Skip to main content
F1000Research logoLink to F1000Research
. 2026 Sep 12;14:1209. Originally published 2025 Nov 5. [Version 4] doi: 10.12688/f1000research.172015.4

Closing the Gap in Early Mathematics: Domain and Cognitive Insights from TIMSS 2023 in South Africa and Singapore

Mathelela Steyn Mokgwathi 1,a
PMCID: PMC13587798  PMID: 42763458

Version Changes

Revised. Amendments from Version 3

This version has been substantially revised in response to peer review. The Introduction and Literature Review have been strengthened by clarifying the rationale for benchmarking South African Grade 5 learners against Singaporean Grade 4 learners within the common TIMSS Grade 4 mathematics framework and by positioning Singapore as an analytical benchmark rather than a model for direct policy transfer. The manuscript now provides greater contextualisation of South Africa's mathematics achievement, including persistent educational inequalities, post-apartheid educational progress, and the potential influence of the COVID-19 pandemic. Methodological descriptions have been expanded to improve transparency regarding the use of plausible values, sampling weights, replicate-based variance estimation, and effect-size estimation. The Results section has been revised to distinguish descriptive reporting from interpretation, while the Discussion has been strengthened through more critical engagement with the international literature and by removing unsupported causal language. The conclusion and limitations have also been revised to ensure that all interpretations are supported by the evidence, to acknowledge the descriptive and comparative nature of the study, and to clarify the scope and limitations of the cross-national comparison.

Abstract

Background

South Africa continues to underperform in primary mathematics, yet overall mean comparisons obscure where achievement gaps are most concentrated. This study examines domain-specific patterns of mathematics achievement using data from the Trends in International Mathematics and Science Study (TIMSS) 2023, benchmarking South African Grade 5 learners against Singaporean Grade 4 learners assessed on the same TIMSS Grade 4 mathematics framework.

Methods

A quantitative secondary analysis was conducted using nationally representative TIMSS 2023 datasets comprising 10,424 South African learners from 285 schools and 6,530 Singaporean learners from 181 schools. South Africa assessed learners in Grade 5 using the TIMSS Grade 4 mathematics instruments, consistent with TIMSS procedures where curriculum exposure aligns with the benchmark framework. Mathematics achievement was analysed across content domains (Number, Measurement and Geometry, Data) and cognitive domains (Knowing, Applying, Reasoning). Weighted estimates, replicate-based variance estimation, and effect sizes (Cohen’s d) were computed in accordance with IEA technical guidelines. The analysis is descriptive and comparative rather than causal.

Results

South African learners performed substantially below the international centre point across all domains. The largest content-domain gap relative to Singapore was observed in Measurement and Geometry, indicating pronounced weaknesses in spatial reasoning. The largest cognitive-domain gap occurred in Knowing, reflecting fragile foundational fluency. Although Applying was relatively stronger within the South African profile, it remained substantially below benchmark levels. Achievement gaps were therefore concentrated in foundational and spatial domains rather than uniformly distributed.

Conclusions

Concentrated achievement gaps in foundational knowledge and geometry characterise the underperformance of primary mathematics in South Africa. The findings provide a structured diagnostic profile of domain-specific achievement disparities, highlighting the concentration of gaps in foundational knowledge and spatial reasoning.

Keywords: Trends in International Mathematics and Science Study (TIMSS), Mathematics Achievement, Content and Cognitive Domains, South Africa, Singapore

1. Introduction

Mathematics achievement at the primary school level has profound implications for learners’ future participation in education, work, and society. Basic mathematics skills support higher-order thinking, problem-solving, and lifelong learning, all of which are central to academic success and broader social participation, as consistently demonstrated in international large-scale assessment research ( Mullis et al., 2020; von Davier et al., 2024). TIMSS has become a critical assessment instrument for benchmarking mathematical learner achievement internationally, providing comparative evidence of mathematics achievement across education systems with different curricular and instructional contexts. Since the advent of democracy, South Africa has expanded access to schooling and implemented curriculum reforms aimed at promoting more equitable learning opportunities. South Africa has participated in TIMSS since 1995 and, despite modest gains over successive cycles, continues to record one of the lowest average mathematics achievement scores among participating education systems ( Mullis et al., 2020; Reddy & Hannan, 2019; Zuze et al., 2018). National average performance, however, masks substantial variation associated with socioeconomic background, language, school contexts, and other structural inequalities. Much of the existing research focuses on Grade 9 data, emphasising long-term learning deficits and systemic inequities ( Mensah & Baidoo-Anu, 2022; Reddy & Hannan, 2019). Far less attention has been given to the primary phase, particularly Grade 5, where learners consolidate fundamental numeracy and begin to transition from concrete to abstract reasoning.

Despite extensive research on mathematics achievement in South Africa, much of the existing literature has concentrated on overall performance trends or on secondary school outcomes, particularly at the Grade 9 level, where cumulative learning deficits become more visible. Comparatively limited attention has been given to domain-specific achievement patterns at the primary level, where foundational competencies are established. Furthermore, while international benchmarking studies highlight performance disparities, fewer studies systematically disaggregate these gaps across both content and cognitive domains using a unified analytical framework. This limits the ability to identify where learning deficits are most concentrated and how these deficits are distributed across different areas of mathematics learning. The present study addresses this gap by providing a structured domain-level analysis of Grade 5 mathematics achievement using TIMSS 2023 data.

TIMSS 2023 results indicate that South African Grade 5 learners achieved an average score of 362 compared to Singapore’s Grade 4 average of 615 ( von Davier et al., 2024). This gap, despite Singaporean learners being a grade lower than South African learners, indicates substantial differences in average mathematics achievement between the two education systems. More specifically, this disparity reflects structured achievement gaps across both mathematical content domains and cognitive domains, rather than a uniform decline across all aspects of mathematics learning. Preliminary TIMSS 2023 reporting shows that the lowest average performance occurred in Measurement and Geometry and in the Knowing cognitive domain. These patterns are consistent with comparatively weaker performance in spatial reasoning and procedural fluency.

In TIMSS 2023, while the international benchmarking grades are 4 and 8, some education systems assess an adjacent grade using the same internationally calibrated instruments when curriculum alignment indicates that the benchmark framework better matches learners’ opportunity to learn. In this study, South Africa’s Grade 5 cohort is analysed because it was assessed using the TIMSS Grade 4 mathematics assessment instruments and framework, providing a common assessment framework for comparison with Singapore’s Grade 4 results ( Mullis et al., 2020; von Davier et al., 2024). Accordingly, the comparison undertaken in this study is not between different curricula, but between two cohorts assessed on the same TIMSS Grade 4 mathematics framework and reporting scale, thereby ensuring measurement equivalence. Singapore was selected as the benchmark because it consistently demonstrates the highest average mathematics achievement in TIMSS. The comparison is intended to provide an analytical benchmark for examining domain-specific achievement gaps rather than to imply equivalence between the two education systems or to advocate direct transfer of educational policies and practices. Given the substantial historical, cultural, socioeconomic, and educational differences between South Africa and Singapore, the findings are interpreted to inform context-sensitive curriculum and pedagogical improvement rather than direct policy adoption.

National assessments such as the Annual National Assessments (ANA), which is now discontinued, and systemic evaluation reports have highlighted low achievement levels in mathematics among learners, but they rarely explore how learners perform across specific content and cognitive domains. Yet TIMSS distinguishes between three content domains (Numbers, Measurement and Geometry, and Data) and three cognitive domains (Knowing, Applying, and Reasoning). Assessing learner achievement through these lenses allows for a diagnostic understanding of learners’ strengths and weaknesses. In this study, the term “achievement gap” refers to measurable differences in scale scores relative to the TIMSS international centre point and relative to a high-performing benchmark system on the same assessment framework. The focus is therefore diagnostic and comparative rather than causal. Research from high-performing education systems such as Singapore describes curriculum alignment, spiral progression, and scaffolded teaching and learning as features associated with balanced development of knowledge, application, and reasoning skills ( Choy & Dindyal, 2024; Low & Wong, 2021; Morony, 2023; Mullis et al., 2020). These features provide valuable comparative insights; however, educational practices are embedded within specific historical, cultural, and institutional contexts and therefore require careful contextual adaptation rather than direct transfer. Conversely, South African studies identify persistent challenges in geometry, reasoning, and teacher content knowledge ( Maqoqa, 2024; Taylor, 2021), compounded by curriculum overload and large class sizes, which are associated with fewer opportunities for formative assessment and conceptual engagement . Improving mathematics learner achievement has been associated with more than curriculum reform; it depends on strengthening the consistency between curriculum design and development, teacher professional development, and classroom practice. Examining how learners perform across content and cognitive domains helps identify the areas requiring targeted instructional support and pedagogical innovation.

This study therefore aims to examine South African Grade 5 learners’ mathematics achievement in TIMSS 2023, disaggregated by content and cognitive domains, and benchmarked against Singapore’s Grade 4 performance on the same TIMSS Grade 4 mathematics framework. By identifying where achievement gaps are most pronounced, the study seeks to provide a domain-specific diagnostic profile to inform curriculum alignment and pedagogical reform. It contributes in three key ways. Firstly, it focuses on the under-researched area of upper primary mathematics learning, where learners consolidate foundational numeracy and transition from concrete to more abstract mathematical reasoning. Secondly, it breaks down performance by content and cognitive areas to identify specific patterns of strength and weakness. Thirdly, it uses Singapore as an analytical benchmark to contextualise South Africa’s domain-specific achievement profile within a common international assessment framework. The study was guided by the following research questions:

1.1. Research Questions

  • 1.

    What is the achievement profile of South Africa’s Grade 5 learners across the TIMSS 2023 Grade 4 mathematics content domains (Number, Measurement and Geometry, and Data)?

  • 2.

    What is the achievement profile of South Africa’s Grade 5 learners across the TIMSS 2023 Grade 4 mathematics cognitive domains (Knowing, Applying, and Reasoning)?

1.2. Literature Review: Benchmarking South Africa’s Grade 5 Results on the TIMSS Primary Mathematics Assessment against Singapore’s Grade 4

Benchmarking with TIMSS 2023

The Trends in International Mathematics and Science Study (TIMSS) provides an internationally comparable assessment of mathematics and science achievement at the primary and secondary school levels ( Mullis et al., 2020; von Davier et al., 2024). Singapore consistently records the highest average mathematics achievement among participating education systems. In TIMSS 2023, Singapore’s Grade 4 learners achieved an average score of 615, while South Africa’s Grade 5 learners achieved an average score of 362 ( von Davier et al., 2024). Although Singaporean learners were assessed one grade lower than South African learners, both cohorts completed the internationally calibrated TIMSS Grade 4 mathematics assessment, providing a common framework for comparison. The comparison therefore serves as an analytical benchmark for examining domain-specific achievement patterns rather than as evidence that the two education systems are directly comparable in all respects. Given the substantial historical, cultural, socioeconomic, and educational differences between the two countries, differences in average achievement should be interpreted cautiously and within their respective educational contexts. The magnitude of the achievement gap highlights the value of examining performance across the TIMSS content domains (Number, Measurement and Geometry, and Data) and cognitive domains (Knowing, Applying, and Reasoning) to better understand domain-specific patterns of learner achievement.

Curriculum Alignment and Content Domains

Singapore’s mathematics curriculum is internationally recognised for its coherence and spiral structure, with mathematical concepts revisited at progressively higher levels of complexity. Previous research describes this curriculum structure as supporting continuity in learners’ mathematical development ( Choy & Dindyal, 2024; Low & Wong, 2021). In TIMSS 2023, Singapore scored 613 in Number, 619 in Measurement and Geometry, and 616 in Data, while South Africa scored 362, 353, and 362, respectively ( von Davier et al., 2024). These results indicate substantial differences in average achievement across all three content domains, with the largest difference observed in Measurement and Geometry. Maqoqa (2024) and Tachie (2020) similarly identify Measurement and Geometry as an area in which South African learners experience persistent learning challenges. The convergence between the TIMSS 2023 findings and previous South African research suggests that Measurement and Geometry remains an important area for curriculum and instructional attention. Although Singapore consistently records higher achievement across all content domains, these differences should be interpreted within the broader educational contexts of the two countries rather than as evidence that any single curricular feature explains the observed performance differences.

