This chapter examines students' confidence in 21st-century mathematics and their level of exposure to these specific tasks. It explores both the relationship between exposure and students' confidence, and between confidence and the use of specific learning strategies and motivation to learn. The chapter delves into the analysis of four specific key areas that together enhance students' ability to apply mathematical concepts in diverse and practical ways, preparing them for the complexity of modern challenges: representing a situation mathematically, extracting mathematical information, interpreting mathematical solutions in the context of a real-life challenge, and programming computers, which involves writing code to solve mathematical problems or simulate real-world phenomena.
PISA 2022 Results (Volume V)
Abstract
For Australia*, Canada*, Denmark*, Hong Kong (China)*, Ireland*, Jamaica*, Latvia*, the Netherlands*, New Zealand*, Panama*, the United Kingdom* and the United States*, caution is advised when interpreting estimates because one or more PISA sampling standards were not met (see Reader’s Guide, Annexes A2 and A4).
Previous PISA analyses revealed a significant decline in student performance over the past decade in science and reading, and a large drop between 2018 and 2022 in mathematics (OECD, 2023[1]). The decline suggests an erosion of skills that are essential for personal and professional development and which constitute the cornerstone of further learning. The downward trend has significant implications for basic education and lifelong learning, and calls for a reassessment of educational priorities and methodologies to meet the demands of the 21st century.
As mathematics was the main domain of assessment in PISA 2022, concepts and analyses in this report are directed towards understanding how students learn mathematics. This allows us to explore learning strategies, motivations, and self-beliefs related to mathematics education across education systems. This chapter focuses on how prepared students are to acquire the skills necessary for lifelong learning in the 21st century, providing a comprehensive view of how mathematical proficiency supports future learning and adaptation in a rapidly evolving world.
While mathematics is just one part of the larger picture, this report’s findings can feed into policies that encompass broader lifelong learning contexts. It is imperative that policy makers and education systems prioritise the development of a robust foundation upon which future learning can be built. This adaptation involves not only revamping curricula but adopting innovative teaching methodologies that incorporate strategies for sustained learning in the 21st century (OECD, 2019[2]). In this way, education systems can better equip students to face new and unforeseen challenges, and ensure that they have the skills and knowledge necessary for lifelong learning (OECD, 2021[3]; OECD, 2019[4]).
Students who engage in proactive behaviour such as relating new material to previously learned content, who diligently check that they have understood what is being taught, and who report being cognitively activated are most likely to be confident in their 21st-century mathematics skills.
However, frequency of exposure to 21st-century mathematics tasks is, overall, low. Less than a third of students frequently represent situations mathematically, for instance, and only one in five apply mathematical solutions to real-life situations. The mathematics task students are least exposed to is coding or programming computers. Education systems who want to prepare students for the technological demands of the modern workforce may want to look into these aspects.
Simply increasing the frequency of tasks is unlikely to be enough to build confidence. Other aspects are also important, including the motivation to learn, as in enjoying challenging schoolwork. This kind of motivation to learn could be a powerful component of confidence in 21st-century mathematics.
Student confidence in 21st-century mathematics varies across countries and economies, with around 70% or more of students being confident in extracting or representing mathematical information in Canada*, France and the United States* but no more than 40% in Japan and Thailand. Because of its positive relationship to learning strategies and outcomes, the integration of real-world applications and technological literacy into 21st-century mathematics education can help students develop a more robust and applicable skill set for the future.
Practices that promote cognitive activation, such as encouraging students to think about different ways of solving problems and explaining their reasoning, are strongly associated with overall confidence in 21st-century mathematics (including confidence in 10 different mathematics tasks). Confident students are more likely to be exposed to these practices. It is likely that teachers who use more cognitive activation techniques in the classroom build confidence while deepening students' understanding of mathematics.
Yet, when looking at individual practices, in particular, representing, extracting and interpreting mathematical information, including in real-life situations, it is more proactive behaviour, such as linking what students learn to prior knowledge, that shows the strongest relationship with confidence. Cognitive activation practices together with asking questions when uncertain are two other strategies for sustained learning that can buttress confidence in these essential 21st-century mathematics skills.
Social and emotional skills such as persistence, curiosity and stress resistance show strong and positive relationships with confidence in these four key areas of 21st-century mathematics. Denmark* and New Zealand* stand out as having students who are often likely to feel confident in mathematics when reporting higher social and emotional skills.
PISA data show disparities in reading fluency between socio-economically disadvantaged and advantaged students. To better equip students to be confident 21st-century learners, such disparities need to be addressed. Reading fluency is an important component of confidence in key 21st-century skills and learning outcomes such as performance in mathematics.
In addition to test questions on mathematics, reading and science students answered in PISA 2022, students were asked about their confidence in dealing with a range of 21st-century mathematics tasks. These include interpretation and analysis of mathematical data; real-world application; statistical reasoning; mathematical modelling; computational and technological literacy; and geometry and measurement. These tasks represent the diverse and essential skills needed for success in the data-driven, technologically advanced learning environments and workplaces of the 21st century. (OECD, 2023[5]). These responses have been integrated into the index of mathematics self-efficacy: mathematical reasoning and 21st-century mathematics1.
Confidence in 21st-century mathematics differs across various domains and countries/economies
Students in OECD countries feel most confident extracting mathematical information from diagrams, graphs or simulations (64%); representing a situation mathematically (using variables, symbols or diagrams) (56%); and interpreting mathematical solutions in the context of a real-life challenge (52%). In France, students are most confident extracting mathematical information and interpreting solutions to real-life challenges (79% and 69%, respectively). In Canada* and the United States*, around 70% or more of students feel most confident about representing situations mathematically. At the opposite end of the spectrum, only 36% of students in Thailand feel confident extracting information and in Brunei Darussalam and Japan, 30% or less of students feel confident interpreting mathematical solutions to real-life challenges. When it comes to representing situations mathematically, less than 40% of students in Japan and Thailand feel confident (Table V.B1.8.6).
