner, M., Preckel, F., Goetz, T., and Martin, R. (2016). Developmental dynamics of general and school-subject-specific components of academic self-concept, academic interest, and academic anxiety. Front. Psychol. 7:356. doi: 10.3389/fpsyg.2016.00356
Hansford, B. C., and Hattie, J. A. (1982). The relationship between self and achievement/ performance measures. Rev. Educ. Res. 52, 123–142. doi: 10.3102/00346543052001123
Harker, R., and Tymms, P. (2004). The effects of student composition on school outcomes. Sch. Eff. Sch. Improv. 15, 177–199. doi: 10.1076/sesi.15.2.177.30432
Hoferichter, F., Bakadorova, O., Raufelder, D., and Francisco, M. B. (2018). A comparison of Russian, and Philippine secondary school students on their socio-motivational relationships in school – a motivation typology. Int. J. Sch. Educ. Psychol. doi: 10.1080/21683603.2018.1446373
Huguet, P., Dumas, F., Marsh, H., Régner, I., Wheeler, L., Suls, J., et al. (2009). Clarifying the role of social comparison in the big-fish-little-pond effect (BFLPE): an integrative study. J. Pers. Soc. Psychol. Bull. 97, 156–170. doi: 10.1037/a0015558
Jackman, K., Wilson, I. G., Seaton, M., and Craven, R. G. (2011). Big Fish in a Big Pond: a study of academic self-concept in first year medical students. BMC Med. Educ. 11:48. doi: 10.1186/1472-6920-11-48
Kock, N., and Lynn, G. S. (2012). Lateral collinearity and misleading results in variance-based SEM: an illustration and recommendations. J. Assoc. Inform. Syst. 13, 546–580. doi: 10.17705/1jais.00302
Köller, O., and Baumert, J. (2001). Leistungsgruppierungen in der Sekundarstufe I. Ihre Konsequenzen für die Mathematikleistung und das mathematische Selbstkonzept der Begabung [Ability grouping at secondary Level 1: consequences for mathematics achievement and the self-concept of mathematical ability]. Z. Pädagog. Psychol. 15, 99–110. doi: 10.1024//1010-0652.15.2.99
Köller, O., Trautwein, U., Lüdtke, O., and Baumert, J. (2006). Zum Zusammenspiel von schulischer Leistung, Selbstkonzept und Interesse in der gymnasialen Oberstufe [The interplay of academic achievement, self-concept and interest in the upper secondary school]. Z. Pädagog. Psychol. 20, 27–39. doi: 10.1024/1010-0652.20.12.27
Koo, T. K., and Li, M. Y. (2016). A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J. Chiropr. Med. 15, 155–163. doi: 10.1016/j.jcm.2016.02.012
Kreft, I. G., De Leeuw, J., and Aiken, L. S. (1995). The effect of different forms of centering in hierarchical linear models. Multivariate Behav. Res. 30, 1–21. doi: 10.1207/s15327906mbr3001_1
Lee, K. M., Lee, J., Chung, C. Y., Ahn, S., Sung, K. H., Kim, T. W., et al. (2012). Pitfalls and important issues in testing reliability using intraclass correlation coefficients in orthopaedic research. Clin. Orthop. Surg. 4, 149–155. doi: 10.4055/cios.2012.4.2.149
Lent, R. W., Brown, S. D., and Gore, P. A. Jr. (1997). Discriminant and predictive validity of academic self-concept, academic self-efficacy, and mathematics-specific self-efficacy. J. Couns. Psychol. 44, 307–315. doi: 10.1037/0022-0167.44.3.307
Lüdtke, O., Marsh, H. W., Robitzsch, A., and Trautwein, U. (2011). A 2 × 2 taxonomy of multilevel latent contextual models: accuracy–bias trade-offs in full and partial error correction models. Psychol. Methods 16, 444–467. doi: 10.1037/a0024376
Lüdtke, O., Marsh, H. W., Robitzsch, A., Trautwein, U., Asparouhov, T., and Muthén, B. O. (2008). The multilevel latent covariate model: a new, more reliable approach to group-level effects in contextual studies. Psychol. Methods 13, 203–229. doi: 10.1037/a0012869
Lüdtke, O., Robitzsch, A., Trautwein, U., and Kunter, M. (2009). Assessing the impact of learning environments: how to use student ratings of classroom or school characteristics in multilevel modeling. Contemp. Educ. Psychol. 34, 120–131. doi: 10.1016/j.cedpsych.2008.12.001
Markus, H., and Wurf, E. (1987). The dynamic self-concept: A social psychological perspective. Annu. Rev. Psychol. 38, 299–337. doi: 10.1146/annurev.ps.38.020187.001503
Marsh, H., and Craven, R. (1997). “Academic self-concept beyond the dustbowl,” in Handbook of Classroom Assessment: Learning, Achievement, and Adjustment, ed. G. D. Phye (San Diego, CA: Academic Press), 131–193.
