OECD Skills Outlook 2025

Reader’s guide

Assessment data

This report uses various data sources. The most frequently used data source is the 2023 Survey of Adult Skills, a product of the OECD Programme for the International Assessment of Adult Competencies (PIAAC).

The 2023 Survey of Adult Skills (PIAAC) covers the following countries and economies: Austria, Canada, Chile, Croatia, Czechia, Denmark, England (United Kingdom), Estonia, Finland, the Flemish Region (Belgium), France, Germany, Hungary, Ireland, Israel, Italy, Japan, Korea, Latvia, Lithuania, the Netherlands, New Zealand, Norway, Poland, Portugal, Singapore, the Slovak Republic, Spain, Sweden, Switzerland, and the United States.

Although two of the participating entities are economies (England [United Kingdom]) and the Flemish Region [Belgium]), for the sake of brevity, we refer to participants as countries rather than countries and economies throughout the report.

Other international large-scale assessment data used in this report include:

• Programme for International Student Assessment (PISA)

• International Civic and Citizenship Education Study (ICCS)

• International Computer and Information Literacy Study (ICILS)

• Trends in International Mathematics and Science Study (TIMSS)

• Progress in International Reading Literacy Study (PIRLS).

International averages and country-level results

The report presents international averages for the OECD (“OECD average”), the European Union (“EU average”) and all countries participating in the 2023 Survey of Adult Skills (PIAAC) (“PIAAC 2023 average”).

The international average for the OECD is based on data from all available OECD member countries and economies, while the international average for the European Union is based on data from all available European Union member countries and economies. The international average for PIAAC is based on all participating (OECD and non-OECD) countries and economies in the 2023 Survey of Adult Skills (PIAAC).

The OECD average, the EU average and the PIAAC 2023 average correspond to the arithmetic mean for the specific countries for which data is available of a specific data source.

For ease of exposition, many results provided in the figures and tables of this report display international averages. Underlying country averages are provided in the annex of each chapter.

Measurement of social and emotional skills

The conceptual framework for the measurement of social and emotional skills in the 2023 Survey of Adult Skills (PIAAC) is the Big Five model of personality (Soto and John, 2017[1]). The Big Five model distinguishes five broad personality domains: extraversion, agreeableness, conscientiousness, emotional stability and open-mindedness.

This report presents results from the 15-item extra-short form of the Big Five Inventory-2 (BFI-2-XS), as it is available for all 29 countries that chose to include the section on social and emotional skills in the background questionnaire.

In the BFI-2-XS, each personality domain is measured through three different facets (i.e. components of the personality domain), while each facet is defined by one item (i.e. a statement to which individuals respond to on a five-point Likert scale, from “strongly agree” to “strongly disagree”). Therefore, measures of individuals’ social and emotional skills are based on individuals’ self-reports across 15 items, one per facet. Table 1 provides details on the facets and items that make up the personality domains.

Table 1. Description of the social and emotional skills measures

Personality domain	Facet	Items
Extraversion	Assertiveness	I am dominant, act as a leader
	Energy level	I am full of energy
	Sociability	* I tend to be quiet
Agreeableness	Compassion	I am compassionate, have a soft heart
	Respectfulness	* I am sometimes rude to others
	Trust	I assume the best about people
Conscientiousness	Organisation	* I tend to be disorganised
	Productiveness	* I have difficulty getting started on tasks
	Responsibility	I am reliable, can always be counted on
Emotional stability	Anxiety	* I worry a lot
	Depression	* I tend to feel depressed, blue
	Emotional volatility	I am emotionally stable, not easily upset
Open-mindedness	Aesthetic sensitivity	I am fascinated by art, music or literature
	Intellectual curiosity	* I have little interest in abstract ideas
	Creative imagination	I am original, come up with new ideas

Note: The table represents the 15-item extra-short form of the Big Five Inventory-2 (BFI-2-XS), which measures each of the five domains with three items, one per facet. These short versions of the Inventory were derived from the BFI-2 by Soto and John (2017[1]; 2017[2]), a contemporary revision of the BFI-2-S by John et al (1991[3]).

Items marked with an asterisk * are reverse coded – respondents who agree less with these items score higher on the respective personality domain.

Source: OECD (2025[4]), Survey of Adult Skills 2023 Technical Report, https://doi.org/10.1787/80d9f692-en.

In the 2023 Survey of Adult Skills, scores for each personality domain (extraversion, agreeableness, conscientiousness, emotional stability, open-mindedness) were achieved by calculating the mean response per respondent for a particular domain [see (OECD, 2025[4])].

Comparison between groups and countries across different skills

The possibility of making comparisons between countries, and the meaning of such comparisons, depends not only on the extent to which comparable instruments and administration conditions are implemented, but also on whether such instruments measure the same underlying skills. In the presence of full comparability (or “scalar invariance”), the relationship (or “loading”) between each question, its underlying concept (“skill”), and the overall starting point (or “intercept”) are comparable across countries. As a result, scales can be constructed using between-country centring, meaning that the same mean and standard deviation are common across all countries.

By contrast, when only the relationship (or “loading”) between each question and its underlying concept are comparable across countries, individuals’ position in different countries cannot be compared directly, although it is possible to compare disparities between groups. Therefore, within-country centring was used in PIAAC to ensure that no between-country comparisons would be conducted for instruments that did not achieve full invariance.