Finally, on average across OECD countries, students feel the least confident about coding and programming computers (33%). In Uzbekistan and Albania, over 60% of students feel confident in this area while less than 25% feel confident in Brunei Darussalam, Japan, Singapore and Chinese Taipei (Table V.B1.8.6).
Mean score in mathematics
Notes: Only countries and economies with available data are shown.
Students who are less confident (more confident) are those in the bottom (top) quarter of the index of confidence in 21st-century mathematics skills in their own country/economy.
Score-point difference between students in top and those in bottom quarters of the index of confidence in 21st-century mathematics skills that are statistically significant are shown in a darker tone (see Annex A3).
Countries and economies are ranked in ascending order of the difference in mathematics score related to confidence in 21st-century mathematics skills (more confident minus less confident students).
Source: OECD, PISA 2022 Database, Table V.B1.8.32. See Table V.8.1 for StatLink at the end of this chapter.
Understanding the reasons for these differences is important because confidence in these tasks has a positive relationship with learning outcomes. Confidence in performing 21st-century mathematics tasks is positively related to using strategies for sustained lifelong learning, as will be shown next. But it is also positively related to other learning outcomes. In most countries and economies, this kind of confidence is positively related to performance in mathematics – there is a positive change in average mathematics scores on the PISA test with a one-unit increase in the index of confidence in the 21st-century skill of mathematical reasoning in all but one country for which information is available (the higher the index, the more confidence students reported). Students who feel confident about their 21st-century mathematics skills (in the top quarter) outperformed their less confident peers (in the bottom quarter) by 56 score points, on average across OECD countries. This gap is greater than 20 score points, which is equivalent to about one year of schooling in most countries and economies (Avvisati, 2021[6]). In Korea, where the gap is the widest (over 100 score points), students in the top quarter of the index can typically complete Level 4 mathematics tasks. Conversely, less confident students struggle with Level 2 tasks, indicating a much more basic understanding of mathematics and a more limited set of lifelong learning skills (Figure V.8.1 and Table V.B1.8.8).
In conclusion, fostering confidence in 21st-century mathematics skills is essential as it is associated with better learning outcomes and stronger performance in mathematics. This highlights the importance of developing teaching and learning strategies that build student confidence in these critical areas.
How much do strategies for sustained lifelong learning matter in preparing students for future challenges? Much of this depends on how they relate to students’ confidence in their 21st-century mathematics skills. PISA 2022 finds that proactive learning behaviours, cognitive activation practices, and critical thinking are particularly related.
Proactive behaviours such as connecting new material to what has been learned in the past relate positively to student confidence
Confident 21st-century learners try to connect new material to what they have learned in previous mathematics lessons significantly more than their less confident peers. The gap between the two is 32 percentage points, ranging from 63% of confident learners to 31% of less confident learners, on average across OECD countries. The gap is large and positive across all countries and economies, ranging from 19 percentage points in Poland to at least 45 in Albania and Baku (Azerbaijan) (Figure V.8.2).
Such proactive behaviours show strong relationships with confidence, on average and across countries and economies, after accounting for students’ and schools’ socio-economic profile. Yet it is important to consider that the relationship is driven by performance in mathematics. While it is not possible to establish directionality, given the relevance of both, this suggests that proactivity and confidence are elements to be fostered together (Figure V.8.3).
Percentage of students who frequently try to connect new material to what they have learned in previous mathematics lessons, by confidence in 21st-century mathematics skills
Notes: Only countries and economies with available data are shown.
Students who are less confident (more confident) are those in the bottom (top) quarter of the index of confidence in 21st-century mathematics skills in their own country/economy.
Percentage-point difference between students in top and those in bottom quarters of the index of confidence in 21st-century mathematics skills are all statistically significant (see Annex A3).
Countries and economies are ranked in ascending order of the percentage-point difference of students who frequently try to connect new material to what they have learned in previous mathematics lessons related to confidence in 21st-century mathematics skills (more confident minus less confident students).
Source: OECD, PISA 2022 Database, Tables V.B1.2.21 and V.B1.8.10. See Table V.8.1 for StatLink at the end of this chapter.
Confidence in one’s mathematics skills does not exclude double-checking one’s understanding
While confidence in one's abilities is an important factor in academic success, a meticulous approach to learning is also key. Confidence and carefulness complement each other in enhancing students' overall learning.
There is a positive relationship between students’ confidence in their 21st-century mathematics skills and controlling one’s own learning; in other words, confidence does not exclude double-checking. A characteristic of confident students seems to be to ask questions when they do not understand the material being taught (see Chapter 3 on the relationship of this practice with growth mindset). This strategy shows a strong relationship with confidence in one's modern mathematical competence (Figure V.8.3).
Across all participating countries and economies, the gap in the proportion of students who reported asking questions is significant. On average across OECD countries, 60% of confident students ask questions when in doubt while 35% of the least confident students do so. This average gap of 25 percentage points is the smallest in Belgium, Italy and Spain (with at least 17 percentage points) and largest in Baku (Azerbaijan) (44 percentage points) (Table V.B1.8.10).
Confident 21st-century learners reported more cognitive activation
Interestingly, cognitive activation practices such as teachers encouraging students to think about how to solve mathematics problems in different ways and asking students to explain their reasoning when solving a mathematics problem are positively and strongly related to confidence in 21st-century mathematics skills (Figure V.8.3).
Explaining the chain of reasoning involved in solving a mathematics problem is driven by students’ performance in mathematics in some countries but remains positive for all participants in PISA 2022. This is something that over half of confident learners reported being exposed to (54%) compared to only 38% of non-confident learners, across OECD countries. The gap between the two groups is the largest in Albania and the Dominican Republic, where it is at least 30 percentage points, and the smallest in Hungary, Japan, the Netherlands* and the Slovak Republic, where it is about 10 percentage points or less (Table V.B1.8.10).
