The main purpose of the current proposal is to investigate the big-fish-little-pond effect (BFLPE) in student language and math self-concept (LSC/MSC). In particular, the difference between LSC and MSC was of interest. Academic self-beliefs are recognized to influence students’ daily life, behaviour, and achievement, as well as their motivation and well-being (Marsh et al., 2019). In educational research of academic self-beliefs, the two most commonly used constructs, academic self-concept and self-efficacy (ASC/ASE), have both been found important for students’ academic attainments (Bandura et al., 1996; Marsh & O’Mara, 2008). It has been demonstrated that students’ beliefs of their academic ability are highly affected by social comparison with school peers (Marsh & Hau, 2003). This phenomenon, referred to as the big-fish-little-pond effect (BFLPE), was first introduced by Marsh and Parker (1984) to describe the process of how school-average achievement could affect students’ academic self-concepts negatively, while individual achievement affected self-concept positively. The BFLPE thus predicts that students attending schools with higher average ability, believe their academic ability to be lower compared to equally abled students who attend schools with lower average ability (Trautwein et al., 2009).

The BFLPE research typically involves ASC to measure academic self-beliefs, and not ASE. One of the reasons for this is that ASC is a construct more prone to be influenced by social comparisons, compared to ASE. However, Marsh et al. (2019) discuss the notion of self-concept-like constructs, which include generalized self-efficacy items. In the present study, we draw on their notion by empirically investigating the function of language and math self-concept-like measures in a BFLPE model, using Swedish data.

When studying the BFLPE in reading, science, and math, Guo et al. (2018) found that the effect was stronger in math than in reading and science. That is, math self-concept was more negatively affected by school-average achievement. According to the internal/external frame-of-reference (I/E) model first discussed by Shavelson et al. (1976), students internally compare their ability in one subject with that in another subject. This leads to negative relations between different subject self-concepts, for example in reading and math. One of the reasons for the stronger BFLP effect in math than in reading and science could be that students perceived their abilities in math to be weaker than in other subjects. Reading or language self-concept (LSC) might not be as sensitive to social comparison, because it to some extent is based on knowledge found in a broader context, not only in school. When studying possible differences in BFLPE sizes, individual student cognitive ability becomes relevant to control for, since it might weigh up for such differences in social comparison sensitivity. It is possible that students depend more on their math ability to form more positive self-concepts, than in other subjects. Research on how student ability relates to the BFLPE have shown varied results. Some studies suggest that the BFLPE is stable across ability levels (Marsh & Hau, 2003; Marsh & O’Mara, 2008), while others suggest that high-ability students are not as affected by the negative BFLPE as students with lower ability (Trautwein et al., 2009).

Against this background, the overall purpose of the study was to investigate the BFLPE for Swedish grade 9 students (ages 15-16), as well as to see whether the BFLPE affected students with different levels of ability differently. More specifically, what effect does school-average achievement have on individual LSC and MSC? And does school-average achievement affect students’ LSC and MSC differently depending on their individual ability?

The study draws on data from the Swedish longitudinal project Evaluation through Follow-up (UGU; Härnqvist, 2000). The sampling was a two-step stratified procedure. First, municipalities were selected and after that, schools were selected. The samples comprise 10 % of the birth cohorts and are nationally representative of their respective population in Sweden. UGU comprises register, survey, and test data. Three age cohorts were used, with grade 9 students born 1992, 1998, and 2004 (N = approximately 10 000 students in each cohort). Information of cognitive ability, gender and parental education was included in the models to control for possible moderation effects. To measure student achievement in the BFLPE models, grade point average (GPA) was used. The GPA was based on eleven theoretical final grades from Grade 9. Additionally, student cognitive ability was operationalized with three tests measuring students’ verbal, spatial, and inductive abilities, all tested in grade 6. MSC was measured with one traditional self-concept item and two more general self-efficacy items, such as “How good do you think you are in math?” and “How well are you able to solve mathematical problems?” This formed a self-concept-like measure as discussed by (Marsh et al., 2019), but with “self-concept-like self-efficacy items.” LSC was constructed in a similar way, with items such as “How good do you think you are in Swedish” and “How well are you able to read and understand a text in Swedish.” The data were processed in IBM SPSS Statistics 27, and modelled in Mplus 8 (Muthén & Muthén, 1998-2017). Confirmatory factor analysis (CFA) was used to construct measurement models, and because three cohorts were used the models will be tested for measurement invariance. Multilevel structural equation models (MSEM) were estimated on student and school level. The BFLPE is a contextual effect (βc), specified as the difference between the group-level relation between GPA and self-concept (βb) and the individual-level relation (βw) (Lüdtke et al., 2008). The BFLPE (βc = βb  βw ) was thus estimated as the expected difference in self-concept between students with the same individual GPA, who attended schools differing by one unit in average GPA. Missing data were taken into account by using full information maximum likelihood (FIML) missing data modeling in Mplus (Muthén & Muthén, 1998-2017). Besides this, calibration weights estimated by Statistics Sweden were used in the analyses, to control for bias due to non-participation.

