Mindsets, attitudes and learning

2. Differences in growth mindset across student groups

Psychological constructs such as growth mindset rarely operate in isolation. Their influence is shaped by students’ prior experiences and backgrounds (Yeager et al., 2019[14]; Eccles and Wigfield, 2002[17]), and is filtered through students’ opportunities, expectations and social contexts. Disadvantaged students tend to encounter more obstacles to academic success, meaning that holding a growth mindset might not always be matched by actual learning opportunities (Claro, Paunesku and Dweck, 2016[15]). Similarly, boys and girls may internalise different social norms about their own ability and effort (Good, Rattan and Dweck, 2012[18]). Students who are struggling may perceive effort differently to those already excelling, and top performers may also avoid challenges to maintain approval within their comfort zones.

Examining growth mindset across student groups helps to reveal how beliefs might be internalised differently or how beliefs relate to outcomes differently. These distinctions matter for policy purposes – if growth mindset is unequally distributed, or if its benefits are uneven (for example, if they are smaller or larger for socio-economically disadvantaged students), approaches must carefully consider the context.

Findings from PISA 2022

As in 2018, when PISA first measured growth mindset, a majority of students reported a growth mindset again in 2022. On average across OECD countries, 58% of 15-year-old students disagreed or strongly disagreed with the statement that “intelligence is something that cannot change very much”. However, as before, there was wide variation across countries/economies. In Austria, Estonia, Germany, Ireland*, Japan, Kazakhstan, Sweden and the United States*, more than two-thirds of students reported having a growth mindset. In contrast, in Albania, Kosovo and Moldova, this figure was closer to one-third (Figure 1).

Figure 1. Growth mindset: Change between PISA 2018 and PISA 2022 results

Change between 2018 and 2022 in the percentage of students disagreeing or strongly disagreeing that "your intelligence is something about you that you can't change very much"

Note: Only countries and economies with available data are shown.

Statistically significant differences between PISA 2018 and PISA 2022 are shown in a darker tone (see Reader’s Guide).

Countries and economies are ranked in descending order of the difference between PISA 2018 and PISA 2022 (PISA 2022 minus PISA 2018).

Source: OECD, PISA 2022 Database, Table B.5.

PISA data show a decline between 2018 and 2022 (Figure 1); on average, the proportion of students in OECD countries reporting a growth mindset fell by five percentage points (Table B.5). The decrease exceeded 10 percentage points in Albania, Colombia, Denmark*, Iceland, Lithuania, Latvia*, Portugal and Türkiye. Overall, 36 countries and economies reported a decline, compared to 15 where the opposite trend was observed. Notably, in most of the countries/economies where growth mindset increased, baseline levels in 2018 were relatively low (below 50%), for example, in North Macedonia, Panama* or Morocco. The only systems where a majority of students had a growth mindset in 2018 and the share increased by 2022 were Japan, Kazakhstan and Sweden (see Box 1).

The shift in fixed and growth mindsets among students between the two PISA cycles may reflect the broader context in which they learned during this period. The 2022 cycle took place after the period of the COVID-19 pandemic, which was marked by prolonged school closures, social isolation and learning disruption in many education systems globally. These conditions could have impacted students' perceptions of their ability to progress academically.

Box 1. Learning characteristics that distinguish systems with increased growth mindset

While the proportion of students with a growth mindset declined in most education systems between 2018 and 2022, three systems – Japan, Kazakhstan and Sweden – stood out by showing increases during this period despite already having over 50% of students with a growth mindset in 2018.

Analyses of PISA 2022 data show that these systems share several characteristics (OECD, 2024[3]). Students in the three systems report exposure to challenging teaching, such as linking new knowledge to previous learning and explicitly explaining their mathematical reasoning, above the OECD average. In two of the three systems (Sweden and Kazakhstan), students also exhibit stronger proactive learning behaviours. Motivation to learn is comparatively higher in Japan (instrumental) and Kazakhstan (both intrinsic and instrumental) than the OECD average, while exposure and confidence in interpreting mathematical information is higher in Sweden and Kazakhstan (OECD, 2024[3]). These indicators suggest that, among other things, these learning environments combine challenge, agency and motivation – conditions that may foster positive beliefs among students about their capacity for improvement. While these patterns are by no means exhaustive, the coexistence of comparatively high levels of these learning attitudes alongside a widespread growth mindset among students suggests that such beliefs are more likely to persist where effort and other attitudes towards learning are integral to everyday learning practices.