Finally, critical-thinking (perspective-taking) strategies, including considering others’ perspectives and seeing issues from different angles, are also positively related to confidence in 21st-century mathematics skills, even if they are more weakly related for the most part. Confident learners reported that they consider others' perspectives before taking a position to a greater extent than their less confident peers, with an average difference of 10 percentage points across OECD countries, rising to 23 in Albania, Saudi Arabia and the United Arab Emirates. Only in Chile and Latvia* is the difference between the two groups of students not significant. The same difference is significant in all countries/economies when looking at students’ reports on viewing things from different angles. For this second perspective-taking, the range is also wide, from more than 25 percentage points in Albania, Singapore and the United Arab Emirates to 6 percentage points in Spain (OECD average difference of 15 percentage points). The only exception is thinking that there is only one correct position in a disagreement, whose relationship with confidence in 21st-century mathematics skills is not statistically significant in most countries and economies (Tables V.B1.8.10 and V.B1.8.33).
Change in the index of confidence in 21st-century mathematics skills after accounting for students' and schools' socio-economic profile when students reported the following learning strategies; OECD average
1. Students doing the corresponding statement more than half of the time.
2. Students doing the corresponding statement more than half of the lessons.
3. Students agree or strongly agree with the corresponding statement.
4. Students disagree or strongly disagree with the corresponding statement.
Notes: Changes in the index of confidence in 21st-century mathematics skills are all statistically significant (see Annex A3).
The socio-economic profile is measured by the PISA index of economic, social and cultural status (ESCS).
Items are ranked by kind of learning strategies and then in descending order of the change in the index of confidence in 21st-century mathematics skills, after accounting for mathematics performance.
Source: OECD, PISA 2022 Database, Table V.B1.8.33. See Table V.8.1 for StatLink at the end of this chapter.
Looking at the relationship between confidence and strategies for sustained learning more closely, four key areas are analysed in detail. Together, they enhance students' ability to apply mathematical concepts in diverse and practical ways, preparing them for the complexity of modern challenges:
Representing a situation mathematically using variables, symbols, or diagrams, which involves translating real-world scenarios into mathematical language
Extracting mathematical information from diagrams, graphs, or simulations, involving examining visual representations to identify and gather relevant data
Interpreting mathematical solutions in the context of a real-life challenge, requiring students to apply their mathematical findings to practical situations, thereby making abstract concepts more concrete and relevant
Programming computers, which entails writing code to solve mathematical problems or simulate real-world phenomena
Students who try to connect new material to what they already know feel more confident about their 21st-century skills
Students who frequently try to link what they learn to prior knowledge are, on average, more than twice as likely to feel confident representing, extracting and interpreting mathematical information in real-life situations than those who do not make those connections frequently, even after accounting for performance in mathematics. There is also a positive relationship with programming but it is weaker, on average. The relationship between this particular strategy and confidence in these four skills is the strongest, on average and in most countries and economies. While there are variations depending on the country or economy, programming shows the weakest relationship, with the sole exceptions of Baku (Azerbaijan), Belgium, Brunei Darussalam, the Dominican Republic, Jordan, Mongolia, Morocco, the Netherlands*, the Palestinian Authority and Poland (Table V.B1.8.11).
Similarly, asking questions when not understanding the material being taught is strongly related to confidence in these four specific skills. Students who do this are, on average, twice as likely to be confident representing situations as well as extracting information and interpreting solutions in real life. Even after accounting for performance in mathematics, the likelihood still remains at 1.9. Although this positive relationship is evident across all countries and economies, the strength of the association varies. Again, confidence in programming shows a weaker association, except in Brunei Darussalam, Hungary, Jordan, Morocco, Poland and Chinese Taipei (Table V.B1.8.12).
Interestingly, cognitive activation tasks that require students to think about how to solve problems differently from what is demonstrated in class show a positive relationship, on average, but the relationship seems weaker than with the previously analysed strategies and behaviours. While it is most strongly related to confidence in interpreting solutions in real life, it also shows positive relations, on average, with confidence in the other three tasks. Nevertheless, in Baku (Azerbaijan), the Dominican Republic, Denmark*, Morocco, Ukrainian regions (18 out of 27), Uzbekistan and Sweden, students are at least twice as likely to feel confident about interpreting solutions in real life if they reported that teachers encourage them to think about how to solve problems differently from what is demonstrated in class, after accounting for students’ and schools’ socio-economic profile (Table V.B1.8.13).
Finally, aspects of critical thinking like trying to consider everybody’s perspective before taking a position and seeing things from different angles are mostly positively related to confidence in these 21st-century tasks but the relationships are weaker than for the previous strategies and attitudes. These perspective-taking attitudes are, on average, most strongly related to extracting information although the relationship varies across countries and economies. Interpreting solutions in real life also shows a positive relationship across countries but is not significant for the two perspective-taking strategies in Chile, France, Latvia*, Malta and the Slovak Republic. Likewise, representing a situation is mostly significant except in Baku (Azerbaijan), Czechia, Germany, Greece, Jordan and Latvia*. The relationship between these perspective-taking positions and confidence in programming is largely not significant across countries and economies, especially for considering everybody’s perspective before taking a position (Tables V.B1.8.14 and V.B1.8.15).
In conclusion, fostering strategies that encourage students to connect new knowledge with prior learning, asking questions when uncertain, and engaging in cognitive activation tasks could support developing confidence in essential 21st-century skills. While there are variations across countries and economies, their overall positive relationship with confidence underscores their importance in sustained lifelong learning.
The relationships between the five social and emotional skills (SES) considered in this report (stress resistance, co-operation, emotional control, curiosity and persistence) and confidence in 21st-century mathematics tasks is positive and significant in most countries and economies with available data, even after controlling for students’ and schools’ socio-economic profile and students’ performance in mathematics.
Persistent, stress-resistant and curious students are, on average, the most confident in their 21st-century mathematics skills. In all three cases the relationship holds after accounting for performance in mathematics, which positively influences the relationship (Figure V.8.4).
Change in the index of confidence in 21st-century mathematics skills after accounting for students' and schools' socio-economic profile with a one-unit increase in the following indices; OECD average
Notes: Changes in the index of confidence in 21st-century mathematics skills are all statistically significant (see Annex A3).