Cognitive Demands: Knowing, Applying, and Reasoning

TIMSS distinguishes between three cognitive domains: Knowing, Applying, and Reasoning, which assess learners’ ability to recall and apply mathematical knowledge and to solve increasingly complex problems ( Mullis et al., 2020). Singapore’s Grade 4 learners achieved average scores of 624 in Knowing, 615 in Applying, and 609 in Reasoning, whereas South Africa’s Grade 5 learners scored 357, 366, and 363, respectively ( von Davier et al., 2024). These results indicate substantial differences in average achievement across all three cognitive domains, with the largest difference observed in the Knowing domain (267 points). Within the South African cohort, the Applying domain recorded the highest average score, while the Knowing domain recorded the lowest average score. This pattern suggests variation in learners’ performance across the cognitive domains rather than uniform performance across all aspects of mathematical cognition. Previous studies have associated strong performance in the Knowing domain with secure foundational mathematical knowledge, while the Applying and Reasoning domains require learners to transfer and extend that knowledge in increasingly complex contexts ( Mullis et al., 2020; Peduk & Ateş, 2019). The comparatively lower performance observed in the Knowing domain therefore highlights an area that warrants further attention when interpreting South Africa’s overall mathematics achievement profile.

1.3. Instructional and Structural Factors

Several systemic factors have been associated with differences in mathematics achievement across education systems. According to Meier and West (2020), South African classrooms often experience overcrowding, with class sizes averaging over 50 learners, which may reduce opportunities for formative feedback and individualised learner support. Teacher content knowledge and pedagogical content knowledge also vary across schools, particularly in Measurement and Geometry ( Bhagwonparsadh & Pule, 2024; Taylor, 2021). These factors have been identified in the South African literature as important contextual conditions that influence the teaching and learning of mathematics.

Studies of Singapore’s education system describe sustained teacher professional development, comparatively smaller class sizes, and strong instructional leadership as important features of its mathematics education system ( Choy & Dindyal, 2024; Low & Wong, 2021). The literature further describes the Concrete–Pictorial–Abstract (CPA) approach as a structured instructional strategy that supports learners’ progression from concrete representations to increasingly abstract mathematical reasoning ( Leong et al., 2015; Lutfi & Dasari, 2024). Although these instructional characteristics provide useful comparative insights, they operate within Singapore’s broader educational, cultural, and institutional context and should not be interpreted as direct explanations for differences observed in TIMSS achievement. Furthermore, although South Africa’s curriculum aims for comprehensive coverage, research has identified concerns regarding curriculum overload, with some studies suggesting that limited instructional time may constrain opportunities for learners to consolidate foundational mathematical knowledge ( Milne & Mhlolo, 2021). Collectively, this body of literature provides contextual evidence for interpreting domain-specific achievement patterns rather than establishing causal explanations for the differences observed between South Africa and Singapore.

1.4. Curriculum–Cognitive Alignment in International Research

International evidence further illustrates the relationship between curriculum objectives, classroom practices, and mathematics learner achievement. Yılmaz et al. (2021) found that, while mathematics curricula emphasised reasoning, textbook activities placed greater emphasis on application, creating a mismatch between intended and enacted cognitive emphases. Similarly, Bulut and Taşpınar-Şener (2023) reported that the Applying cognitive domain received greater emphasis than the Knowing and Reasoning domains within the mathematics curriculum, although this emphasis varied across grade levels. In primary schools, Pertiwi and Wahidin (2020) showed that fourth-grade assessments placed greater emphasis on number-related content, with comparatively less attention given to geometry and data.

These results align with the TIMSS assessment framework, where Knowing involves factual recall and procedural fluency, Applying requires learners to use mathematical knowledge in familiar contexts, and Reasoning involves non-routine problem-solving and critical thinking ( Mullis et al., 2020; Peduk & Ateş, 2019). International studies have also reported that curriculum alignment across content and cognitive domains is associated with stronger mathematics achievement, although the strength of these relationships varies across educational contexts. Viewed alongside this international evidence, South Africa’s TIMSS 2023 mathematics achievement profile shows comparatively higher average performance in the Applying domain than in the Knowing domain, while substantial achievement differences remain across all three cognitive domains. These patterns are consistent with findings reported in several international studies, although the present study does not examine the classroom or curricular mechanisms underlying these differences. In contrast, Singapore’s balanced performance across the three cognitive domains illustrates the level of cognitive proficiency demonstrated by a high-performing education system assessed on the same TIMSS framework, providing a useful benchmark for interpreting South Africa’s domain-specific achievement profile rather than evidence of direct curricular effects.

1.5. TIMSS: A Diagnostic Instrument

TIMSS provides a structured empirical basis for analysing domain-specific performance patterns across participating education systems using a common assessment framework, beyond broad structural explanations. The comparison with Singapore indicates that, although the two education systems are assessed on the same TIMSS Grade 4 mathematics framework, their average achievement profiles differ across both content and cognitive domains. These differences should be interpreted within the broader historical, cultural, socioeconomic, and educational contexts of the two countries rather than as evidence that any single curricular or instructional characteristic explains the observed achievement patterns. The substantially higher average achievement of Singapore’s Grade 4 learners, relative to South Africa’s Grade 5 learners, provides an analytical benchmark for examining domain-specific differences in mathematics achievement. Rather than attributing these differences to learner age or curriculum exposure alone, the comparison highlights the value of examining curriculum coherence, cognitive demands, instructional practices, and teacher development within their respective educational contexts, as documented in previous research.

The TIMSS 2023 results highlight not only the scale of South Africa’s average mathematics achievement relative to the benchmark, but also its domain-specific and cognitive achievement profile. While Singapore provides an example of a high-performing education system characterised in the literature by coherent curriculum design and sustained teacher professional development, South Africa’s mathematics achievement profile indicates comparatively lower average performance in foundational knowledge and Measurement and Geometry, alongside broader systemic challenges identified in the South African literature. Accordingly, the comparison supports context-sensitive reflection on curriculum and instructional priorities rather than direct adoption of another education system’s practices.

1.6. Conceptual Framework

This study is guided primarily by the TIMSS conceptual model, which distinguishes between mathematical content domains (Number, Measurement and Geometry, and Data) and cognitive domains (Knowing, Applying, and Reasoning) ( Mullis et al., 2020). Together, these dimensions provide a diagnostic lens for examining not only what learners are expected to know but also how they engage with mathematical tasks at increasing levels of cognitive demand. Rather than relying solely on overall average achievement, the framework enables the examination of domain-specific performance patterns across content and cognitive domains. Within this study, the TIMSS framework functions as the primary analytical structure, guiding the disaggregation of achievement results and the identification of domain-specific achievement gaps.

The study further employs Curriculum Alignment Theory, which emphasises the coherence between curricular intentions, classroom enactment, and assessment demands ( Porter, 2002). Curriculum Alignment Theory proposes that meaningful learning is more likely when the intended curriculum, the implemented curriculum, and the attained curriculum operate in concert. In this study, Curriculum Alignment Theory serves as the principal interpretive lens, supporting the interpretation of domain-specific achievement patterns identified through the TIMSS framework. The theory therefore provides a conceptual basis for examining how alignment between curriculum, instruction, and assessment is reflected in the existing literature, rather than for establishing causal explanations for the observed achievement differences.

International evidence from high-performing education systems such as Singapore describes strong curriculum alignment through a spiral curriculum that revisits mathematical concepts at progressively higher levels of complexity, supported by the Concrete–Pictorial–Abstract (CPA) approach, which scaffolds learning from concrete representations to abstract reasoning ( Leong et al., 2015; Lutfi & Dasari, 2024). In contrast, research on South African primary mathematics identifies an overloaded curriculum, limited instructional time for mastery of foundational domains, and persistent challenges in geometry and spatial reasoning, together with variation in teacher content knowledge and pedagogical confidence ( Maqoqa, 2024; Taylor, 2021). Viewed through the combined TIMSS and Curriculum Alignment frameworks, the comparison between South Africa’s Grade 5 and Singapore’s Grade 4 results provides a structured basis for interpreting differences in domain-specific mathematics achievement within a common international assessment framework.

To maintain conceptual clarity and parsimony, the study does not introduce additional theoretical models beyond these two complementary frameworks. The TIMSS model structures the measurement and domain disaggregation, while Curriculum Alignment Theory guides the interpretation of domain-specific achievement patterns. This dual framework ensures both diagnostic precision and interpretive coherence without theoretical fragmentation.

Figure 1 illustrates the integrated conceptual framework underpinning this study. The TIMSS framework structures the analysis of content and cognitive domains, while Curriculum Alignment Theory guides the interpretation of the resulting achievement patterns. Together, these complementary frameworks provide a diagnostic and interpretive basis for examining South Africa’s Grade 5 mathematics achievement in relation to the TIMSS 2023 assessment framework.

Figure 1. Conceptual framework integrating the TIMSS domain structure and Curriculum Alignment Theory guiding the diagnostic and interpretive analysis.

 Figure 1.

2. Methodology

2.1. Research Design

This study employed a quantitative secondary data analysis design, using data from the Trends in International Mathematics and Science Study (TIMSS) 2023. This design is particularly appropriate for investigating the mathematics achievement of South African Grade 5 learners across content and cognitive domains for several reasons. First, TIMSS provides a large, internationally standardised dataset that is both rigorous in design and nationally representative, making it suitable for examining learners’ performance patterns with a high degree of reliability. Second, secondary analysis enables the use of TIMSS’s robust psychometric procedures, including item response theory scaling and plausible value methodology, which strengthen the validity of inferences about learners’ achievement. Third, the TIMSS framework allows for meaningful international benchmarking, making it possible to situate South Africa’s performance in relation to high-performing education systems such as Singapore.

Guided by the TIMSS assessment framework, the analysis focused on three content domains (Number, Measurement and Geometry, and Data) and three cognitive domains (Knowing, Applying, and Reasoning). This two-fold focus enabled a diagnostic assessment of learners’ domain-specific achievement profiles within the South African context and supported structured benchmarking against an international reference system. Importantly, the South African sample reported here reflects the TIMSS 2023 Grade 5 administration using the Grade 4 mathematics assessment instruments and framework. Within TIMSS procedures, some education systems assess an adjacent grade when curriculum alignment indicates that the benchmark framework better reflects learners’ opportunity to learn. Consequently, the comparison undertaken in this study is not between different grade curricula, but between two cohorts assessed on the same TIMSS Grade 4 mathematics framework and common international reporting scale. The analysis therefore benchmarks Grade 5 in South Africa against Grade 4 in Singapore on an equivalent instrument, enabling comparison of performance patterns across TIMSS content and cognitive domains on the same scale metric ( Mullis et al., 2020; von Davier et al., 2024). It is important to emphasise that this design is descriptive and comparative rather than causal. The study identifies structured achievement gaps across domains but does not infer grade-level learning gains or causal mechanisms.

2.2. Participants

The South African TIMSS 2023 primary grade sample comprised 10,424 Grade 5 learners from 285 schools who were assessed using the TIMSS Grade 4 mathematics instruments, while Singapore assessed 6,530 Grade 4 learners from 181 schools using the same Grade 4 mathematics assessment framework ( Department of Basic Education, 2024; von Davier et al., 2024). The International Association for the Evaluation of Educational Achievement (IEA), in collaboration with Statistics Canada, used a two-stage stratified cluster sampling methodology to guarantee nationally representative estimates ( Siegel & Foy, 2024). In the first stage, schools were selected with probabilities proportional to their size, and in the second stage, intact Grade 5 classes were sampled. The stratification variables included the school sector (public or private), the language of instruction, the geographic region, socioeconomic indicators, the degree of urbanisation, and prior academic achievements (Ibid.). The use of stratified cluster sampling and population weights ensures that reported statistics represent national achievement distributions rather than sample-level estimates.

2.3. Data Collection and Analysis

Data for this study were drawn from the TIMSS 2023 mathematics assessment and associated contextual background questionnaires administered to participating learners, teachers, and schools. The TIMSS mathematics achievement is reported on an internationally standardised scale with a centre point of 500 and a standard deviation of 100, enabling valid comparisons across countries and education systems. The mathematics assessment comprised 183 items distributed across three content domains, namely Number (94 items), Measurement and Geometry (49 items), and Data (40 items), as well as three cognitive domains, namely Knowing (58 items), Applying (85 items), and Reasoning (40 items) ( Reynolds, 2024). Each learner completed one assessment booklet, with achievement estimates derived using item response theory and reported as plausible values. Plausible values represent multiple imputed proficiency estimates for each learner and are designed to produce unbiased population-level statistics rather than individual scores ( von Davier, 2020). This design supports reliable population-level estimation while accounting for measurement uncertainty inherent in large-scale assessment data.

