Conference Program

This paper provides a preliminary overview of the impact of the pandemic on K-12 student
learning losses. The authors summarize some of the more salient trends documented globally in
the aftermath of the pandemic such as deteriorations in cognitive and non-cognitive student
outcomes such as socioemotional learning, mental and physical health. Examples of prominent
education policies and interventions that have been adopted to address these cognitive and non-
cognitive student outcomes will be highlighted. The authors describe the types of evidence that
currently exist, and the strengths and limitations of the evidence used to enact pandemic related
education policies. The introduction also explains the rationale for the selection of educational
jurisdictions that are represented in a recently published Routledge volume. The introduction
situates the importance of the volume within a broader global context, arguing that the pandemic
and resulting research provides an illustration of the educational impacts of unexpected,
devastating events that suddenly change the learning environment, and the differential impacts
on these sudden changes on vulnerable student populations.

Educational inequalities continue to represent a structural and persistent challenge for contemporary school systems. This issue is especially pronounced in upper secondary education in Italy, where marked disparities in student performance persist across institutions and educational tracks. In this context, learning outcomes are strongly shaped by family background characteristics, such as socioeconomic status, cultural capital, and parental educational attainment, as well as by school composition factors (Giancola & Salmieri, 2020).

In this analysis, educational inequalities are interpreted as the outcome of the interaction between individual and school-level factors: on the one hand, the mechanisms of reproduction of familial cultural capital through selective school processes (Bourdieu, 1971); on the other, differences in social and relational capital across school contexts (Coleman, 1988, 1990).

The aim of this study is to examine the evolution of these inequalities over time by using OECD-PISA test scores in Italian (reading literacy) and mathematics as proxy indicators of learning outcomes. The analysis draws on data from the 2018 and 2022 assessment cycles, corresponding respectively to the pre- and post-pandemic periods, in order to evaluate the effects associated with school closures and the economic crisis following COVID-19. This intertemporal approach makes it possible to move beyond a static perspective and to observe changes in the mechanisms through which educational inequalities are produced over time.

To analyze the effects of transformations over time, an intertemporal dataset was constructed by integrating the original database with time variables and interaction terms. Following a preliminary descriptive analysis, multilevel models and multiple linear regression models (four for reading and four for mathematics) were estimated in order to assess changes over time in the impact of family- and school-level factors (Jaccard & Turrisi, 2003; Giancola & Colarusso, 2020), through the progressive inclusion of variables. The baseline model included the main individual characteristics (gender and migrant background), as well as family and school characteristics, together with territorial controls. In subsequent models, interaction terms between family background, school composition, and time were introduced, first separately and then jointly, so as to estimate variations in their effects across the two waves.

The empirical findings reveal a significant increase in the influence of both family-level factors (e.g., socioeconomic status) and school-level factors (e.g., the sociodemographic composition of classes) on educational inequalities, alongside a widening gap between advantaged and disadvantaged students. The pandemic appears to have intensified these dynamics, generating differential effects for more vulnerable groups.

Overall, the evidence points to the need for targeted educational policies aimed at reducing structural inequalities within the Italian school system, promoting interventions grounded in a stronger equity-oriented perspective.

The effects of peer abilities in education are central to understanding how the
individual educational pathway is shaped by social environments. Although tradi-
tional theories emphasize positive spillovers from high-achieving peers, alternative
perspectives such as social contrast hypotheses highlight potential negative conse-
quences for student motivation and self-concept. In addition, much attention has
been devoted to the important role of parental socioeconomic background, but less
attention has been paid to its interaction with ability-peer effects. This study aims
to investigate how the effects of peer ability and the socioeconomic composition
of classmates influence the enrollment of students on the path to university when
they switch to lower secondary schools in Italy. The results show that the average effect of
peer ability is negative, but the socioeconomic composition of the classroom plays a
compensatory role. In addition to that it comes out that low achievers are more hit by the peer comparison.

Attainment gaps are a key indicator of educational and societal inequalities. The analysis and reporting of educational attainment are often based on cross-sectional data and take the form of simple breakdowns of students’ achievement by selected characteristics. These descriptive statistics, although relevant, are potentially misleading. They fail to account for the interplay between students’ characteristics and, crucially, for information that is only available when longitudinal data is used.

