Keywords

9.1 Introduction

Numerous studies have investigated factors that address socioeconomic and ethnic disparities in academic achievement. One area of research, often referred to as opportunity gaps, highlights the importance of teaching quality and practices in promoting educational equity (Akiba, et al., 2007; Klieme et al., 2009; Nilsen & Gustafsson, 2016; Seidel & Shavelson, 2007).

However, a lack of consensus makes defining measures for teaching quality and assessment practices challenging. A widely accepted framework (the three-dimension conceptualization framework) attempting to do so, defines teaching quality as a three-dimensional construct encompassing classroom management, supportive climate, and cognitive activation (Baumert et al., 2010; Creemers & Kyriakides, 2007; Lazarides & Ittel, 2012; Sogunro, 2017). Adopting this framework, this study aims to examine the impact of teaching quality on students’ performance across different contexts, thereby identifying effective strategies to reduce educational disparities.

In recent decades, the education systems in the Nordic countries have evolved towards a high degree of autonomy, privatization, and marketization, leading to increased school segregation in terms of students’ intake, educational resources, teacher and teaching quality, and academic achievement (Fjellman et al., 2018; Yang Hansen & Gustafsson, 2016). In this context, this chapter explores the potential influence of teaching and assessment practices on socioeconomic and ethnic equity in mathematics achievement across Nordic classrooms. The study examines how differences in teaching practices, including teaching quality, formative assessment practices, and teachers’ emphasis on academic success, contribute to educational equity considering the students’ socioeconomic and language backgrounds.

9.2 Review of Previous Research

9.2.1 Students’ Socioeconomic and Ethnic Backgrounds and Their Academic Achievement

Socioeconomic background is perhaps the best known and most extensively studied predictor of educational outcomes (see e.g., Sirin, 2005; Strietholt et al., 2019), and is fundamental in determining an individual’s life chances (Pinquart & Sörensen, 2000; Tan et al., 2020). As a collective endeavor, schooling is an arena in which students’ performance is influenced by their peers. The composition of the student body within a school and a classroom has a profound impact on individual students’ achievement, with schools that admit more socioeconomically advantaged students typically displaying stronger academic outcomes (e.g., Agirdag et al., 2012).

Multiple classroom composition effects have been documented as predictors of achievement. Firstly, studying alongside more socioeconomically advantaged peers has been established as a predictor of student outcomes (e.g., van Ewijk & Sleegers, 2010; Yang Hansen et al., 2022). Secondly, increased ethnolinguistic diversity in a classroom is associated with lower average performance in assessments of the national language of the school system (Seuring et al., 2020). Lastly, students benefit from studying alongside high-performing classmates, a phenomenon which particularly benefits high-achieving students (e.g., Lavrijsen et al., 2022).

Domestic studies have observed notable segregation between schools across the Nordic region (e.g., Bernelius & Vilkama, 2019; Rangvid, 2007; Rogne et al., 2021; Yang Hansen & Gustafsson, 2016). These local studies are reflected in pervasive achievement gaps in the Nordic countries across multiple international studies targeting different age groups, including IEA’s Trends in International Mathematics and Science Study (TIMSS) 2015 and 2019, and the OECD’s Programme for International Student Assessment (PISA) 2015 and 2018 (Mullis et al., 2020; OECD, 2016). Pertinently to this study, composition effects can be observed in the international results for mathematics in TIMSS 2019 grade four, which show variation in achievement scores between schools with differing language use populations and socioeconomic composition (Mullis et al., 2020).

9.2.2 School Emphasis on Academic Success of Teachers and Student’s Academic Achievement

While teachers and students are key actors within the classroom, they are not the sole stakeholders in schools as learning communities. Administrators and parents also play an essential role in building the culture within the school. While autonomy to create a school culture varies between countries, one common aspect of school culture that exists across educational systems is the school’s emphasis on promoting academic success, which has been shown to relate to student learning (i.e., Bryk & Schneider, 2003; Hoy et al., 2006; Kythreotis et al., 2010; Martin et al., 2013).