The analysis focused on South Africa’s Grade 5 results, and Singapore’s Grade 4 performance was used as an international benchmark to contextualise domain-specific patterns of mathematics achievement. This comparison is consistent with TIMSS procedures, as both cohorts were assessed using the same Grade 4 mathematics framework and instruments, calibrated on a common international scale. Weighted descriptive statistics were computed in accordance with International Association for the Evaluation of Educational Achievement (IEA) guidelines to account for TIMSS’s two-stage stratified cluster sampling design. Sampling weights were applied to ensure nationally representative estimates and to correct for unequal probabilities of selection and non-response, thereby reducing bias in cross-national comparisons ( Siegel & Foy, 2024). All analyses incorporated the full set of plausible values, and estimates were combined using Rubin’s rules in accordance with TIMSS technical procedures to ensure accurate estimation of means and variability. In addition to significance testing, effect sizes (Cohen’s d) were calculated to evaluate the practical magnitude of observed differences between domains and between South Africa and Singapore. Effect sizes above 2.5 indicate extremely large educational disparities, reflecting substantial differences in performance between the two systems. Variance estimation incorporated the TIMSS replicate weights to account for the two-stage stratified cluster sampling design. Standard errors and confidence intervals were computed using these replicate weights, ensuring appropriate estimation of sampling variability. The combined use of statistical significance and effect size estimation enabled a more substantively meaningful interpretation of performance gaps, beyond reliance on mean differences alone.

Although TIMSS 2023 collects extensive contextual information through learner, teacher, school, and curriculum questionnaires, the present study prioritised a structured diagnostic comparison of domain-specific achievement profiles. Consequently, contextual variables such as socioeconomic status, language of instruction, school resources, and teacher characteristics were not incorporated into multivariate or multilevel models in the main analysis. The exclusion of these variables from inferential modelling was deliberate, given that the primary objective was to identify and quantify achievement gaps across content and cognitive domains rather than to estimate explanatory or causal effects. Instead, contextual factors are referenced interpretively in the discussion to situate the observed performance patterns within broader structural and policy contexts.

This analytic decision reflects a staged research logic: descriptive domain disaggregation precedes explanatory modelling. Such an approach is consistent with established practice in large-scale assessment research, where diagnostic profiling and causal modelling are treated as sequential and complementary phases of inquiry rather than simultaneous analytical requirements ( OECD, 2019; Rutkowski & Delandshere, 2016).

To ensure appropriate estimation of statistical uncertainty, all achievement comparisons were based on the full set of TIMSS plausible values. Estimates were combined using Rubin’s rules, as prescribed in IEA technical documentation, to account for measurement imputation variability inherent in plausible value methodology. Variance estimation was conducted in accordance with IEA technical guidelines for complex survey data. Weighted means were calculated using student sampling weights, and standard errors were derived using the TIMSS replicate weights to account for the two-stage stratified cluster sampling design.

Statistical inference was undertaken using these replicate-based variance estimates. Effect sizes (Cohen’s d) were reported alongside significance tests to provide an educationally meaningful interpretation of observed differences beyond statistical significance alone. Confidence intervals are reported in the supplementary materials to enhance transparency and facilitate replication of domain-level comparisons.

2.4. Ethical Considerations

This study used publicly available TIMSS 2023 datasets provided by the IEA. The data contain no personal identifiers and were collected under strict international ethical protocols during the original administration. Because this research involved secondary analysis of anonymised data, no institutional ethics approval was required. There was no formal request to use the dataset from the IEA since the data is available in the public domain, and all analyses adhered to its guidelines for responsible data use.

3. Results

This section presents the empirical results addressing Research Questions 1 and 2. Research Question 1 examines South Africa’s achievement profile across TIMSS mathematics content domains, while Research Question 2 examines achievement across cognitive domains. For the purposes of this study, an “achievement gap” is operationally defined as (a) the scale score distance from the TIMSS international centre point of 500, and (b) the benchmark difference between South African Grade 5 learners and Singaporean Grade 4 learners assessed on the same TIMSS Grade 4 mathematics framework. Effect sizes (Cohen’s d) are used to indicate the magnitude of these disparities.

3.1. Content Domain Achievement

In response to Research Question 1, the analysis reveals that South African Grade 5 learners scored substantially lower average mathematics achievement than Singaporean Grade 4 learners across all three TIMSS content domains: Number, Measurement and Geometry, and Data. Relative to the TIMSS international centre point (500), South Africa’s average achievement in each content domain remained well below the international reference point. These results indicate consistently lower average achievement across all three content domains rather than isolated weaknesses in a single area. Among the three domains, Measurement and Geometry recorded the largest benchmark gap, with the greatest scale score difference between South Africa and Singapore and very large effect sizes (d = 2.66), indicating an extremely large practical difference. Table 1 presents the weighted mean scores and effect sizes for South African Grade 5 learners and Singaporean Grade 4 learners across the three TIMSS mathematics content domains.

Table 1. Mean Scale Scores and Effect Sizes in Mathematics Content Domains, TIMSS 2023.

Content domain Singapore (Grade 4) South Africa (Grade 5) Gap Cohen’s d
Numbers 613 362 251 2.51
Measurement & Geometry 619 353 266 2.66
Data 616 362 254 2.54

Note. All estimates are weighted. Standard errors and confidence intervals were computed using TIMSS plausible values and replicate weights and are reported in the supplementary materials.

All three content domains produced significant large effect sizes (d > 2.5), demonstrating substantial benchmark differences between South African and Singaporean learners. Although Measurement and Geometry recorded the largest observed benchmark gap, all three domains showed pronounced differences in average achievement. Within the South African achievement profile, Measurement and Geometry recorded the lowest average score, whereas Number and Data recorded slightly higher, but still substantially lower, average achievement relative to the Singapore benchmark.

3.2. Cognitive Domain Achievement

Addressing Research Question 2, the results indicate that South African Grade 5 learners scored substantially lower average mathematics achievement than Singaporean Grade 4 learners across all three TIMSS cognitive domains: Knowing, Applying, and Reasoning. Among the three cognitive domains, Knowing recorded the largest benchmark gap, with a scale score difference of 267 points and the largest effect size (d = 2.67). Applying and Reasoning also recorded substantial benchmark differences, with effect sizes of 2.49 and 2.46, respectively. Table 2 presents the weighted mean scores and effect sizes for South African Grade 5 learners and Singaporean Grade 4 learners across the TIMSS mathematics cognitive domains.

Table 2. Mean Scale Scores and Effect Sizes in Mathematics Cognitive Domains, TIMSS 2023.

Cognitive domain Singapore (Grade 4) South Africa (Grade 5) Gap Cohen’s d
Knowing 624 357 267 2.67
Applying 615 366 249 2.49
Reasoning 609 363 246 2.46

Note. All estimates are weighted. Standard errors and confidence intervals were computed using TIMSS plausible values and replicate weights and are reported in the supplementary materials.

Within the South African achievement profile, Applying recorded the highest average score (366), followed by Reasoning (363) and Knowing (357). Despite these relative differences, average achievement across all three cognitive domains remained substantially below both the TIMSS international centre point and the Singapore benchmark. Although Applying recorded the highest average achievement within the South African profile, the benchmark differences remained substantial across all cognitive domains . The Knowing domain recorded the largest observed benchmark gap, while Applying and Reasoning also demonstrated pronounced differences relative to Singapore.

3.3. Item-Level Illustrations

To further illustrate the domain-specific patterns identified in Research Questions 1 and 2, selected released TIMSS items are used to demonstrate how content and cognitive differences are reflected at the task level. In the Knowing domain (Number), more than 90% of Singaporean learners correctly answered an item requiring the recall of basic multiplication facts, compared with fewer than 40% of South African learners. This example is consistent with the comparatively larger benchmark gap observed in the Knowing domain. In the Applying domain (Data), routine tasks involving the interpretation of a simple bar chart were answered correctly by a greater proportion of South African learners than items in the Knowing and Reasoning domains, consistent with the relatively higher average score recorded for Applying within the South African achievement profile. In the Reasoning domain (Measurement and Geometry), items requiring multi-step reasoning involving angle relationships were answered correctly by fewer than 20% of South African learners, compared with a majority of Singaporean learners. This example reflects the substantial benchmark difference observed in tasks requiring mathematical reasoning within geometric contexts. Collectively, these released items illustrate how the benchmark differences identified in the scale scores are reflected across tasks representing different mathematics content and cognitive domains. The examples provide illustrative support for the quantitative results by demonstrating variation in learners’ success across tasks with different content and cognitive demands.

3.4. Visualising the Gaps

Figure 2 illustrates the comparative performance of South African Grade 5 and Singaporean Grade 4 learners across the three TIMSS mathematics content domains. The figure shows consistently lower average achievement for South African learners across Number, Measurement and Geometry, and Data. Among the three content domains, Measurement and Geometry recorded the largest benchmark gap, although substantial differences were evident across all domains. The graphical presentation complements the results reported in Table 1 by illustrating the magnitude and distribution of the benchmark differences across the content domains.

Figure 2. Comparative performance of South African Grade 5 and Singaporean Grade 4 learners across TIMSS content domains (Number, Measurement & Geometry, and Data).

 Figure 2.

Figure 3 illustrates the comparative performance of South African Grade 5 and Singaporean Grade 4 learners across the three TIMSS cognitive domains. The figure shows that the Knowing domain recorded the largest benchmark gap, while Applying and Reasoning also exhibited substantial benchmark differences. Although Applying recorded the highest average score within the South African cognitive profile, average achievement remained substantially below both the TIMSS international centre point and the Singapore benchmark. Together, Figures 2 and 3 visually reinforce the quantitative findings presented in Tables 1 and 2 by illustrating the distribution of benchmark differences across the mathematics content and cognitive domains.

Figure 3. Comparative performance of South African Grade 5 and Singaporean Grade 4 learners across TIMSS mathematics cognitive domains (Knowing, Applying, and Reasoning).

 Figure 3.

3.5. Summary of Results

Synthesising the findings for Research Questions 1 and 2, South African Grade 5 learners recorded substantially lower average mathematics achievement than Singaporean Grade 4 learners across all TIMSS mathematics content and cognitive domains. Among the content domains, Measurement and Geometry recorded the largest benchmark gap, while the Knowing domain recorded the largest benchmark gap among the cognitive domains. Although Applying recorded the highest average score within the South African cognitive profile, substantial benchmark differences remained evident across all three cognitive domains. Overall, the results demonstrate that benchmark differences were observed across all mathematics content and cognitive domains, with comparatively larger differences in Measurement and Geometry and Knowing.

4. Discussion

This section interprets the domain-specific and cognitive achievement patterns identified in Section 3 in relation to curriculum alignment and learning progression. The TIMSS 2023 results indicate that South African Grade 5 learners achieved an average score of 362, substantially below the international centre point of 500. However, the analytical emphasis of this study lies not only in the overall average score, but in the concentration and distribution of achievement gaps across specific content and cognitive domains. Disaggregating achievement reveals that the largest content-domain gap occurs in Measurement and Geometry, while the largest cognitive-domain gap occurs in Knowing. Although substantial achievement differences were observed across all content and cognitive domains, these two domains exhibited the largest benchmark gaps. These domain-specific disparities provide a more precise diagnostic basis for interpreting patterns of mathematics learning and progression in the primary phase.

4.1. Content-Domain Patterns and Foundational Gaps

The largest benchmark gap was observed in Measurement and Geometry, where effect sizes indicated an extremely large disparity relative to Singapore. Although Measurement and Geometry exhibited the largest benchmark gap, substantial differences were evident across all three content domains. This finding indicates that the South African mathematics achievement profile reflects broad domain-specific disparities rather than weakness confined to a single content area. While performance in Number and Data also remains substantially below international benchmark levels, the relative ordering of domain means indicates that achievement varies across content areas. The comparatively larger gap in Measurement and Geometry aligns with national research identifying geometry as a persistent area of difficulty in South African classrooms, where studies have reported variation in teacher content knowledge, pedagogical content knowledge, and opportunities for conceptual engagement ( Maqoqa, 2024; Taylor, 2021).