The aim of this presentation is to show the advantage of using longitudinal data for the systematic reporting of attainment gaps. From a methodological perspective, we will demonstrate that, despite their complexity, educational inequalities can be systematically measured and reported. More specifically, we will outline an analytical research strategy that makes use of micro-data gathered through schooling paths to account for the interplay between students’ characteristics and other factors affecting their academic performance, most notably attainment at prior stages of education. From a more substantive perspective, we will present evidence on attainment gaps across different groups of secondary school students in a way that facilitates the identification of over-time trends in educational inequalities.

We used the National Pupil Database, a rich administrative linked micro-dataset, maintained by the UK Department for Education. The dataset covers eight cohorts of students in England completing their lower- and upper- secondary education between 2018 and 2025. For all students, it features a broad set of characteristics at individual- and school-level recorded during their primary and secondary education. At student level, the dataset includes standardised measures of educational attainment, specifically at the end of primary, lower secondary and upper secondary education. It also includes socio-demographic characteristics (eg, gender, ethnicity, first language spoken at home, special educational needs), and socio-economic background. At school level, we used information on the type of schools attended, their geographical location, and a measure of deprivation. A multivariate analysis based on a regression approach was employed to measure the impact of each characteristic, once other factors were held fixed. To account for school-level effects, a multi-level regression approach was used. Each year’s data was analysed separately, using the longitudinal component of the data as a crucial aspect of modelling academic performance. To interpret findings, a set of criteria was considered to flag changes over time in attainment gaps that were considered ‘notable’.

We will present evidence on attainment gaps for students taking different education pathways in England. We will focus on the dynamics of educational inequalities between 2018 and 2025, drawing conclusions on the impact of the assessment arrangement put in place in 2020 and 2021 as a response to the pandemic and the return to a ‘new normal’. We will show the mitigating (or sometimes exacerbating) effects that the longitudinal information on prior attainment and the school-level characteristics can have on the estimation of attainment gaps. In this way, it will be possible to argue that collecting and making available rich longitudinal micro-data is key to producing evidence on the true dynamics of educational attainment.

School dropout is a complex and persistent phenomenon with significant and concerning repercussions at both the individual and social levels. At the individual level, it affects employment, health and well-being, while at the social level, it fosters greater economic and social inequalities and hinders the development of a more equitable and inclusive society. Given the potential consequences of school dropout for individuals and society, preventive intervention is not only a priority, but also an ethical necessity. However, although research shows that dropping out is not a sudden act, but rather the result of multiple factors, the data currently available at national and international levels is often inaccurate, fragmented and «ex post», i.e. collected months or even years later.

Consequently, longitudinal models that use Big Data to monitor students’ pathways over time are needed. This will allow us to identify situations of student hardship early on and intervene before they result in school dropout.

Based on these premises, the present study stems from a doctoral project carried out in collaboration with the municipality of X in Italy. The aim was to build a shared system for data collection and management among the various local educational agencies to improve monitoring of school dropout through longitudinal, continuous and updated data. In this context, the models used by five educational agencies in the X area were examined (the Ministry of Education, the Province, and the Vocational Education and Training department).

A comparative and documentary analysis of the surveyed data and systems highlights significant heterogeneity in the methods used to collect, manage and measure the school dropout rate. While the sources demonstrate some operational differences, they also exhibit structural issues that hinder an integrated interpretation of the phenomenon. In general, there is a lack of clear definition of the construct, a problem that has already been highlighted in the literature, which leads different agencies to measure the phenomenon using different and often non-overlapping variables, such as dropouts, withdrawals and transfers.

Similarly, most models use a data collection system at specific moments (e.g. the end of the school year or transition between cycles) which does not allow for timely intervention with students, but which, as previously mentioned, provides data referring to months or even years earlier.

Some differences instead concern the scope of coverage: these range from the Ministry of Education and Merit, which is responsible for schooling and ensures coverage across the entire territory, to three education and training systems (therefore relating to a more limited student population), up to the Province, which, despite having access to a large amount of data, responds more to planning purposes than evaluative ones and therefore does not deal with analyzing the phenomenon.

This fragmentation confirms what has emerged in national and international literature and makes the need for shared data systems among agencies operating in the same territory, including both education and training, even more urgent. From this perspective, data integration is essential for the effective and timely prevention of school dropout.

Taking advantage of the possibility offered by INVALSI data to link students’ performance over time, this study investigates the educational trajectories of students with a migrant background in the Italian school system, focusing on the evolution of achievement gaps. Specifically, it examines whether second-generation students’ academic outcomes progressively converge toward those of native students, as predicted by classic theories of generational progress, while first-generation students experience a more persistent disadvantage (Portes & Zhou, 1992; Alba & Nee, 1997).