Marsh, H., and Craven, R. (2002). “The pivotal role of frame of reference in academic self- concept formation: The big-fish-little-pond effect,” in Academic Motivation of Adolescents, eds F. Pajares and R. Urdan (Greenwich, CT: Information Age), 83–123.
Marsh, H., and Rowe, K. (1996). The effects of single-sex and mixed-sex mathematics classes within a coeducational school: a reanalysis and comment. Aust. J. Educ. 40, 147–162. doi: 10.1177/000494419604000203
Marsh, H., Trautwein, U., Lüdtke, O., and Koller, O. (2007). The big-fish-littlepond effect: persistent negative effects of selective high schools on self-concept after graduation. Am. Educ. Res. J. 44, 631–669. doi: 10.3102/0002831207306728
Marsh, H. W. (1987). The big-fish-little-pond effect on academic self-concept. J. Educ. Psychol. 79, 280–295. doi: 10.1037/0022-0663.79.3.280
Marsh, H. W. (1990a). Causal ordering of academic self-concept and academic achievement: a multi-wave, longitudinal panel analysis. J. Educ. Psychol. 82, 646–656. doi: 10.1037/0022-0663.82.4.646
Marsh, H. W. (1990b). Self-Description Questionnaire-II Manual. Campbelltown: University of Western Sydney.
Marsh, H. W., Chessor, D., Craven, R. G., and Roche, L. (1995). The effects of gifted and talented programs on academic self-concept: the big fish strikes again. Am. Educ. Res. J. 32, 285–319. doi: 10.3102/00028312032002285
Marsh, H. W., and Hattie, J. (1996). “Theoretical perspectives on the structure of self-concept,” in Handbook of Self-Concept, ed. B. A. Bracken (New York, NY: Wiley), 38–90.
Marsh, H. W., Köller, O., and Baumert, J. (2001). Reunification of East and West German school systems: longitudinal multilevel modeling study of the Big-Fish-Little-Pond Effect on academic self-concept. Am. Educ. Res. J. 38, 321–350. doi: 10.3102/00028312038002321
Marsh, H. W., Kong, C.-K., and Hau, K.-T. (2000). Longitudinal multilevel models of the big- fish–little-pond effect on academic self-concept: counterbalancing contrast and reflected-glory effects in Hong Kong schools. J. Pers. Soc. Psychol. 78, 337–349. doi: 10.1037/0022-3514.78.2.337
Marsh, H. W., Lüdtke, O., Robitzsch, A., Trautwein, U., Asparouhov, T., Muthén, B., et al. (2009). Doubly-latent models of school contextual effects: Integrating multilevel and structural equation approaches to control measurement and sampling error. Multivariate Behav. Res. 44, 764–802. doi: 10.1080/00273170903333665
Marsh, H. W., and Martin, A. J. (2011). Academic self-concept and academic achievement: relations and causal ordering. Br. J. Educ. Psychol. 81, 59–77. doi: 10.1348/000709910X503501
Marsh, H. W., Seaton, M., Trautwein, U., Lüdtke, O., Hau, K. T., O’Mara, A. J., et al. (2008). The big-fish-little-pond effect stands up to critical scrutiny: implications for theory, methodology, and future research. Educ. Psychol. Rev. 20, 319–350. doi: 10.1007/s10648-008-9075-6
McFarland, C., and Buehler, R. (1995). Collective self-esteem as a moderator of the frog-pond effect in reactions to performance feedback. J. Pers. Soc. Psychol. 68, 1055–1070. doi: 10.1037/0022-3514.68.6.1055
Muthén, L. K., and Muthén, B. O. (1998–2012). Mplus User’s Guide, 7th Edn. Los Angeles, CA: Muthén & Muthén.
Nagengast, B., and Marsh, H. W. (2012). Big fish in little pond aspire more: mediation and cross-cultural generalizability of school-average ability effects on self-concept and career aspirations in science. J. Educ. Psychol. 104, 1033–1053. doi: 10.1037/a0027697
Paccagnella, O. (2006). Centering or not centering in multilevel models? The role of the group mean and the assessment of group effects. Eval. Rev. 30, 66–85. doi: 10.1177/0193841X05275649
Paulhus, D. L., Robins, R. W., Trzesniewski, K. H., and Tracy, J. J. (2004). Two replicable suppressor situations in personality research. Multivariate Behav. Res. 39, 303–328. doi: 10.1207/s15327906mbr3902_7
Portney, L. G., and Watkins, M. P. (1993). Foundations of Clinical Research: Application TO Practice. Stamford, CT: Appleton & Lange.