The performance profiles of the three countries differ. Japan has a top-performing system with a low proportion of low performers (12%), while Sweden has a moderate proportion (27%) and Kazakhstan has a higher proportion (50%) (OECD, 2023[1]). This shows that maintaining or expanding a growth mindset is not limited to high-performing systems. In all cases, the prevalence or persistence of a growth mindset coexists with stronger-than-average learning behaviours and attitudes, including challenging teaching, proactive learning, motivation and confidence.

Taken together, these findings suggest that, despite an average decline in growth mindset, sustaining it may not be tied to any single performance level or country profile, but rather to the breadth and consistency of supportive learning conditions. However, these patterns should be taken as illustrative rather than exhaustive. Some other education systems (Austria, Australia*, Canada* and Singapore) share similar combinations of these learning attitudes without a corresponding increase in growth mindset between 2018 and 2022. This suggests that while such conditions are conducive to sustaining adaptive learning beliefs, their influence may depend on broader teaching and learning practices and cultures.

Despite this downward shift, the 2022 results confirm earlier findings that across most countries/economies, a growth mindset is positively associated with academic performance (Figure 2). On average across OECD countries, students who disagreed or strongly disagreed with the statement that intelligence is fixed scored 18 points higher in mathematics, and 22 points higher in both reading and science than their peers reporting a fixed mindset, even after accounting for the socio-economic profile of students and schools – standardised effect sizes associated with an increase by one standard deviation in growth mindset were 0.20 for mathematics, 0.23 for reading and 0.22 for science (Tables B.12 and B.13). The performance gap in mathematics between students with a growth mindset and those with a fixed mindset was especially large (at least 40 score points) in New Zealand*, Australia* and the United States*. Only in 16 countries/economies was no significant relationship observed; and in Hong Kong (China)* and Moldova, the pattern was reversed – students with a growth mindset scored lower than those with a fixed mindset, by 8 and 7 points respectively.

Figure 2. Growth mindset and performance in mathematics

Change in mathematics performance when students disagreed or strongly disagreed that "your intelligence is something about you that you can’t change very much"

¹The socio-economic profile is measured by the PISA index of economic, social and cultural status (ESCS).

Note: Only countries and economies with available data are shown.

Coefficients that are statistically significant are shown in a darker tone (see Reader’s Guide).

Countries and economies are ranked in descending order of the gap in mathematics performance related to general growth mindset, before accounting for students' and schools' socio-economic profile.

Source: OECD, PISA 2022 Database, Table B.12.

The cross-country correlation between the growth mindset and performance associations in 2018 and 2022 (r=0.54) indicates moderate stability, suggesting that systems with relatively stronger or weaker associations in 2018 tended to remain with them in 2022, though some changes across countries/economies are evident.

Taken together, these results suggest that although a growth mindset remains widespread, its prevalence has declined in many education systems, possibly reflecting broader contextual challenges. At the same time, the growth mindset-performance link is not a fixed system characteristic, but it changed over time in some contexts, suggesting that school practices and policy changes may shape its alignment in a number of education systems.

Box 2. Differences in growth mindset by gender

PISA results suggest that gender differences in growth mindset are generally small, but not absent. Across OECD countries, girls are only marginally more likely than boys to report a growth mindset (by less than one percentage point) (Table B.10). However, variation across countries is considerable (see Figure 3. Growth mindset by gender) – in some, such as the Dominican Republic, Qatar and Uruguay, boys are more likely to report having a growth mindset, whereas in Finland, Germany, Portugal and Switzerland, the opposite holds.

Figure 3. Growth mindset by gender

Percentage of students disagreeing or strongly disagreeing that "your intelligence is something about you that you can't change very much", by gender

Note: Only countries and economies with available data are shown.

Statistically significant differences between boys and girls are shown in a darker tone (see Reader’s Guide).

Countries and economies are ranked in descending order of the difference between boys and girls.

Source: OECD, PISA 2022 Database, Table B.10.