International studies have also associated the development of spatial reasoning with coherent instructional progression and sustained conceptual engagement ( Choy & Dindyal, 2024). Viewed alongside this body of literature, the TIMSS findings suggest that Measurement and Geometry remains an important area for further curriculum and instructional attention in South Africa. Importantly, these results do not suggest an absence of geometry instruction but rather highlight patterns that are consistent with possible differences between intended curriculum, classroom enactment, and assessment expectations. Consistent with Curriculum Alignment Theory ( Porter, 2002), differences between these dimensions may be reflected in learners’ achievement profiles across content domains. The present study, however, does not directly examine curriculum implementation or classroom practice. Consequently, the comparatively larger disparity observed in Measurement and Geometry should be interpreted as a diagnostic indicator within the broader mathematics achievement profile rather than as evidence of a single underlying cause. Together with the substantial differences observed in Number and Data, these findings highlight the importance of strengthening learners’ mathematical development across all content domains while recognising the comparatively greater challenge presented by Measurement and Geometry.

4.2. Cognitive-Domain Performance and Learning Progression

The largest cognitive achievement gap was observed in the Knowing domain. South African learners scored substantially lower than Singaporean learners (357 versus 624), with a very large effect size. Although achievement differences were evident across all three cognitive domains, the Knowing domain exhibited the largest benchmark gap. Because the Knowing domain captures factual recall and procedural fluency ( Mullis et al., 2020), the comparatively lower performance in this domain indicates that foundational mathematical knowledge warrants particular attention within South Africa’s overall achievement profile. Within the TIMSS framework, foundational fluency provides the basis for progression to Applying and Reasoning tasks. Viewed alongside this framework, the comparatively lower achievement in Knowing is consistent with the broader pattern of performance observed across the cognitive domains. Although Applying represents a relatively stronger domain within the South African distribution, it remains substantially below international benchmark levels. This pattern indicates variation in achievement across the cognitive domains rather than uniformly low performance.

Previous research has associated secure foundational knowledge with learners’ ability to engage successfully in increasingly complex mathematical tasks ( Mullis & Martin, 2017). Consistent with this literature, the present findings suggest that strengthening foundational mathematical knowledge may support learners’ progression across higher cognitive demands. However, the present study does not directly examine the instructional or curricular processes through which such progression occurs. Similar patterns have been reported in large-scale assessment studies, where learners demonstrate partial procedural competence but experience greater difficulty with tasks requiring conceptual integration and multi-step reasoning. Taken together, these findings describe a cognitive achievement profile characterised by comparatively lower performance in foundational knowledge alongside substantial differences across all three cognitive domains. Rather than identifying causal mechanisms, the results provide a diagnostic profile that may inform future research and curriculum development aimed at strengthening mathematical learning progression.

4.3. Curriculum Alignment and Cross-National Insights

The benchmark comparison with Singapore provides a useful context for interpreting domain-specific achievement patterns through the lens of Curriculum Alignment Theory. The literature describes Singapore’s mathematics curriculum as characterised by a tightly sequenced spiral structure and the systematic use of the Concrete–Pictorial–Abstract (CPA) progression, which supports continuity in learners’ mathematical development across content and cognitive domains ( Leong et al., 2015). Within the Singaporean context, these curriculum features have been associated with coherent learning progression and balanced achievement across mathematical domains. Conversely, research on South African primary mathematics has identified concerns regarding curriculum breadth, limited instructional time for consolidating foundational concepts, and persistent challenges in domains such as Measurement and Geometry ( Maqoqa, 2024; Taylor, 2021). These contextual factors have been discussed in the literature as potential influences on mathematics teaching and learning; however, the present study does not directly examine their effects on learner achievement.

Curriculum Alignment Theory ( Porter, 2002) proposes that meaningful learning is more likely when the intended, implemented, and assessed curricula are coherently aligned. Viewed through this theoretical perspective, the comparatively larger achievement gaps observed in Knowing and Measurement and Geometry are consistent with possible differences in curriculum alignment across these dimensions. However, because the present study analyses secondary TIMSS data, it cannot determine the extent to which curriculum alignment or classroom implementation contributed to the observed achievement patterns. Importantly, the comparison should not be interpreted as evidence that Singapore’s curriculum or instructional approaches directly account for its higher achievement. Rather, Singapore provides an analytical benchmark against which South Africa’s domain-specific achievement profile can be interpreted within a common international assessment framework. Accordingly, the comparison offers comparative insight into achievement patterns while recognising the distinct historical, cultural, socioeconomic, and educational contexts within which the two education systems operate.

4.4. Interpreting Singapore as a Benchmark

While Singapore’s performance provides a valuable benchmark for examining curriculum coherence and learning progression, it is essential to recognise the contextual conditions within which this performance is situated. Singapore’s education system operates within a highly structured and competitive environment, characterised by strong central curriculum control, selective teacher recruitment, intensive professional preparation, and high societal expectations regarding academic achievement ( Ng, 2017). The literature associates these systemic characteristics with consistent instructional practices and sustained learner engagement, alongside high levels of mathematics achievement across content and cognitive domains. At the same time, these features are accompanied by early differentiation and sustained academic pressure, which, although associated with high achievement, may also generate stress and equity concerns that are not fully captured in large-scale assessment data ( Tan, 2018). This broader context highlights the importance of interpreting benchmark comparisons within their socio-educational settings, rather than attributing performance differences solely to instructional or curricular factors.

The value of Singapore as a comparator lies in providing a well-established international reference point against which South Africa’s achievement profile can be interpreted, rather than serving as a model for direct policy transfer ( Deng, 2013). Within the context of this study, Singapore functions as an analytical benchmark that helps contextualise the domain-specific achievement patterns observed in South Africa. For South Africa, the findings highlight the importance of considering curriculum coherence, focused content progression, and pedagogical scaffolding in ways that are responsive to the country’s educational context, systemic capacity, linguistic diversity, and resource constraints ( Spaull & Kotze, 2015). Accordingly, the comparison supports context-sensitive reflection on mathematics curriculum and instructional priorities rather than wholesale adoption of practices from another education system. The benchmark therefore functions as a comparative reference point that helps to situate the scale and distribution of domain-specific disparities, while recognising the distinct educational contexts in which the two systems operate.

4.5. Implications for Policy and Practice

The results identify a dual pattern in South African learners’ mathematics achievement. First, substantial achievement differences are evident across all content and cognitive domains, with comparatively larger gaps observed in foundational knowledge and Measurement and Geometry . Second, the relatively stronger performance in the Applying domain indicates comparatively higher average achievement within the South African profile, although performance remains substantially below the international benchmark. Together, these findings indicate variation in achievement across domains rather than uniform mathematics underperformance. From a Curriculum Alignment perspective, these domain-specific patterns are consistent with the possibility of differences between curriculum expectations, instructional practices, and assessment demands, as discussed in the literature. The present study, however, does not directly examine curriculum implementation or classroom practice and therefore cannot determine the extent to which these factors contributed to the observed achievement patterns. Within this context, the findings highlight the importance of strengthening curriculum coherence, instructional sequencing, and opportunities for developing both foundational knowledge and conceptual understanding.

The findings also suggest that efforts to improve mathematics achievement should extend across all content and cognitive domains, while recognising the comparatively larger disparities identified in Measurement and Geometry and the Knowing domain. Accordingly, the results support a balanced approach to curriculum and instructional improvement rather than prioritising a single domain in isolation. More broadly, South Africa has made important progress in expanding access to education and improving average mathematics achievement over successive TIMSS cycles. Nevertheless, substantial inequalities associated with socioeconomic conditions, language, school resources, and educational opportunities continue to shape learners’ mathematics achievement. In addition, the educational disruption associated with the COVID-19 pandemic may have influenced learning opportunities for the cohort assessed in TIMSS 2023 and should therefore be considered when interpreting these findings.

Taken together, these findings indicate that South African mathematics achievement is characterised by substantial domain-specific variation rather than uniformly low performance. The diagnostic achievement profile presented in this study provides evidence that may inform curriculum review, teacher professional development, and future research aimed at strengthening mathematics learning in the primary phase. Interpreted through Curriculum Alignment Theory, these patterns are consistent with possible differences between intended curriculum, classroom enactment, and assessment expectations. Rather than identifying causal mechanisms, the study provides a structured analytical basis for understanding how achievement varies across content and cognitive domains within the TIMSS 2023 assessment framework.

5. Conclusion

This study examined domain-specific mathematics achievement among South African Grade 5 learners using TIMSS 2023 data by comparing performance across mathematics content and cognitive domains with Singapore as an analytical benchmark. By disaggregating achievement across these domains, the analysis demonstrated that substantial benchmark differences were evident across all mathematics content and cognitive domains, with comparatively larger differences observed in Measurement and Geometry and the Knowing domain. Rather than relying solely on overall mathematics scores, the study provides a structured diagnostic profile that identifies how achievement varies across specific areas of mathematical knowledge and cognitive demand.

The findings contribute to the literature by demonstrating the value of domain-specific analyses for understanding patterns of mathematics achievement within an internationally comparable assessment framework. Interpreted through Curriculum Alignment Theory, the results are consistent with the importance of coherence between curriculum expectations, instructional practice, and assessment in supporting mathematics learning. However, the study does not identify the underlying causes of the observed benchmark differences. Instead, it provides an empirical basis for informing curriculum review, teacher professional development, and future research aimed at strengthening mathematics learning within the South African context. By positioning Singapore as an analytical benchmark rather than a model for direct policy transfer, the study contributes context-sensitive evidence to comparative mathematics education research and highlights the value of domain-level analyses for informing educational improvement.

6. Limitations and Future Research

This study provides a structured diagnostic profile of domain-specific mathematics achievement using TIMSS 2023 data; however, several limitations should be acknowledged. First, the analysis is based on cross-sectional secondary data and therefore supports descriptive and comparative interpretation rather than causal inference. The benchmark differences reported in this study describe patterns of achievement between South African Grade 5 learners and Singaporean Grade 4 learners assessed using the common TIMSS Grade 4 mathematics framework, but they do not explain the processes that produced these differences.

Second, the mathematics content and cognitive domain subscales contain fewer assessment items than the overall mathematics scale, which may reduce the precision of domain-level estimates. Although the use of plausible values and replicate-based variance estimation enhances the robustness of the analyses, statistical uncertainty remains inherent in domain-level comparisons and should be considered when interpreting the reported effect sizes.

Third, TIMSS is designed to measure learner achievement rather than classroom processes or curriculum implementation. Consequently, the study cannot directly examine instructional practices, curriculum enactment, teacher quality, learner engagement, or other contextual factors that may be associated with domain-specific achievement patterns. Interpretations informed by Curriculum Alignment Theory therefore remain theoretical rather than empirically tested within the present study.

Finally, although TIMSS employs nationally representative sampling procedures, certain educational contexts may remain under-represented, particularly small, remote, or multigrade schools. In addition, differences in national exclusion rates should be considered when interpreting benchmark comparisons, as the assessed populations in South Africa and Singapore may not be directly equivalent. Furthermore, the comparison with Singapore should be interpreted as an analytical benchmark within a common international assessment framework rather than as evidence supporting direct policy transfer between education systems.

Future research would benefit from longitudinal, multilevel, and mixed-methods approaches that combine large-scale assessment data with classroom observations, curriculum analyses, teacher data, and learner experiences to examine how domain-specific achievement patterns develop over time and how they may be addressed through contextually appropriate instructional and curriculum interventions.

Author Contributions

The author, Mathelela Steyn Mokgwathi, is responsible for the conceptualisation, data curation, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualisation, and writing of the original draft, as well as the writing, review, and editing processes.

Declaration of AI Use

The author affirms that no generative artificial intelligence tools (such as ChatGPT or similar models) were used to produce the academic content, analysis, or interpretations presented in this manuscript. QuillBot (premium) was employed solely for grammar and spelling checks. The author personally reviewed and edited the final manuscript and takes full responsibility for its content and conclusions.

