Introduction

In countries around the world, educational attainment is one of the most important predictors of societal progress across a range of outcomes, from health outcomes to economic development to prosocial factors such as civic and social engagement1. The proportion of a population attaining key levels of education within many countries has been studied by international organizations such as the Organisation for Economic Co-operation and Development (OECD) and the United Nations Educational, Scientific and Cultural Organization (UNESCO)2,3. The current work will compare estimates of educational attainment in the Global Flourishing Study (GFS), a survey of over 200,000 individuals in 22 countries weighted to be nationally representative, with existing sources and provide more nuanced information about the distribution of educational attainment across countries.

The current work extends the findings from OECD and UNESCO reports on national trends in educational attainment in several ways. First, the current study enables a consistent comparison of education attainment across countries based on a common reporting system (such as the International Standard Classification of Education, ISCED). A commonly chosen cutoff point for evaluating educational attainment is at least a “tertiary” level of education (ISCED5 or greater), which is prone to definitional differences across organizations, leading to difficulties in comparing estimates across sources when a common data collection method is used. In the GFS data, educational attainment is primarily assessed using what Gallup Inc (the primary data collectors) provided a variable harmonizing educational levels across countries (Up to 8 years, 9 to 15 years, and 16 + years of education) to align with their existing cross-national surveys such as the Gallup World Poll4. The highest level of achievement (16 + years of education) is not exactly equal to the ISCED5 (2–3 years of post-secondary education) or ISCED6 (bachelor’s degree or equivalent leading to a mismatch in our data with existing sources from the OECD2. The threshold of 16 + years of education for attaining a tertiary education is more extreme than the ISCED5 (2–3-year short cycle), resulting in an expected lower educational attainment nationally compared to the general definition of tertiary education. While the misalignment in definitions limits the current paper’s generalizability cross-nationally, the results we present on variation in attainment within subgroups across countries can still be enlightening about the current state of educational attainment within the countries included in the GFS.

The current paper contributes to expanding the population for which educational attainment is evaluated. In the OECD report5, a focus is limited to the 25–64-year-old population, which is reasonable considering the population of adults commonly considered part of the workforce. However, this excludes the 18–24 and 65 + portion of the adult population across countries. While the 18–24-year-old population is less likely to have completed tertiary education, the 65 + population does not have this constraint but is more likely to be retired. Our current work incorporates the entirety of the adult population, thus expanding the breadth of generalizability of our results beyond the age restrictions of existing sources; though, we will evaluate an age-restricted sub-subsample to evaluate the consistency of our results with existing sources.

Further, the GFS incorporates countries not included in previous analyses. Thus, this work presents complementary data to fill gaps in our understanding of educational attainment around the world. For example, the OECD5 report includes estimates of educational attainment for 16 of the 22 countries included in the GFS, with the GFS contributing new data for Egypt, Hong Kong (S.A.R. of China), Kenya, Nigeria, Philippines, and Tanzania. Extending the breadth of the world population in which we have educational attainment information provides important insights into educational equity across the global population.

Moreover, the data from the GFS now provide an opportunity to examine the context in which educational attainment is achieved for different populations as well. The data reported in this study are weighted to be nationally representative of each country, providing insights into the population characteristics of subgroups6,7. Describing the demographic groups for which individuals have attained greater education is a major descriptive advance of the current work beyond the presently available country-level estimates of educational attainment. Descriptive statistics of educational attainment by subgroups across 22 countries is a major contribution of this work. The greater nuance of the context, or variation in educational attainment across demographic characteristics and countries, provides a useful backdrop for potential new avenues of exploration of educational attainment.

Variation in educational attainment

To better promote equity in health and well-being, it is essential to understand the distribution of social determinants of health and well-being8. Education is a social determinant of many key health and well-being outcomes9,10. Educational attainment varies on average across countries, but an important consideration in these differences is where the differences lie within a country. For instance, differences in the qualifications of migrants can be crucial in how individuals become integrated into a new system11,12. Other contextual factors include family income, familial wealth, and parents’ occupation, which shape the context in which an individual’s education is attained, and these contextual nuances can lead to differential outcomes in current income and occupation13. Furthermore, persistent achievement gaps between students from affluent and disadvantaged backgrounds reflect entrenched socio-economic educational disparities, for example, within the United States14,15. Identifying how educational attainment varies across other sociodemographic characteristics will similarly provide insights into potential disparities in a wider cross-national context.