The socio-economic profile is measured by the PISA index of economic, social and cultural status (ESCS).
Items are ranked in descending order of the change in the index of confidence in 21st-century mathematics skills, after accounting for mathematics performance.
Source: OECD, PISA 2022 Database, Table V.B1.8.34. See Table V.8.1 for StatLink at the end of this chapter.
Moreover, persistent students are, above all, more likely to report feeling confident extracting information, on average. This is consistent across countries showing the strongest relationship in Australia*, Denmark* and New Zealand*. To a lesser extent, they are also more likely to be confident representing situations mathematically and interpreting solutions to real-life challenges, with Denmark* and New Zealand* again showing the strongest relationships in both cases. And although the relationship with confidence in programming is the weakest, persistent students are the most likely to report confidence in this area in OECD countries. The relationship is the strongest in Korea and Chinese Taipei, and is not significant only in Jamaica* (Figure V.8.4 and Table V.B1.8.17).
Likelihood to report confidence in the following 21st-century mathematics skills when persistent, curious or stress-resistant after accounting for students' and schools' socio-economic profile; OECD average
Note: All odds ratio coefficients are statistically significant (see Annex A3).
Source: OECD, PISA 2022 Database, Tables V.B1.8.17, V.B1.8.18 and V.B1.8.19. See Table V.8.1 for StatLink at the end of this chapter.
While curious students are likely to be confident extracting information, they are, on average across OECD countries, as confident about interpreting solutions to real-life challenges. In both cases, the relationship is the strongest in Denmark* and New Zealand*, together with Macao (China) and the Ukrainian regions (18 out of 27), respectively. They are also more likely than non-curious students to report confidence in representing situations mathematically and, to a lesser extent, programming. While in Denmark*, students are more likely to feel confident representing situations mathematically per one-unit increase in the index of curiosity, it is in Korea that students are the most likely to feel confident in programming. This pattern is observed across most countries and economies (Figure V.8.5 and Table V.B1.8.18).
Finally, students who report most stress-resistance are, on average, most likely to report confidence in interpreting mathematical solutions to real-life challenges, with Denmark* showing again the strongest relationship. Interestingly, these students are the most likely to report confidence in programming, on average, and across countries and economies. In Hong Kong (China)*, stress-resistant students are most likely to feel confident programming. The relationship between stress resistance and confidence in extracting information and representing situations mathematically is also positive and strong, on average and across countries, with the Netherlands*, and Finland and Iceland showing the strongest relationships, respectively (Figure V.8.5 and Table V.B1.8.19).
In all cases, the positive relationships described hold after accounting for students’ and schools’ socio-economic profile, and students’ performance in mathematics.
To better understand students’ opportunities to learn, acquire and develop essential skills for the future through mathematics instruction, PISA asked students about their exposure to 21st-century mathematics topics and tasks (see Box V.8.2). Analysing students' responses provides valuable insights into how effectively schools are preparing students for the future.
Exposure to 21st-century mathematics tasks is important for student confidence but there are other aspects at play too
PISA 2022 data suggest that student confidence and frequency of exposure to 21st-century mathematics tasks are positively related2. However, frequent exposure does not guarantee confidence, at least not in every education system. PISA data reveal a statistically significant but moderate correlation between the frequency of exposure to 21st-century mathematics tasks and students' confidence in completing such tasks. This suggests that exposure alone does not substantially boost confidence and that other aspects are at play.
Being motivated to learn likely plays a role in the relationship between exposure and confidence. For example, analyses show that intrinsic motivations such as enjoying learning new things at school and challenging schoolwork have an indirect effect on the relationship between frequency of exposure to tasks and students’ confidence. On average, about 4% to 6% of the positive relationship between frequency of exposure and confidence can be indirectly attributed to differences in such intrinsic motivations (Table V.B1.8.9). However, results across systems vary. Detailed analyses in each system can shed further light on what the levers are for increasing student confidence.
Enjoying challenging schoolwork can be a strong component of confidence in 21st-century mathematics
Previous sections have highlighted that intrinsic motivations such as enjoying learning new things in school show strong relationships with students’ learning outcomes, including performance in mathematics. When examining the relationship with confidence in 21st-century tasks, enjoyment of challenging schoolwork stands out as the strongest related motivation. While more instrumental motivations such as wanting to do well in mathematics class also show a strong relationship, they are, unsurprisingly, strongly driven by performance in mathematics. Similarly, seeing school as a place that teaches useful skills for future jobs is positively related to confidence in 21st-century tasks, but to a lesser extent (Figure V.8.6).
At the country level, the patterns are consistent. Both enjoying challenging schoolwork and wanting to do well in mathematics class relate to confidence in 21st-century tasks. Challenging schoolwork is strongly related to confidence in Hong Kong (China)* while the relationship is the weakest, albeit positive, in Italy and Spain, after accounting for students’ and schools’ socio-economic profile and students’ performance in mathematics. Viewing school as a place that teaches useful skills for future jobs, while positive and significant, is the least relevant of these four in most countries after accounting for students’ and schools’ socio-economic profile and students’ performance in mathematics. While in the Dominican Republic we find the strongest relationship, it is not significant in Colombia and Indonesia. (Figure V.8.6).
The relationship between students’ opportunities to learn through exposure to mathematics tasks and their confidence in 21st-century mathematics suggests that there are other aspects at play in how effectively students acquire and apply such skills. One of these is motivation to learn. Fostering intrinsic motivations such as the enjoyment of challenging schoolwork alongside more instrumental motivations such as wanting to excel in mathematics class can be important in increasing students' confidence in 21st-century mathematics. These motivations are associated not only with greater confidence but other learning outcomes such as mathematics performance. By emphasising these motivational components, educators can better prepare students for the complexity of modern mathematical challenges and ensure they are equipped with the confidence and skills needed for lifelong learning in the 21st century.