Data Availability

The data that support the results of this article are derived from the publicly accessible Trends in International Mathematics and Science Study (TIMSS) 2023 database, available through the International Association for the Evaluation of Educational Achievement (IEA) at https://timssandpirls.bc.edu. Derived and processed data underlying the statistical analyses, including the variables used to generate the tables, figures, and descriptive statistics presented in this article, are openly available in the Zenodo repository ( Mokgwathi, 2025) at https://doi.org/10.5281/zenodo.17379412, under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence. These files include anonymised values underlying the means, standard deviations, and effect sizes, as well as detailed variable descriptions and methodological notes supporting replication of the comparative analyses between South Africa (Grade 5) and Singapore (Grade 4). No ethical approval or participant consent was required, as all data originate from secondary sources that are publicly available through the IEA TIMSS 2023 repository.

Extended Data

Supplementary materials supporting this article are available in the Zenodo repository at https://doi.org/10.5281/zenodo.17379412 ( Mokgwathi, 2025), under the CC BY 4.0 license. These include:

TIMSS2023_SGvZA_Derived_Dataset.xlsx (domain-level statistics and effect sizes),

TIMSS2023_Variable_Descriptions.xlsx (variable definitions and sources), and

TIMSS2023_Supplementary_Notes.pdf (methodological documentation).

Acknowledgements

The author acknowledges the International Association for the Evaluation of Educational Achievement (IEA) for making the Trends in International Mathematics and Science Study (TIMSS) 2023 data publicly available and appreciates its continued commitment to advancing educational research globally.

Funding Statement

The author(s) declared that no grants were involved in supporting this work.

[version 4; peer review: 1 approved

References

  1. Bhagwonparsadh Y, Pule K: The effects of educators’ mathematics pedagogical content knowledge on the mentoring of Grade 12 students using problem-based learning. J Culture Values Edu. 2024;7(1):99–117. 10.46303/jcve.2024.7 [DOI] [Google Scholar]
  2. Bulut AS, Taşpınar-Şener Z: Analysis of Secondary School Mathematics Curriculum Learning Outcomes by TIMSS-2019 Cognitive Domain Skills. Türk Akademik Yayınlar Dergisi (TAY Journal). 2023;7(1):303–328. 10.29329/tayjournal.2023.537.14 Reference Source [DOI] [Google Scholar]
  3. Choy BH, Dindyal J: The Singapore mathematics curriculum: Influences and confluences.In Rata E, editor. Research Handbook on Curriculum and Education. Edward Elgar Publishing;2024; pp.534–549. 10.4337/9781802208542.00048 [DOI] [Google Scholar]
  4. Deng Z: The Singapore school curriculum: Unpacking the system, building the nation. Asia Pacif J Educ. 2013;33(3):324–337. 10.1080/02188791.2013.786679 [DOI] [Google Scholar]
  5. Department of Basic Education: South Africa 2023 Trends in International Mathematics and Science Study (TIMSS) Highlights Report. Pretoria: Department of Basic Education;2024. [Google Scholar]
  6. Grundy S: Beyond guaranteed outcomes: Creating a discourse for educational praxis. Aust J Educ. 1992;36(2):157–169. 10.1177/000494419203600204 [DOI] [Google Scholar]
  7. Kilpatrick J, Swafford J, Findell B: Adding it up: Helping children learn mathematics. National Academy Press;2001. 10.17226/9822 [DOI] [Google Scholar]
  8. Leong YH, Ho WK, Cheng LP: Concrete-Pictorial-Abstract: Surveying its origins and charting its future. Math Educ. 2015;16(1):1–18. Reference Source [Google Scholar]
  9. Low L, Wong LF: Using connecting mathematical tasks for coherence, connections and continuity.In Toh TL, Choy BH, editors. Mathematics—Connection and Beyond: Yearbook 2020 Association of Mathematics Educators. World Scientific Publishing;2021; pp.95–120. 10.1142/9789811236983_0006 [DOI] [Google Scholar]
  10. Lutfi JS, Dasari D: Research trends on learning mathematics with the CPA (concrete-pictorial-abstract) approach. PRISMA. 2024;13(1):103–112. 10.35194/jp.v13i1.3947 [DOI] [Google Scholar]
  11. Mabena N, Mokgosi PN, Ramapela SS: Factors contributing to poor learner performance in mathematics: A case of selected schools in Mpumalanga province, South Africa. Probl Educ 21st Cent. 2021;79(3):451–466. 10.33225/pec/21.79.451 [DOI] [Google Scholar]
  12. Maqoqa T: An exploration of learners’ understanding of Euclidean geometric concepts: A case study of secondary schools in the OR Tambo Inland District of the Eastern Cape. E-J Human Arts Soc Sci. 2024;5(5):658–675. 10.38159/ehass.2024557 [DOI] [Google Scholar]
  13. Meier C, West J: Overcrowded classrooms – The Achilles heel of South African education? South African J Childhood Educ. 2020;10(1):1–10. 10.4102/sajce.v10i1.617 [DOI] [Google Scholar]
  14. Mensah E, Baidoo-Anu D: Towards quality and equitable education in South Africa: Unpacking the relationship between teacher factors, students’ socioeconomic background and mathematics achievements. Res Math. 2022;9(1). 10.1080/27684830.2022.2088645 [DOI] [Google Scholar]
  15. Milne A, Mhlolo M: Lessons for South Africa from Singapore’s gifted education – A comparative study. S. Afr. J. Educ. 2021;41(1):1–8. 10.15700/saje.v41n1a1839 [DOI] [Google Scholar]
  16. Mokgwathi MS: Derived Data and Extended Tables: TIMSS 2023 South Africa–Singapore Mathematics Comparison (1.0).[Data set]. Zenodo. 2025. 10.5281/zenodo.17379412 [DOI]
  17. Morony W: Coherence in a range of mathematics curriculum reforms.In Shimizu Y, Vithal R, editors. Mathematics curriculum reforms around the world: The 24th ICMI Study. Springer International Publishing;2023; pp.127–150. 10.1007/978-3-031-13548-4_10 [DOI] [Google Scholar]
  18. Mullis IVS, Martin MO, editors: TIMSS 2019 Assessment Frameworks. TIMSS & PIRLS International Study Center;2017. Retrieved from Boston College. Reference Source [Google Scholar]
  19. Mullis IVS, Martin MO, Foy P, et al. : TIMSS 2019 International Results in Mathematics and Science. TIMSS & PIRLS Int Study Center;2020. Reference Source [Google Scholar]
  20. Ng PT: Learning from Singapore: The power of paradoxes. Phi Delta Kappan. 2017;98(8):43–48. 10.1177/0022057417712323 [DOI] [Google Scholar]
  21. OECD : PISA 2018 Assessment and Analytical Framework. Paris: PISA, OECD Publishing;2019. 10.1787/b25efab8-en [DOI] [Google Scholar]
  22. Peduk B, Ateş Ö: Analysis of the science course curriculum objectives and high school entrance examination questions according to TIMSS framework. Int J Educ Methodol. 2019;5(3):433–449. 10.12973/ijem.5.3.433 [DOI] [Google Scholar]
  23. Pertiwi A, Wahidin W: Are the mathematics textbooks for eighth-grade meet the Trends in International Mathematics and Science Study (TIMSS) 2019 mathematics framework? Edum J Riset Pendidikan Matematika. 2020;3(2):129–135. 10.32939/ejrpm.v3i2.623 [DOI] [Google Scholar]
  24. Porter AC: Measuring the Content of Instruction: Uses in Research and Practice. Educ Res. 2002;31(7):3–14. 10.3102/0013189X031007003 [DOI] [Google Scholar]
  25. Reddy V, Hannan S: TIMSS in South Africa: Making global research locally meaningful. Human Sci Res Council. 2019. [Google Scholar]
  26. Reynolds KA: Updating the TIMSS instruments for describing contexts of student learning.In Davier M, Fishbein B, Kennedy A, editors. TIMSS 2023 technical report (methods and procedures). Boston College, TIMSS & PIRLS International Study Center;2024; pp.2.1–2.8. Reference Source [Google Scholar]
  27. Rutkowski D, Delandshere G: Causal inferences with large-scale assessment data: Using a validity framework. Large-scale Assessm Educ. 2016;4(1):6. 10.1186/s40536-016-0019-1 [DOI] [Google Scholar]
  28. Siegel P, Foy P: TIMSS sample design.In Davier M, Fishbein B, Kennedy A, editors. TIMSS 2023 Technical Report (Methods and Procedures). Boston College, TIMSS & PIRLS International Study Center;2024; pp.3.1–3.30. 10.6017/lse.tpisc.timss.rs7952 [DOI] [Google Scholar]
  29. Spaull N, Kotze J: Starting behind and staying behind in South Africa: The case of insurmountable learning deficits in mathematics. Int J Educ Dev. 2015;41:13–24. 10.1016/j.ijedudev.2015.01.002 [DOI] [Google Scholar]
  30. Tachie SA: The challenges of South African teachers in teaching Euclidean geometry. Int J Learn Teach Educ Res. 2020;19(8):297–312. 10.26803/ijlter.19.8.16 [DOI] [Google Scholar]
  31. Tan C: Learning from Shanghai, learning from Singapore: The paradox of Singapore’s curriculum reforms. J Curric Stud. 2018;50(5):659–677. 10.1177/1478210317742224 [DOI] [Google Scholar]
  32. Taylor N: The dream of Sisyphus: Mathematics education in South Africa. South African J Childhood Educ. 2021;11(1):1–12. 10.4102/sajce.v11i1.911 [DOI] [Google Scholar]
  33. Davier M: TIMSS 2019 scaling methodology: Item Response Theory, population models, and linking across modes.In Martin MO, Davier M, Mullis IVS, editors. Methods and procedures: TIMSS 2019 Technical Report. TIMSS & PIRLS International Study Center;2020; pp.11.1–11.25. Reference Source [Google Scholar]
  34. Davier M, Kennedy A, Reynolds K, et al. : TIMSS 2023 international results in mathematics and science. Boston College: TIMISS & PIRLS International Study Center;2024. [Google Scholar]
  35. Yılmaz N, Ay Z, Aydın Ş: An investigation of tasks in the mathematics textbooks and objectives in mathematics curriculum from 4th to 8th grade related with data content domain according to TIMSS 2019 cognitive domains. Cukurova Univ Faculty Educ J. 2021;50(2):1397–1436. 10.14812/cuefd.745164 [DOI] [Google Scholar]
  36. Zuze L, Reddy V, Visser M, et al. : TIMSS 2015 Grade 9 national report: Understanding mathematics and achievement amongst Grade 9 learners in South Africa. Human Sci Res Council. 2018. Reference Source [Google Scholar]
F1000Res. 2026 Sep 15. doi: 10.5256/f1000research.207186.r520589

Reviewer response for version 4

MARGUERITE KHAKASA MIHESO O'CONNOR 1

The authors have responded satisfactorily to the suggestions and have made commendable efforts to improve the manuscript. The revisions have enhanced the overall coherence and clarity of the article. While the use of generic terms to explain the gaps identified in the results would have been strengthened by incorporating more empirical evidence, this limitation does not substantially detract from the overall quality and coherence of the revised manuscript. I am therefore satisfied with the authors’ responses and recommend the article for Approval status.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Yes

Are the conclusions drawn adequately supported by the results?

Yes

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Mathematics Education and teacher Education

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

F1000Res. 2026 Jul 1. doi: 10.5256/f1000research.196841.r490388

Reviewer response for version 3

Jennie Golding 1

This paper addresses the very, very important issue of South African mathematics performance at Grade 5, as reflected in TIMSS 2023 data and particularly in comparison with that of Grade 4 in the highest-performing system in TIMSS 2023, Singapore. While aiming high is usually commendable, it would be good to have a rationale for aiming quite so high when, as the author notes, the Singapore education system context is very different from that prevalent in South Africa (Indeed, additionally, several authors suggest the prevailing Confucian ethos adds significantly to that context).

It is valuable to have analysis of South Africa’s TIMSS 2023 performance by content and by cognitive domain: South Africa does not appear to include that in their national report at   TIMSS 2024.pdf whereas a number of other systems do, and e.g.   https://tinyurl.com/yx622ea5 shows performance in Grade 4 content and cognitive domains across a number of different countries. (Note, though, that the centre point and SD of TIMSS scores were those established as mean and SD in the first iteration of TIMSS, and the SD varies each cycle, though the centrepoint remains fixed, and separate from the mean).

