Knowledge and Skills are about what people know and can do. This chapter discusses the results of the OECD’s PISA tests of cognitive skills in maths, reading and science at age 15; and adult numeracy and literacy, as assessed through the OECD’s PIAAC study. Over the last decade or so, average scores in maths, reading and science for students at 15 have fallen in around one-quarter of OECD countries. Around 1 in every 8 students has a very low score in all three PISA subjects, and around 16% of adults have very low scores in both literacy and numeracy. Among both youths and adults, men perform better than women in mathematics, while girls tend to outperform boys in reading. There are large inequalities in skills at age 15 by socio-economic background. Older adults (aged 45-65) fare worse in literacy and numeracy tests compared to younger cohorts (aged 16-44).
How's Life? 2020
6. Knowledge and Skills
Abstract
Figure 6.1. Knowledge and Skills snapshot: current levels, and direction of change over the last decade or so
Note: The snapshot depicts data for 2018, or the latest available year, for each indicator. The colour of the circle indicates the direction of change, relative to 2010, or the closest available year: improvement is shown in blue, deterioration in orange, no clear or consistent change in grey, and insufficient time series to determine trends in white. For each indicator, the OECD country with the lowest (on the left) and highest (on the right) well-being level are labelled, along with the OECD average. For full details of the methodology, see the Reader’s Guide.
Source: Data drawn from the OECD Programme for International Student Assessments (PISA) in reading, mathematics and science, and the OECD Program for the International Assessment of Adult Competencies (PIAAC) assessments in literacy and numeracy.
Cognitive skills at age 15: PISA scores in maths, reading and science
The OECD Programme for International Student Assessment (PISA) tests the abilities of 15-year-old students in mathematics, reading and science. Scores are measured on a scale that is standardised to 500 for the OECD average. This standardisation is established in the first year a subject is introduced as the major testing domain (e.g. 2003 for mathematics), to enable comparisons over time. For this reason, the OECD average may not exactly equal 500 in any given year. Also, the reference year used for assessing changes over time varies by subject.
PISA mathematics scores
Japanese students have the highest average mathematics scores in the OECD, followed by Korea, Estonia and the Netherlands (Figure 6.2). At the other end of the scale, Colombia has the lowest average score, with Mexico and Chile just above. Since 2003, the average maths score of students aged 15 has significantly improved in just 8 OECD countries, but worsened in 13. The largest gains occurred in Israel (over 6 points), while Finland experienced the largest falls (almost 10 points).
Figure 6.2. Maths skills of students aged 15 have declined in more than one-third of OECD countries since 2003
Note: † indicates that the country falls below the OECD average, * indicates that the country is above average. Countries with no accompanying mark are not statistically different from the OECD average. The PISA mathematics scores are measured on a scale that is normalised to 500 for the OECD average. Normalisation is established in the first year a subject is introduced as the major testing domain – 2003 for mathematics – to allow for year-on-year comparability. For this reason, the OECD average may not exactly equal 500 in any given year. The trend is reported only for countries that have recorded significant improvements or deteriorations since 2003, i.e. no starting point is shown for countries whose change in test scores is not significant. See Box 6.1 for more details on the calculation of average trends.
Source: (OECD, 2019[1]), PISA 2018 Results (Volume I): What Students Know and Can Do, PISA, OECD Publishing, Paris, https://doi.org/10.1787/5f07c754-en.
PISA reading scores
In Estonia, Canada, Finland and Ireland, 15-year-old students have the highest average PISA reading scores among OECD countries, followed very closely by Korea and Poland (Figure 6.3). As in the case of maths, Colombia, Mexico and Chile have the lowest average scores. Since 2000, students’ average reading scores fell significantly in 8 OECD countries, but increased in 7. The largest increases occurred in countries falling below the OECD average (Chile, Colombia, Israel and the Russian Federation), as well as Estonia and Germany, whose performance is above average. Significant declines in reading performance primarily occurred in countries already faring relatively well (e.g. Finland, Australia and the Netherlands).