Raufelder, D., Jagenow, D., Drury, K., and Hoferichter, F. (2013). Social relationships and motivation in secondary school: Four different motivation types. Learn. Individ. Differ. 24, 89–95. doi: 10.1016/j.lindif.2012.12.002
Richter, T., and Naumann, J. (2002). Mehrebenenanalysen mit hierarchisch-linearen Modellen [Multilevel-analysis with hierarchic-linear models]. Z. Medienpsychol. 14, 155–159. doi: 10.1026//1617-6383.14.4.155
Rindermann, H., and Heller, K. A. (2005). The benefit of gifted classes and talent schools for developing students’ competences and enhancing academic self-concept. Z. Pädagog. Psychol. 19, 133–136. doi: 10.1024/1010-0652.19.3.133
Rogers, C. R. (1947). Some observations on the organization of personality. Am. Psychol. 2, 358–368. doi: 10.1037/h0060883
Schöne, C., Dickhaeuser, O., Spinath, B., and Stiensmeier-Pelster, J. (2002). Skalen zur Erfassung des Schulischen Selbstkonzepts [Scales to Assess School Self-Concept]. Goettingen: Hogrefe.
Seaton, M., Marsh, H. W., and Craven, R. G. (2009). Earning its place as a pan-human theory: universality of the big-fish-little-pond effect across 41 culturally diverse countries. J. Educ. Psychol. 101, 403–419. doi: 10.1037/a0013838
Stäbler, F., Dumont, H., Becker, M., and Baumert, J. (2017). What happens to the fish’s achievement in a little pond? A simultaneous analysis of class-average achievement effects on achievement and academic self-concept. J. Educ. Psychol. 2, 191–207. doi: 10.1037/edu0000135
Steele, C. M. (1988). “The psychology of self-affirmation: sustaining the integrity of the self,” in Advances in Experimental Social Psychology, Vol. 21, ed. L. Berkowitz (New York, NY: Academic Press), 261–302.
Suls, J., Martin, R., and Wheeler, L. (2002). Social comparison: Why, with whom, and with what effect? Curr. Dir. Psychol. Sci. 11, 159–163. doi: 10.1111/1467-8721.00191
Swann, W. B. Jr. (1997). The Trouble with change: self-verification and allegiance to the self. Psychol. Sci. 8, 177–180. doi: 10.1111/j.1467-9280.1997.tb00407.x
Trautwein, U., Lüdtke, O., Köller, O., and Baumert, J. (2006a). Self-esteem, academic self- concept, and achievement: How the learning environment moderates the dynamics of self-concept. J. Pers. Soc. Psychol. 90, 334–349.
Trautwein, U., Lüdtke, O., Marsh, H. W., Köller, O., and Baumert, J. (2006b). Tracking, grading, and student motivation: using group composition and status to predict self-concept and interest in ninth grade mathematics. J. Educ. Psychol. 98, 788–806. doi: 10.1037/0022-0663.98.4.788
Trautwein, U., Lüdtke, O., Marsh, H., and Nagy, G. (2009). Within-school social comparison: How students perceive the standing of their class predicts academic self-concept. J. Educ. Psychol. 101, 853–866. doi: 10.1037/a0016306
Valentine, J. C., DuBois, D. L., and Cooper, H. (2004). The relation between self-beliefs and academic achievement: a meta-analytic review. Educ. Psychol. 39, 111–133. doi: 10.1207/s15326985ep3902_3
van Ewijk, R., and Sleegers, P. J. C. (2010). Peer ethnicity and achievement: a meta-analysis into the compositional effect. Sch. Effect. Sch. Improv. 21, 237–265. doi: 10.1080/09243451003612671
West, C. K., Fish, J. A., and Stevens, R. J. (1980). General self-concept, self-concept of academic ability and school achievement: implications for “causes” of self-concept. Aust. J. Educ. 24, 194–213. doi: 10.1177/000494418002400207
Zeinz, H., and Köller, O. (2006). “Noten, soziale Vergleiche und Selbstkonzepte in der Grundschule,” in Risikofaktoren kindlicher Entwicklung: Migration, Leistungsangst und Schulübergang, ed. A. Schründer-Lenzen (Wiesbaden: VS), 177–190. doi: 10.1007/978-3-531-90075-9_9
Keywords: big-fish-little-pond-effect, social, criterial, absolute, individual academic self-concept (SESSKO), high-ability tracked students
Citation: Hoferichter F, Lätsch A, Lazarides R and Raufelder D (2018) The Big-Fish-Little-Pond Effect on the Four Facets of Academic Self-Concept. Front. Psychol. 9:1247. doi: 10.3389/fpsyg.2018.01247
Received: 24 March 2018; Accepted: 29 June 2018;
Published: 20 July 2018.
Edited by:
Michael S. Dempsey, Boston University, United StatesReviewed by:
Ronny Scherer, University of Oslo, NorwayStefan Sütterlin, Østfold University College, Norway
Copyright © 2018 Hoferichter, Lätsch, Lazarides and Raufelder. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Frances Hoferichter, frances.hoferichter@uni-greifswald.de