Cultural attitudes and societal norms are additional aspects of individuals’ context that influence educational attainment by shaping their perceptions of the value of education and their aspirations for the future16,17. The habits and dispositions within a culture can profoundly impact educational outcomes, potentially perpetuating social stratification within different cultures18. Additionally, Silventoninen et al.19 provided evidence suggesting that many differences in educational attainment can be directly explained by environmental factors. Social stratification and cultural context can shape educational aspirations and attainment levels across societies20.

Moreover, governmental policies and educational systems play a key role in shaping the educational landscape, with varying approaches yielding disparate outcomes21. Different political contexts and social welfare systems may also lead to variations in the population distribution of educational attainment and related conditions across countries. For example, a consistent benefit of higher educational attainment is the potential for greater earnings. Adults (25–64-year-olds) in OECD countries with upper secondary or post-secondary non-tertiary attainment working full-time and for the full year earn, on average, about 25% more than those without such qualifications. The difference is over 40% between Chile, Colombia, the Czech Republic, and Germany.

In contrast, in Finland, workers with upper secondary or post-secondary non-tertiary attainment earn almost the same as those below upper secondary attainment5. The current work will provide additional insights into how higher educational attainment is related to higher earnings within 22 culturally and geographically diverse countries. The results of the present study will help shed light on how educational attainment varies in a cross-national context, leading to insights into where additional effort to improve educational outcomes may be needed within the countries included in the GFS.

Research questions and hypotheses (Pre-registered)

The current study is primarily descriptive. We have three pre-registered study research questions:

Research Question #1: What are the distributions and descriptive statistics of key demographic factors (age, gender, marital status, employment, immigration status, and income) in our diverse, international sample across 22 countries?

Hypothesis #1: The distributions and descriptive statistics of key demographic features (age, gender, marital status, employment, immigration status) will reveal diverse patterns across our international sample from 22 countries.

Research Question #2: How does the proportion of individuals in a country achieving up to a tertiary level of educational attainment order across different countries?

Hypothesis #2: The mean levels of educational attainment will vary meaningfully across different countries.

Research Question #3: How does the proportion of individuals in a country achieving a tertiary level of educational attainment vary across different demographic categories such as age, gender, marital status, employment, immigration status, and income?

Hypothesis #3: Educational attainment will vary across different demographic categories, such as age, gender, marital status, employment, and immigration status. These differences will also vary by country.

The assessment of income varied by country, leading to different categories and labels appropriate for each country (see GFS Codebook, https://osf.io/cg76b). We utilized the country-specific income brackets to break down how the proportion of educational attainment (16 + years) varied among these income brackets. We will comment on these country-specific analyses, but we did not aggregate over countries for these proportions due to the different income brackets.

Research questions and hypotheses (Post-registration)

After exploring the data, we developed the following questions and analyses to help provide either additional insight into the GFS data (RQ#4) or sensitivity analyses of the conclusions regarding the methods and cut-points used in the country-specific educational categories (RQ#5).

Research Question #4: Do the country-level estimates of educational attainment obtained from the GFS align with existing estimates of same-country educational attainment obtained from the OECD12?

Hypothesis #4: Once the samples are appropriately conditioned to the same age range and level of education, the GFS estimates will be closely aligned with existing estimates of population-level educational attainment.

Research Question #5: Are the country-level estimates of educational attainment evaluated in RQ2 robust to alternative approaches to estimating country-level educational attainment? Alternative methods include estimating the latent educational attainment distribution based on the methods of Reardon and colleagues22 and using a re-defined education variable to obtain a high-school equivalence measure for each country instead of at a tertiary level.

Hypothesis #5: The rank-ordering of countries will be consistent across approaches to estimating educational attainment.

Methods

The methods described below have been adapted from VanderWeele et al.23. Further methodological detail is available elsewhere6,7,24,25,26,27,28,29.