When considering performance, the score differences associated with reporting a growth mindset are broadly similar for boys and girls (Table B.17). Yet, where differences are observed, they tend to favour girls. Across OECD countries, the average difference is 27 score points for girls and 22 for boys. In Moldova, the Slovak Republic and Sweden, the gap in favour of girls exceeds 20 score points. Peru is the only system in which the relationship between growth mindset and performance in mathematics is stronger for boys than for girls, by 11 score points. In addition, in the two countries/economies where a growth mindset is negatively related to performance (Moldova and Hong Kong (China)*), this holds for boys, while for girls the relationship is either positive (Moldova) or not statistically significant (Hong Kong (China)*). Similarly, in most countries where the relationship is overall not significant, the score gap for girls alone remains statistically significant.

While gender differences can be context-dependent, these results suggest that the performance advantage in mathematics linked with growth mindset is at least as strong – and in some contexts stronger – for girls.

Distribution of growth mindset by performance group

Differences are visible across the student performance distribution. PISA data show nuances in how growth mindset is distributed among low, mid and top performers (the level of performance in PISA is measured according to six proficiency levels in mathematics1, see the Reader’s Guide for more details). Students at mid and top levels are more likely to report a growth mindset than low performers (Figure 4 and Table B.1). Among low performers, fewer than half report a growth mindset, on average across OECD countries – 49% report a growth mindset and 51% report a fixed mindset. In most countries and economies (55), fewer than half of low-performing students report a growth mindset. Only in a few countries – Austria, Estonia, Germany, Japan, Kazakhstan and Sweden – do more than 60% of low performers hold a growth mindset. In contrast, 61% of mid-performing students report a growth mindset on average; also, in most countries/economies, over half of them report a growth mindset. In some cases, such as Brazil, Estonia, Ireland*, Germany and the United States*, over three-quarters of mid-performers report it. Among top performers, the pattern is clearest – on average across OECD countries, 67% report a growth mindset, and in 15 countries/economies more than three-quarters do so. In Australia*, Brazil, Estonia, Ireland*, Kazakhstan, New Zealand* and the United States*, over 80% of top performers report a growth mindset.

Figure 4. Growth mindset by level of performance in mathematics

Percentage of students disagreeing or strongly disagreeing that "your intelligence is something about you that you can't change very much", by level of performance in mathematics

Note: On average across the OECD, the percentage of students reporting a growth mindset is 49% among low performers, 61% among mid-performers, and 67% among top performers.

Only countries and economies with available data are shown.

Countries and economies are ranked in descending order of the percentage of students disagreeing or strongly disagreeing that "your intelligence is something about you that you can't change very much" among medium performers in mathematics.

Source: OECD, PISA 2022 Database, Table B.1.

While prevalence of growth mindset increases with performance level, the score gaps associated with growth mindset are not uniform across the mathematics performance distribution. As shown in Figure 5, students in the middle (who scored at proficiency Level 2, 3 or 4) show the largest and most consistent gaps, around 8 score points, after accounting for students’ and schools’ socio-economic profile, on average (Table B.14). These mid-level gaps are particularly large (over 19 score points) in Ireland* and the United Arab Emirates. For top performers (who scored at Level 5 or above), however, score gaps linked with a growth mindset are not statistically significant on average across OECD countries. Among low performers (who scored below Level 2), the relationship is less consistent across countries/economies and of smaller magnitude, on average (1.6 score points). These results are confirmed by the standardised effect sizes associated with an increase by one standard deviation in growth mindset, which are 0.04 for low performers, 0.17 for mid-performers and with no association for top performers (Table B.15). While growth mindset is more common among higher-performing students, in a number of systems the association between mindset and achievement is most relevant for students in the middle levels of proficiency, where performance trajectories can shift in either direction.

Figure 5. Growth mindset and mathematics performance, by student performance

Change in mathematics performance when students disagreed or strongly disagreed that "your intelligence is something about you that you can’t change very much", by level of performance in mathematics; OECD average

¹The socio-economic profile is measured by the PISA index of economic, social and cultural status (ESCS).

Note: On average across the OECD, the percentage of students reporting a growth mindset is 49% among low performers, 61% among mid-performers, and 67% among top performers.

Coefficients are all statistically significant (see Reader’s Guide).

Source: OECD, PISA 2022 Database, Table B.14.

