The Big-Fish-Little-Pond Effect for Reading Self-Beliefs: A Cross-National Exploration with PISA 2018

Figures & data

Table 1. Pearson’s correlations for all student-level study variables.

		1	2	3	4	5	6	7	8	9	10
1	Reading Competence
2	Reading Difficulty	−.319
3	PISA Test Difficulty	−.310	.493
4	Student-Level Reading Achievement	.259	−.264	−.370
5	Age	.013	−.004	−.015	.033
6	Gender	−.078	.027	−.006	−.138	.001
7	SES	.186	−.153	−.257	.387	−.007	.011
8	Modal Grade	.084	−.064	−.102	.242	.255	−.057	.210
9	Grade Repetition	−.102	.087	.114	−.254	.008	.061	−.183	−.447
10	Immigrant Status	.030	.007	.007	.014	.011	.008	.056	.009	.059
11	Language Spoken at Home	−.025	.062	.080	−.121	.000	.013	−.036	.024	.060	.303

Correlations in italics are not statistically significant (i.e., p > .05). Gender: 1 = females and 2 = males; grade repetition: 0 = no, 1 = yes; immigrant status: 1 = natives, 2 = second-generation immigrants, 3 = first-generation immigrants; language spoken at home: 0 = language of testing and 1 = not language of testing.

Table 2. Results from survey-weighted multilevel models predicting reading self-concept subscales and PISA test difficulty rating.

	M1: SC Competence		M2: SC Difficulty		M3: PISA Test Difficulty
	β	SE	β	SE	β	SE
Fixed Effects
Intercept	−.028	.031	−.034	.024	.002	.031
Student-Lvl Reading Ach	.348*	.010	−.317*	.008	−.340*	.010
School-Avg Reading Ach	−.198*	.011	.119*	.010	.028*	.011
Covariates
Age	.006*	.001	.000	.002	−.004*	.001
Gender	−.041*	.001	−.008*	.001	−.044*	.001
SES	.130*	.002	−.073*	.002	−.110*	.002
Modal Grade	−.016*	.002	.013*	.002	.001	.002
Repeated Year	−.021*	.002	.010*	.002	.011*	.002
Immigration Status	.020*	.002	−.005*	.002	.004*	.002
Language Spoken at Home	−.018*	.002	.016*	.002	.025*	.002
Random Effects	SD		SD		SD
School-Lvl Intercept	.148		.123		.130
Country-Lvl Intercept	.266		.206		.204
Student-Lvl Reading Ach| Country	.084		.066		.069
School-Avg Reading Ach | Country	.087		.071		.067
Residual	.901		.931		.888

Parameter estimates (β) for fixed effects that differ from zero by more than two standard errors (SE) are statistically significant (p < .05), and are indicated by *. Random effects are presented in standard deviations (SD).

Figure 1. Country-by-country variation in beta estimates (with 95% confidence intervals) for school-average achievement predicting reading self-beliefs.

NOTE: Beta estimates are taken from 2-level models run separately in each country, for each dependent variable, controlling for student-level reading achievement and covariates. Black triangles indicates significant estimates (95% CI line does not cross zero), while gray circles indicates non-significant resultsestimates (95% CI line crosses 0).