Data

The Global Flourishing Study (GFS) is a study of 202,898 participants from 22 geographically and culturally diverse countries, with nationally representative sampling within each country, concerning the distribution of determinants of well-being. Wave 1 of the data included the following countries and territories: Argentina, Australia, Brazil, Egypt, Germany, Hong Kong, India, Indonesia, Israel, Japan, Kenya, Mexico, Nigeria, the Philippines, Poland, South Africa, Spain, Sweden, Tanzania, Turkey, United Kingdom, and the United States. The countries were selected to (a) maximize coverage of the world’s population, (b) ensure geographic, cultural, and religious diversity, and (c) prioritize feasibility and existing data collection infrastructure. Data collection was carried out by Gallup Inc. Data for Wave 1 were collected principally during 2023, with some countries beginning data collection in 2022 and exact dates varying by country6. Sample sizes vary by country based on the availability of local resources and the incorporation of existing panel studies, and additional information on sample sizes and data collection is provided elsewhere by Padgett and colleagues7. Four additional waves of panel data on the participants will be collected annually from 2024–2027. Survey items included aspects of well-being such as happiness, health, meaning, character, relationships, and financial stability30, along with other demographic, social, economic, political, religious, personality, childhood, community, health, and well-being variables. The data are publicly available through the Center for Open Science (COS, https://www.cos.io/gfs). During the translation process, Gallup adhered to the TRAPD model (translation, review, adjudication, pretesting, and documentation) for cross-cultural survey research (ccsg.isr.umich.edu/chapters/translation/overview). Ethical approval was granted by the institutional review boards at Baylor University and Gallup, all participants provided informed consent, and all methods were performed in accordance with the relevant guidelines and regulations.

Sampling design

The precise sampling design to ensure nationally representative samples varied by country6. In most countries, local field partners were guided in implementing a probability-based face‑to-face or telephone methodology to recruit panel members. Recruitment involved an intake survey that mainly gathered basic demographics and information for recontact. Shortly following recruitment, participants received invitations to participate in the annual survey via phone or online. Three major sampling frames were used for recruitment in the GFS, namely a probability-based sample, a non-probability-based sample, or a combination of the two. Post-stratification and nonresponse adjustments were carried out within each country separately using census data or a reliable secondary source7. Additional information about the sampling design is available in Ritter et al.6 and Padgett et al.7.

Measures

Demographics variables

Continuous age was classified as 18–24, 25–29, 30–39, 40–49, 50–59, 60–69, 70–79, and 80 or older. Gender was assessed as male, female, or other. Marital status was assessed as single/never married, married, separated, divorced, widowed, and domestic partner. Employment was assessed as employed, self-employed, retired, student, homemaker, unemployed and searching, and other. Service attendance was assessed as more than once/week, once/week, one to three times/month, a few times/year, or never. Immigration status was dichotomously assessed with: “Were you born in this country or not?” Religious tradition/affiliation with categories of Christianity, Islam, Hinduism, Buddhism, Judaism, Sikhism, Baha’i, Jainism, Shinto, Taoism, Confucianism, Primal/Animist/Folk religion, Spiritism, African-Derived, some other religion, or no religion/atheist/agnostic; precise response categories varied by country31. Racial/ethnic identity was assessed in some, but not all, countries, with response categories varying by country. Household income was assessed using country-specific income brackets to align with local currency. For additional assessment details, see the COS GFS codebook or Crabtree et al.24,25.

Outcome variable

As assessed within the GFS, educational attainment utilized country-specific education categories24,25. These categories are reported on in our online supplement. For example, in India, years of education were assessed using the categories: Illiterate; Below SSC; SSC/HSC; Some college but did not graduate; Graduate/Post Graduate—General; or Graduate/Post Graduate – Professional. Gallup, the managers of data collection for the GFS, provided a simplified coarsened variable of educational attainment that aligns with their Gallup World Poll annual global survey that collapsed the country-specific categories down to three ordered categories: up to 8 years, 9–15 years, and 16 + years of education. In our primary analyses, as preregistered, we dichotomized educational attainment at 16 + years of education (i.e., a tertiary education) versus lower levels of attainment.

Additionally, as part of research question 6, we used the Gallup-provided country-specific educational categories to create a high-school equivalence completion indicator of educational attainment. The categories selected for each country were: Argentina: Complete secondary school; Australia: Year 12 (Higher School Certificate HSC or equivalent) (usually 18 years old); Brazil: High school complete/Media education complete; Egypt: Completed secondary; Germany: High school diploma/A-levels, age 11–18/19, grades 5–12/13—qualifies for college/university; Hong Kong: Completed senior/higher secondary (GCE O Level (Form 5), GCE A Level (Form 7 in old structure), DSE Level; India: SSC/HSC; Indonesia: Diploma; Israel: 12 years with matriculation; Japan: Higher prof or Junior college; Kenya: Sixth year of secondary education; Mexico: High school/technical career complete; Nigeria: Sixth year of secondary education; Philippines: High school; Poland: Secondary degree, Liceum Technikum; South Africa: Fifth year of secondary education (Grade 12); Spain: Second level of EGB, Secondary School Graduate or ESO complete (Certificate of success in EGB course); Sweden: Upper secondary education (ISCED 3); Tanzania: Sixth year of secondary education; Turkey: High school / Vocational school at high school level; United Kingdom: Upper Secondary School; and United States: High school graduate (Grade 12 with diploma or GED certificate). The selected country-specific categories are highlighted in our Online Supplement in the context of the other categories within each country.