It is good also to have technical aspects of TIMSS assessments explained, though sources should always be quoted.

However, if performances are to be compared then the differences in testing should also be acknowledged: the TIMSS 2023 International Results at https://www.iea.nl/publications/timss-2023-international-report state that South African learners completed paper-based assessment comprised of TIMSS 2019 trend items in at least one grade (so no curriculum-matched scores ae available, though the same source states that there is a 100% South Africa curriculum match with the relevant TIMSS Grade 4 mathematics assessment framework); and that South Africa assessed students in their fifth and ninth years of schooling to better match curricula and maintain trend measurement (which clarifies the rationale for Grade 5 participation). 

Similarly, the paper very largely refers to average performance (and including that word ‘average’ each time would strengthen rigour). The national report already identifies significant disparities by gender (with girls significantly outperforming boys, on average) and an average 111-point gap between those who “Always” speak the LOLT at home (405) and those who “Never” do (294), as well as very large gaps in average scores by socio-economic background. Note also that Singapore’s exclusion rate was >> that in South Africa. Taken together, these considerations   contribute to a very different distribution of scores in South Africa than the distribution seen among the (non-excluded) assessed pupils in Singapore.

However, the comparison across the two countries remains fairly simple, focusing on gaps between average scores in each (content and cognitive domain). The patterns seen in South Africa (relatively low scores in Measurement and Geometry, and in Knowing) are not uncommon across education systems: see e.g. https://tinyurl.com/yx622ea5  - and yes, probably at least in part reflect a mismatch between the intended curriculum and the enacted, or received, curriculum. However, in each domain, the difference between focus average performance in South Africa and that in Singapore amounts of 2.5-2.7, using Cohen’s d to 1 dp, which is a very substantial difference of a magnitude broadly comparable across domains.

In terms of response by that measure, the focus for action needs to be across the curriculum, rather than primarily in just one content and just one cognitive domain (and of course there is a swathe of evidence from across the globe that points to the desirability of learning development across content and cognitive domains).

Given the starkness of the contrast between reported average scores in South Africa and those in Singapore, it would be useful to report also the statistical significance of the differences across domains within South Africa in terms of each of content and cognition, which might give a more realistic steer as to the urgency of a particular focus within mathematics learning.

The author very rightly points to the necessity of adapting apparently more successful approaches from other contexts, but greater criticality in the use of international literature would further strengthen the paper, e.g. curriculum papers are usually focused on one particular curriculum, set of curriculum materials, classroom practices etc. – so cannot unproblematically be transferred across very different cultural and contextual systems. It’s also important to avoid implying causation without due regard to rigour: for example ‘shaping’ suggests causality and that needs to be justified if it is claimed. Further, all claims such as those around curriculum system responses should be referenced (together with critique of the related evidence in relation to transferability to the South African context).

Finally, the focus of this paper is the comparison with Singapore, pointing to significant potential for enhancement of learning in South Africa. However, it is also important to acknowledge the enormous progress made in recent years, post-apartheid, to give more equitable access to a broad and balanced mathematics curriculum across young people in South Africa, with the related challenges to resources and in particular, the equipping of teachers in all communities. TIMSS 2023 assessments also took place in the shadow of the global pandemic. The TIMSS 2024.pdf national report recognises the sustained and significant impact of the pandemic on learning across much of South Africa, not least because of very limited digital infrastructure to support continued learning. Learners in Singapore also suffered pandemic impact, as evidenced by the drop in average scores from TIMSS 2019, but evidence from Singapore suggests a pivot to digital or hybrid learning was more widely achieved.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Yes

Reviewer Expertise:

International large scale studies in mathematics and science; mathematics education policy-practice interface; mathematics teacher education.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

F1000Res. 2026 Jul 23.
Mathelela Steyn Mokgwathi 1

Dear Dr Golding,

I would like to sincerely thank you for your comprehensive, thoughtful, and constructive review of this manuscript. Your comments have been invaluable in strengthening the study, particularly regarding its comparative framing, methodological clarity, and interpretation of the findings. I greatly appreciate the time and expertise you devoted to reviewing the manuscript.

In response to your recommendations, I have substantially strengthened the rationale for benchmarking South African Grade 5 learners against Singaporean Grade 4 learners by explaining the TIMSS procedures that permit assessment of adjacent grades using the common TIMSS Grade 4 mathematics framework. The manuscript now makes it clear that Singapore is used as an analytical benchmark within a common international assessment framework rather than as a model for direct policy transfer. Throughout the manuscript, I have also acknowledged the considerable historical, cultural, socioeconomic, and educational differences between the two countries and emphasised that the findings should be interpreted within their respective contexts.

The Introduction and Literature Review have been revised to provide a more balanced discussion of South Africa's mathematics achievement, recognising the significant progress made since the advent of democracy while also acknowledging the persistent inequalities associated with socioeconomic conditions, language, school resources, and educational opportunities. I have also incorporated discussion of the potential influence of the COVID-19 pandemic on the TIMSS 2023 cohort to provide a more comprehensive contextual interpretation of the findings.

The manuscript has been carefully revised to enhance methodological precision and interpretive rigour. References to mathematics performance have been standardised to average mathematics achievement where appropriate, and causal language has been removed or moderated throughout the manuscript. Expressions that could imply causal inference have been replaced with more appropriate terminology that reflects the descriptive and comparative nature of the study. In addition, the Methods section has been expanded to clarify the analytical procedures employed, including the use of plausible values, sampling weights, replicate-based variance estimation, and effect-size estimation in accordance with IEA technical guidelines.

The Results, Discussion, and Conclusion have also been substantially revised. The Results section now focuses exclusively on descriptive reporting of the empirical findings, while interpretation has been reserved for the Discussion. The manuscript now consistently emphasises that substantial benchmark differences were observed across all mathematics content and cognitive domains, with comparatively larger differences in Measurement and Geometry and the Knowing domain, thereby avoiding undue emphasis on a single domain. The Discussion has been strengthened by critically engaging with the international literature and by positioning Curriculum Alignment Theory as an interpretive framework rather than as evidence of causal relationships. The Conclusion has likewise been revised to ensure that all conclusions are directly supported by the reported findings.

Finally, the Limitations section has been expanded to acknowledge the descriptive nature of the study, the constraints of cross-sectional secondary data, the limitations inherent in domain-level comparisons, differences in national exclusion rates, and the need to interpret the comparison with Singapore as an analytical benchmark rather than evidence supporting direct policy transfer.

I trust that these revisions have addressed the concerns raised and have strengthened the manuscript in terms of methodological transparency, interpretive balance, and its contribution to comparative mathematics education research.

Thank you once again for your insightful and constructive review.

Kind regards,

Dr Mathelela Steyn Mokgwathi

F1000Res. 2026 Apr 13. doi: 10.5256/f1000research.196841.r472940

Reviewer response for version 3

Fidele Ukobizaba 1,2

The authors have made notable improvements to the manuscript compared to the previous version. Most of the earlier concerns have been addressed, and the manuscript is now generally readable. However, some issues still require revision:

  1. The third research objective is overly lengthy and lacks sufficient focus. Furthermore, it is unclear how this objective is operationalized in the results section, as no corresponding findings are presented. Considering this and given the absence of aligned results, it is recommended that this objective be removed.

  2. Following the revision of Comment 1, the statement in the first paragraph of results section: “Research Question 3, concerning curriculum and pedagogical implications, is addressed in the Discussion section” should also be deleted.

  3. The discussion section should explicitly interpret and synthesize the study findings in relation to relevant literature. Accordingly, Sections 4.6 and 4.7 should be removed, as they do not directly engage with or interpret the study results.

  4. A strong conclusion should sound clear, concise, and synthesis-oriented. It should not introduce new ideas, nor discuss the results, and should not include citations. Instead, it should bring the study to a logical close by highlighting meaning, contribution, and implications.

  5. The limitations section appears excessively extensive, which may weaken the perceived robustness of the study. It is recommended that the authors consolidate this section into two concise paragraphs.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Yes

Are all the source data underlying the results available to ensure full reproducibility?

Partly

Is the study design appropriate and is the work technically sound?

No

Are the conclusions drawn adequately supported by the results?

No

Are sufficient details of methods and analysis provided to allow replication by others?

No

Reviewer Expertise:

Mathematics education.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

F1000Res. 2026 Apr 15.
Mathelela Steyn Mokgwathi 1

Dear Dr Ukobizaba,

I would like to thank you for your constructive feedback and for recognising the improvements made in the manuscript. In this revision, I have carefully addressed the remaining concerns you raised. The third research question has been removed to ensure alignment between the study objectives and the empirical findings, and all references to it in the Results section have been deleted.

The Discussion section has been substantially revised to strengthen the interpretation and synthesis of findings in relation to relevant literature. Sections 4.6 and 4.7 have been removed, and the discussion now focuses on analytically engaging with the results rather than presenting prescriptive content. In addition, the Conclusion has been rewritten to ensure that it is concise, synthesis-orientated, and free from citations or new arguments.

Finally, the Limitations section has been consolidated into two concise paragraphs to improve clarity and balance while retaining key methodological considerations.

I trust that these revisions address the remaining concerns and further strengthen the clarity, coherence, and scholarly contribution of the manuscript.

Kind regards,

F1000Res. 2026 Feb 17. doi: 10.5256/f1000research.194776.r453643

Reviewer response for version 2

Fidele Ukobizaba 1,2

The study addresses an important and timely issue, “Closing the gap in early mathematics”, which is commendable. However, the manuscript has several substantial concerns that should be addressed to strengthen its quality and clarity.

The introduction does not clearly identify or elaborate on the specific gaps in early mathematics among South African students. As a result, the purpose and direction of the study are not sufficiently clear to the reader. The focus and motivation of the study should be made explicit earlier in the manuscript. In this regard, think about the first line of the conclusion (which may be the aim of the study), where it is written: “This study examined South African Grade 5 learners’ mathematics achievement in TIMSS 2023, disaggregated by content and cognitive domains and benchmarked against Singapore’s Grade 4 performance.”

The research questions are currently too broad and appear to target multiple aspects simultaneously, resulting in a lack of focus. For greater coherence and rigor, the research questions should be more specific and precisely formulated.

The manuscript draws on three theoretical frameworks. This may dilute the conceptual focus of the study. The authors are encouraged to consider using one framework, or at most two that clearly complement each other, to provide a more coherent theoretical grounding.

Research design raises concerns. The rationale for comparing Grade 5 South African students with Grade 4 Singaporean students is not adequately justified, as these groups are not at the same educational level. Comparisons should ideally be made between equivalent grade levels. Moreover, comparing two groups’ results alone is insufficient to identify learning gaps in mathematics. The authors should consider an alternative design or provide a stronger justification for their approach.

The results section primarily reports differences in performance between the two groups (effect size) but does not clearly identify or analyze “gaps” in learning. Given the title, “Closing the gap in early mathematics…,” the manuscript should explicitly identify where the gaps exist, describe their nature, explore possible reasons for them, and suggest ways these gaps could be addressed.

In the discussion section, it is written: “This section addresses Research Question 3 by interpreting domain-specific and cognitive performance patterns in relation to curriculum alignment, pedagogy, and teacher professional development and outlining implications for improving mathematics achievement in South Africa.” However, this section should be discussing the results in line with the research questions. Additionally, this section should more explicitly connect the findings to existing literature and prior research to situate the study within the broader scholarly conversation.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Yes

Are all the source data underlying the results available to ensure full reproducibility?

Partly

Is the study design appropriate and is the work technically sound?

No

Are the conclusions drawn adequately supported by the results?

No

Are sufficient details of methods and analysis provided to allow replication by others?

No

Reviewer Expertise:

Mathematics education.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

F1000Res. 2026 Feb 19.
Mathelela Steyn Mokgwathi 1

Reviewer Comments

Response to Comments

The introduction does not clearly identify or elaborate on the specific gaps in early mathematics among South African students. The purpose and direction of the study are not sufficiently clear.

The introduction has been revised to explicitly state the purpose of the study earlier and to clearly identify the specific achievement gaps motivating the analysis. The revised text now highlights that the largest content-domain gap occurs in Measurement and Geometry and the largest cognitive-domain gap occurs in Knowing. The study aim is explicitly articulated in the opening section, clarifying the diagnostic intent of disaggregating achievement across content and cognitive domains.