In TIMSS, the belief in group success within the school context is conceptualized as the school emphasis on academic success (SEAS). SEAS is compiled from the teacher and principal responses to questionnaire items indicating a supportive environment for academic success (Martin et al., 2013), and a moderate relationship between SEAS and achievement across mathematics, reading, and science has been found among fourth graders. Further, Martin et al.’s (2013) SEAS model has been demonstrated to have high construct validity and be applicable across multiple national contexts. It is also a strong predictor of science achievement in countries like Norway (Nilsen & Gustafsson, 2014). SEAS is strongly associated with classroom SES composition, with teachers reporting higher levels of SEAS in schools with higher SES classrooms (Rolfe et al., 2022). However, while SEAS is highly related to SES, it has been found not to predict achievement when modelling teacher quality and opportunity to learn in Sweden (Rolfe et al., 2022).

9.2.3 Teaching Quality and Student’s Academic Achievement

High quality teaching is an established explanatory factor for student achievement (e.g., Blömeke et al., 2016; Darling-Hammond, 2000). However, there is no fixed consensus on what defines teacher quality. Goe (2007) proposes a framework in which various indicators of teacher quality can be grouped as inputs, processes, or outcomes. The present study focuses on the processes and outcomes of teacher quality, particularly on the quality of teaching and assessment practices and student mathematics achievement.

In terms of teaching processes as indicators of teacher quality, the present study examines cognitive activation, classroom management, and supportive climate in line with the three-dimension conceptualization framework of teaching quality (see Baumert et al., 2010; Creemers & Kyriakides, 2007; Lazarides & Ittel, 2012; Sogunro, 2017). Teaching processes involve complex and interweaving sets of behaviors that teachers embody through their professional practice. Many of the findings discussed in the literature which illuminate these processes are the results of locally and regionally administered survey studies (see Goe, 2007).

Cognitive activation, as summarized by Lipowsky et al (2009), involves developing conceptual understanding between new and old content, selecting activities which operate on progressively higher cognitive levels, and engaging in quality discourse with students. Cognitive activation is a significant predictor of student achievement (e.g., Baumert et al., 2010; Li et al., 2021; Lipowsky et al, 2009). This relationship may be related to compositional effects, with Le Donné et al. (2016) proposing a stronger predictive relationship in more socioeconomically advantaged schools.

The previously discussed concept of SEAS is distinct from the concept of classroom climate. SEAS focuses on the prioritization of learning and achievement by multiple stakeholders, including students, teachers, parents, and school leadership–all of whom contribute to the school having a good climate for success (e.g., Nilsen & Gustafsson, 2014). In contrast, a supportive climate for learning can be summarized as allowing individual students exposure to ideas and feedback, supporting them in self-reflection, and promoting behavior modification to facilitate learning (Gibb, 1958).

Established research on the interrelationships between school climate, SES, and achievement is somewhat contradictory (Berkowitz et al., 2017). A positive supportive climate has been suggested as a moderator of the SES-achievement relationship (see, e.g., Cheema & Kitsantas, 2014). School climate is, in turn, seen as a compensatory mechanism for low SES (Brand et al., 2003; Schagen & Hutchison, 2003), or as a phenomenon influenced by the SES of the student body. For example, schools with lower SES students experiencing higher out-of-school social risks may struggle to establish the type of supportive climate, which can affect student outcomes (McCoy et al., 2013).

Additionally, it is worth noting that there is some disagreement in the literature regarding the presence of ethnic bias in teacher ratings of student behavior (see Mason et al., 2014). However, as much of the research in this field is situated in the American context and uses race as its conceptualization of ethnicity, the findings of these studies may not be directly applicable to the Nordic educational context.

9.2.4 Assessment Practices and Students’ Academic Achievement

Assessment plays a crucial role in classroom practices, as it informs teachers about students’ progress and can be used as a tool to motivate students’ learning (e.g., Broadfoot et al., 2002; Gronlund, 2006). Extensive research has explored the impact of teacher assessment practices on student behaviors and outcomes, including learning depth, self-motivation, and achievement (Crooks, 1988). Notably, frequent use of tests in the classroom has a moderate effect size on student attainment (Bangert-Drowns et al., 1991; Yang et al., 2021). However, teachers’ perceptions of student achievement may be clouded by implicit biases, such as student ethnicity and socioeconomic background (Darling-Hammond, 1995) or gender (e.g., Guez et al., 2020).