Analysis

Descriptive statistics for the full sample, weighted to be nationally representative within each country, were estimated for each demographic variable. Nationally representative proportions for educational attainment (16 + years) were estimated separately for each country and ordered from highest to lowest, along with 95% confidence intervals. Variations in proportions in educational attainment (16 + years) across demographic categories were estimated, with all analyses initially conducted by country (online supplement). Primary results consisted of random effects meta-analyses of country-specific educational attainment proportions (16 + years) within each specific demographic category32,33,34. The estimated mean proportion with 95% confidence intervals, standard errors, lower and upper limits of a 95% prediction interval across countries, heterogeneity (τ), and I2 for evidence concerning variation within a particular demographic variable across countries was reported35. Forest plots of estimates are available in the online supplement. All meta-analyses were conducted in R36 using the metafor package37. Within each country, a global test of variation of outcome across levels of each particular demographic variable was conducted, and a pooled p-value38 across countries was reported concerning evidence for variation within any country. Bonferroni corrected p-value thresholds are provided based on the number of demographic variables39,40. Religious affiliation/tradition and race/ethnicity were used, when available, as control variables within country, but were not included in the meta-analyses since the availability of these response categories varied by country. As a supplementary analysis, population weighted meta-analyses were also conducted. The population-weighted meta-analysis effectively treats each person in the 22 countries equally, rather than treating each of the 22 countries equally as the random effects meta-analysis does. All analyses were pre-registered with COS before data access (https://doi.org/https://doi.org/10.17605/osf.io/p479h), and all code to reproduce analyses is openly available in an online repository27.

Missing data

Missing data on all variables was imputed using multivariate imputation by chained equations, and five imputed datasets were used41,42. To account for variation in the assessment of certain variables across countries (e.g., religious affiliation/tradition and race/ethnicity), the imputation process was conducted separately in each country. This within-country imputation approach ensured that the imputation models accurately reflected country-specific contexts and assessment methods. Sampling weights were included in the imputation models to account for specific variable missingness that may have been related to the probability of inclusion in the study.

Accounting for complex sampling design

The GFS used different sampling schemes across countries based on the availability of existing panels and recruitment needs6. All analyses accounted for the complex survey design components, including weights, primary sampling units, and strata. Additional methodological detail, including accounting for the complex sampling design, is provided elsewhere7,26.

Additional exploratory analyses (Added post pre-registration)

Additional analyses were conducted after examining these data (i.e., post–pre-registration). 1) Comparing the country-level estimates of educational attainment with OECD estimates of same-country educational attainment obtained from OECD12. 2) The recategorized educational attainment variable was used as part of a sensitivity analysis of the country orderings. The alternative ordering of countries based on our created high-school equivalency indicator is provided in our online supplement (Table S24), and the meta-analysis of the demographic characteristics is reported in Table S25. 3) Estimates of the latent educational attainment distribution means and variances using the trichotomized levels of education based on the heteroskedastic ordered probit model22 are also presented in our online supplement (Table S26), along with additional technical details.

Results

Descriptive statistics

Table 1 describes the sample characteristics of the full sample, including the combined 22 countries. The estimates provided in Table 1 are, on average, nationally representative of each country but not meant to be an internationally representative estimate (i.e., representative of the population of all 22 countries). The sample had a balanced representation of female (51%) and male (48%), and a small proportion of other gender (< 1%). Most participants were married (52%), employed by an employer (39%), had 9–15 years of education (57%), never attended religious services (37%), and were native-born (94%). Sample sizes in each country ranged from 1,473 (Turkey) to 38,312 (United States). Participant characteristics for each country were shown in Supplementary Table S1A to S22A. Online supplement Table S23 provides the population-weighted estimated proportions that are more reflective of an internationally representative estimate of the population characteristics.