Figure 6.3. Reading skills of students aged 15 have declined in one-quarter of OECD countries since 2000
Note: † indicates that the country falls below the OECD average, * indicates that the country is above. Countries with no accompanying mark are not statistically different from the OECD average. The PISA reading scores are measured on a scale that is normalised to 500 for the OECD average. Normalisation is established the in the first year a subject is introduced as the major testing domain – 2000 for reading – to allow for year-on-year comparability. For this reason, the OECD average may not exactly equal 500 in any given year. The OECD average does not include Spain, due to missing data. The trend is reported only for countries that have recorded significant improvements or deteriorations since 2000, i.e. no starting point is shown for countries whose change in test scores is not significant. See Box 6.1 for more details on the calculation of average trends.
Source: (OECD, 2019[1]), PISA 2018 Results (Volume I): What Students Know and Can Do, PISA, OECD Publishing, Paris, https://doi.org/10.1787/5f07c754-en.
PISA science scores
For science, average PISA scores are highest in Estonia, Japan, Finland, Korea and Canada, and lowest in Colombia, Mexico and Chile (Figure 6.4). Since 2006, science scores have increased in only three OECD countries (Colombia, Portugal and Turkey), while they have fallen in around half, a pattern that mirrors (with greater intensity) the one already observed for maths and reading.
Overall, only two OECD countries – Colombia and Portugal – have improved their scores in all three subjects over the past decade or so. An additional four (Estonia, Israel, Poland and the Russian Federation) have improved their scores in both reading and maths. Seven countries have seen their scores deteriorate in all three subjects: Australia, Finland, Iceland, Korea, the Netherlands, New Zealand and the Slovak Republic.
Figure 6.4. Science skills of students aged 15 have declined in 18 OECD countries and improved in 3, since 2006
Note: † indicates that the country falls below the OECD average, * indicates that the country is above. Countries with no accompanying mark are not statistically different from the OECD average. The PISA science scores are measured on a scale that is normalised to 500 for the OECD average. Normalisation is established the in the first year a subject is introduced as the major testing domain – 2006 for science – to enable comparisons over time. For this reason, the OECD average may not exactly equal 500 in any given year. The trend is reported only for countries that have recorded significant improvements or deteriorations since 2006, i.e. no starting point is shown for countries whose change in test scores is insignificant. See Box 6.1 for more details on the calculation of average trends.
Source: (OECD, 2019[1]), PISA 2018 Results (Volume I): What Students Know and Can Do, PISA, OECD Publishing, Paris, https://doi.org/10.1787/5f07c754-en.
The distribution of cognitive skills among students at age 15
On average, across OECD countries, the top-performing students (those who score in the 90th percentile) have PISA scores more than 60% higher than the lowest-performing ones (those in the 10th percentile) (Figure 6.5). Inequalities are typically larger in OECD countries with comparatively poor average performance in all subject areas. Looking across reading, maths and science tests combined, Israel has the highest inequalities between high and low achievers in the OECD (followed by Luxembourg and the Slovak Republic), while several countries with strong performances on average skills (e.g. Estonia, Ireland, Denmark, Finland Japan, Poland and Canada) all have below average inequalities. Korea is an exception, with strong average performance, but also larger inequalities than in other countries performing at that level (Figure 6.6). In Colombia and Mexico, over 30% of students have low scores in all fields, while only 4.2% of students in Estonia do.
Figure 6.5. On average, 15-year-old students at the 90th percentile have cognitive skills around 65% higher than those at the 10th
Note: Vertical inequalities are measured by the ratio of cognitive skills among top performers (those above the 90th percentile) to bottom performers (those below the 10th percentile), for each of the three PISA subject areas. The closer the ratio is to 1, the lower the gap between top and bottom students. The OECD average for reading excludes Spain, due to missing data; the OECD average for maths and science includes all 37 OECD countries.
Source: (OECD, 2019[1]), PISA 2018 Results (Volume I): What Students Know and Can Do, PISA, OECD Publishing, Paris, https://doi.org/10.1787/5f07c754-en.
