Explaining Wealth-Based Disparities in Higher Education Attendance: The Role of Societal Factors

Abstract

This article examines factors associated with wealth-based inequalities in higher education attendance at the national level. We draw on data from 99 countries to calculate two distinct country-level indicators for the extent of wealth-based inequality in higher education attendance, namely the dissimilarity index (D-Index) and the Human Opportunity Index (HOI). We then examine each indicator’s association with country-level factors using a series of regression models. We find that secondary completion rates, national wealth, economic inequality and the extent of political egalitarianism are all associated with wealth-based disparities in higher education access. However, there are important differences between indicators. Economic inequality is associated with disparities in access but not the level of overall access. In contrast, politically egalitarianism is associated with expanded educational access, but not wealth-based disparities alone. The study suggests that both economic and political equality are associated with higher educational outcomes. Yet, it also cautions that how we conceptualize and measure educational equity can shape our interpretations of the extent of a country’s educational equity.

1. Introduction

Wealth-based disparities in access to education exist at all levels of education and around the world (Buchmann and Hannum 2001) and these inequalities persist despite significant expansion of higher education globally over the past few decades. Substantial research on the extent and nature of inequality in educational access and completion has documented how ascribed characteristics, such as parental socio-economic status and family wealth, shape an individual’s expected educational attainment (Alon 2009; Shavit et al. 2007). That said, much of this work focuses on high-income countries (Jerrim et al. 2015; Jerrim and Vignoles 2015), where extensive and reliable data infrastructures exist. Despite some notable exceptions (see Ilie and Rose 2016), research on inequalities in access to higher education in lower- and middle-income countries is rarer, in part due to a lack of data availability and research infrastructure.

In this article, we build on recent scholarship that has developed measures of inequality in access to higher education at the country-level, which permits comparison across diverse national contexts (De Barros et al. 2009). Building on the work of Buckner and Abdelaziz (2023), we calculate two distinct indicators of wealth-based inequalities in HE attendance cross-nationally, namely the Dissimilarity Index (D-Index) and the Human Opportunity Index (HOI) and then explore societal-level factors associated with country-level inequalities. Specifically, we examine the extent to which economic and political inequality are associated with higher education inequality.

We find that both economic inequality and political egalitarianism are associated with inequality in higher education access, although our findings also suggest that the way inequality is operationalized matters in terms of interpretation. Economic inequality, as measured by the Gini index, is associated with the extent of wealth-based disparities, but not with overall access, while political egalitarianism appears to have a stronger association with overall access to higher education than wealth-based disparities. Ultimately, our findings bring new ways of quantifying inequalities in higher education access at the national level, and our regression analyses support the argument that inequalities in access to higher education are closely related to broader societal processes. Equality of higher educational opportunities are not only a feature of a country’s educational system but reflect and likely contribute to how equal economic and political structures are in a given society.

2. Literature and Background

Over the past several decades, access to higher education has expanded significantly across the globe, with gross tertiary enrollment rates rising from under 10% in 1970 to nearly 40% by 2018 (UNESCO Institute for Statistics 2020; Buckner and Abdelaziz 2023). This trend, often called massification in the field of higher education (Marginson 2016), is an explicit policy goal of many governments, as expanded access to higher education is associated with human capital, economic development, social mobility, and innovation.

Yet despite this global expansion, wealth-based disparities in higher education access and attendance remain entrenched, and in many contexts, are more pronounced than disparities based on gender or race (Buckner and Abdelaziz 2023; Ilie and Rose 2016). Students from higher wealth quintiles are substantially more likely to access and complete tertiary education, attend elite institutions, pursue selective specializations and transition into high-status occupations, compared to their peers from more disadvantaged backgrounds. This persistent gap calls into question the equity of opportunity often assumed to accompany expansion and shifts the focus from overall enrollment trends to the distribution of educational opportunity within national contexts.

Decades of research on access to higher education have documented the many factors shape an individual’s access to higher education, including individual and societal factors. At the individual level, an extensive literature has documented how individual, family, school and community characteristics all shape students’ educational trajectories and outcomes, including whether they attend higher education, what they study, and whether they graduate (Armstrong and Hamilton 2013; Hout and DiPrete 2006).

Various conceptual and theoretical frameworks have been developed to explain why wealth continues to exert a powerful influence on higher education access. These perspectives vary in emphasis and assumptions, but collectively, point to the important role that material and sociocultural advantages associated with wealth play in reproducing educational privilege across generations. For example, from a human capital perspective, higher education is viewed as an investment with a high expected return (Becker 1962). Wealthier families, with more disposable resources and lower opportunity costs, are better positioned to finance tuition, support extended schooling, and access high-return fields or institutions, permitting their children to attend higher education at higher rates (Marginson 2016). Rational choice models (e.g., Breen and Goldthorpe 1997) further suggest that even when lower-wealth families value education, the perceived costs and uncertainties over long-term outcomes may deter participation, particularly at key transition points and when considering selective or expensive institutions.

In contrast to human capital theory, social reproduction theory emphasizes the intergenerational transmission of inequality through both economic capital (e.g., financial resources) and cultural and social capital. Bourdieu (2011) argued that wealthier families are not only better resourced materially but also are socialized with the taken-for-granted norms associated with dominant academic practices and selective admissions processes. As a result, they are more likely to navigate complex educational systems successfully and gain access to university when access is not universal, and to access more prestigious universities or programs when admission is selective.