Change in the index of confidence in 21st-century mathematics skills after accounting for students' and schools' socio-economic profile when students agree or strongly agree with the following motivations
Notes: Only countries and economies with available data for the four motivations are shown.
Changes in the index of confidence in 21st-century mathematics skills that are statistically significant are shown in a darker tone (see Annex A3).
Changes in the index of confidence in 21st-century mathematics skills when students agree or strongly agree with the statements "I love learning new things in school" and "I like school work that is challenging" are all statistically significant (see Annex A3).
The socio-economic profile is measured by the PISA index of economic, social and cultural status (ESCS).
Countries and economies are ranked in ascending order of the average change in the index of confidence in 21st-century mathematics skills when students agree or strongly agree with the statement "I love learning new things in school".
Source: OECD, PISA 2022 Database, Table V.B1.8.22. See Table V.8.1 for StatLink at the end of this chapter.
The PISA 2022 framework emphasises the need for students to be exposed to relevant mathematics content, focusing on content exposure. Content exposure considers the time allocated for and dedicated to instruction as well as the depth of teaching provided (OECD, 2023[5]).
This concept was developed within the framework of opportunity to learn (OTL), which refers to the extent to which students’ learning is influenced by the time they spend engaged in it. Essentially, students cannot be expected to learn effectively unless they have sufficient time to engage in the learning process (Carroll, 1963[7]). The concept of OTL has expanded over time to encompass all contextual aspects that capture the cumulative learning opportunities a student encounters (Bertling, Marksteiner and Kyllonen, 2016[8]).
In PISA 2022, students were asked about their exposure to various 21st-century mathematics tasks in class and about their confidence in dealing with such tasks. Answers were integrated into two indices: the index of exposure to mathematical reasoning and 21st-century mathematics tasks1, and the index of mathematics self-efficacy: mathematical reasoning and 21st-century mathematics2.
In the 21st century, the role of mathematics extends beyond traditional calculations and theoretical problems to include practical applications and interdisciplinary integration (Boaler, 2022[9]). The areas covered by these two indices include extracting mathematical information from visual representations, applying mathematical solutions to real-world contexts, using statistical concepts for decision-making, identifying mathematical components in real-world problems, understanding the foundations of mathematical modelling, representing situations mathematically, evaluating data patterns, engaging in coding and computer programming, using mathematical software tools, and calculating the geometric properties of complex shapes. These tasks reflect a comprehensive approach to mathematics education that emphasises not only computational skills but also critical thinking, problem-solving, and the application of mathematics in different contexts.
Moreover, the emphasis on coding, computer programming, and the use of mathematical software tools reflects the integration of technology into mathematics education. These skills are not only relevant for careers in technology (STEM) but also for understanding and solving complex problems in various disciplines through mathematical modelling and simulations (Weintrop et al., 2015[10]).
In conclusion, the domains of mathematics addressed in PISA 2022 are integral to developing a well-rounded mathematical education that prepares students for the complexities of the 21st century.
1. The index comprises students’ frequency ratings of how often they encounter a range of different types of mathematics tasks related to mathematical reasoning and 21st-century mathematics tasks at school (e.g. “Extracting mathematical information from diagrams, graphs, or simulations”, “Using the concept of statistical variation to make a decision”). Each of the 10 items included in this scale had four response options (“Frequently”, “Sometimes”, “Rarely”, “Never”).
2. This index includes students’ ratings of how confident they felt about having to do a range of mathematical reasoning and 21st-century mathematics task (e.g. “Extracting mathematical information from diagrams, graphs, or simulations”, “Using the concept of statistical variation to make a decision”). Each of the 10 items included in this scale had four response options (“Not at all confident”, “Not very confident”, “Confident”, “Very confident”) s.
Across countries and economies, students reported the highest exposure to tasks that involve extracting mathematical information, with just over a third of students across the OECD (35%). In some education systems such as in Canada*, Denmark*, Kazakhstan, the Netherlands*, the United Kingdom* and Singapore, about half of students reported frequent exposure to this task. Conversely, in Czechia and Slovenia, fewer than one in five students did (Figure V.8.7 and Table V.B1.8.1).
Frequency of exposure to 21st-century mathematics tasks; OECD average
Source: OECD, PISA 2022 Database, Table V.B1.8.1. See Table V.8.1 for StatLink at the end of this chapter.
Fewer than one-third of students are frequently engaged in representing situations mathematically
Representing situations mathematically, reported by just under a third of students (31%), is crucial for translating real-world problems into a mathematical framework, enabling the effective analysis, solution, and communication of complex situations. In Canada*, the United States* and Singapore, about half of students reported exposure. In Estonia, Finland, Iceland and Poland, however, less than one in five students reported exposure (Figure V.8.7 and Table V.B1.8.1).
Furthermore, an essential aspect of 21st-century mathematics is the ability to use mathematics to solve problems in real-world contexts. These contexts represent different aspects of an individual's environment in which problems arise. The choice of appropriate mathematical strategies and representations often depends on the context of the problem (OECD, 2023[5]).
One in five students frequently interpret mathematical solutions in real-life contexts in class
On average, only about 20% of students reported being frequently asked to interpret mathematical solutions in the context of a real-life challenge. This percentage is notably low in Czechia, Hong Kong (China)*, Korea, Macao (China) and Poland, where only about 11% of students reported being exposed to such tasks. In contrast, over 40% of students in Uzbekistan did (Table V.B1.8.1).
Other 21st-century mathematics tasks are reported on average by about one in five students or less. Notably, coding/programming computers is the least reported skill, with less than 10% of students, on average across OECD countries, indicating frequent exposure. This falls to around 6% or less in countries and economies such as Australia*, Estonia, Germany, Hong Kong (China)*, Ireland*, the Netherlands*, Portugal, Singapore and Chinese Taipei. The low exposure to coding tasks highlights a significant gap in preparing students for the technological demands of the modern workforce (Figure V.8.7 and Table V.B1.8.1).