Figure 6.6. In the average OECD country, 1 in 8 students have low scores on all 3 PISA subjects
Note: Low achievers are those with cognitive skills below Level 2 in all three subjects. The OECD average does not include Spain, due to missing reading score data.
Source: (OECD, 2019[1]), PISA 2018 Results (Volume I): What Students Know and Can Do, PISA, OECD Publishing, Paris, https://doi.org/10.1787/5f07c754-en.
Cognitive skills of adults: PIAAC mean scores in literacy and numeracy
The OECD Programme for the International Assessment of Adult Competencies (PIAAC) assesses the cognitive skills of adults in numeracy, literacy and problem-solving. Unlike PISA, PIAAC results are not standardised to a fixed OECD average level, but measured on a scale from 0 to 500.
Adult literacy and numeracy
The first (and latest available) wave of the OECD Adult Skills Survey was fielded in around 2012. Numeracy scores among adults were highest in Japan, followed by Finland, Belgium (Flanders) and the Netherlands. Chile, Mexico and Turkey have the lowest scores among OECD countries. Cross-country differences in literacy scores across OECD countries generally mirror those in numeracy (Figure 6.7).
As in the case of students’ cognitive skills, OECD countries with high average numeracy scores among adults also have a more equal distribution of scores, i.e. the gap between the top (90th percentile) and bottom (10th percentile) performers is smaller. Japan, for example, has both the highest mean numeracy score among adults and the lowest gap between high and low performers. Conversely, countries with low mean scores – such as Chile and Turkey – have high levels of inequality.
The gap between the top (90th percentile) and bottom (10th percentile) performers is smaller in literacy than it is for numeracy, but the general cross-country pattern holds, and countries with lower mean literacy scores also have larger gaps in performance between top and bottom achievers, as in the case of Chile, Mexico and Turkey. Israel has the third-highest inequality in adult skills, following only Chile and Turkey. Japan has both the highest average literacy score and the lowest level of inequality.
Deprivations in adult literacy and numeracy
In OECD countries, on average, 16% of the adult population have very low levels of literacy and numeracy, defined as scoring at Level 1 or below on both literacy and numeracy assessments (Figure 6.8). High-performing countries tend to have low levels of deprivation – for example, Japanese adults have both the highest average scores in the OECD for literacy and numeracy, and the smallest share of low achievers, at only 3.9%. Similarly, Chile, Mexico and Turkey have some of the lowest average test scores among OECD countries and some of the largest shares of low achievers – 48.2%, 46.1% and 39%.
Figure 6.7. Differences in literacy scores across OECD countries generally mirror those in numeracy
Note: * indicates that the PIAAC score is significantly above the OECD average; † indicates that the PIAAC score is significantly below the OECD average. Countries with no mark are not statistically different from the OECD average. Data refer to 2011-12 for Australia, Austria, Belgium, Canada, the Czech Republic, Denmark, Estonia, Finland, Germany, Ireland, Italy, Japan, Korea, the Netherlands, Poland, the Russian Federation, the Slovak Republic, Spain, Sweden, the United Kingdom and the United States; 2012 for France; and 2014-15 for Chile, Greece, Israel, Lithuania, New Zealand, Slovenia and Turkey. Data for Belgium refer to Flanders; those for the United Kingdom distinguish between England and Northern Ireland, and those for the Russian Federation exclude the Moscow municipal area. The OECD average excludes Colombia, Iceland, Latvia, Luxembourg, Portugal, and Switzerland, due to a lack of data. England and Northern Ireland are both included in the OECD average, as is a simple average of both United States time series (2012/14 and 2018).
Source: (OECD, 2016[2]), Skills Matter: Further Results from the Survey of Adult Skills, OECD Skills Studies, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264258051-en.
Figure 6.8. Almost 50% of the adult population performs at or below level 1 in the worst-performing OECD countries
Note: The OECD average excludes Colombia, Iceland, Latvia, Luxembourg, Portugal, and Switzerland, due to a lack of data. England and Northern Ireland are both included in the OECD average, as is a simple average of both United States time series (2012/14 and 2018).