Although much of the research on inequality in access to higher education has historically focused on the role of individual characteristics, comparing individuals within the same country, there is also an important tradition of comparative work that has investigated cross-national patterns (Shavit et al. 2007). This body of scholarship has consistently emphasized the mediating role of national institutions and policies in structuring how wealth translates into educational advantage. For example, countries with strong public education sectors and universal financial aid systems tend to have more equitable educational outcomes, whereas market-oriented systems often reinforce wealth disparities through higher tuition and private providers (Shavit et al. 2007; Hout and DiPrete 2006; Shavit and Blossfeld 1993). The key insight of this body of work is that macro-level structures shape the educational opportunities available to individuals and can either amplify or buffer the effects of family wealth.

Early comparative work on inequalities in higher education used indicators such as parental education to operationalize class or socioeconomic status (e.g., Shavit and Blossfeld 1993) and tended to focus on inter-generational differences. Earlier studies were also limited by data availability to a certain number of countries, typically industrialized or high-income (Shavit and Blossfeld 1993). Yet, over the last decade, a number of studies have sought to measure wealth-based inequalities in access to higher education directly and for a broader range of countries.

For example, De Barros et al. (2009), in a major World Bank report on inequality of opportunity in Latin America, employed the dissimilarity index (D-Index) to assess education outcomes and compare disparities across countries. Building on this methodology, Krafft and Alawode (2018) analyzed nationally representative youth surveys from Egypt, Jordan, and Tunisia to construct a D-Index that captures inequality in access to higher education, which they argue is the first international comparison of higher education inequality using a D-index. Building on this approach, Buckner and Abdelaziz (2023) draw on household wealth quintile data from 117 countries in the World Inequality Database on Education (WIDE), to calculate and compare disparities in higher education attendance across low-, middle-, and high-income contexts. They introduce four distinct indicators to measure inequality in higher education attendance, namely a Wealth Parity Index, Odds Ratio, Dissimilarity Index, and Human Opportunity Index. Like other studies, they show that wealth-based disparities in access to higher education are significant, finding them larger in higher education than other levels of schooling. Nonetheless, their work remains primarily descriptive; they do not analyze what factors are associated with wealth-based disparities in access. In this article, we seek to build on Buckner and Abdelaziz (2023) to investigate the role of societal structures in shaping the extent of educational inequality. Drawing on the insights of prior work regarding the role that societal level factors play in shaping opportunity structures for individuals, we focus on two well-researched societal factors, namely: economic equality and political egalitarianism.

We expect that economic inequality will be positively associated with inequality in higher education access. Decades of research in the sociology of education have documented how greater economic inequality in society translates into wealth-based disparities in higher education access through various often reinforcing mechanisms at the family, institutional and political levels (Hout and DiPrete 2006). At the national level, public investment in education is often regressive or unevenly distributed, resulting in lower-quality primary and secondary schooling for low-income, rural or otherwise marginalized groups, which limits academic performance and subsequent eligibility for higher education. Second, private expenditures on education, which include not only tuition at private schools but also tutoring, test preparation permit students from wealthier families to succeed at all levels of education, including in the transition to higher education. These expenditures allow affluent families to navigate selective admissions processes, secure access to prestigious institutions, and invest in social capital-building experiences (e.g., extracurriculars, study abroad). Third, economic inequality in a society often correlates with residential segregation and regional disparities in wealth, which concentrate disadvantage in under-resourced areas with limited postsecondary options or weak school-to-university pipelines. Similarly, in many lower and middle-income countries, universities, particularly new and demand-absorbing private ones, are located primarily in urban areas to serve wealthier families whose children do not obtain spots in subsidized public universities (Buckner 2013). Geographic proximity facilitates access for wealthier urban families, while making those from more rural regions travel long distances to attend university (Chankseliani 2013). Collectively, these mechanisms mean that as wealth inequality rises, typically so too does the stratification of educational opportunity, with access to higher education highly contingent on a family’s available economic resources. In contrast, greater economic equality likely means that families across the wealth spectrum are better able to afford the direct and indirect costs associated with higher education. Specific policy instruments such as means-tested financial aid or the provision of educational tax credits may play a mediating role by offsetting these costs, and while outside the scope of this analysis, should be examined in future studies.

H1. 

Economic inequality will be associated with inequality in higher education access, as measured by the D-Index.

In addition to economic inequality, we also hypothesize that political structures shape access to higher education and specifically that political egalitarianism will be associated with the equality of opportunity in a society. We expect that the more egalitarian political representation is across groups in a society, the more that different groups can advocate for access to opportunity structures, like education. Specifically, we expect political egalitarianism to be linked to more equitable access and outcomes at the primary and secondary level, thereby expanding eligibility for higher education and reducing inequalities. At the tertiary level, political egalitarianism, through the enfranchisement of diverse communities, is likely to be associated with greater public investment in higher education generally and reflected in access-oriented policies. These may include lower tuition and fees, availability of scholarships, bursaries, and loans, affirmative action policies, and less geographic concentration of universities. Combined, we expect that policies such as these will make higher education more accessible, geographically and financially, and will lead to expanded overall access. In contrast, in societies where political representation is unequal, political support for redistributive policies such as need-based financial aid or tuition subsidies may be weaker, leading to the privatization of educational costs and a heavier burden on low-income students. Across various political systems, we hypothesize that the more open the political process and the more access to political representation and decision-making that different groups have, the more equitable social services, including education, will be. We hypothesize that political egalitarianism will be associated with expanded access to higher education generally, which will lead to greater equality in higher education access as measured by the HOI, but not the D-Index.

H2. 

Political egalitarianism will be associated with expanded access to higher education generally, and reduced inequalities in access to higher education as measured by the HOI but not the D-Index.