In practice, between-country centring was used in the 2023 Survey of Adult Skills for information- processing skills (also referred to as core 21st-century skills in this report), while within-country centring was used for social and emotional skills. Because willingness to delay gratification was measured using a single item, it is considered as an observed scale rather than a latent construct, because no invariance testing could be conducted. Figure 1 illustrates for three theoretical countries, A, B and C, the consequences of adopting a between- and a within-country centring approach.

Figure 1. Between- and within-country centring

Note: The left side of the illustration, “Between-country centring”, illustrates an example where skills are comparable across countries, and mean scores and standard deviations are calculated using information across all countries. The right side of the illustration, “Within-country centring”, illustrates an example where skills are not comparable across countries, and averages and standard deviation are calculated separately for each country.

Information-processing skills and willingness to delay gratification are between-country centred

PIAAC ensures the comparability of assessment questions across countries and population groups through qualitative review of items, prescribed procedures for translation, and quantitative indicators aimed at identifying systematic deviations from expected results given the assumption of full comparability Information-processing skills are therefore standardised using a common mean and a common standard deviation across all countries and economies participating in the 2023 Survey of Adult Skills. This means that the scores of the entire participating population are pooled together when deriving means and standard deviations (OECD, 2025[4]). It is therefore possible to identify if a group of individuals living in country A have higher literacy achievement than a group of individuals living in country B and if the variation in literacy skills in the population in country A is more dispersed than in the population in country B.

Social and emotional skills are within-country centred

Invariance analyses of PIAAC data reveal that scalar invariance was not achieved by the PIAAC social and emotional skills instruments, whereas metric invariance was achieved. This likely reflects the fact that how people express themselves varies across countries (Cheng, Cheung and Montasem, 2014[5]). An extrovert in the cultural/social context of country A may only be considered moderately outgoing in country C when compared to those around them, despite behaving in the exact same way in both countries. Moreover, the value of high sociability and how it is interpreted by people in a community depends on broader social values, which differ across countries.

To account for the impact of cultural/social contexts when deriving measures of social and emotional skills, the scores of individuals were standardised in relation to others in the country in which they took the survey. As a result, whereas differences between groups (so called “effect sizes”) can be compared, it is not possible to compare averages (i.e. mean scores of different groups) across countries (OECD, 2025[4]). The population scores of each PIAAC-participating country and economy were considered separately when deriving the measures of standard deviation and mean score of the population such that each country’s mean and standard deviation are the same irrespective of average level of responses in the underlying questions in different countries [see (OECD, 2025[4])].

Interpreting the magnitude of differences between groups

To interpret the magnitude of standardised differences between groups for information-processing skills and delayed gratification (so called “standardised differences”) and in social and emotional skills (so called “effect sizes”) this report relies on “Cohen’s d” benchmarks. Standardised differences are interpreted as small effects if Cohen’s d (the standardised difference) is below 0.2, as medium for values between 0.2 and 0.5, as large for valued of 0.5 and greater. A different way to evaluate differences is to evaluate these in light of differences resulting from educational interventions. For example, the standardised difference in achievement over the course of a school year among participants in PISA (around 0.20 SD) (Avvisati and Givord, 2023[6]) and the majority of educational interventions have effects sizes on standardised achievement that are below the 0.2 SD threshold (Kraft, 2020[7]).

Interpretation of country-level results by immigrant background

Immigrant populations differ widely across countries and economies in terms of their size, composition, and migration histories. Readers are therefore encouraged to consult the country-level results provided in the annex tables of each chapter when interpreting findings by immigrant background. Caution is also advised when comparing results across countries and economies, since differences in composition and migration histories could not be fully accounted for in the analyses.

References

[6] Avvisati, F. and P. Givord (2023), “The learning gain over one school year among 15-year-olds: An international comparison based on PISA”, Labour Economics, Vol. 84, p. 102365, https://doi.org/10.1016/j.labeco.2023.102365.

[5] Cheng, C., M. Cheung and A. Montasem (2014), “Explaining differences in subjective well‐being across 33 nations using multilevel models: Universal personality, cultural relativity, and national income”, Journal of Personality, Vol. 84/1, pp. 46-58, https://doi.org/10.1111/jopy.12136.

[3] John, O., E. Donahue and R. Kentle (1991), “Big Five Inventory”, PsycTESTS Dataset (database), https://doi.org/10.1037/t07550-000 (accessed on 6 March 2025).

[7] Kraft, M. (2020), “Interpreting effect sizes of education interventions”, Educational Researcher, Vol. 49/4, pp. 241-253, https://doi.org/10.3102/0013189x20912798.

[4] OECD (2025), Survey of Adult Skills 2023 Technical Report, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/80d9f692-en.

[1] Soto, C. and O. John (2017), “Short and extra-short forms of the Big Five Inventory–2: The BFI-2-S and BFI-2-XS”, Journal of Research in Personality, Vol. 68, pp. 69-81, https://doi.org/10.1016/j.jrp.2017.02.004.

[2] Soto, C. and O. John (2017), “The next Big Five Inventory (BFI-2): Developing and assessing a hierarchical model with 15 facets to enhance bandwidth, fidelity, and predictive power.”, Journal of Personality and Social Psychology, Vol. 113/1, pp. 117-143, https://doi.org/10.1037/pspp0000096.