Brookhart (1997) introduced the classroom assessment model, which views the assessment environment as a communal experience for students. In this model, teachers define assessment purposes, set tasks and criteria, provide feedback, and monitor outcomes (see also Brookhart, 2001). The classroom assessment environment influences cognitive and non-cognitive outcomes of students. As students progress through middle school, teachers reported an increased use of assessment tools considered more informative for grading (Martínez et al., 2009). Research conducted in socioeconomically disadvantaged schools has demonstrated that implementing evidence-based instructional and behavior management strategies leads to improved student mathematics knowledge during the elementary and middle years (Reddy et al., 2020). This highlights the importance of effective instructional practices in fostering student progress.

9.2.5 Opportunity to Learn and Students’ Academic Achievement

Opportunity to learn (OTL) is a concept which rests on the assumption that an individual will not perform well on tests covering content they have not been taught (e.g., Eggen et al., 1987). While some scholars emphasize the importance of time dedicated to covering content (e.g., Carroll, 1963), as elaborated on in Chap. 2, OTL in TIMSS is considered as the alignment between the intended curriculum (formal curricula documents based on national or regional standards), the implemented curriculum (what teachers have taught), and the attained curriculum (what students have learned) (see Schmidt et al., 1997). However, as noted in Chap. 4, the mathematics curricula in the Nordic countries are structured in such a way that expected learning is not expressed in grade-specific knowledge, but across multi-year phases, which do not align with the grades assessed by TIMSS. This may potentially impact the alignment between the intended and implemented curricula measured through the OTL construct for the national samples in this study.

The relationship between OTL and student achievement has been extensively studied at both the individual (e.g., Schmidt et al., 2013, 2015) and collective levels (see Scheerens et al., 2007; Seidel & Shavelson, 2007). Existing evidence indicates that OTL is a positive predictor of achievement and a mediator of the SES-achievement relationship (e.g., Schmidt et al., 2015). From secondary analyses of ILSA data, it appeared that the strength of this relationship varies depending on the study (PISA vs TIMSS) and the formulation of the construct (Luyten, 2017; Rolfe et al., 2021; Schmidt et al., 2015; Yang Hansen & Streitholt, 2018) and the subject, being evident in mathematics but not science (Luyten, 2017; Rolfe et al., 2021).

9.3 The Hypothesis Model and Research Questions

In summary, the relationships between teaching quality, assessment practices, content coverage, and students’ achievement in mathematics are complex and interrelated. It can be hypothesized that teaching quality and assessment practices influence the delivery of curriculum content, which in turn affects students’ achievement in a specific subject. It can also be hypothesized that such a conditional classroom mechanism may help to reduce the achievement inequality among students due to their family socioeconomic and ethnic backgrounds (see Appendix 1 for the hypothesis model).

Applying data from TIMSS 2019, the present study aims to address the following research questions by testing the hypothesized conditional mechanism:

  1. 1.

    What are the differences in students’ mathematics achievement, socioeconomic status, and ethnic composition across classrooms?

  2. 2.

    Is there socioeconomic and ethnic inequality in students’ mathematics achievement?

  3. 3.

    Do socioeconomic and ethnic inequalities in students’ mathematics achievement differ significantly across different classrooms?

  4. 4.

    Do teachers’ instructional and assessment practices as well as content coverage impact student’s mathematics achievement and socioeconomic and ethnic inequality in their achievement?

Considering the institutional differences in the Nordic education systems, the conditional mechanism may vary. This chapter adopts a comparative perspective in exploring the above-mentioned research questions.

9.4 Method

9.4.1 Samples

The current study used TIMSS 2019 data from four Nordic countries, i.e., Denmark, Finland, Norway, and Sweden. Table 9.1 shows the number of individual students and classrooms in the four samples. In total, there are 15,873 students and 966 classrooms to facilitate the current analyses. Finland holds the largest number of sampled classrooms (316) and students (4730), while Denmark has the lowest numbers with 195 classrooms and 3227 students. Norway and Sweden have rather similar sample sizes. It should be noted that Norway participated in TIMSS 2019 with grade five students and for the rest of the analyzed Nordic countries, grade four students were included in the samples.

