An Exploration of Factors Driving School Success in Diverse Students Through Meta-Analytic and Value-Added Modeling

In Luxembourg, the student population is diversifying regarding socioeconomic and language background (Klein & Peltier, 2022; STATEC, 2021). This surge in diversity gave rise to educational inequality in the trilingual educational system already in primary school (see Hadjar et al., 2018; Hoffmann et al., 2018; Sonnleitner et al., 2021). While increased educational inequalities would naturally deteriorate performance in international comparisons, Luxembourg remains stable in its test results, for example, in the Programme for International Student Assessment (PISA; Weis et al., 2018). These stable test results suggest several schools must use strategies to effectively address educational inequalities and support their students against the odds. Motivated by the narrative literature review in the Theoretical Considerations section, the present thesis explores factors driving school success such as teaching quality, teacher-student relationships (TSRs), or cognitive functions, in diverse students internationally and specifically in Luxembourg. The Grand Duchy can be seen as a living laboratory for the multilingual and diverse future of educational systems in a globalized world.
Thus, the present thesis aims to identify effective educational psychological practices to address educational inequalities in a diverse, multilingual student population in the host of potential variables, schools, and stakeholders’ perspectives. Two interrelated Research Strands follow this aim using open science practices. Research Strand 1 explores general social and cognitive learning processes in the international literature through meta-analytic methods in Studies 1 and 2. Study 1 delves into the link between TSRs and student outcomes. Study 2 aims to deepen our understanding of executive functions (EFs) and their relation to mathematics skills. Research Strand 2 focuses on concrete educational effectiveness measures and educational psychological practices in Luxembourg to help all students succeed. As such, our objective was to identify primary schools with stable performance against the odds (Study 3) and compare educational psychological practices in such schools to successfully address educational inequalities (Study 4).
Specifically, Study 1 (Research Strand 1) presents a preregistered systematic review of meta-analyses plus original second-order meta-analyses (SOMAs). We synthesized over 70 years of research on TSRs by aggregating 24 meta-analyses encompassing a total of 116 effect sizes based on more than 2 million prekindergarten and K-12 students. Several three-level SOMAs indicated that TSRs had similarly strong significant relations with eight clusters of outcomes: academic achievement, academic emotions, appropriate student behavior, behavior problems, EFs and self-control, motivation, school belonging and engagement, and student well-being. Age, gender, and informant (i.e., student, peer, or teacher report) were the most frequently examined moderators in prior research, and our original moderator analyses suggested student grade level and social minority status as moderators. We further identified meaningful differences in quality between the meta-analyses, and these differences were not associated with the TSR-outcome links. Mapping the field of TSR research, Study 1 indicates how TSRs could contribute to improving student outcomes via relationship building.
Study 2 (Research Strand 1) utilized data extracted from 47 preschool studies (363 effect sizes, 30,481 participants) from 2000 to 2021. We found that, overall, EFs are significantly related to math intelligence ("r" ̅ = .34, 95% CI [.31, .37]). This link holds for all three EF subdimensions, that is, inhibition ("r" ̅ = .30, 95% CI [.25, .35]), shifting ("r" ̅ = .32, 95% CI [.25, .38]), and updating ("r" ̅ = .36, 95% CI [.31, .40]). Key measurement characteristics of EFs (e.g., task type), but neither children’s age nor gender, moderated this relation. These findings indicate a positive link between EFs and math intelligence in preschool children and emphasize the importance of measurement characteristics. Further, we examined the joint relations between EFs and math intelligence with meta-analytic structural equation modeling. Evaluating different models and representations of EFs, we found no support for the expectation that the three EF subdimensions are differentially related to math intelligence.
Study 3 (Research Strand 2) examined the stability of value-added (VA) scores over time for mathematics and language learning, as VA models are widely used for accountability purposes. We drew on representative, large-scale, and longitudinal data from two cohorts of standardized achievement tests in Luxembourg (N = 7,016 students in 151 schools). We found that only 34-38% of the schools showed stable VA scores over time, with moderate rank correlations of VA scores from 2017 to 2019 of r = .34 for mathematics and r = .37 for language learning. Although they showed insufficient stability over time for high-stakes decision-making, school VA scores could be employed to identify teaching or school practices that are genuinely effective—especially in heterogeneous student populations.
Study 4 (Research Strand 2) compared what schools with high VA scores do more successfully than schools with medium or low VA scores. Based on the results of Study 3, we selected 16 schools with stable high, medium, or low VA scores as promising target schools for this comparison. The assessed variables included, for example, instructional quality (Klieme et al., 2001), school climate (Wang & Degol, 2016), TSR, boredom, and collective teacher self-efficacy (Hattie, 2008; Waack, 2018). Additionally, we measured previously identified Luxembourg specificities, such as language use and the perceived role of the school president. In a multi-perspective, mixed-methods data collection in 49 classrooms, we conducted observations and collected questionnaire data on 511 students in Grade 2, their 410 parents, 191 classroom and subject teachers, 14 school presidents, and 13 regional directors. Our sample, roughly 10 % of all primary schools and Grade 2 students, is somewhat representative of the general student population and showed similar means in instructional quality, TSR, and school climate as in other European countries. While schools with a stable high VA score did not differ from the other two groups on most variables, we found some evidence that teachers’ acknowledgment and inclusion of the student's home language could be one of the drivers of school differences and thus help reduce educational inequities.
In conclusion, the four studies identified several factors driving school success in diverse student populations using meta-analytic and VA approaches. Teachers should strengthen positive and avoid negative TSR because they correlate with crucial student variables in students of any age. EFs are already related to math skills in preschoolers but must be assessed and practiced separately, as they are distinct. Especially in multilingual classrooms, teachers should use their students' home languages to support them following the lesson, activate them cognitively, and create a welcoming learning climate. Limitations of the current thesis's meta-analytic and VA approaches indicate the need for future research to advance the field and find further the factors driving school success in diverse students.