3. Materials and Methods

This analysis draws on the World Inequality Database on Education (WIDE), an open-access database developed by UNESCO’s Global Education Monitoring Report (GEMR) to track progress toward Education for All and the Sustainable Development Goals. The database compiles information from nationally representative demographic, health, and labor force surveys such as the Demographic and Health Surveys (DHS) and the Multiple Indicator Cluster Surveys (MICS) to monitor disparities in education. For this study, we use data from 99 countries spanning 2003 to 2021. While WIDE provides global and cross-income coverage, the dataset has limited representation of high-income and industrialized countries. As a result, our analyses focus primarily on countries in lower and middle-income countries, which are generally not as well studied as those in North America and Europe.

Table 1. Total Number of Countries by Region.

Region	Countries (N)	% Total Countries
East Asia & Pacific	14	14.14
Europe & Central Asia	16	16.16
Latin America & Caribbean	17	17.17
Middle East & North Africa	8	8.08
South Asia	6	6.06
Sub-Saharan Africa	38	38.38
Total	99

provides an overview of the countries in the dataset, by world region, and

Table 2. Total Number of Countries by Country Income Group Classification.

Country Income Group	Countries (N)	% Total Countries
High Income	4	4.04
Upper middle	32	32.32
Lower middle	38	38.38
Low Income	25	25.25
Total	99

shows the countries classified by country income group.

To conceptualize wealth, we disaggregate educational outcomes by young people’s family wealth background. We use WIDE’s data on wealth income quintiles (WIQs), where WIQ1 represents the poorest 20%, and WIQ5 represents the wealthiest 20% of the population. Each indicator conceptualizes inequality slightly differently, shedding light on nuances in unequal attendance rates. Many of these indicators have previously been applied to examine inequalities in primary and secondary education, as well as in other areas such as health. In our study, we further disaggregate our analysis by country income levels using the World Bank’s income classification, which is based on Gross National Income (GNI) per capita (converted to U.S. dollars), grouping countries into four categories: low-income, lower-middle-income, upper-middle-income, and high-income (World Bank n.d.).

Our analysis centers on higher education, which in most national contexts encompasses both two-year and four-year postsecondary programs leading to a diploma or degree (ISCED levels 5 and 6). A commonly used metric in cross-national research is the Gross Tertiary Enrollment Ratio (GTER), reported by the UNESCO Institute for Statistics (UIS). While widely available, the GTER is an administrative measure that does not capture differences in tertiary access across social or demographic groups. Household survey data, by contrast, offer the advantage of including individual background characteristics, enabling more detailed disaggregation. For this reason, we draw on an indicator from WIDE measuring current higher education participation: the percentage of individuals aged 18–22 attending higher education at the time of the survey. This measure does not distinguish between those who dropped out and those who never enrolled, and its narrow age band excludes over-age students. Consequently, it is best understood as a proxy for a country’s net tertiary attendance rate.

3.1. Outcome Variables: Indicators of Educational Inequality

At the national level, we measure wealth-related disparities in access to higher education using two indicators: the D-Index and the Human Opportunity Index, following the approach of Buckner and Abdelaziz (2023). These indicators have previously been applied to examine inequalities in primary and secondary schooling, as well as in other areas such as health care.

Dissimilarity Index (D-Index). The D-Index captures the share of opportunities that would need to be reallocated from the most privileged groups to the least privileged in order to achieve equal access across all groups (De Barros et al. 2009; Krafft and Alawode 2018). Its values range from 0 to 1, with 0 representing complete equality of opportunity and 1 representing maximum inequality. In our analysis, each quintile represents 20% of the population; thus, under perfect equality, every quintile would occupy exactly 20% of available tertiary education seats. Such a distribution corresponds to a D-Index value of zero.

$$\mathrm{D}={\displaystyle \frac{1}{2\mathrm{C}}}\sum _{\mathrm{k}=1}^{\mathrm{m}}{\mathsf{\alpha }}_{\mathrm{k}}\left|\mathrm{C}-{\mathrm{C}}_{\mathrm{k}}\right|,\mathrm{m}=\mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{c}\mathrm{i}\mathrm{r}\mathrm{c}\mathrm{u}\mathrm{m}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e} \mathrm{g}\mathrm{r}\mathrm{o}\mathrm{u}\mathrm{p}\mathrm{s}.$$

(1)

The D-index has previously been used to quantify the extent of cross-national wealth-based disparities in higher education access (Buckner and Abdelaziz 2023; Krafft and Alawode 2018). Nonetheless, one of the limitations of the D-Index is that it is not sensitive to absolute levels of access. Because it measures inequality only in relative terms (i.e., each group relative to the country mean), a country can have a low value on the D-Index, indicating low levels of inequality, when overall access is uniformly poor. Therefore, we complement our analysis of the D-Index with the HOI, which also considers overall levels of higher education access.

The Human Opportunity Index (HOI). The human opportunity index (HOI), which is an indicator inspired by Sen’s (1976) capabilities approach. Central to this approach is the premise that individuals should have not only formal rights, but also the real freedoms and capabilities to access basic public services such as education, health care, and housing. Similar to the Dissimilarity Index, the Human Opportunity Index (HOI) rests on the idea that inequality arises when access to opportunities is associated with an individual’s circumstances (such as ascribed characteristics). At the same time, a situation in which no one has access to essential, capability-enhancing services like education represents equal deprivation rather than true equality. The HOI, therefore, integrates both the overall extent of availability of opportunities and the fairness of their distribution. It is scaled from 0 to 1, where 0 reflects universal deprivation and 1 reflects universal access with full equality. As a distribution-sensitive measure, the HOI takes into account how opportunities are shared between disadvantaged and advantaged groups, while also increasing as the total number of available opportunities expands for each group.

$$HOI=\left(1-D\right)\ast C$$

(2)

Previous studies have shown that the HOI is a robust measure for quantifying cross-national wealth-based inequalities in higher education access (Buckner and Abdelaziz 2023). Nonetheless, there are important limitations with both the D-Index and the HOI. One important drawback is that they do not indicate where in the wealth spectrum inequalities are concentrated. For instance, a country with moderate disparities spread across the poorest 60 percent of the population could have the same value on the D-Index as one where only the poorest quintile experiences a high rate of exclusion. These two realities could have the same value on the D-Index despite reflecting very different educational realities that also have different implications for policy. A second important limitation, relevant to our analysis, is that both the D-Index and HOI can be sensitive to how groups are defined. In this study, we use quintiles, but this approach may mask more fine-grained inequalities that would likely emerge if data were disaggregated at the decile or even percentile level.