Table 9.1 The number of cases in each Nordic education system from TIMSS 2019
Full size table

9.4.2 Variables

Appendix 2 presents a comprehensive list of all the variables that were involved in the current study. Student socioeconomic status is measured by the number of books at home, ranging from 0 (none to 10 books) to 4 (more than 200 books). The dummy-coded ethnic background is based on student responses as to how often they speak the language of the test at home. If students responded with “always or almost always speak the language of the test at home”, they are classified as native, whereas if they reported to “sometimes or never speaking the language of the test at home”, they are classified as immigrant. The aggregated cluster mean of SES and ethnicity are used as indicators of socioeconomic and ethnic compositions of classrooms. Teachers’ instructional quality (i.e., cognitive activation and supportive climate), assessment practices, their emphasis on academic success (SEAS), and OTL in terms of the percentage of content coverage for the three-mathematics content domains in TIMSS were measured using the teacher questionnaire data (see Appendix 2 for detailed information on these teacher-related constructs). Finally, students’ mathematics achievement was captured using five plausible values of the test score and used as the outcome variable in the analysis (see Chap. 3 Analytical Framework for further details).

9.4.3 Analytical Method and Process

Data in TIMSS 2019 were collected using a stratified two-stage cluster sampling design, resulting in a hierarchical data structure where students were nested within classrooms and schools (LaRoche et al., 2020). A two-level structural equation modelling technique (Hox et al., 2017) was therefore required to decompose the total variance of an outcome into individual-level and classroom-level variance components, allowing for accurate standard error estimation based on the correct sources of variation and avoiding type I error in statistical inference. The Intraclass Correlation Coefficient (ICC) quantifies the proportion of total variance in an outcome variable that can be ascribed to variations among students belonging to different classrooms. It serves as a measure of the extent of heterogeneity among students across classrooms, with a higher ICC indicating greater disparities in student outcomes across different classrooms.

Table 9.2 provides information on the ICCs of SES, ethnicity, and mathematics achievement in the four Nordic countries. Sweden held the highest between-classroom differences in mathematics (0.18), SES (0.17), and ethnicity (0.24), indicating a higher level of classroom segregation. Finland also has a relatively high proportion of cross-classroom differences in mathematics scores (0.17) and ethnicity (0.19), but low SES differences. Denmark has the most homogenous classrooms, with the lowest ICCs in mathematics (0.10), SES and ethnicity (both 0.08). The ICCs in Norway were at the intermediate level among the four Nordic countries. In general, cross-classroom differences in the Nordic countries are low when compared internationally, especially in Denmark. However, from an educational inequality perspective, these differences are still substantial and need to be further explained by the teacher-level relevant factors.

Table 9.2 Intraclass correlation coefficients of mathematics score, socioeconomic status, and ethnic background
Full size table

The analyses were conducted in multiple steps. Firstly, the sub-dimensions of teaching quality (i.e., cognitive activation and supportive climate) and SEAS were tested for measurement invariance to ensure the comparability of the constructs across the Nordic countries. Metric invariance level was successfully achieved, assuring the comparison of the relationship among the constructs across the four Nordic countries (see Appendix 3). Secondly, a principal component factor score was estimated for these three constructs and used as manifest variables in the two-level models. The average values for formative assessment practices in mathematics classrooms and mathematics OTL were computed, based on the corresponding indicators of the constructs. Subsequently, a series of two-level models at the student and classroom levels were estimated in each of the Nordic countries.

To address research questions 1 and 2, model 1 tested the relationship between SES and ethnicity with mathematics achievement in a two-level structural model for each country, where socioeconomic and ethnic inequality in achievement and classroom segregation can be estimated. Model 2 comprised a set of two-level random slope models, examining whether the socioeconomic and ethnic inequality in mathematics achievement varied significantly across different classrooms in the Nordic countries. The result from model 2 provides the answer to research question 3. Finally, model 3 examined whether teaching and assessment practices, teacher emphasis on academic success, and OTL account for the variation in mathematics achievement and socioeconomic and ethnic inequality in achievement (i.e., the random slopes), which will unveil the results of the research question 4 (see Appendix 1).