Differences in exposure to essential 21st-century mathematical skills across different education systems can have profound implications for lifelong learning. However, exposure should take into account not only the frequency with which students are exposed to these tasks and skills but the quality and content of exposure. Ensuring that students are frequently exposed to rich and relevant mathematical content during their formative years is key. When students are provided with the right content and appropriate support, they can be better equipped to pursue further education and navigate the labour market successfully (OECD, 2019[4]).
In today’s information-rich world, text, whether printed or spoken, serves as the main carrier of content, meaning and context. This extends to texts on social networks, media articles, advertisements, etc. The ability to understand and work with text is crucial not only for general literacy but mathematical literacy. Text comprehension, in particular, reading fluency, is fundamental to students’ excelling in various areas, including mathematics (OECD, 2023[5]).
Reading fluency involves the accurate and automatic decoding of words, allowing readers to devote more cognitive resources to comprehension rather than the mechanical aspects of reading (Kuhn and Stahl, 2003[11]). When students struggle with decoding, they use a significant portion of their cognitive capacity on basic reading tasks, leaving fewer resources for comprehension and problem solving. Fluent readers who can decode words effortlessly are better able to grasp the full meaning of a text (Ehri, 2005[12]). This enhances their learning in all subject areas (OECD, 2023[5]).
In the 2022 PISA assessment, about 78% of students across OECD countries were classified as fluent readers while 15% were either slow or inaccurate readers (Tables V.B1.8.35 and V.B1.8.36). The remaining 7% were students who did not engage meaningfully with the reading fluency test1. This highlights the significant number of students who may be disadvantaged in their learning due to suboptimal reading fluency.
PISA data show that fluent readers are generally more confident extracting mathematical information from diagrams, graphs, and simulations than their slow and inaccurate reading peers. This is particularly evident in countries such as France, the United Arab Emirates and the United States*, where fluent readers are more than twice as likely as slow and inaccurate readers to report confidence in this 21st-century mathematics task (Table V.B1.8.40). The ability to quickly understand and integrate textual labels, legends, and annotations with visual information is critical for the accurate extraction of information (Holsanova, Holmberg and Holmqvist, 2008[13]) – a task that is key for sustained learning throughout life. This is also relevant given that, as shown in this chapter, confidence in extracting mathematical information relates positively and strongly with mathematics performance (Table V.B1.8.8).
However, increased confidence extracting information from visual data does not necessarily extend to other mathematical reasoning tasks, such as interpreting solutions in real-life situations or representing situations mathematically. These tasks may require abstract reasoning that is less directly related to reading fluency.
When the socio-economic profile of students and schools is taken into account, the relationship between reading fluency and confidence in extracting mathematical information becomes not significant in most countries and economies (Table V.B1.8.40). This suggests that socio-economic factors play a critical role in both the development of reading fluency and students’ confidence applying 21st-century mathematical skills (Tables V.B1.8.7 and V.B1.8.36). As discussed earlier in this report, students' socio-economic status is an important component in their attitudes and use of strategies for sustained learning. This is also the case for reading fluency, as these analyses reiterate the importance of addressing socio-economic inequalities in education.
Likelihood of agreeing or strongly agreeing to wanting to do well in mathematics class when students are fluent readers
1. The socio-economic profile is measured by the PISA index of economic, social and cultural status (ESCS).
Notes: Only countries and economies with available data are shown.
Odds ratio coefficients that are statistically significant are shown in a darker tone (see Annex A3).
Students classified as "hasty readers" are excluded from this analysis.
Countries and economies are ranked in descending order of the odds ratio, after accounting for students' and schools' socio-economic profile.
Source: OECD, PISA 2022 Database, Table V.B1.8.40. See Table V.8.1 for StatLink at the end of this chapter.
Fluent readers are, across OECD countries and PISA 2022 participating countries and economies, more likely to report being motivated to do well in mathematics class than slow and inaccurate readers. This relationship holds even after accounting for students’ and schools’ socio-economic profile. Motivation to do well in class is crucial, as it has the second strongest association with confidence in 21st-century mathematics tasks and in some countries/economies it is the most strongly related (Figure V.8.8 and Table V.B1.8.22). While fluency alone does not entirely explain all the difference in motivation, the relationship between fluency and motivation remains, regardless of students’ and school’s socio-economic profile. One possible interpretation is that fluent readers are more likely to see the value in mathematics tasks because they can understand and complete them more efficiently. Fluent readers who find it easier to cope with the reading aspects of mathematical problems are more likely to recognise and appreciate the value of these tasks, and feel more motivated to do well in mathematics.
While fluent readers may be more motivated to do well in mathematics class, this does not necessarily translate into a greater intrinsic love of learning in all subjects or a stronger belief in the practical utility of schooling for future employment (Table V.B1.8.40). Intrinsic motivation is influenced by a wider range of factors, including personal interests, curiosity and the learning environment while instrumental motivation can often be shaped by other factors including career aspirations and socio-economic profile (Deci and Ryan, 1985[14]; Wigfield and Eccles, 2000[15]).
Reading fluency also shows an important relationship with performance in mathematics. Across most language groups represented in PISA 2022, between 7% and over 20% of the variation in mathematics performance is accounted for by students’ reading fluency2 (Table V.B1.8.39). Students who score at the lower proficiency levels in PISA are likely to read at a significantly slower rate or be more inaccurate than higher-performing students. In some countries, such as Jamaica*, Panama*, and Peru, more than 30% of low-performing students are slow readers compared to an OECD average of 13% (Table V.B1.8.38).
When the proportion of inaccurate readers is added, over a half of low mathematics performers in Jamaica*, Morocco, Peru and the Philippines turn out to be either slow or inaccurate readers – 25% on average across the OECD (Table V.B1.8.38). At the other end of the scale, Finland has the smallest proportion of slow readers among low performers in mathematics and when the proportion of inaccurate readers is added, the total does not exceed 17% (Table V.B1.8.38). For comparison, most skilled performers students (i.e. who scored at Level 3 or above) are fluent in reading (89% on average across the OECD) (Table V.B1.8.38 and Figure V.8.9a online).