3.2. Key Predictor Variables

Economic inequality. The Gini index is a widely accepted measure of national-level income inequality (Charles et al. 2022; Jenkins 1991; Qazi et al. 2018; Sitthiyot and Holasut 2020). We use the Gini index to measure economic inequality, logged to reduce skewness. The Gini index quantifies the deviation of income distribution among individuals within a given economy from a perfectly equal income distribution. It ranges from 0 to 100, zero indicating perfect equality and 100 indicating perfect inequality. Data on the Gini index were extracted from the World Development Indicators database, which is compiled by the World Bank.

Political egalitarianism. For the purposes of this study, we use the Egalitarian Component Index (ECI) from the Varieties of Democracy (V-Dem) project as a measure of political egalitarianism. The ECI is designed to capture the extent to which the egalitarian principle of democracy is realized, namely, that all individuals, regardless of their social group, have equal capabilities to participate in political and civic life. This index draws on three interrelated dimensions: (1) equal protection of rights and freedom among all social groups; (2) equal distribution of resources; (3) equal distribution of access to power by social group, gender, and socioeconomic status (Coppedge et al. 2022, p. 55). ECI ranges from 0 (low) to 1 (high).

The ECI is a core component of the broader Egalitarian Democracy Index (EDI), the EDI includes additional elements related to electoral democracy such as free and fair elections, suffrage, and pluralism that extend beyond our focus on distributive equality. As the V-Dem Methodology Handbook notes, “gross inequalities of health, education, or income are understood to inhibit the exercise of political power and the de facto enjoyment of political rights” (Coppedge et al. 2024, p. 4). For this reason, we rely on the ECI, which provides a more conceptually precise measure aligned with our concern for structural opportunity conditions, rather than electoral institutions.

Although the ECI has been widely used in fields such as international relations (Altman et al. 2023; Kasuya and Reilly 2023), sociology (e.g., Calnitsky and Wannamaker 2024), and political science (Awada 2018), it remains underutilized in comparative higher education research. To our knowledge, its application in this domain has been limited to studies of academic freedom, which is a distinct V-Dem component not included in our analysis (e.g., Spannagel and Kinzelbach 2023). Our study instead examines country-level structural conditions shaping opportunity structures within higher education. Finally, our use of V-Dem data is supported by Boese (2019), whose comparative analysis finds that V-Dem offers the most valid and reliable cross-national indicators of democracy, particularly for multidimensional and normative concepts such as egalitarianism.

3.3. Control Variables

Prior research has found that GDP per capita and secondary completion rates were strongly associated with our outcome variables, the Human Opportunity Index (HOI) and the D-index (Buckner and Abdelaziz 2023). This finding is further supported by our current correlation matrix (see robustness checks section). Accordingly, we include both secondary completion rate and GDP per capita as control variables in our analysis. Our data for the secondary completion rate comes from WIDE, which is calculated from household survey data as the share of young people who are between three and five years older than the expected graduation age and who have completed upper secondary education. Similar to tertiary attendance indicators, this measure relies on specific age cohorts and excludes students who are older than the defined range, which may limit cross-country comparability. GDP per capita data were extracted from the World Development Indicators.

3.4. Data Analysis

To examine the relationship between our predictor and outcome variables, we conducted a descriptive analysis by categorizing our key predictors, the Gini index and the Egalitarian Component Index (ECI), into terciles (i.e., low, medium, and high levels). We then calculated the mean values of our two outcome variables, the D-index and the Human Opportunity Index (HOI), within each tercile group. This allowed us to assess how varying levels of economic inequality and political egalitarianism relate to opportunity and inequality outcomes.

We then carry out a series of panel regression analyses to examine country-level factors associated with inequality and universal access in higher education attainment, one for each of our two dependent variables. While WIDE data is drawn from cross-national household surveys, a key limitation is that the timing of data collection varies across countries. After data cleaning, most countries have one or two survey years, but not in the same periods. This irregularity limits our ability to estimate temporal trends within countries. However, the structure of the dataset still meets the criteria for panel data, as it includes multiple observations for a given country over time, albeit limited. Since our objective is to examine how country-level factors are associated with inequality and universal access in higher education attainment globally, we employ a panel regression model, which is particularly suited to datasets with both cross-sectional and time-series dimensions (Zhang 2010). Furthermore, panel regression is particularly well suited for modelling global trends in higher education access, as it captures both within-country and between-country variation (e.g., Sánchez and Singh 2018; Yang and McCall 2014). We apply random effects model, which assumes that the variation across countries is uncorrelated with the predictors in the model. Such an assumption is reasonable given the structure and goals of our analysis.

4. Results

In our descriptive analyses, we synthesize disparities in higher education attainment and completion rates into two indicators of inequality to facilitate comparisons across country contexts and educational levels.

Table 3. Descriptive Statistics of Inequality Indicators by Country Income Group.