Modelling was done in Mplus (Muthén & Muthén, 1998–2017) with MLR (maximum likelihood estimator with robust mean and variance) for model 1, where missing data was handled using the expectation–maximization algorithm. The Bayesian estimator was applied for models 2 and 3 with random slopes. For all models, five plausible values for mathematics achievement were used.

9.5 Results

9.5.1 Socioeconomic and Ethnic Inequality in Mathematics Achievement at Student and Classroom Levels

Socioeconomic and ethnic inequality in mathematics achievement is measured by the relationship between students’ SES and respective ethnic backgrounds with their mathematics achievement. Table 9.3 shows that at the individual level, the relationship between SES and mathematics was positive and significant in all Nordic countries, and the beta coefficients were rather even, ranging from 0.24 in Denmark and 0.28 in Norway and Sweden. The same is true for the relationship between students’ ethnic backgrounds and mathematics achievement, indicating that native students generally have higher achievement than students with migration backgrounds. However, the regression coefficients were rather small after controlling for SES.

Table 9.3 Relationship between socioeconomic status, ethnic background, and mathematics achievement at individual and classroom levels
Full size table

At the classroom level, the relationship between classroom SES composition and average mathematics achievement was much higher, compared to those at the individual level. The beta coefficient ranges from around to above 0.70. This implies that about or over half of the variation in mathematics achievement across classrooms can be explained by the differences in the SES composition of the student intake. Finland had the highest SES contextual relationship at 0.79, while Norway had the lowest at 0.66. It is important to note that the ethnic contextual relationship with average mathematics achievement was not significant in Finland and Norway. While in Denmark and Sweden, an additional ethnic contextual relationship was found, being 0.25 and 0.22, respectively. In total, the SES and ethnic context of classrooms explained 62% of the mathematics achievement variation in Finland, followed by 58 percent in Denmark, 52 percent in Sweden and 44% in Norway.

9.5.2 Testing Random Slopes

The socioeconomic and ethnic inequalities in mathematics achievement among students in the current analysis were captured by the regression coefficients (i.e., the slopes) of mathematics achievement on SES or ethnicity. Previous research indicates that the prevalence of neoliberal ideology has resulted in a global trend towards market-like school systems that emphasize school choice and education provision on demand (e.g., Blossing et al., 2014). Unfortunately, this trend has led to greater social and ethnic segregation as well as quality differences between schools in many countries (e.g., Bonal & Bellei, 2018; OECD, 2012). It can be assumed that some schools and classrooms can effectively compensate for students’ disadvantages in sociodemographic backgrounds and help to reduce the SES- or ethnicity-achievement relationship, while others may fail to fulfil their compensatory mission for disadvantaged students. This mechanism is reflected by the so-called random slopes, meaning that the two slopes vary depending on which school or classroom the student attends.

Table 9.4 shows the estimated mean and variance of the two random slopes. The estimated mean of the slopes S1 (SES-mathematics relationship) and S2 (ethnicity-mathematics relationship) were found to be positive and significant for all Nordic countries. Albeit small in effect, the impact of students’ family SES and ethnic background on their mathematics achievement once again was confirmed (e.g., Rolfe & Yang Hansen, 2021). The variance of the two slopes was also statistically significant, implying that the impact of SES and ethnicity of children on their mathematics achievement varies significantly across different classrooms. It should be interesting to explore the classroom-level factors that may be important in accounting for the variation.

Table 9.4 Estimated mean and variance of the random slopes
Full size table

9.5.3 Impacts of Teacher-Related Factors on Mathematics Achievement and Inequalities

Two-level path analysis with cross-level interaction and random slopes was conducted to test the hypothesis model (see Sect. 9.3). In that, teaching quality and assessment practices are allowed to affect the delivery of the contents in mathematics, in turn, affect mathematics achievement and SES-achievement and ethnicity-achievement relationships at the individual level. This mechanism was also conditional by teachers’ emphasis on academic success, classroom SES and ethnic composition. This is a saturated model with perfect model fit. Figures 9.1, 9.2, 9.3 and 9.4 present the model results of the interrelationships. Note that the non-significant paths are not included in the figures.