Although evidence on generational progress is substantial, it largely refers to traditional immigration countries. Research on the Italian context has often been limited by the historically smaller presence of second-generation students (Azzolini et al., 2012) and by the lack of longitudinal data (Lovaglio et al., 2018).

This study exploits population-based standardized assessments of reading and numeracy administered by INVALSI. The analysis follows a cohort of students who attended Grade 5 in the 2018/19 school year through lower secondary school (Grade 8 in 2022/23) and into the second year of upper secondary education (2023/24).

The study addresses two main research questions. First, do second-generation students display greater convergence toward native peers than first-generation students over the course of their school trajectories? Second, to what extent are these differences explained by socio-economic background?

To answer these questions, the analysis reconstructs the evolution of achievement gaps between migrant generations and native students across school stages. The empirical analysis relies on OLS models controlling for survey wave to estimate the relative position of first- and second-generation students compared with native peers at different points in their school careers.

Overall, the study contributes to the understanding of migration-related educational inequalities by providing longitudinal evidence on how achievements gaps between migrant generations and native students evolves across different stages of school careers.

Introduction

The transition from lower- to upper-secondary education represents a decisive juncture in students’ educational trajectories, with long-term consequences for academic achievement, labour-market outcomes, and social mobility. In Italy, this transition occurs at the end of Grade 8, when students choose whether to enrol in an academic, technical, or vocational track. While a growing body of research has documented learning losses associated with the COVID-19 pandemic, less attention has been paid to its influence on educational decision-making at key branching points (Aalto et al., 2023; van de Werfhorst et al., 2023). Disruptions to in-person schooling, counselling activities, and teacher-family interactions, combined with economic uncertainty, may have altered how families evaluated the risks and returns of upper-secondary tracks.

Ethnic inequalities in Italian education are well documented (Azzolini & Barone, 2013). Even net of academic performance, students with a migration background are less likely than natives to enrol in academic tracks and more likely to enter technical or vocational programmes (Ferrara & Brunori, 2024). The pandemic may therefore have reinforced employment-oriented decisions among disadvantaged families, potentially widening pre-existing ethnic gaps. This study examines whether the COVID-19 pandemic influenced upper-secondary track choices and whether these changes differed between native children and children of immigrants.

Data and methods

The analysis uses population data from INVALSI-SNV, longitudinally linked to follow students from Grade 8 into Grade 10. Two cohorts are observed: a pre-Covid cohort in Grade 8 in 2016/17 and Grade 10 in 2018/19, and a Covid cohort in Grade 8 in 2020/21 and Grade 10 in 2022/23. The cohorts are pooled and analysed through a multilevel multinomial logistic model with school-level random intercepts, random slopes for the Covid cohort, and provincial fixed effects.

The outcome variable is upper-secondary track choice (academic, technical, vocational). The main independent variable is migration background, distinguishing native, mixed-origin, second-generation and first-generation students. An interaction term between migration background and cohort is added. Four models are estimated. Model 1 controls for gender, regularity of school progression, attendance of early childhood education. Model 2 additionally includes parental socioeconomic status. Models 3 and 4 further control for prior academic achievement, measured using students’ lower-secondary grades in Italian and mathematics in Grade 8 (Model 3) and standardised test scores in the same subjects (Model 4).

Preliminary findings

Results show persistent ethnic stratification in track choice. In Model 1, second- and first-generation students are more likely than natives to enrol in technical and vocational tracks rather than academic ones, with stronger shifts towards non-academic tracks in the Covid cohort. In Model 2, controlling for parental socioeconomic background reduces these differences, but they do not disappear. The key findings emerge in Models 3 and 4, which account for prior achievement using alternative measures. When achievement is measured through lower-secondary grades (Model 3), Covid-by-migration interactions become statistically insignificant. However, when achievement is measured using standardised test scores (Model 4), significant post-Covid interactions persist, particularly for second-generation students, who show an increased likelihood of entering technical and vocational tracks rather than academic ones.

What is the purpose of education?

The answer to this question can go in very different directions, and mostly depends on the context an educational effort is engrained into and the goals it pursues, explicitly and implicitly. Defining a context and the sets of goals pursued there is also not an easy task. For sure the context - whatever metaphysics characterises it - influences education greatly.

The opposite is also true, just less seen: education influences the context, mostly in the effort - in some cases consciously, in some probably not - of reproducing or consolidating it.