	Cross-National Average		High Income		Upper Middle-Income		Lower Middle-Income		Low Income
	M	SD	M	SD	M	SD	M	SD	M	SD
Secondary Completion
D-Index	0.29	0.16	0.10	0.14	0.15	0.09	0.29	0.14	0.43	0.12
HOI	0.29	0.26	0.71	0.25	0.50	0.21	0.25	0.23	0.09	0.10
Tertiary Attendance
D-Index	0.39	0.16	0.22	0.15	0.28	0.12	0.38	0.14	0.51	0.13
HOI	0.09	0.11	0.27	0.14	0.17	0.13	0.08	0.10	0.02	0.03

presents the mean value for both of our inequality indicators, the dissimilarity index and human opportunity index, at the secondary and tertiary levels. The table also disaggregates the index by the country’s level of economic development by classifying countries into four country-income groups. The table shows that the magnitude of inequality tends to increase with each successive level of education, with secondary completion rates being relatively more equal than tertiary attendance rates. However, the educational level at which family wealth most strongly shapes access to higher education tends to vary by a country’s level of economic development. In low-income countries, large inequalities in secondary completion contribute directly to unequal access to tertiary education. In high-income and middle-income countries, secondary completion is more evenly distributed compared to tertiary attendance rates. This finding suggests that wealth plays a decisive role in the transition and entry to higher education, primarily benefiting students from the top income quintile. That said, it is important to remember that our dataset includes very few high-income countries, and future iterations will hopefully be able to account better for patterns in high-income countries.

We also examined the relationship between our two predictors of interest—economic inequality and political egalitarianism—and our dependent variables of interest, as shown in

Table 4. Descriptive Statistics of D-Index and HOI by Gini Level.

	D-Index		HOI
	M	SD	M	SD
Low Econ. Ineq. (GI < 36)	0.31	0.16	0.14	0.14
Med. Econ. Ineq. (GI 37–61)	0.45	0.15	0.06	0.11
High Econ Ineq. (GI ≥ 62)	0.43	0.15	0.08	0.10

Notes. Each Gini index group presents one-third of our data.

and

Table 5. Descriptive Statistics of D-Index and HOI by ECI Level.

	D-Index		HOI
	M	SD	M	SD
Low Pol. Equal. (ECI < 0.47)	0.43	0.14	0.05	0.05
Med Pol. Equal. (ECI 0.47–0.64)	0.42	0.17	0.07	0.12
High Pol. Equal. (ECI > 0.64)	0.33	0.16	0.13	0.14

Notes. Each ECI group presents one-third of our data.

, respectively. For each predictor, we created three categorical groups (low medium, high), by splitting the sample evenly into three groups, and then examined the mean value of the inequality indicators for countries at that level of economic or political inequality. Table 4 shows a clear relationship between economic inequality as measured by Gini index, and the two inequality indicators. As economic inequality increases, the D-index, our measure of inequality, also rises, while the Human Opportunity Index (HOI) declines, indicating lower equity in opportunity. Notably, the most significant gains in equality appear below a Gini of 36.40. Beyond this point, differences between countries with middle and high inequality are less pronounced, possibly due to a ceiling effect or limited variation within these higher tiers.

Table 5 presents the results for the Egalitarian Component Index (ECI) in relation to our outcome variables. As the level of adherence to egalitarian principles increases, we observe a decline in inequality as measured by the D-index and a corresponding rise in the Human Opportunity Index (HOI). This suggests that greater commitment to political egalitarianism is associated with more equitable access to opportunities and reduced disparity. However, as with economic inequality shown in Table 4, the descriptive analysis does not suggest linear trends; rather, the impact of political egalitarianism on higher education attendance seems to be strongest among those with high levels of political egalitarianism.

4.1. Regression Findings for Dissimilarity Index (D-Index)

In order to test Hypothesis 1, we ran a series of panel regression models for the D-Index, as shown in

Table 6. D-Index Panel Regression Analysis.

	M0	M1	M2	M3
GDP per Capita (logged)	−0.06 ***	−0.06 ***	−0.07 ***	−0.08 ***
	(0.02)	(0.02)	(0.02)	(0.02)
Secondary Completion Rate	−0.28 ***	−0.26 ***	−0.23 ***	−0.24 ***
	(0.04)	(0.04)	(0.04)	(0.04)
Gini Index (logged)		0.003 *	0.003 *	0.003 *
		(0.001)	(0.001)	(0.001)
ECI			−0.01
			(0.05)
ECI × Income Group
High income & Upper Middle				−0.05
				(0.06)
Lower Middle				−0.02
				(0.05)
Low Income				0.01
				(0.07)
Constant	0.95 ***	0.91 ***	0.94 ***	1.01 ***
	(0.12)	(0.11)	(0.11)	(0.0.17)
Observations	205.00	185.00	179.00	179.00

Notes. The Gini Index (from World Bank Data) and the Egalitarian Component Index (from Varieties of Democracy) were extrapolated to align with the WIDE dataset. p < 0.05 *, p < 0.001 ***.

. Our findings indicate that countries’ secondary completion rates, national wealth, and economic inequality are all associated with inequality in higher education attendance, as measured by the D-index. The findings from the random effects models indicate that secondary completion rates are negatively associated with D-index while the Gini index is positively associated with D-index, and these results hold even after controlling for the country’s wealth and level of political egalitarianism. Yet, a country’s political egalitarianism is not associated with inequality in higher education attendance, and this is true even after we disaggregate our analysis by country income group. Overall, secondary completion rates and a country’s national wealth as measured by logged GDP per capita remain highly associated with inequality in higher education attainment.

Model 0 shows that logged GDP per capita and secondary completion rate are negatively associated with inequality in higher education attendance as captured by D -index (see Table 6). More specifically, a one-unit increase in logged GDP per capita is associated with a 6% decrease in the D-index after controlling for secondary completion (p < 0.001). The association in secondary completion is even stronger: a one-unit increase corresponds to a 28% decrease in the D-index (p < 0.001).