Table 1 Nationally representative descriptive statistics of the observed sample.
Full size table

countries ordered by proportions of educational attainment

The overall country-level estimated proportion of the population attaining a tertiary level (16 + years) of education is shown in Table 2. The estimated proportion ranges from 0.01 in India and Tanzania to 0.53 in Israel. Figure 1 complements Table 2 by ordering countries by the estimated proportion of the population attaining only up to 8 years of education, with the middle category also reported on for completeness. Although, based on these data, Israel has the highest estimated proportion of the population attaining a high level of education, a greater proportion of the population only obtained up to 8 years of education relative to several other countries (see Fig. 1). Additionally, countries with the highest proportions of the population completing 16 + years of education were predominantly European and North American countries, except Israel, Australia, and Japan. The countries with lower proportions of the population attaining 16 + years of education included India and African nations such as Tanzania, Nigeria, Kenya, South Africa, Indonesia, and the Philippines. Geographically close countries such as Mexico and the United States sometimes differed by large margins (10 points), and some geographically distant countries sometimes had similar proportions (e.g., South Africa and the Philippines). These results provide interesting avenues for comparison with existing estimates of educational attainment (discussed below).

Table 2 Ordered proportions of educational attainment (16 + years).
Full size table
Fig. 1
figure 1

Educational attainment (3 categories) across 22 countries. Note Values along left and right boarder represent the estimated percent of the population within country at Up to 8 Years of Education (left), At least 16+ Years of Education (right), and the middle percent is labeled at the midpoint of the 9-15 Years of Education bar. Countries are ordered by the percent of population that are in the ‘Up to 8 Years of Education’ category.

Full size image

Variation in educational attainment

The results of the random-effects meta-analyses of the proportions within each demographic category attaining 16 + years of education are shown in Table 3. The estimated heterogeneity (\({\varvec{\tau}}\)) of proportion is of significant interest and is backtransformed from the logit scale. The heterogeneity estimates provide evidence for how variable the proportion is across countries and are useful for evaluating relative variability across variable categories. The means of the proportion can be highly variable across countries even if, on average, the proportion is quite low. Our online supplement forest plots illustrate the high degree of heterogeneity for each meta-analyzed proportion (see Figures S1-S31).

Table 3 Random effects meta-analysis of educational attainment (16 + yrs) proportions by demographic category.
Full size table

The prediction intervals similarly provide an avenue for understanding the heterogeneity of the meta-analyzed proportions. The prediction interval estimates the likely range of a new proportion for a randomly selected country. Several wide prediction intervals provided additional evidence of educational attainment variability within a demographic category. The global p-value also evidences the variability in proportions across countries. The p-values were statistically significant for all demographic characteristics, rejecting the null hypothesis of no difference in proportions of attaining 16 + years of education among categories of a demographic variable in all countries. Implying that, even after controlling for multiple tests, there is a statistically significant difference in the proportion among categories in at least one country.

Education was related to household income in nearly all countries (see Tables S1-22b). Evidence of an association was weak in South Africa (p = 0.022), where the proportion of the population attaining 16 + years of education ranged from 0.09 (95% CI: 0.00, 0.17) for those reporting a monthly income of 200 or less rand and the proportion was 0.24 (95% CI: 0.09,0.40) for those reporting a monthly income of 20,001—30,000 rand. Across all countries, income (monthly or annual) was generally related to educational attainment, and the proportions were frequently at or above 0.50 but rarely above 0.75 for the highest income brackets within countries. This provides evidence that although those reporting higher income are likely to have higher levels of education on average, there is still a significant proportion (> 0.25) of individuals with fewer years of education, even for top incomes.

We found that income was generally positively related to attaining 16 + years of education, and this trend was even stronger based on the reclassified educational categories of high school completion. Thus, we provide evidence that completing at least a high school equivalency is strongly related to being a top earner across countries.

Additional exploratory analyses

The following results are based on post–pre-registration analyses. These exploratory results supplement the findings above by helping to contextualize the pre-registered analyses with comparability to existing estimates of country-level educational attainment and alternative rankings of countries based on latent means and variances of educational attainment.

Comparability of estimates to existing sources

The primary analyses above used educational attainment dichotomized as obtaining at least a tertiary level of education (16 + years of education) versus not. The dichotomization was preregistered27 and provided an avenue for directly comparing country-level results with existing estimates of educational attainment from the OECD12 report for the population of 24–64-year-olds. Table 4 provides the estimated proportion of each country’s population that has obtained a tertiary level of education based on the OECD report and those from the current GFS study. A valuable contribution of the GFS to education research is providing an estimate of educational attainment for several countries and territories not in the OECD report, such as Egypt, Hong Kong (S.A.R of China), Kenya, Nigeria, the Philippines, and Tanzania. Of the 16 countries with data available to compare educational attainment (16 + years) estimates in both the GFS data and the OECD report on tertiary education (excluding short-cycle tertiary education), the results were quite similar, as shown in Fig. 2. The comparable estimates correlated at r = 0.83 (95% CI: 0.57, 0.94).