Source: (OECD, 2016[2]), Skills Matter: Further Results from the Survey of Adult Skills, OECD Skills Studies, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264258051-en.
Knowledge and Skills inequalities: gaps between population groups
There are persistent gender differences in knowledge and skills
In the large majority of OECD countries, the average PISA mathematics score for boys is higher than for girls. Although this difference is not statistically significant in many countries, girls are underrepresented among high-achieving maths students, especially in the lowest-performing countries overall (Breda, Jouini and Napp, 2018[3]). Conversely, average reading scores are consistently higher for girls than for boys (Figure 6.9) – though this gap has been narrowing over time, due to deteriorations in the average scores of girls rather than to improvements in the average scores of boys (OECD, 2019[4]).
Figure 6.9. Girls outperform boys on reading in all OECD countries
Note: Gender ratios with values above 1 indicate better outcomes for girls. The OECD average does not include Spain, due to missing data.
Source: (OECD, 2019[4]), PISA 2018 Results (Volume II): Where All Students Can Succeed, PISA, OECD Publishing, Paris, https://doi.org/10.1787/b5fd1b8f-en.
The gender differences in cognitive skills among adults tell a similar, though not identical, story. Men’s numeracy scores (268) exceed those for women (256) in all OECD countries, with gender gaps that are usually much larger than those for 15-year-old students. Inequalities in adult literacy are more diverse: the OECD average scores for both men and women are very similar (267 and 265, respectively); women’s scores exceed those of men by at least five points only in Poland, while men’s scores exceed those of women by at least five points in seven countries.
Older adults do less well on literacy and numeracy tests
Older adults (aged 45 to 65) perform worse on numeracy assessments (with an OECD average of 251 points) than either middle-aged (25 to 44) or youth (16 to 24) cohorts (who score 270 and 266, respectively) (Figure 6.10). A similar pattern holds for the literacy tests.
Parents’ educational attainment is associated with cognitive skills at age 15
Fifteen year-old students whose parents have only attained a primary level of education perform worse on PISA assessments in reading (OECD average of 417), as compared to their classmates whose parents have a secondary (463) or tertiary (489) education (Figure 6.11). This pattern of results also holds for maths and science performance.
Figure 6.10. Older adults perform worse on numeracy than their younger peers in all OECD countries
Note: The OECD average excludes Colombia, Iceland, Latvia, Luxembourg, Portugal, and Switzerland, due to a lack of data. England and Northern Ireland are both included in the OECD average, as is a simple average of both United States time series (2012/14 and 2018).
Source: (OECD, 2016[2]), Skills Matter: Further Results from the Survey of Adult Skills, OECD Skills Studies, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264258051-en.
Figure 6.11. Fifteen year-old students with primary educated parents perform worse than their peers with better educated parents
Note: Parental education is classified according to the International Standard Classification of Education (ISCED) framework: primary education includes no education, ISCED 1 and ISCED 2; secondary education includes ISCED 3B and ISCED 3A, 4; tertiary education includes ISCED 5B and ISCED 5A/6. The OECD average does not include Spain due to missing data.
Source: (OECD, 2019[4]), PISA 2018 Results (Volume II): Where All Students Can Succeed, PISA, OECD Publishing, Paris, https://doi.org/10.1787/b5fd1b8f-en.
Box 6.1. Measurement and the statistical agenda ahead
Knowledge and skills are about what people know and can do. Literacy and numeracy are foundational skills that enable full participation in daily activities such as work and leisure, but other skills such as science and digital skills are increasingly becoming a basic requirement for inclusion in economic and social activities. Beyond these core building blocks, the range of knowledge and skills that can contribute to well-being is wide, from job-specific skills to parenting. Non-cognitive abilities, such as social and emotional skills – including resourcefulness, perseverance, adaptability and team-working – can also be considered as essential competencies. The indicators used in this chapter (Table 6.1) are limited to cognitive skills; an important priority for future statistical work is to assess additional aspects of people’s knowledge and skills (below).