In Model 1, we add the Gini index as a measure of countries’ economic inequality to the baseline model to control for variations in country income inequality across countries. We find that a country’s economic inequality is positively associated with inequality in higher education attendance. We find that a one-unit increase in a country’s economic inequality increases D-Index by 0.3% (p < 0.05) after controlling for other variables.

In Model 2, we add the Egalitarian Component Index (ECI) as a measure of a country’s political egalitarianism. Surprisingly, we find that political egalitarianism is not associated with inequality in higher education attendance. However, we find that the magnitude of association of secondary completion on inequality in higher education attendance decreases by 5%, which might imply that egalitarian systems are associated with lower inequality in secondary completion rates or that it facilitates higher transition from secondary to higher education which in return leads to lower inequality in higher education attendance. We also notice that the national wealth coefficient magnitude increases by 1% hinting at the potential impact of national wealth on political egalitarianism policies.

To untangle this further, Model 3 includes an interaction between ECI and country income groups, based on the hypothesis that egalitarian systems vary by a country’s income level. In other words, we expect stronger egalitarian systems and thus lower higher education inequality in high-income and upper-middle-income countries, and weaker systems in low-income contexts. Consistent with model 2, we find no association between the interaction terms and D-index.

4.2. Regression Findings on Human Opportunity Index

In order to test Hypothesis 2, we ran a series of panel regression models on the HOI, as shown in

Table 7. HOI Panel Regression Analysis.

	M0	M1	M2	M3
GDP per Capita (logged)	0.03 ***	0.04 ***	0.04 ***	0.04 †
	(0.01)	(0.01)	(0.01)	(0.02)
Secondary Completion Rate	0.19 ***	0.19 ***	0.18 ***	0.18 ***
	(0.05)	(0.05)	(0.05)	(0.05)
Gini Index (logged)		−0.001 *	−0.002 *	−0.002 *
		(0.001)	(0.001)	(0.001)
ECI			0.10 *
			(0.04)
ECI × Income Group
High income & Upper Middle				0.12 *
				(0.06)
Lower Middle				0.07 †
				(0.04)
Low Income				0.10 *
				(0.06)
Constant	−0.25 ***	−0.26 **	−0.28 ***	−0.25 †
	(0.07)	(0.08)	(0.08)	(0.15)
Observations	205.00	185.00	179.00	179.00

Notes. The Gini Index (from World Bank Data) and the Egalitarian Component Index (from Varieties of Democracy) were extrapolated to align with the WIDE dataset. p < 0.10 ^†, p < 0.05 *, p < 0.01 **, p < 0.001 ***.

. The table shows that countries’ secondary completion rates, national wealth, economic inequality, and political egalitarianism are all associated with inequality in higher education attainment as measured by the HOI. Specifically, we found that both logged GDP per capita and secondary completion rates are positively associated with HOI, while the Gini index is negatively associated with HOI. We also find that in high-income countries and upper-middle-income countries, an increase in the country’s political egalitarianism (ECI) increases HOI after controlling for all other variables.

Model 0 shows that logged GDP per capita and secondary completion rate are positively associated with HOI, thus reducing overall disparities among wealth groups in higher education attainment across countries (see Table 7). We find that a one-unit increase in logged GDP per capita increases HOI by 3% (p < 0.001) after controlling for the secondary completion rates. Moreover, a one-unit increase in secondary completion rate increases HOI by 19% (p < 0.001) after controlling for logged GDP per capita. The model shows that increasing both secondary completion rates and a country’s national wealth are associated with decreasing inequality (D-Index) and increasing equal distribution among wealth groups in higher education access (HOI).

In Model 1, we add the Gini Index as a measure of the country’s economic inequality, which we expect to be negatively associated with HOI. We find that a one-unit decrease in economic inequality increases HOI by 0.01% (p < 0.05). In Model 2, we find that countries with higher political equality as captured by ECI are positively associated with HOI. Specifically, a one-unit increase in ECI increases HOI by 10% (p < 0.05) after controlling for other variables. In Model 3, we find that at all levels of country income groups, political egalitarianism is marginally associated with HOI. In high-income countries and upper-middle income countries, a one-unit increase in ECI increases HOI by 12% (p < 0.05) after controlling for other variables. In lower-middle income countries, a one-unit increase in ECI increases HOI by 7% (p < 0.10) after controlling for other variables. However, the model is only marginally associated. Finally, in low-income countries a one-unit increase in ECI increases HOI by 10% (p < 0.05) after controlling for other variables. Moreover, we find that in this model the logged GDP per capita becomes only marginally associated with HOI, indicating that at the global level, secondary completion rates and political egalitarianism are more important in increasing HOI in higher education attainment than a country’s national wealth after controlling for other variables. It might also show that for lower-middle and low-income countries, where secondary completion rates are lower on average among all groups, increasing secondary completion rates are more important for transition into higher education.

4.3. Robustness Checks

To ensure the robustness of our findings, we conducted several robustness check analyses. First, we examined correlations between our various independent variables, and the results indicate weak correlations, except for a strong positive correlation between GDP per capita and secondary completion rates (see

Table A1. Correlation Matrix.

	n	M	SD	1	2	3	4
Log GDP per Capita	221	8.61	0.95	1.00
Secondary Completion rates	223	0.36	0.27	0.84	1.00
Gini Index	201	40.3	7.68	−0.04	−0.18	1.00
Egalitarian component Index	213	0. 31	0.16	0.33	0.22	0.21	1.00

Notes. The Gini Index (from World Bank Data) and the Egalitarian Component Index (from Varieties of Democracy) were extrapolated to align with the WIDE dataset.

in the Appendix A). However, as these are both significant in regression models, we keep both in our preferred models.