The research questions are too broad and lack focus.

The research questions were refined for precision and coherence. Research Questions 1 and 2 now focus specifically on content-domain and cognitive-domain profiling, while Research Question 3 is limited to curriculum and pedagogical implications derived from the empirical findings. This ensures a clear alignment between Results (RQ1 and RQ2) and Discussion (RQ3).

The manuscript draws on three theoretical frameworks, which may dilute conceptual focus.

The conceptual framework has been streamlined to retain two complementary frameworks: the TIMSS domain framework and Curriculum Alignment Theory (Porter, 2002). Additional theoretical layering was removed to improve conceptual clarity. Figure 1 was redesigned to reflect this streamlined structure.

The rationale for comparing Grade 5 South African learners with Grade 4 Singaporean learners is not adequately justified. Comparisons should ideally be between equivalent grade levels.

The methodology section now clearly explains that both cohorts were assessed using the same TIMSS Grade 4 mathematics framework and reporting scale. The comparison is instrument-equivalent rather than curriculum-equivalent. The manuscript explicitly clarifies that the design is descriptive and comparative, not causal, and is intended to benchmark performance patterns rather than infer grade-level learning differences.

Comparing two groups alone is insufficient to identify learning gaps.

The Results section now operationalises “achievement gap” explicitly as (a) deviation from the TIMSS international centre point and (b) benchmark scale-score differences with associated effect sizes. The analysis identifies the location, magnitude, and concentration of domain-specific gaps rather than merely reporting mean differences.

The Results section reports effect sizes but does not clearly analyse gaps in learning.

The Results section was restructured to explicitly identify where gaps exist, describe their magnitude, and distinguish between uniform underperformance and concentrated domain-specific weaknesses. Item-level illustrations were included to demonstrate how gaps manifest in foundational fluency and spatial reasoning tasks.

The Discussion should more explicitly connect findings to the research questions and prior literature.

The Discussion was reorganised to mirror the empirical findings and to explicitly anchor each subsection in the identified content and cognitive gaps. Connections to prior South African and international research were strengthened, and interpretations are framed within the retained theoretical structure.

Is the study design appropriate and technically sound? (Reviewer indicated “No”)

The Method section was strengthened by explicitly detailing the use of plausible values, Rubin’s rules for combining estimates, replicate weights for variance estimation, complex sampling adjustments, and effect size interpretation. The manuscript now clearly states that the analysis is descriptive and non-causal.

Are sufficient details provided to allow replication? (Reviewer indicated “No”)

Additional methodological transparency was incorporated. The manuscript now specifies estimation procedures, variance computation methods, and reporting of confidence intervals. Derived datasets and methodological documentation have been deposited in an open-access repository to support replication.

Are the conclusions adequately supported by the results? (Reviewer indicated “No”)

The Conclusion was revised to align strictly with the empirical findings. Causal overreach was removed, reform priorities are directly grounded in identified domain-specific gaps, and claims are framed proportionately to the descriptive comparative design.

F1000Res. 2026 Apr 15.
Mathelela Steyn Mokgwathi 1

Dear Dr Ukobizaba,

I would like to thank you for your detailed and constructive feedback, which has significantly strengthened this manuscript. In this revision, I have clarified the research gap in the Introduction by explicitly identifying the limited focus on domain-specific mathematics achievement at the primary level and by stating the purpose of the study more clearly at an earlier stage. The research questions have been refined and reduced to ensure greater focus and alignment with the study design. In addition, the conceptual framework has been streamlined by adopting curriculum alignment theory as the primary analytical lens, with the TIMSS framework providing the structure for domain-level analysis.

I have also strengthened the justification for comparing South African Grade 5 learners with Singaporean Grade 4 learners by explaining the use of the TIMSS Grade 4 assessment framework and ensuring comparability on a common measurement scale. The Results section has been revised to explicitly identify and analyse achievement gaps across content and cognitive domains, rather than focusing only on differences in performance.

Furthermore, the Discussion section has been restructured to align directly with the research questions, remove references to the previously included third research question, and strengthen the integration of findings with existing literature. The Methods section has been expanded to provide clearer detail on analytical procedures, including the handling of plausible values, sampling weights, and variance estimation, to enhance transparency and reproducibility. Finally, the conclusion has been rewritten to ensure that it is fully supported by the findings and provides a clear and concise synthesis of the study’s contribution.

I trust that these revisions address the concerns raised and improve the overall clarity, methodological rigour, and contribution of the study.

Kind regards,

F1000Res. 2025 Dec 30. doi: 10.5256/f1000research.189692.r435305

Reviewer response for version 1

MARGUERITE KHAKASA MIHESO O'CONNOR 1

 Review  report

Introduction

The article addresses an important and timely question in mathematics education: domain-specific patterns of achievement in TIMSS 2023, comparing South African Grade 5 learners with Singaporean Grade 4 learners. The topic is relevant and well grounded providing an opportunity for policy and practice impact in South Africa and the wider Sub-Saharan region.

The research purpose and alignment with TIMSS conceptual frameworks are quite clear. Descriptive statistical analysis and accurate identification of domain-specific weaknesses is evident.  There is observed an effective use of TIMSS as an international benchmark to highlight systemic performance gaps. The integration of literature from both high-performing (Singapore) and lower-performing (South Africa) systems is provided in a coherent narrative that links content and cognitive deficits to curriculum and instructional factors.

However several conceptual and methodological limitations  which currently reduce the manuscript’s rigor. have been identified. Revisions are required, particularly in theoretical framing, TIMSS- specific methodology, and the depth of interpretation and contextualization. Some minor corrections have also been identified for authors attention for proof reading

The  following observations have been made and related. action points  suggested for consideration

A): Presentation

The article  is generally well structured, articulate, and contextually grounded. The introduction to the article provides a clear rationale for focusing on South African primary mathematics achievement using TIMSS 2023 and situates the work within persistent national under performance. The performance gap (- 250 points) between South African Grade 5 and Singaporean Grade 4 learners is clearly stated and framed as a systemic issue rather than a purely learner-level deficit. The need to move beyond overall scores to content and cognitive domain analysis is well argued and aligned with TIMSS reporting structures. In addition, the contribution of the study, especially the focus on Grade 5 as an under-researched phase relative to Grade 9, is well articulated

The following issues were noted in the presentation

The rationale  of benchmarking South African Grade 5 learners against Singaporean Grade 4 learners though  methodologically common in TIMSS is not well explained. The clarification of why Singapore Grade 4 is used as a comparator is necessary . Secondly ,the article introduces and sustains a misaligned reference to Early Childhood Education (ECE). This introduction at one point frames the work as focusing on Early Childhood Education, which is conceptually inaccurate for TIMSS Grade 5. Grade 5 sits in the primary/intermediate phase, not ECE.. In addition, the conceptual rationale for the  theoretical framing of content and cognitive domains did not recognize frameworks such as: TIMSS assessment framework, Kilpatrick et al.’s strands of mathematical proficiency and  theories of cognitive demand and learning progression,. Finally, there is need to clarify  the research gap.  It is stated that Grade 5 and domain-specific analyses are under-researched, but this gap is not supported by specific citations and a brief synthesis of existing work.. 

As a result, the author is encouraged  to Action as follows

  • Clearly justify the benchmarking of South Africa Grade 5 against Singapore Grade 4.

  • Remove or revise all references to ECE; and  use more accurate terms  

  • Strengthen theoretical framing by linking content and cognitive domains to recognized frameworks (e.g., mathematical proficiency, cognitive demand, learning trajectories).

  • Clarify the research gap with citations that show the relative neglect of Grade 5 and domain-specific analyses.

  • Streamline the introduction by reducing redundant statements and improving paragraph transitions

 B: Study Design and Data Sources.

The use of quantitative secondary analysis of TIMSS 2023 data is appropriate and well aligned with the research questions. The sample sizes and the national representativeness of the TIMSS samples justifies the generalizability of the findings. The recognition of  TIMSS as a well-established international assessment enables valid cross-national comparisons. The description of the test design is broadly accurate and shows familiarity with large-scale assessment methodology.

 The following issues were observed

The  description of TIMSS technical procedures in the article is incomplete.  Although TIMSS design is mentioned, the article does not explicitly state how plausible values (PVs) were handled in the analysis. These details will help to confirm that TIMSS technical standards were followed.

The analysis does not explain whether or how contextual variables collected during the study (SES, language of instruction, teacher qualifications, opportunity-to-learn indicators) were used. Given TIMSS’s strong emphasis on context, this is a missed opportunity for explanatory depth.

For the psychometric and reliability considerations, there is need for reference to internal consistency, While TIMSS provides validated items, secondary analyses typically reference psychometric adequacy or refer explicitly to IEA technical documentation. Inclusion of some justification of Grade-Level Comparison is missing .   A rationale for South Africa testing at Grade 5 (rather than Grade 4) for the primary cycle should be explicitly noted to strengthen the conceptual legitimacy of the comparison with Singapore Grade 4.

The following action points are suggested

  • Ex plicitly describe the use of plausible values (number of PVs, which domains, how handled in analysis).

  • Detail the weighting strategy , including sampling weights and replicate weights, and explain how the two-stage sampling structure was accounted for.

  • Indicate whether contextual variables  (SES, language, school resources, teacher characteristics) were included in any extended analysis; if not, explain this limitation.

  • Reference psychometric evidence  (e.g., IEA technical reports) to reassure readers about reliability and validity of domain scores.

  • Clarify the grade-level choice  for South Africa and how it affects comparability with Singapore Grade 4.

 C: Methodology and Statistical Analysis The categorization of content domains (Number, Measurement & Geometry, Data) and cognitive domains (Knowing, Applying, Reasoning) is appropriate and consistent with TIMSS. The use of weighted estimates, mean differences, and effect sizes (Cohen’s d) is methodologically  suitable for characterizing performance gaps.. 

The following issues were observed:

To avoid biased estimates, domain-level comparisons in TIMSS require appropriate PV handling and replication methods. It is not clear whether the analysis uses domain-specific plausible values correctly 

For the variance estimation and significance testing, it is does not specify whether standard errors, confidence intervals, or p-values were computed. Descriptive statistics without uncertainty estimates weaken inferential robustness. The author had limited use of advanced modelling. The analysis is largely descriptive (means and effect sizes)..Given the richness of TIMSS data, multivariate methods (e.g., regression, multilevel models, mediation with contextual variables, DIF analyses) could substantially deepen the insights. While Cohen’s D  is reported, the interpretation of effect sizes in context is limited. There is little discussion of what constitutes a small/medium/large effect in educational terms., 

Suggested action points

  • Specify analytical tools and procedures used to handle PVs and weights (software, settings, treatment of domains).

  • Report standard errors and/or confidence intervals for key estimates and clarify whether hypothesis tests were performed.

  • Where feasible, incorporate multivariate or multilevel analyses to examine how contextual variables relate to domain-specific performance.

  • Interpret effect sizes in educationally meaningful terms, not just numerically.

D: Interpretation of Findings, Discussion, and Conclusion

The discussion is comprehensive and analytically rich. Content and cognitive domain findings and linking them to curriculum, pedagogy, and systemic factors is integrated well. Content-level weaknesses  and cognitive weaknesses are clearly interpreted and consistent with previous research .Policy and practice implications including pillars of reform around foundational knowledge, geometry/spatial reasoning, and leveraging application skills to cultivate reasoning is well structured. The conclusion offers a strong synthesis, links to broader goals such as SDG 4, and emphasizes curriculum-pedagogy-assessment coherence through Curriculum Alignment Theory

Issues observed included  the need for greater engagement with learning progression theory, to deepen the explanatory power.

Some  claims require evidence or qualification statements such as “procedural accuracy over conceptual exploration” or “curriculum spreads content thinly” need either citation or more cautious phrasing.

The claim that learners are “stronger in applying” requires quantification and statistical validation .

ECE References  in this section should also be expunged and corrected .

 To maintain academic balance, the discussion on using  Singapore as a model should acknowledge potential constraints such as high academic pressure, streaming, selective teacher recruitment, and cultural expectations.

Some weaknesses (e.g., geometry and knowing) are repeated in multiple sections affecting  structure   In Pillar(3) three  repetition  is noted, The discussion is not using empirical evidence and relies of vague quantifiers. The structure of the discussion could be sharpened with clearer subheadings and more concise phrasing.