Table 6.1. Knowledge and skills indicators considered in this chapter
Student cognitive skills are measured using the 2018 OECD Programme for International Student Assessment (PISA) test scores. PISA assessments are conducted once every three years, with the focal subject cycling between mathematics, reading and science. The most recent PISA round focused on reading. In 2018, PISA tested around 600 000 15-year-old students, representing 32 million students across 79 countries. PISA assessments are normalised such that the OECD average is 500 points, with a standard deviation of 100 points. The normalisation is done in the first year a subject is a focal subject, implying that the value of the OECD average in any given year may not be equal to 500. PISA trends over time are measured as an average of each three-year period (given that assessments are implemented every three years), from the first time a given subject was the focal subject to the present-day assessment. Therefore, for PISA 2018, trends over time are calculated as the average trend from 2003 for mathematics, from 2006 for science and from 2000 for reading. Because PISA assessments are conducted within schools, they capture the cognitive ability only of 15-year-olds who are currently enrolled in school. These tests thus do not include drop-outs, or home-schooled students.
Adult cognitive skills are measured using the OECD Programme for the International Assessment of Adult Competencies (PIAAC) assessments in literacy and numeracy. The first cycle of PIAAC comprised three rounds, running from 2011 to 2017, covering over 220 000 adults in 38 countries. Adults are administered assessments of numeracy, literacy and problem-solving skills, with possible scores ranging from 0 to 500 (unlike PISA, PIAAC results are not normalised, meaning that the highest possible score is 500). At present, no time series is available for adult skills: Cycle 2 of PIAAC is planned for 2021-22, with results expected in 2023. Data for Belgium are limited to Flanders, and those for the United Kingdom to England and Northern Ireland.
Correlations among Knowledge and Skills indicators
Knowledge and skills are highly correlated across subjects, and across age groups: countries with higher levels of maths, reading and science for students aged 15 also have higher literacy and numeracy among adults (Table 6.2). Correlations are particularly high (above 0.94) among maths, science and reading skills at age 15, and between numeracy and literacy in adulthood. The weakest association (0.64) is between reading skills at age 15 and adult numeracy.
Table 6.2. Knowledge and Skills indicators are strongly correlated
Bivariate correlation coefficients among the Knowledge and Skills indicators
Note: Table shows the bivariate Pearson’s correlation coefficient; values in parentheses refer to the number of observations (countries). * Indicates that correlations are significant at the p<0.10 level, ** that they are significant at the p<0.05 level, and *** at the p<0.01 level.
Statistical agenda ahead
In the current indicator set, Knowledge and Skills for both 15-year-old students and adults are primarily measured through literacy and numeracy measures. However, there are a number of other measures that capture knowledge and skills – including ability to problem solve, logical reasoning and non-cognitive skills – that are not captured by the indicators used in this chapter. PIAAC has rolled out an adaptive problem-solving component of its assessments that will be included in forthcoming rounds. The OECD Study on Social and Emotional Skills (SSES) aims to capture non-cognitive abilities in childhood and adolescence: the project began in mid-2017, thus data are not yet available for this publication.
References
[3] Breda, T., E. Jouini and C. Napp (2018), “Societal inequalities amplify gender gaps in math”, Science, Vol. 359/6381, pp. 1219-1220, http://dx.doi.org/10.1126/science.aar2307.
[1] OECD (2019), PISA 2018 Results (Volume I): What Students Know and Can Do, PISA, OECD Publishing, Paris, https://dx.doi.org/10.1787/5f07c754-en.
[4] OECD (2019), PISA 2018 Results (Volume II): Where All Students Can Succeed, PISA, OECD Publishing, Paris, https://dx.doi.org/10.1787/b5fd1b8f-en.
[2] OECD (2016), Skills Matter: Further Results from the Survey of Adult Skills, OECD Skills Studies, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264258051-en.