Second, we ran the regression analysis without the secondary completion rates to understand whether the SCR was driving the association between GDP per capita and HOI or D-Index. Removing the secondary completion rates from the D-index regression model resulted in a substantial increase in the GDP per capita coefficient across all models. For example, in the baseline model, the coefficient shifts from −0.06 to −0.16. This suggests that secondary completion rates moderate the relationship between GDP per capita and the D-index (see

Table A2. D-Index Panel Regression Analysis without Secondary Completion Rates—Robustness Check.

	M0	M1	M2	M3
GDP per Capita (logged)	−0.16 ***	−0.12 ***	−0.12 ***	−0.13 ***
	(0.02)	(0.02)	(0.02)	(0.02)
Gini Index (logged)		0.004 *	0.004 *	0.004 *
		(0.001)	(0.001)	(0.001)
ECI			−0.01
			(0.06)
ECI × Income Group
High income & Upper Middle				−0.01
				(0.07)
Lower Middle				−0.04
				(0.06)
Low Income				0.03
				(0.08)
Constant	1.37 ***	1.25 ***	1.25 ***	1.33 ***
	(0.10)	(0.11)	(0.11)	(0.17)
Observations	205.00	185.00	179.00	179.00

Notes. The Gini Index (from World Bank Data) and the Egalitarian Component Index (from Varieties of Democracy) were extrapolated to align with the WIDE dataset. p < 0.05 *, p < 0.001 ***.

in the Appendix A). The coefficients and standard errors for the remaining independent variables show minimal variation, with no changes in the direction or significance of associations. However, notable differences appear in the interaction model (Model 3): the coefficient for high- and upper-middle-income countries decreases from −0.05 to −0.01, while that for low-income countries increases from 0.01 to 0.03. These shifts highlight the important role of secondary education in reducing inequality in tertiary education access, as captured by the D-index, regardless of a country’s levels of political inequality. Interestingly, for lower-middle-income countries, the coefficient becomes more negative (from −0.02 to −0.04), suggesting that other factors may be driving reductions in inequality in this group, beyond secondary completion rates.

Similarly, we exclude secondary completion rates from the HOI regression model, which leads to a slight increase in the GDP per capita coefficient across all models. In the baseline model, for instance, the coefficient rises from −0.03 to −0.07 (see

Table A3. HOI Panel Regression Analysis without Secondary Completion Rates—Robustness Check.

	M0	M1	M2	M3
GDP per Capita (logged)	0.07 ***	0.08 ***	0.08 ***	0.07 **
	(0.01)	(0.01)	(0.01)	(0.02)
Gini Index (logged)		−0.003 ***	−0.003 **	−0.003 ***
		(0.001)	(0.001)	(0.001)
ECI			0.13 **
			(0.05)
ECI × Income Group
High income & Upper Middle				0.16 *
				(0.06)
Lower Middle				0.09 †
				(0.05)
Low Income				0.13 *
				(0.06)
Constant	−0.52 ***	−0.48 ***	−0.50 ***	−0.44 ***
	(0.09)	(0.09)	(0.09)	(0.17)
Observations	205.00	185.00	179.00	179.00

Notes. The Gini Index (from World Bank Data) and the Egalitarian Component Index (from Varieties of Democracy) were extrapolated to align with the WIDE dataset. p < 0.10 ^†, p < 0.05 *, p < 0.01 **, p < 0.001 ***.

in the Appendix A). Gini coefficients increase by 1 to 2 points across models and become highly significant, despite the low correlation between the Gini index and secondary completion rates (r = −0.04). Similarly, the effect of political egalitarianism strengthens and reaches significance at p < 0.01 in Model 2, further indicating that secondary completion rates contribute to reducing inequality in tertiary attendance as measured by HOI. Finally, in the interaction models, the coefficients and standard errors increase slightly but remain significant at the same levels, reinforcing the robustness of the observed relationships.

Third, given the limited number of high-income countries (N = 4), we grouped them with upper-middle-income countries in our interaction models. Thus, we conducted a final robustness check to ensure that our models remain robust after grouping. Comparing the results in

Table A4. D-Index Panel Regression Analysis—Grouping Robustness Check.

	M0	M1	M2	M3
GDP per Capita (logged)	−0.05 ***	−0.06 ***	−0.07 ***	−0.08 ***
	(0.02)	(0.02)	(0.01)	0.02)
Secondary Completion Rate	−0.28 ***	−0.26 ***	−0.23 ***	−0.25 ***
	(0.04)	(0.04)	(0.04)	(0.04)
Gini Index (logged)		0.003 *	0.003 *	0.003 *
		(0.001)	(0.001)	(0.001)
ECI			−0.01
			(0.05)
ECI × Income Group
High income				−0.02
				(0.09)
Upper Middle				0.05
				(0.06)
Lower Middle				−0.02
				(0.05)
Low Income				0.01
				(0.08)
Constant	0.95 ***	0.91 ***	0.94 ***	0.99
	(0.12)	(0.11)	(0.11)	(0.0.17)
Observations	205.00	185.00	179.00	179.00

Notes. The Gini Index (from World Bank Data) and the Egalitarian Component Index (from Varieties of Democracy) were extrapolated to align with the WIDE dataset. p < 0.05 *, p < 0.001 ***.

in the Appendix A (before grouping) and Table 6 (after grouping), we find that grouping high-income countries with upper-middle-income countries does not substantially alter the model coefficients in the D-index regression model. The only notable difference is a slight change in Model 0, where the coefficient for GDP per capita shifts from −0.05 to −0.06 after grouping. In the interaction model, the standard error for the grouped category (0.06) is lower than that of high-income countries alone (0.09), likely due to the limited number of observations. This grouping also affects the direction of the interaction term: while upper-middle-income countries show a positive association in the original model, they display a negative association when combined with high-income countries, driven by the latter’s effect. However, the interaction is not significant.