Table 4 Comparing GFS estimates of educational attainment (proportion of Tertiary or 16 + years of education) with OECD.
Full size table

Nevertheless, there were a few prominent exceptions for which estimates were considerably more discrepant, such as Argentina, Germany, India, and Israel. These results suggest that the GFS estimates are generally similar to existing sources, but the estimates based on the GFS may underestimate the rates of tertiary education in some countries. The utility of these GFS data to provide unique information about cross-national educational attainment may, therefore, be partially limited when examining data from the national level. Notably, the GFS can provide more nuanced descriptive information on education than most existing sources by providing respondent-level data for subgroup analyses, as shown in Table 3.

Fig. 2
figure 2

Congruence between GFS estimates of tertiary educational attainment and OECD estimates of tertiary educational attainment (excluding short-cycle). Note Each point represents a country; the bold black line represent the line of best fit (simple linear regression), and the Pearson product-moment correlation is reported along with a 95% confidence interval. Estimated correlation was obtained in R using the cor.test(.) function.

Full size image

Discussion

Education level by country

Social science studies of wellbeing, particularly in psychology, have been criticized for focusing on so-called WEIRD (Western, Educated, Industrialized, Rich, and Democratic) nations43. The present study explores a diversity of countries and regions of the world. A notable contribution of this work is providing insights into the state of education in non-OECD nations/territories, including Egypt, Hong Kong (S.A.R of China), Kenya, Nigeria, the Philippines, and Tanzania. Together with previous analyses, this gives us a greater view of educational contexts worldwide. Moreover, while our comparison of countries showed variations in education as expected, particularly among non-Western nations compared to Western nations, there were interesting variations in nations nestled in various geographies around the world.

The difference between the most and least educated nations was much greater than expected based on the OECD estimates for these countries and the general definition of tertiary education used for both estimates. For example, within the GFS, the educational attainment estimate in India was only 1%, whereas it was 13% in the report by the OECD. The largest differences in estimates were greater than 10 points for Argentina, India, Japan, Spain, Sweden, and the United States. The estimates could be different due to the GFS using a limited sample in each country to approximate the population characteristics. One limiting factor in aligning the GFS’s estimates of educational attainment to the OCED is the methods and target source used in raking/post-stratification adjustments of the sampling weights7. Using alternative sources as target population counts could potentially explain some of these differences. However, these differences were generally smaller when the OECD estimates excluded the short-cycle tertiary category. For example, the difference in estimates for the United States was reduced from 15 to 4 points, and for Japan, the difference went from 37 to 16 points. The conflation of levels of education and the exact matching of categories between the GFS and the ISCED categories makes direct comparison difficult. The mismatch in definitions of tertiary education is one possible explanation for the diverging estimates for countries. In general, the definition we have for the 16 + tertiary education provided by Gallup is a lower bound relative to the OCED estimates. However, several instances may have led to discrepancies in the other direction (i.e., the GFS estimates are greater than the OECD estimates), such as for Germany, Israel, and Turkey.

One possible explanation for overestimation is how the weights were calibrated. In Germany, only the marginal distribution of educational attainment was available and not the cross-classified cells with regard to other key demographic characteristics such as age and sex, which can lead to poorly calibrated weights in some cases7. A similar argument could be made for the Turkey discrepancy. The estimates for Israel are relatively close in terms of the total tertiary, but once the short cycle is excluded, the estimates diverge drastically. When we investigate the high level of education attained using the country-specific categories for Israel (see Table S9a), we see an interesting picture start to emerge where 20% of the sample were in Israel: 13–14 years (non-academic, like technician, practical engineer, nurse) category, 24% of the sample were in Israel: 15–16 years (first degree, such as BA, BSC) category, 7.4% of the sample were in Israel: 17 + years (second degree, such as MA, MSc) category, and 0.6% of the sample were in the Israel: Ph.D. category. Once we exclude the 13–14 years (non-academic, like technician, practical engineer, nurse) category, the proportion is approximately 32% of the sample in a tertiary level of educational attainment, which would then get the estimate for Israel closer to the OECD tertiary excluding short-cycle education estimate of 39%. This tells us that the collapsed education 3-level categorization provided by Gallup may be misaligned to match the ISCED classifications. Instead of a direct estimate, the estimated proportion of the population in each country attaining a tertiary level of education using these GFS data may be reasonable as a lower bound for total tertiary education (ISCED classifications 5 +) for countries where no other estimates are available.