Authors like John Dewey, those of the so-called "Frankfurter Schule" and Martha Nussbaum centred many of their reflections on understanding and describing these peculiar dynamics.

This contribution to this panel aims to illustrate a pedagogical practice developed on one assumption: Democracies are not a given, and their constitutions are perishable products of history if the promises of equality of opportunities for every citizen and therefore of freedom - both positive and negative - made at their origins are not taken seriously.

There are many factors that play a significant role in whether the foundational promise of equality, freedom and justice are kept or not, and to what degree. Education is one of these factors, and it can and should do more to make democracies and their promises attainable and sustainable. This is the reason why promoting equity of opportunities, justice, is key.

Promoting educational justice in disadvantaged contexts is the mission of Teach For Italy. The practice I'll be presenting was designed and implemented to help the classrooms where Teach For Italy fellows teach in being places that foster student leadership, well-being and inclusion, while centering on "Student Voice" as one of the elements that constitute a Monitoring, Evaluation and Learning system.

A Monitoring, Evaluation and Learning system that has the peculiarity of being an instrument for collecting data (quantitative and qualitative evidence) generated by students, as well as their parents and teachers, which serves the principle of a collective hermeneutic circle. All involved stakeholders are actively participating not just in the process of generating data, but also - and equally essentially - in interpreting them, making sense of them for the purpose of enhancing dialogue, sense of belonging, reflection, wellbeing - advanced citizenry capabilities.

The paper ultimately argues that longitudinal data acquire democratic value when they enable institutions to recognize patterns of inequality, deliberate collectively about their meaning, and act intentionally to transform them. In this sense, tracing inequalities becomes a concrete pathway toward fostering democratic education.

Education systems have been reshaped in response to the social challenges of globalisation, including changing labour markets, migration dynamics, and processes of individualization (van de Werfhorst 2014). In this context, many countries have adopted quasi-market reforms in education (Ichilov 2012), inspired by economic theory (Friedman 1962; Chubb and Moe 1990) and by international organisations (Sellar and Lingard 2013). Neoliberal education policies have promoted school autonomy, privatisation, and accountability (Levačić 1995; Clark 2009). These reforms have intensified competition among schools, encouraging strategies to attract or select higher-performing students. Among these strategies, school priority rules for admission have become increasingly important. Schools may use criteria such as prior achievement, recommendations from schools or parents, catchment-area requirements, and interest in specific programmes. However, the consequences of these rules for socio-economic segregation between schools remain insufficiently understood.

This study examines the diffusion of school priority rules across countries, distinguishing between performance-based and residence-based criteria, and investigates their effects on socio-economic segregation between schools. It addresses four research questions: (RQ1) Are schools increasingly adopting performance-based and residence-based priority rules? (RQ2) Is the diffusion of these rules greater in countries where they were initially less widespread, suggesting convergence over time? (RQ3) Does the adoption of these policies increase socio-economic segregation between schools, and which type has the strongest effect? (RQ4) Do performance-based and residence-based rules have cumulative effects when adopted jointly?

We use data from all eight waves of OECD-PISA, from 2000 to 2022. The analytical sample includes more than 122,000 schools in 72 countries, corresponding to 465 country-year observations. A school is defined as adopting priority rules when the principal reports that residence in an area or students’ academic record is sometimes or always used as an admission criterion. To address RQ1, we describe the share of schools adopting each rule across PISA waves. For RQ2, we estimate regression models in which the change in the share of schools adopting these policies between T and T−1 is regressed on the share observed at T−1. For RQ3 and RQ4, we regress the level of segregation between schools on the share of schools adopting residence-based and performance-based rules. Segregation is measured using the dissimilarity index (Duncan and Duncan 1955) for students’ ESCS, gender, and migrant background. These models include country fixed effects. Preliminary findings are presented for RQ1 and RQ2; analyses for RQ3 and RQ4 will be completed before the conference.

Preliminary results show different diffusion patterns for the two rules. Residence-based rules remained stable, increasing from an average of 55% of schools in 2000-2009 to 60% in 2012-2022. By contrast, performance-based rules grew substantially, from 48–52% in 2003–2009 to 65% in 2015, before declining to around 55% in 2018 and 2022. Results for RQ2 indicate that both types of rules tend to spread more rapidly where prior adoption was lower, while growth is more limited where adoption is already widespread. This pattern is robust to nonlinear specifications, suggesting cross-country convergence in the diffusion of priority rules. Analyses will explore ceiling effects and geographical heterogeneity.