In the HOI regression results (

Table A5. HOI Panel Regression Analysis—Grouping Robustness Check.

	M0	M1	M2	M3
GDP per Capita (logged)	0.03 ***	0.04 ***	0.4 ***	0.03 †
	(0.01)	(0.01)	(0.01)	(0.02)
Secondary Completion Rate	0.19 ***	0.19 ***	0.18 ***	0.18 ***
	(0.05)	(0.05)	(0.05)	(0.05)
Gini Index (logged)		−0.001 *	−0.002 *	−0.002 *
		(0.001)	(0.001)	(0.001)
ECI			0.10 *
			(0.04)
ECI × Income Group
High income				0.16 †
				(0.10)
Upper Middle				0.11 †
				(0.06)
Lower Middle				0.07
				(0.04)
Low Income				0.10 †
				(0.06)
Constant	−0.25 ***	−0.26 **	−0.28 ***	−0.24
	(0.07)	(0.08)	(0.08)	(0.15)
Observations	205.00	185.00	179.00	179.00

Notes. The Gini Index (from World Bank Data) and the Egalitarian Component Index (from Varieties of Democracy) were extrapolated to align with the WIDE dataset. p < 0.10 ^†, p < 0.05 *, p < 0.01 **, p < 0.001 ***.

, before grouping; Table 7, after grouping), we find that combining high-income and upper-middle-income countries does not substantially affect the model’s coefficients. The only notable change appears in Model 3, where the GDP per capita coefficient increases slightly from 0.03 to 0.04 after grouping. In the interaction model, the standard error for the grouped high- and upper-middle-income category drops from 0.10 (for high-income alone) to 0.06, likely reflecting the limited number of observations in the high-income group. This grouping also strengthens the significance of the interaction term: while high- and upper-middle-income countries were only marginally significant in the original model, the grouped category becomes statistically significant at the p < 0.05 level. Additionally, the coefficients for lower-middle-income countries become marginally significant, and low-income countries shift from marginal to statistically significant, with no change in standard errors. These results suggest that earlier instability was primarily driven by the small sample size within the high-income group.

5. Discussion, Limitations and Conclusions

Our regression analyses suggest that societal-level economic inequality and political inequality are both associated with educational inequalities, in the direction expected. This finding substantiates the hypothesis that educational structures, which are inherently embedded with particular social, cultural and political contexts, are likely affected by the broader opportunity structures of their societies. Our findings suggest that there are positive relationships between the extent of political egalitarianism and the overall access to higher education, as measured by the human opportunity index; the more egalitarian the political structures, the more expanded higher education is likely to be. Similarly, the more equally wealth is distributed in a particular society, the more likely higher educational opportunities are to be equally distributed. In short, our findings support the idea that both economic equality and political egalitarianism are both positively associated with more equitable access to higher education, even after controlling for secondary completion rates. That said, our findings cannot point to directionality—it is possible that the relationships we identify are bi-directional, such that more equal access to higher education is both a result of and a contributing factor to greater economic equality and political egalitarianism.

Our study is one of the first we know of to use the ECI from V-DEM as a predictor variable in analyses of educational access and equity. While other data and indices from V-DEM have been used in political science and related fields to examine various political processes, our findings show that it is also associated with equitable access to higher education. Although some prior work has used V-DEM to look at academic freedom in higher education (Kratou and Laakso 2021; Spannagel and Kinzelbach 2023), our findings show that the dataset may be useful to scholars interested in access and inequality as well, suggesting its relevance to broader audiences.

Another important implication of our regression findings concerns the use of metrics. In our analysis, we use two distinct indicators of wealth-based inequality in higher education attendance rates and find that each is correlated with different societal-level factors. This suggests that different indicators capture distinct dimensions of educational inequalities and that precise metrics are important for uncovering specific relationships.

Importantly, the indicators of inequality we use capture inequality across the wealth spectrum; this approach is different than other common indicators of inequality, which compare education attendance rates of only the top and bottom quintiles. Given that higher education access is expanding primarily among the middle classes in many parts of the world, we argue that inequality measures should capture differentiation across the full wealth spectrum, among the most and least advantaged. Examining the full wealth spectrum and validating prior research on the role of societal structures is one of the important contributions of our study to cross-national research on higher education inequality.

Finally, it is worth noting that our findings show that secondary completion rates and GDP per capita are both associated with higher education inequalities. This is perhaps not surprising as both factors have been found to be strong predictors of overall access to higher education in other studies (Buckner and Khoramshahi 2021), but our analysis supports these findings with a novel cross-national dataset that focuses specifically on inequalities.

Future research can and should build on our findings by exploring the relationship between inequality and other country-level factors. For example, studies should examine the relationship between public expenditures on tertiary education and the extent of privatization in national higher education systems and educational inequalities, as we expect both to be associated with inequality. In addition, although investigating the impact of specific policies was outside the scope of this analysis, numerous and varied policy mechanisms likely play a role in mediating the relationship between economic and political inequality and higher education access, and that these relationships warrant further study in the future. Comparative research could also examine how democratic processes mediate the relationship between political egalitarianism and higher education access, and how this relationship varies across different economic and social systems. For example, future research can investigate how social democratic and more market-oriented capitalist models shape the extent and form of redistributive policies and their impact on higher education access and equity.

Additionally, we recognize that our study relies on only one measure, the ECI, to capture political egalitarianism. This index is based on expert assessments and reflects only a small set of political features. Future work could test the robustness of our findings using alternative indicators of political structures, such as measures of electoral democracy or regime type, which may offer additional insights into how political structures shape educational opportunities.