Fig. 9.1
figure 1

Footnote

For all the figures in the chapter, the variable abbreviations are denoted as following: SESB = class-level socioeconomic composition; EthnicB = class-level ethnic composition; SEAS = teacher’s emphasis on student’s academic success; CogAct = Cognitive activation practices; ASSESS = assessment practices; SupClim = Supportive climate in the classroom; OTL = content coverage reflecting opportunity to learn; Math = classroom mathematics achievement; S1= random slope between student’s socioeconomic status and their mathematics achievement; S2 = random slope between student’s ethnic background and mathematics achievement.

Classroom-level results in Denmark

Full size image
Fig. 9.2
figure 2

Classroom-level results in Finland

Full size image
Fig. 9.3
figure 3

Classroom-level results in Norway

Full size image
Fig. 9.4
figure 4

Classroom-level results in Sweden

Full size image

Denmark

Based on the results depicted in Fig. 9.1, it is apparent that the teaching quality and teacher assessment practices did not demonstrate any significant relationship with the coverage of content in the TIMSS mathematics content domains, nor with mathematics achievement or random slopes. However, contextual factors within the classroom, such as the teacher’s emphasis on academic success and the ethnic composition of the classroom, did yield significant impacts. In particular, the ethnic composition of the classroom was positively correlated with the average mathematics score of the classroom (0.49) and served to offset the relationship between ethnicity and individual-level achievement. This suggests that attending a classroom with a greater number of native students mitigates the impact of one’s own ethnic background on mathematics achievement (− 0.44), holding other classroom practices and conditions constant. Moreover, the socioeconomic composition of the classroom demonstrated a significant influence on the coverage of content in mathematics domains (11.5) as well as on the teacher’s emphasis on academic success (1.12), which in turn influenced the supportive climate of the classroom (0.11).

Finland

Similar to Denmark, there was no significant association between teaching quality and assessment practices and the coverage of content in the content domains of mathematics, nor with mathematics achievement or random slopes in Finland. However, the classroom SES composition had a significant impact on the average mathematics score of the classroom (0.48) and mitigated the relationship between students’ family SES and their mathematics achievement (− 0.15). In addition, SES composition was significantly linked to teachers’ emphasis on academic success (0.62), which, in turn, influenced cognitive activation (0.19) and the supportive climate (0.13) in the classroom.

Norway

In Norway, a similar pattern of relationship was observed between classroom SES composition and average mathematics achievement (0.48), as well as SES-mathematics random slope (− 0.15). This suggests that a higher SES composition not only has an impact on classroom mathematics achievement but also reduces the influence of students’ family SES on their mathematics achievement. Additionally, classroom SES composition had an indirect effect on the ethnic-mathematics random slope through teachers’ emphasis on academic success (0.79) and assessment practices (− 0.31). High SES composition classrooms reduce the impact of ethnic composition on students’ mathematics achievement through teachers’ emphasis on academic success and assessment practices (− 0.26). Furthermore, classroom SES composition indirectly affected cognitive activation (0.13) and supportive climate (0.20) through teachers’ emphasis on academic success.

Sweden

Both ethnic and SES classroom compositions had an impact on teachers’ emphasis on academic success (1.04 and 1.15, respectively), which in turn, had a positive relationship with supportive climate (0.25), cognitive activation (0.10), and content coverage (3.60). No significant relationship was found between any teacher or classroom-level variables and the two random slopes. Furthermore, the SES composition had a positive effect on the average mathematics achievement of the class (0.20).

9.6 Discussion and Conclusions

The global prevalence of neoliberal ideology, especially market mechanisms, such as school choice, autonomy, and competition, in recent decades has significantly transformed the unified and egalitarian Nordic model into systems with increasingly diverse educational practices across systems and intensified segregation along socioeconomic and ethnic lines (e.g., Blossing et al., 2014; Yang Hansen & Gustafsson, 2019).

