Inequality is measured by comparing how resources, opportunities or outcomes are distributed across individuals, households or groups. The appropriate measure depends on what form of inequality is being studied, because income inequality, wealth inequality, educational inequality, health inequality and inequality of opportunity are not interchangeable. Researchers commonly examine the distribution of income or consumption using statistics such as the Gini coefficient, income shares, percentile ratios and measures of dispersion. Wealth inequality requires data on assets and liabilities, while inequality in education may involve differences in attainment or access. Measurements can be presented as absolute differences, ratios, shares or summary indices. A single indicator therefore cannot provide a complete picture of social inequality.
Income inequality is often measured by arranging households or individuals from lowest to highest income and examining how income is distributed across that ranking. Researchers may calculate the share of total income received by the bottom, middle or top portions of the population, such as the lowest tenth or highest tenth. Percentile ratios compare income at different points in the distribution, allowing researchers to describe gaps between groups.
Median income represents the middle of the distribution, while the mean represents the arithmetic average and can be strongly influenced by very high incomes. Analysts therefore need to specify whether they are using gross income, disposable income, earnings, equivalised household income or another concept. Taxes and benefits can substantially alter measured inequality, so comparisons before and after redistribution can reveal the effect of fiscal policy.
Wealth inequality is measured differently because wealth is a stock accumulated over time rather than a flow received during a particular period. It can include property, savings, investments, pensions, businesses and other assets, minus debts. Researchers may calculate the proportion of total net wealth owned by different parts of the population or compare wealth percentiles.
Wealth data are often harder to collect accurately than income data because assets can be difficult to value, privately held businesses may be opaque and very wealthy households can be underrepresented in surveys. Inheritance and capital gains also create measurement challenges because wealth can change without appearing as conventional income. For these reasons, wealth inequality estimates often involve greater uncertainty than standard income statistics.
Inequality can also be measured through ratios and distributional comparisons. The income share of the richest tenth compared with the poorest tenth can provide an intuitive description of relative concentration, while percentile ratios such as the ratio between the ninetieth and tenth percentiles summarise differences across the distribution. The Palma ratio compares the income share of the richest tenth with that of the poorest four-tenths and is sometimes used because it focuses on the distributional ends. These measures answer different questions and can produce different impressions of the same distribution. Researchers should therefore choose measures according to the phenomenon being studied rather than assuming that one statistic captures every dimension of inequality.
Group-based inequality examines differences between categories such as regions, occupations, genders, ethnic groups, age groups or educational categories. Researchers may compare employment rates, earnings, educational attainment, health outcomes or wealth between groups. These comparisons can identify disparities but do not automatically establish their causes. A difference in average earnings between two groups may reflect differences in occupation, hours worked, age, education, experience, discrimination or other factors. Statistical techniques can control for some observable characteristics, but causal interpretation requires careful research design. Group averages can also conceal substantial variation within each group, so both between-group and within-group inequality can matter.
Inequality measurement requires decisions about the population and unit of analysis. A comparison of individual incomes can produce a different result from a comparison of household incomes because household members share resources. Household income can also be adjusted for household size and composition through equivalisation, recognising that a larger household requires more resources but may also benefit from economies of scale. Researchers must also specify the time period because annual income, lifetime income and wealth accumulated over decades represent different forms of economic position. Cross-country comparisons require attention to currency, purchasing power, taxation systems, household structures and data quality. Without these specifications, apparently precise inequality statistics can be misleading.
Multidimensional inequality measures extend beyond money. Researchers may examine unequal access to healthcare, education, housing, transport, digital connectivity, environmental quality or political participation. Composite indices can combine several indicators, but combining dimensions requires choices about weighting and comparability. Equality of opportunity is another distinct concept, focusing on whether outcomes depend on circumstances outside an individual's control, such as family background, rather than simply measuring outcome differences. Equality of outcome and equality of opportunity can therefore move differently. A society may reduce some outcome inequalities while leaving substantial differences in opportunities, or it may accept some outcome differences while attempting to equalise access to key institutions.
No inequality measure is universally sufficient because every statistic simplifies a complex distribution. Summary measures such as the Gini coefficient are useful for comparing overall distributions, while income shares, percentile ratios and group comparisons show where differences are concentrated. Researchers should also report definitions, data sources, population coverage and methodological limitations. Survey non-response, under-reporting, missing top incomes, changes in household composition and differences in administrative data can all affect estimates.
The strongest analysis therefore uses several complementary indicators and interprets them alongside information about living standards, poverty, mobility and access to services. Measuring inequality is ultimately not just a matter of calculating one number but of selecting appropriate concepts and methods for the particular inequality being investigated.