Low performers in mathematics who are also slow readers or show high levels of inaccuracy face important challenges that may affect their lifelong learning trajectories. The main issue highlighted by these PISA data is the potential for reduced reading efficiency and comprehension, which can affect their ability to process and understand complex information quickly.
This double disadvantage can hinder their ability to extract the relevant information from a range of support and data sources, which is important academically, professionally and in everyday life. Improving reading fluency can increase students' motivation to engage in class and acquire the necessary skills for the 21st century. Policies and programmes to improve reading fluency can be implemented throughout the school years at different grade levels, and with adapted and concrete objectives (see Box V.8.4 for an example of such programmes in France).
1. The PISA reading fluency test consisted of presenting students with a series of simple sentences, one at a time, and asking them to determine whether each sentence made sense. These sentences were all relatively simple and there was no ambiguity about whether they made sense or not. To determine if low-achieving students can read fluently, language-specific norms for response time on reading fluency tasks were established based on the reading pace for the median response time of Level 3 students (at least 50%), which is the baseline for skilled performers in this report. Based on the accuracy of their sensitivity judgments and the time taken to complete the test, students were categorised into four groups:
Fluent readers: Made no more than one mistake in their sensitivity judgments and took at most twice as long as the language-specific norm.
Slow readers: Took more than twice as long as the norm, regardless of accuracy.
Inaccurate readers: Made more than one mistake but did not complete the test unusually quickly or slowly.
Hasty test-takers: Made more than one mistake and completed the test faster than 99% of their peers who made, at most, one mistake, indicating rapid guessing.
These categories cover the totality of test-takers and are mutually exclusive (i.e. every student falls into one category). Hasty test-takers are not included in the analyses in this box as they did not engage with the reading-fluency test as expected.
2. This analysis only considers language groups within countries/economies with at least 1 000 students.
It is not just about frequency but the nature of exposure to 21st-century mathematics tasks
It is important to note that in most countries/economies, greater exposure to 21st-century mathematics tasks is not automatically associated with better learning outcomes, even after accounting for students’ and schools’ socio-economic profile. The relationship between performance in mathematics and exposure to 21st-century mathematics tasks is positive in only 16 out of all PISA-participating countries. Only in Australia*, the Philippines, and Singapore is the performance gap about 10 score points, after accounting for students’ and schools’ socio-economic profile (Table V.B1.8.3).
Various reasons could explain this. First, students’ reports of their frequent exposure to some of the tasks measured in PISA may be influenced by their understanding of the tasks themselves. This is likely to be reflected in differences in how low and skilled performers reported their exposure to different tasks. For example, in Australia*, Brunei Darussalam, Denmark*, Malta, the Netherlands*, Singapore and the United Kingdom*, over 50% of skilled performers reported being frequently exposed to extracting mathematical information from diagrams, graphs, or simulations while about a third or less of low performers did3 (Table V.B1.8.23). Other significant variations can be seen across countries/economies (Tables V.B1.8.24, V.B1.8.25 and V.B1.8.26).
Second, the relationship between instructional time and learning outcomes has been analysed in previous PISA volumes with data showing that while more hours spent in regular lessons and homework do not always correlate with higher scores, on average, performance in mathematics is positively associated with each additional hour of regular lessons per week up to a certain point (OECD, 2023[1]). This suggests that other factors may be at play, such as teaching approaches and student engagement, and that the relationship between instruction time and learning outcomes is not linear.
In countries where students reported the most exposure to extracting mathematical information, representing a situation mathematically and interpreting mathematical solutions4, associated performance gaps can vary widely, after accounting for students’ and schools’ socio-economic profile. For example, among these countries, exposure to extracting mathematical information is associated with positive performance gaps of at least 40 score points in Denmark*, Singapore, the United Arab Emirates and the United Kingdom*, but is not significant in Uzbekistan. Similarly, it is above 40 score points in the United Arab Emirates for representing situations mathematically and is not significant for interpreting mathematical solutions in Kazakhstan, the Dominican Republic and Saudi Arabia. In a large number of countries and economies, these relationships are not significant or even negative, as can be the case for the OECD average (Tables V.B1.8.3, Table V.B1.8.23, Table V.B1.8.24 and Table V.B1.8.26).
This suggests that, at least in some systems, students who struggle with traditional teaching methods are given more teaching time and innovative approaches. While this tailored support is intended to help these students, it may distort the overall relationship between teaching time and learning outcomes (Hattie, 2008[16]). This is because the extra time compensates for learning difficulties rather than improving learning for all students. In other words, the extra time may be needed simply to bring these students up to a baseline level rather than to take them to higher levels of achievement.
In conclusion, the nuanced relationship between exposure to 21st-century mathematics tasks and learning outcomes underscores the importance of focusing on both the quality and the content of educational experiences. These PISA 2022 results suggest that simply increasing exposure may not be enough: effective learning hinges on other relevant aspects too, including how well these tasks are integrated into the curriculum and how they are taught. They also highlight the need for innovative teaching approaches along with strong student motivation and effective learning strategies to fully maximise the benefits of students’ opportunities to learn. Moreover, the disparity in task exposure between skilled and low performers suggests that more personalised and equitable teaching methods are essential to ensure that all students can acquire a solid foundation in critical mathematical skills. To truly enhance learning outcomes, educational systems must go beyond increasing instruction time and focus on improving the overall learning process.
Fluency tests can be easily implemented in the classroom and do not take much time. Since 2020, all sixth-grade students in France are assessed in their reading fluency at the beginning of the school year. This reading fluency test is part of a 60-minute French test that assesses students on their writing, comprehension, and knowledge. The test provides teachers with an overview of students’ skills. It helps teachers identify students who may need additional help and put in place appropriate remedial measures to aid the transition from elementary to middle school.
During the reading fluency test, students are asked to read aloud text for one minute. The teacher then reports the number of words that were correctly read. This simple test can detect severe reading difficulties very early on in the school year.