Variation in educational attainment

Despite the evidence for the alignment of country-level estimates of the proportion of adults in each country attaining a tertiary level of education, these data still have valuable contribution with regard to diving into the rich contextual information about who has attained 16 + years of education (which does not, in general, align with the ISCED classifications). The meta-analytic estimates described in the results (see Table 3) provided a good baseline for how gender, marital status, employment status, and other demographic characteristics describe the context of educational attainment; however, there is significant cross-national variation in these estimates. The results presented in this article provide a map of patterns of similarities and differences in how contexts shape educational attainment across countries. This lays the foundation for future research into what factors, such as education policies, education practices, cultural attitudes, and societal norms, contribute to these notable similarities and differences in how contexts impact educational attainment across countries.

In the country-specific results of income and trends, we found, unsurprisingly, that attaining higher levels of education was associated with self-reported income where higher income was associated with greater proportion of the population attaining 16 + years of education. Participant employment status was also related to the amount of education attained but varied significantly across categories of self-reported employment status and countries. For those disclosing being unemployed and looking for a job, the proportion of the population with at least 16 + years ranged from 0.02 (95% CI: 0.01,0.04) in Nigeria up to 0.31 (95% CI: 0.21,0.41) in Australia. This proportion tended to be higher in Western and industrial countries (e.g., United States, United Kingdom, etc.), whereas it was lower in African and Asian countries. For those reporting being employed for an employer, the proportion was on average 0.22 (95% CI: 0.15, 0.30), but this proportion was highly varied across countries ranging from 0.01 (95% CI: 0.00, 0.02) in Tanzania up to 0.65 (95% CI: 0.55,0.75) in Israel. Some of the largest differences in education were observed when individuals disclosed being a Homemaker, which could be understood as a stay-at-home parent27. Individuals disclosing being a Homemakers with high levels of educational attainment were from Australia, the United States, Japan, Germany, and Sweden. Still, the proportion was 0.25 (95% CI: 0.16,0.35) for Australia, indicating that a substantial proportion of the population attaining 16 + years of education were currently homemakers instead of actively seeking a job. Educational attainment proportions were slightly higher on average for individuals disclosing being employed for an employer (0.22, 95% CI: 0.15, 0.30) relative to those being self-employed (0.15, 95% CI: 0.09, 0.24). In most countries, this trend held with the proportion higher for those being employed by an employer or not being different. However, in Poland, the proportion of the population attaining 16 + years of education was significantly higher for those being self-employed (0.44, 95% CI: 0.38, 0.49) over being employed for an employer (0.33, 95% CI: 0.31, 0.35). The trend between employment and education is one lens to look at similarities and differences among countries, and next, we looked at differences between countries across all demographic variables.

Looking across demographic variables on trends in educational attainment provides an interesting nuance to these results. We found that the pattern of the estimated proportion of the population attaining 16 + years of education differs most between Brazil and Japan. Older individuals in Brazil tended to have more education, whereas in Japan, younger individuals were more likely to complete 16 + years of education. In Japan, females reported higher rates of 16 + years of education, whereas there was practically no difference in Brazil. Across marital statuses, the proportion of educational attainment was similar (approximately 0.23) among married individuals in both countries. Among separated individuals, the proportion was approximately double in Japan (0.41, 95% CI: 0.31, 0.50) compared to Brazil (0.18, 95% CI: 0.14, 0.23), but the reverse occurred for divorced individuals where the proportion was higher in Brazil (0.26, 95% CI: 0.22, 0.30) than in Japan (0.18, 95% CI: 0.16, 0.20). In contrast to the stark differences in patterns of educational attainment between Japan and Brazil, we observed notable similarities in patterns as well.