Some recommendations (e.g., reducing class sizes) may be politically or financially challenging; indicating alternative or incremental strategies would strengthen the practical relevance.

The conclusion would benefit from a brief statement of study limitations (e.g., reliance on secondary data, limited use of contextual variables, cross-sectional design) and directions for future research.

Action points suggested

  • Deepen theoretical engagement by explicitly drawing on mathematical proficiency and learning progression frameworks.

  • Support key claims with evidence (citations or data) and quantify statements about “relative strength” in applying.

  • Remove or reframe ECE references, using appropriate phase terminology.

  • Provide a more nuanced view of Singapore, noting contextual constraints and non-transferable features.

  • Consolidate repeated material on geometry/knowing weaknesses and improve subheadings and paragraph transitions.

  • Comment on feasibility of policy recommendations and suggest realistic pathways or phases of implementation.

  • Add a short section on study limitations and future research directions

  •  replace vague quantifiers with measurable empirical evidence

 Summary of Key Action Points for Revision

Conceptual and Theoretical

Clarify the benchmarking rationale (Singapore Grade 4 vs South Africa Grade 5).

Correct ECE misalignment and use appropriate phase terminology.

Strengthen theoretical framing using TIMSS, mathematical proficiency, and learning progression frameworks.

Provide a more balanced portrayal of Singapore, acknowledging contextual constraints.

Methodological and Statistical

Explicitly describe handling of plausible values, sampling weights, and replicate weights.

Report standard errors, confidence intervals, and significance tests for key comparisons.

Consider multivariate or multilevel analyses using contextual variables

Interpret effect sizes substantively, not just numerically.

Context and Interpretation

Integrate contextual questionnaire data (SES, language, resources, teacher factors) where possible.

Link performance patterns explicitly to systemic factors (teacher knowledge, curriculum pacing, language of learning and teaching)

Ensure claims are evidence-based and appropriately referenced.

Structure and Presentation  >>minor corrections

Streamline the introduction and discussion, removing redundancy and improving coherence.

Strengthen section headings and transitions for easier navigation.

Include tables or figures with domain-specific means, SEs, and effect sizes

Add a concise section on limitations and future research.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Yes

Are the conclusions drawn adequately supported by the results?

Yes

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Mathematics Education and Teacher Education

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

F1000Res. 2026 Jan 1.
Mathelela Steyn Mokgwathi 1

I would like to thank Reviewer 2 for the detailed and constructive feedback concerning theoretical framing, TIMSS-specific methodology, interpretation of domain-level findings, and the balance and structure of the Discussion and Conclusion. These concerns have been addressed through strengthened integration of mathematical proficiency and learning progression frameworks, expanded methodological transparency regarding plausible values and variance estimation, refined interpretation of effect sizes, removal of misaligned ECE references, and a more nuanced treatment of Singapore as a benchmark. We believe the revisions comprehensively address the issues identified.

F1000Res. 2026 Apr 15.
Mathelela Steyn Mokgwathi 1

Dear Dr O’Connor,

I would like to thank you for your detailed and constructive feedback, which has contributed significantly to strengthening this manuscript. In this revision, I have clarified the benchmarking rationale by providing a detailed explanation of the use of the TIMSS Grade 4 framework and ensuring comparability between South African Grade 5 and Singaporean Grade 4 learners. I have also strengthened the articulation of the research gap, removed misaligned references to Early Childhood Education, and refined the theoretical framing to improve conceptual clarity.

In addition, I have expanded the methodological section to provide clearer explanations of TIMSS procedures, including the handling of plausible values, sampling weights, and variance estimation, and clarified the descriptive and diagnostic focus of the study in relation to the use of contextual variables and modelling approaches. The interpretation of effect sizes has been strengthened, and claims have been revised to ensure they are appropriately supported and cautiously framed.

The discussion has been revised to enhance analytical depth, reduce prescriptive elements, improve structural coherence, and better integrate the findings with existing literature, while also providing a more balanced interpretation of the Singapore comparison. A concise limitations section has also been included to reflect key design and contextual constraints.

I trust that these revisions address the concerns raised and improve the overall clarity, rigour, and contribution of the study.

Kind regards,

F1000Res. 2025 Dec 30. doi: 10.5256/f1000research.189692.r435300

Reviewer response for version 1

Rashmi Khazanchi 1

The manuscript titled “Closing the Gap in Early Mathematics: Domain and Cognitive Insights from TIMSS 2023 in South Africa and Singapore” addresses an important topic by examining gaps in elementary-level mathematics learning through a comparison of TIMSS results from the two countries. While the study’s focus and findings are potentially valuable, the manuscript contains several major methodological and conceptual issues that need to be addressed.

The author compares TIMSS results of Grade 4 learners from Singapore with Grade 5 learners from South Africa; however, TIMSS assesses learners at Grades 4 and 8 only. The manuscript therefore needs to clarify how Grade 5 results were derived from TIMSS data and justify the rationale for benchmarking Grade 4 and Grade 5 learners. The research Question 1 may need to be revised, as TIMSS does not directly assess Grade 5 learners.

Several in-text citations are missing throughout the manuscript as statement should be supported with appropriate citations. On page 3, the title of the literature review (“Comparative analysis of South African and Singaporean Grade 5 mathematics achievement”) requires correction and should be revised consistently throughout the paper, as it implies that the comparison is limited solely to Grade 5 mathematics achievement.

In the conceptual framework section, the author introduces an integrative framework combining a comparative lens, the TIMSS curriculum model, and curriculum theory dimensions; however, the sentence presenting this framework is a paragraph long and reduces clarity. To enhance clarity, the author needs to explain the role of each framework component, articulate the directional relationships among key constructs, and strengthen the narrative link to Figure 1.

In the results section, findings should be discussed in direct relation to the research questions rather than beginning with tables or figures. The discussion section would benefit from further elaboration and stronger integration with the literature reviewed earlier, as well as more explicit connections to the research questions to enhance clarity and readability.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

No

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Artificial Intelligence in Education, AI - based tutoring systems, Mathematics Achievement, Student Engagement.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to state that I do not consider it to be of an acceptable scientific standard, for reasons outlined above.

F1000Res. 2025 Dec 30.
Mathelela Steyn Mokgwathi 1

Dear Reviewer,

Thank you for your careful reading of the manuscript and for the detailed and constructive feedback provided. I appreciate the time and scholarly attention devoted to the review, and I am grateful for the suggestions, which have been invaluable in strengthening the clarity, methodological transparency, and overall quality of the paper. I respond to each of the major points raised below.

With regard to the comment concerning the comparison between Grade 4 learners in Singapore and Grade 5 learners in South Africa, I appreciate the opportunity to clarify this important methodological issue. While TIMSS is internationally benchmarked at Grades 4 and 8, it is also well documented that several participating countries, including South Africa, assess learners at Grade 5 using the Grade 4 TIMSS assessment instruments. This approach is adopted where curriculum alignment indicates that the Grade 4 framework more accurately reflects the instructional exposure of learners, a practice that parallels the assessment of Grade 9 learners using the Grade 8 TIMSS instruments in some contexts. In the revised manuscript, I will make this rationale explicit by clarifying that the South African Grade 5 learners were assessed using the standard TIMSS Grade 4 mathematics framework and instruments, thereby ensuring construct equivalence and international comparability. In response to the reviewer’s suggestion, Research Question 1 will be refined to explicitly reflect this assessment design and to avoid any implication that TIMSS directly assesses a distinct Grade 5 framework.

Concerning the observation that several in-text citations are missing, I acknowledge this limitation. A thorough review of the manuscript is currently underway to ensure that all empirical claims and theoretical statements are appropriately supported by relevant and up-to-date scholarly sources. These revisions will improve academic rigour and align with best practices in scholarly writing.

Regarding the comment on the literature review title on page 3, I appreciate the reviewer’s attention to precision in terminology. The study intentionally focuses on Grade 5 learners in South Africa, assessed using the Grade 4 TIMSS instrument, and compares their performance with Grade 4 learners in Singapore, assessed using the same framework. To avoid ambiguity, the literature review title and related headings will be revised for consistency and clarity, ensuring that they accurately reflect the comparative focus and assessment design of the study, rather than implying a narrow or exclusive emphasis on Grade 5 achievement alone.

I am also grateful for the insightful feedback on the conceptual framework section. I acknowledge that the current presentation of the integrative framework is overly dense. In the revised manuscript, this section will be restructured to clearly explain the role of each component, namely the comparative education lens, the TIMSS curriculum model, and the curriculum theory dimensions. The directional relationships among the key constructs will be explicitly articulated, and the narrative will be more tightly aligned with Figure 1 to enhance conceptual coherence and reader accessibility.

Finally, I appreciate the guidance regarding the results and discussion sections. In response, the results will be reorganised so that the findings are presented explicitly in relation to the research questions before introducing supporting tables and figures. The discussion section will be expanded to provide deeper engagement with the literature reviewed earlier in the manuscript and to draw clearer, more explicit links between the findings, the research questions, and the broader scholarly debates on early mathematics achievement and international benchmarking.

Once again, I sincerely thank the reviewer for these constructive comments. I am confident that addressing these points will substantially strengthen the manuscript and enhance its contribution to the literature on comparative mathematics education.

F1000Res. 2026 Jan 1.
Mathelela Steyn Mokgwathi 1

I would like to thank Reviewer 1 for highlighting critical concerns regarding grade-level comparability, clarity of the benchmarking rationale, citation completeness, and alignment between the research questions, results, and interpretation. These issues have been carefully addressed through clarifications of TIMSS grade participation, revisions to the conceptual framing and research questions, strengthened citation support, and restructuring of the Results and Discussion sections. We believe the revised manuscript now fully addresses the raised concerns.

F1000Res. 2026 Apr 15.
Mathelela Steyn Mokgwathi 1

Dear Dr Khazanchi,

I would like to thank you for your constructive and insightful feedback, which has significantly strengthened this manuscript. In this revision, I have clarified the rationale for benchmarking South African Grade 5 learners with Singaporean Grade 4 learners by explicitly explaining the use of the TIMSS Grade 4 framework and ensuring measurement equivalence. I have also refined the research questions to align with the TIMSS design and streamlined the conceptual framework to improve clarity and coherence.

In addition, I have strengthened the literature support throughout the manuscript, revised section headings for consistency, and restructured the Results and Discussion sections to ensure clear alignment with the research questions and stronger integration with existing literature. The conclusion has been rewritten to provide a concise synthesis fully grounded in the findings.

I trust that these revisions address the concerns raised and improve the overall clarity, rigour, and contribution to the study.

Kind regards,

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Citations

    1. Mokgwathi MS: Derived Data and Extended Tables: TIMSS 2023 South Africa–Singapore Mathematics Comparison (1.0).[Data set]. Zenodo. 2025. 10.5281/zenodo.17379412 [DOI]

    Data Availability Statement

    The data that support the results of this article are derived from the publicly accessible Trends in International Mathematics and Science Study (TIMSS) 2023 database, available through the International Association for the Evaluation of Educational Achievement (IEA) at https://timssandpirls.bc.edu. Derived and processed data underlying the statistical analyses, including the variables used to generate the tables, figures, and descriptive statistics presented in this article, are openly available in the Zenodo repository ( Mokgwathi, 2025) at https://doi.org/10.5281/zenodo.17379412, under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence. These files include anonymised values underlying the means, standard deviations, and effect sizes, as well as detailed variable descriptions and methodological notes supporting replication of the comparative analyses between South Africa (Grade 5) and Singapore (Grade 4). No ethical approval or participant consent was required, as all data originate from secondary sources that are publicly available through the IEA TIMSS 2023 repository.

    Extended Data

    Supplementary materials supporting this article are available in the Zenodo repository at https://doi.org/10.5281/zenodo.17379412 ( Mokgwathi, 2025), under the CC BY 4.0 license. These include:

    TIMSS2023_SGvZA_Derived_Dataset.xlsx (domain-level statistics and effect sizes),

    TIMSS2023_Variable_Descriptions.xlsx (variable definitions and sources), and

    TIMSS2023_Supplementary_Notes.pdf (methodological documentation).


    Articles from F1000Research are provided here courtesy of F1000 Research Ltd

    RESOURCES