Students who require additional support may benefit from programmes and plans by the Ministry of Education, such as the Plan d’Accompagnement Personnalisé (PAP). The PAP enables students to have personalised learning plans, which are put together by a team of educators, parents and professionals, and is revised yearly. Examples of learning accommodations for mathematics lessons include allowing the use of calculators even when prohibited, provision of useful tool sheets such as on definitions and theorem, and the use of different colours (e.g. units are coloured in red, tens in blue, hundreds in green).
References
[6] Avvisati, F. (2021), “How much do 15-year-olds learn over one year of schooling?”, PISA in Focus, No. 115, OECD Publishing, Paris, https://doi.org/10.1787/b837fd6a-en.
[8] Bertling, J., T. Marksteiner and P. Kyllonen (2016), “General Noncognitive Outcomes”, in Methodology of Educational Measurement and Assessment, Assessing Contexts of Learning, Springer International Publishing, Cham, https://doi.org/10.1007/978-3-319-45357-6_10.
[9] Boaler, J. (2022), Mathematical Mindsets: Unleashing Students’ Potential through Creative Math, Inspiring Messages, and Innovative Teaching., Jossey-Bass.
[7] Carroll, J. (1963), “A Model of School Learning”, Teachers College Record, Vol. 64, pp. 723-733.
[14] Deci, E. and R. Ryan (1985), Intrinsic Motivation and Self-Determination in Human Behavior, Springer US, Boston, MA, https://doi.org/10.1007/978-1-4899-2271-7.
[12] Ehri, L. (2005), “Learning to Read Words: Theory, Findings, and Issues”, Scientific Studies of Reading, Vol. 9/2, pp. 167-188, https://doi.org/10.1207/s1532799xssr0902_4.
[16] Hattie, J. (2008), Visible learning: A synthesis of over 800 meta-analyses relating to achievement., Routledge.
[13] Holsanova, J., N. Holmberg and K. Holmqvist (2008), “Reading information graphics: The role of spatial contiguity and dual attentional guidance”, Applied Cognitive Psychology, Vol. 23/9, pp. 1215-1226, https://doi.org/10.1002/acp.1525.
[11] Kuhn, M. and S. Stahl (2003), “Fluency: A review of developmental and remedial practices.”, Journal of Educational Psychology, Vol. 95/1, pp. 3-21, https://doi.org/10.1037/0022-0663.95.1.3.
[20] Ministère de l’Éducation Nationale (2024), Enseigner à des élèves à besoins éducatifs particuliers, https://eduscol.education.fr/3890/enseigner-des-eleves-besoins-educatifs-particuliers (accessed on 24 May 2024).
[18] Ministère de l’Éducation Nationale (2022), Online tests for the national assessment of students in sixth and second grade, https://labo.societenumerique.gouv.fr/en/articles/online-tests-for-national-assessment-of-sixth-and-second-graders/ (accessed on 24 May 2024).
[19] Ministère de l’Éducation Nationale (2022), Test de Fluence En Classe de Sixième, https://eduscol.education.fr/document/41971/download (accessed on 24 May 2024).
[17] Ministère de l’Éducation Nationale (2015), Plan d’accompagnement personnalisé, https://cache.media.education.gouv.fr/file/5/50/4/ensel1296_annexe_plan_daccompagnement_personnalise_386504.pdf (accessed on 24 May 2024).
[5] OECD (2023), PISA 2022 Assessment and Analytical Framework, PISA, OECD Publishing, Paris, https://doi.org/10.1787/dfe0bf9c-en.
[21] OECD (2023), PISA 2022 Results (Volume I): The State of Learning and Equity in Education, PISA, OECD Publishing, Paris, https://doi.org/10.1787/53f23881-en.
[1] OECD (2023), PISA 2022 Results (Volume II): Learning During – and From – Disruption, PISA, OECD Publishing, Paris, https://doi.org/10.1787/a97db61c-en.
[3] OECD (2021), OECD Skills Outlook 2021: Learning for Life, OECD Publishing, Paris, https://doi.org/10.1787/0ae365b4-en.
[4] OECD (2019), OECD Skills Outlook 2019: Thriving in a Digital World, OECD Publishing, Paris, https://doi.org/10.1787/df80bc12-en.
[2] OECD (2019), Trends Shaping Education 2019, OECD Publishing, Paris, https://doi.org/10.1787/trends_edu-2019-en.
[10] Weintrop, D. et al. (2015), “Defining Computational Thinking for Mathematics and Science Classrooms”, Journal of Science Education and Technology, Vol. 25/1, pp. 127-147, https://doi.org/10.1007/s10956-015-9581-5.
[15] Wigfield, A. and J. Eccles (2000), “Expectancy–Value Theory of Achievement Motivation”, Contemporary Educational Psychology, Vol. 25/1, pp. 68-81, https://doi.org/10.1006/ceps.1999.1015.
← 1. This index includes students’ frequency ratings of how often they engage in behaviours indicative of effort and persistence in mathematics (e.g. “I actively participate in group discussions during mathematics class”, “I put effort into my assignments for mathematics class”). Each of the eight items included in this scale had five response options (“Never or almost never”, “Less than half of the time”, “About half of the time”, “More than half of the time”, “All or almost all of the time”).
← 2. The correlation between the two indices is relatively similar when looking at the average across OECD countries (correlation coefficient=0.25), and across all countries and economies participating in PISA 2022 (correlation coefficient=0.21).
← 3. PISA 2022 data show that this type of variation in reports between low and skilled performers holds in both, academically segregated and comprehensive systems alike, as measured by the PISA index of academic inclusion (OECD, 2023[21]). For example, among the countries mentioned above, Denmark* and the Netherlands* are at opposite ends of the index of academic inclusion.
← 4. The countries and economies whose relationships are considered here are:
Extracting mathematical information: countries/economies where more than 45% of students reported exposure to this task.
Representing situations mathematically: countries/economies where at least 40% of students reported exposure to this task.
Interpreting mathematical solutions: countries/economies where at least 33% of students reported exposure to this task.