Using the fourth grade TIMSS 2019 data from Denmark, Finland, Norway, and Sweden, this chapter aimed to identify similarities and disparities in the impact of teaching quality (i.e., cognitive activation and supportive climate), assessment practices, and content coverage on socioeconomic and ethnic inequalities in mathematics achievement across the Nordic countries. The study also analyzed how these relationships are influenced by the sociodemographic context of the classroom, including classroom SES and ethnic compositions, and the teacher's emphasis on academic success.

It was revealed that the socioeconomic and ethnic contexts of the classroom played important roles in students’ mathematics achievement in all Nordic countries analyzed. Furthermore, attending a school or classroom with a high proportion of native students and a high socioeconomic status was found to have a compensatory effect in reducing the effect of family socioeconomic status and ethnic background on students’ mathematics achievement, therefore, beneficial for students’ learning outcomes and social ethnic inequality. Another common feature found in all four Nordic countries studied was that classroom sociodemographic contexts, especially the socioeconomic composition of a classroom, positively associated with teachers’ emphasis on students’ academic success and, in turn, promoted classroom teaching quality. In Sweden, the ethnic composition seemed to be equally important. This may be attributed to the fact that Sweden has been a leader in extensive changes marked by decentralization and significant marketization and privatization (e.g., Lundahl, 2016). This is also evident in the high disparities in mathematics achievement, SES, and ethnic composition across Swedish classrooms/schools.

Lubienski et al. (2022) highlighted that implementing such policies could lead to social segregation, as families might select schools based on non-academic social factors, and schools could adopt practices that restrict enrollment for less favored students. In Sweden, for example, the universal voucher program with free choice of schools led to increased segregation of native and immigrant students, as well as further stratification based on parental education (e.g., Yang Hansen & Gustafsson, 2016, 2019). In many countries, teachers tend to prefer working in schools with a more socioeconomically advantaged student composition or even a higher proportion of ‘white’ students (Bonesrønning et al., 2005; Glassow & Jerrim, 2022; Hansson & Gustafsson, 2016). It is evident that schools with a higher socioeconomic status composition and a higher proportion of native students often form a better school ethos for learning and access well-qualified and motivated teachers with a strong emphasis on student’s academic success (e.g., Akiba et al., 2007; Han, 2018). The differentiated learning environment, peer groups, and teacher resources contribute to the achievement gap between school and student outcomes.

However, the results regarding the impact of teachers’ instructional and assessment practices and content coverage on mathematics achievement were unexpected. None of these factors considered was found to have a significant effect on classroom mathematics achievement or the socioeconomic and ethnic inequality of mathematics achievement, except for Norway. It was demonstrated that mathematics performance differences between native Norwegian children and children with migration backgrounds seem to be reduced by assessment practices. In other words, in a classroom where the teacher applied different types of formative assessment practices more frequently, the mathematics achievement gap between native and non-native students was smaller. The compensatory effect of assessment practices was reinforced by classroom socioeconomic context and teachers’ emphasis on students’ academic success. In Norway, formative assessment practices are often used in the classrooms to provide feedback to students (Gamlem & Smith, 2013; Havnes et al., 2012). This approach is recognized as a useful tool for supporting student learning and guiding teacher instruction (Havnes et al., 2012). By identifying areas of strength and weakness, teachers can adjust their instruction to better support the learning needs of their students. This approach may offer potential benefits for students with a migration background, as it has the potential to address their unique learning needs.

To conclude, the study has highlighted several areas that require attention in both teacher practices and policy innovation. These efforts may contribute to enhancing educational equity and promoting school desegregation.

Promote socioeconomically and ethnically diverse classrooms

Education policies could promote socioeconomically and ethnically diverse classrooms by ensuring that schools are not segregated based on students’ backgrounds. This could be achieved through policies that promote school integration, such as zoning policies that promote diversity in student populations.

Increase emphasis on academic success

Teachers and schools should emphasize academic success and set high expectations for all students, regardless of their socioeconomic or ethnic background. This could include providing additional support for struggling students, setting academic goals, and creating a positive and supportive learning environment.

Improve assessment practices

Education policies and practices should aim to improve assessment practices for learning, including providing additional training for teachers on how to assess student performance fairly and equitably, and ensuring that assessment practices are aligned with curriculum standards and goals. By doing so, education systems can promote a more accurate understanding of student learning and progress, which can help to reduce performance differences between students.