Quantile regressions confirm that growth mindset and performance are not related in a similar way across the performance distribution. The relationship tends to be the strongest in the middle of the distribution and weaker at both extremes, but especially so among students with the lowest mathematics performance (Table B.16). To illustrate this, the analysis in Figure 6 uses the interquartile difference (IQD), defined as the gap between the estimated coefficients at the 75th and 25th percentiles of the performance distribution. A positive IQD indicates stronger associations among higher performers, while a negative IQD indicates stronger associations among lower performers.

As shown in Figure 6, the IQD correlates strongly to the share of low-performing students in a system. Where many students are below baseline proficiency, the association with growth mindset tends to be weak at the bottom but to remain positive at higher levels, resulting in a large positive IQD. By contrast, in high-performing systems with very few low performers, the IQD is small or even negative. For example, in the Dominican Republic, which has the largest share of low performers, the IQD is 20 score points, meaning that growth mindset aligns more strongly with the performance of higher-achieving students than with that of low performers. In Singapore, the IQD is -34 score points, meaning the association is stronger for lower performers (Table B.16).

Figure 6. Interquartile difference of the relationship between growth mindset and performance in mathematics and share of low performers

Note: Only countries and economies with available data are shown.

The interquartile difference (IQD) is defined as the gap between the estimated coefficients at the 75th and 25th percentiles of the performance distribution.

Source: OECD, PISA 2022 Database, Tables I.B1.3.1 and B.16.

This reinforces the importance of understanding growth mindset not only as an individual belief but also in relation to the learning environments in which students develop. Moreover, the strong relationship with the share of low-performing students in a system suggests that the extent to which students can act on what they know, and the learning opportunities available to them, may shape how such beliefs are formed and sustained.

In terms of policy, this means that encouraging growth mindset at scale must go hand in hand with differentiated strategies. Special attention should be paid to low-performing students who could not reach the baseline proficiency, as these students may hold growth beliefs but not see them reflected in their learning experiences. This signals that promoting growth mindset for struggling learners may require careful calibration, ensuring that pedagogical strategies, feedback, supports and learning resources do not inadvertently raise expectations without providing the means to meet them. More broadly, however, the performance gradients observed in the data may also reflect the organisation of learning opportunities within education systems. When low-performing students are grouped together – for instance, through early tracking or distinct pathways – structural separation can exacerbate differences in students' experiences of success and feedback (OECD, 2023[19]). Therefore, addressing these disparities requires classroom-level strategies and system-level policies that foster inclusion and permeability across learning pathways, as well as high expectations for all (OECD, 2023[19]).

Box 3. Differences in growth mindset and performance by socio-economic status

Socio-economic status shows clearer distinctions in differences in growth mindset and performance. Across OECD countries, advantaged students are about nine percentage points more likely to report a growth mindset than their disadvantaged peers (Table B.10). In no education system do disadvantaged students report growth mindset to a larger extent than their advantaged peers. However, in 20 countries and economies there are no differences by students’ socio-economic status. In six of them – Croatia, Germany, Korea, Poland, Chinese Taipei and Uzbekistan – it is possible to argue that having a growth mindset is a representative trait of its student population (i.e. over half of students report a growth mindset), and it is relatively equitably distributed.

On the other hand, performance gaps vary. While the OECD average does not show a clear socio-economic pattern, in many education systems the differences in mathematics score associated with growth mindset are larger among advantaged students. For example, score gaps exceed 30 score points in Thailand, Panama*, Brunei Darussalam, the United Arab Emirates, Malaysia and Qatar (Table B.17). Yet, in others, the inverse is true. In Spain, for example, the score gap is 10 points larger for disadvantaged students, and it is as high as 35 score points in Singapore. In some of the systems where growth mindset is prevalent and it is equitably distributed (i.e. Croatia, Germany, Korea, Poland, Chinese Taipei and Uzbekistan), disadvantaged students even show stronger associations with performance, as in Croatia, Germany and Chinese Taipei (over 15 score points). In Korea, Poland and Uzbekistan, socio-economic status does not seem to make a difference in the relationship between growth mindset and performance.

This suggests that the distribution of growth mindset matters – education systems in which growth mindset is both widespread and evenly distributed are likely to foster more supportive learning environments for a larger proportion of their students.

Note

← 1.  There are six proficiency Levels in mathematics in PISA. Low performers are students who did not attain baseline Level 2 in mathematics; mid-performers are students who attain Levels 2, 3 or 4 in mathematics; top performers are students who attain Levels 5 or 6 in mathematics. For further information see (OECD, 2023[1]).