How reading competence and age of arrival shape educational outcomes

• In Event: Beyond Mere Performance: Educational Equity and Effectiveness in International Large-Scale Assessment Research

Abstract

Objectives and Purpose: According to the 2022 results of the PISA study, non-immigrant students tend to outperform immigrant students in most countries, while Germany is not only one of the countries with the largest proportion of immigrant students, surpassing migration-driven countries such as the USA, but also exhibits one of the largest differences in performance between students with and without migration background. Moreover, German immigrant students are characterized by significantly lower skills in mathematics and reading than those in virtually all other major destination countries (OECD, 2023 & 2024a). Even after adjusting for socio-economic status, a substantial disparity remains. This raises the question of how this discrepancy of competence between immigrant and non-immigrant students can be explained and understood.
Theoretical Framework: Recent migration flows have changed the composition of student populations, with growing numbers of first-generation immigrants facing unique linguistic and educational challenges shaped by their age of arrival. While prior research has highlighted structural inequality and institutional stratification as major drivers of educational outcomes (Mang et al., 2022), there remains a need for more differentiated models that take into account individual-level biographical and linguistic factors (Teltemann & Jude, 2019). Drawing on the findings of second language acquisition research (Gogolin et al., 2020; Bordag & Opitz 2021) and test modalities for the participation of immigrant students in PISA, we responded to this gap and investigated the predictive value of age of arrival and externally rated reading comprehension, compared to traditional predictors such as home language use and socio-economic status.
Data: In our study, we examined predictors of academic success among first-generation immigrant students drawing on PISA data from Germany conducted in 2012, 2015, 2018 and 2022 (N = 1,189). Outcome variables were domain-specific achievement scores in mathematics, reading and science. Predictor variables included age of arrival (continuous; centered years of age at immigration), reading competence (teacher-rated), language use at home and indicators of the student’s socio-economic status.
Methods: The analyses proceeded in three stages. First, we computed descriptive statistics for key variables including age of arrival and domain-specific competence scores. Second, we applied unsupervised machine learning (random forest clustering) to detect non-linear groupings of arrival ages. Third, we estimated multilevel linear regression models with latent outcomes to assess the unique effects of covariates.
Results: We identified (1) age of arrival and (2) external rating of reading competence as the strongest predictors of academic performance surpassing socio-economic factors or well-tried variables such as language use at home, aligning with findings of the state of research in second language acquisition. Language skills alone sufficiently explained the lower performance of first-generation immigrant students. Additionally, our results indicated that current standardized assessment frameworks may confound fairness and comparability.
Significance of the Study: Focusing on the student level, our findings underscore the need for language-sensitive assessments and a reconsideration of reading competence as the primary determinant of academic success in international large-scale assessments. Enhanced measurement approaches are required to ensure educational effectiveness and equitable outcomes for multilingual learners.

Authors

• Lisa Teufele, Technical University of Munich

• Clievins Selva, University of Hildesheim