Similar patterns across demographic characteristics were observed in Australia, Germany, South Africa, Sweden, the United Kingdom, and the United States. In these countries, educational attainment was highest among the 25–29- or 30–39-year-olds. Males and females tended to have similar proportions of educational attainment. Across marital statuses, married or never married individuals tended to have the higher proportions, whereas widowed or divorced individuals tended to have lower proportions of attaining 16 + years of education. However, in Israel, the proportion of the population attaining 16 + years of education was significantly higher for married individuals (0.44, 95% CI: 0.38, 0.49) over those reporting being single/never married (0.33, 95% CI: 0.31, 0.35). Conversely, in Hong Kong (S.A.R. of China), single/never married individuals (0.26, 95% CI: 0.22, 0.29) were more likely to complete 16 + years of education relative to married individuals (0.16, 95% CI: 0.14, 0.19). Across levels of religious service attendance, these countries also tended to have the highest proportions among individuals endorsing attending 1–3/month or a few times a year. Across all of these countries the proportion was higher among individuals born in another country. Other countries had similar patterns within specific demographic characteristics, such as the 30–39-year-old population, which had the highest proportion in Hong Kong (S.A.R of China) and Turkey. The patterns of similarities and differences in the context in which individuals report their educational attainment across countries highlight interesting nuances for future research into what factors, such as government policies or cultural differences, contribute to these similarities or differences.

Higher secondary educational attainment helps individuals be competitive in the labor market5. Higher upper-secondary completion rates help create a more educated workforce and is associated with higher income and greater career possibilities44. We noted the same in this sample, particularly in nations such as Japan. For example, of Japanese participants earning more than a million yen, 38% were highly educated. It would warrant further study to compare these educational outcomes with economic outcomes and explore whether or not these outcomes correlate consistently with Gini coefficients or other indicators of economic disparities across nations. Moreover, higher educational attainment has been associated with many positive wellbeing outcomes45,46,47. Future analyses of the present data will allow us to compare education and human flourishing and see whether there are any causal connections over time.

Limitations and strengths

This study had notable strengths contributing to the cross-national evaluation of education. This is one of the largest international studies with respondent-level data on educational attainment with accompanying background characteristics, and it correlates across a wide span of the population. Further, we used a large and diverse sample that was weighted to be nationally representative, with each of the 22 countries and territories leading to broad population coverage. This coverage expands beyond the WEIRD samples typically evaluated in psychological research43. This allows for future research into the relationship between educational attainment and well-being across various populations.

This study had several limitations. First, the sociodemographic characteristics were limited to the factors available in the dataset, and rankings were limited to the 22 countries within the dataset. Future research with available data from a wider range of countries and background characteristics would provide new insights into variations in educational attainment and the contexts in which education is attained. Second, education was assessed using country-specific categories that were collapsed to be cross-nationally interpretable, and this collapsing of information may lead to more uncertainty in how many years of education are appropriate for direct comparison. A major limitation of these national-level estimates of educational attainment (tertiary education)48 is that the categories are mismatched with the ISCED classification, where the estimates provided by this article may be, at best, a lower bound for the ISCED5 or ISCED6 classifications of tertiary education. Despite this mismatch on a national level, the subgroup estimates of educational attainment across these countries may still be of interest because the use of random-effects meta-analyses to pool estimates of educational attainment does not presume cross-cultural measurement equivalence of the measure of education, only that the measures are representative of the same population of educational attainment7,26. The underlying population is defined within this article as the adult 18 + population in all 22 countries, which can differ from the populations of interest in other studies of educational attainment. Third, since this was a descriptive cross-sectional study utilizing univariate analyses, we cannot infer causality between the included demographic characteristics and educational attainment. Fourth, comparing countries based solely on rankings should be done with caution. As we showed with the alternative rankings in Table S24 (online supplement) based on the HETOP approach22, a country’s ranking can differ quite noticeably depending on how we utilize the harmonized education variable. For example, the United Kingdom moved from the second highest proportion of 16 + years of education to 8th overall on latent means of educational attainment. We caution against solely relying on either set of results when comparing educational attainment because even within a country, the rates of educational attainment differed quite significantly across socio-demographic characteristics. Lastly, due to the vastness of our dataset and space limitations, it became unwieldy to interpret and discuss all facets and country-specific characteristics of educational attainment. However, we present within-country summary statistics of educational attainment using the country-specific categories, and we encourage future researchers to consider using these classifications to study within country facets of how demographic variation in educational attainment.

Concluding remarks

Educational attainment varies significantly across countries and even within countries across demographic characteristics of individuals, providing new insights into the context of educational attainment across the world. Educational attainment contributes to positive outcomes in the labor force47 and enhances well-being in multiple domains of human flourishing, including happiness, social-emotional well-being, and character strengths29,49,50,51.