Results of preliminary analyses suggest that the self-concept-like measures in language and math, which included traditional self-concept items as well as more generalized self-efficacy items, seem to function in the BFLPE models. That is, the measures seem to capture the mechanisms behind self-beliefs in language and math, that to some degrees are formed when students compare themselves to their surrounding peers. The model fit of the measurement models were good on both levels. The initial BFLPE modelling indicate that contextual effects exist in the data, namely that individual LSC and MSC both were positively affected by individual GPA but negatively affected by school-average GPA. Additionally, there seem to be a difference in the BFLPE size between LSC and MSC. The effect of school-average achievement on LSC was -.34 (p < .001), while the effect of school-average achievement on MSC was -.40 (p < .001). However, as LSC and MSC were modelled separately, the difference will be formally tested with the Wald chi-square test to see if it is significant, by including LSC and MSC in the same model and run the Mplus command “Model Test”. Cognitive ability will be included as a cross-level moderator in the BFLPE models, to investigate if school-average GPA affects students’ LSC and MSC differently depending on their individual ability. It is possible that students with higher cognitive ability are not as negatively affected by the BFLPE compared to their relatively lower-achieving peers. It is also possible that cognitive ability mitigates the difference in BFLPE size between LSC and MSC. More analyses will be conducted, for example with gender and parental education as control variables. It is possible, for example, that there are significant gender differences in both language and math that could influence the size of the BFLPE.

Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (1996). Multifaceted impact of self-efficacy beliefs on academic functioning. Child Development, 67(3), 1206–1222. Chen, S., Hwang, F., Yeh, Y., & Lin, S. S. (2012). Cognitive ability, academic achievement and academic self‐concept: Extending the internal/external frame of reference model. British Journal of Educational Psychology, 82(2), 308–326. Guo, J., Marsh, H. W., Parker, P. D., & Dicke, T. (2018). Cross-cultural generalizability of social and dimensional comparison effects on reading, math, and science self-concepts for primary school students using the combined PIRLS and TIMSS data. Learning and Instruction, 58, 210–219. Härnqvist, K. (2000). Evaluation through follow-up. A longitudinal program for studying education and career development. In C.-G. Janson (Ed.), Seven Swedish longitudinal studies in behavioral science (pp. 76–114). Stockholm: Forskningsrådsnämnden. Lüdtke, O., Marsh, H. W., Robitzsch, A., Trautwein, U., Asparouhov, T., & Muthén, B. O. (2008). The multilevel latent covariate model: A new, more reliable approach to group-level effects in contextual studies. Psychological Methods, 13(3), 203–229. Marsh, H. W., & Hau, K. T. (2003). Big-fish-little-pond effect on academic self-concept. A cross-cultural (26-country) test of the negative effects of academically selective schools. American Psychologist, 58(5), 364–376. Marsh, H. W., & O’Mara, A. J. (2008). Reciprocal effects between academic self-concept, self-esteem, achievement, and attainment over seven adolescent years: Unidimensional and multidimensional perspectives of self-concept. Personality and Social Psychology Bulletin, 34, 542–552. Marsh, H. W., & Parker, J. W. (1984). Determinants of Student Self-Concept: Is It Better to Be a Relatively Large Fish in a Small Pond Even if You Don’t Learn to Swim as Well? Journal of Personality and Social Psychology, 41(1), 213–231. Marsh, H. W., Pekrun, R., Parker, P. D., Murayama, K., Guo, J., Dicke, T., & Arens, A. K. (2019). The murky distinction between self-concept and self-efficacy: Beware of lurking jingle-jangle fallacies. Journal of Educational Psychology, 111(2), 331–353. Muthén, L. K., & Muthén, B. O. (1998-2017). Mplus User’s Guide (8th ed.). Los Angeles, CA: Muthén & Muthén. Shavelson, R. J., Hubner, J. J., & Stanton, G. C. (1976). Self-concept: Validation of construct interpretations. Review of Educational Research, 46, 407–441. Trautwein, U., Lüdtke, O., Marsh, H. W., & Nagy, G. (2009). Within-school social comparison: How students perceive the standing of their class predicts academic self-concept. Journal of Educational Psychology, 101(4), 853–866.