Gender, Family & Sexuality: What Is Measured and What Is Argued

In Niger, the average woman has about six children. In South Korea, she has fewer than one. Both facts are true in the same year, 2024, on the World Bank's series, and they describe the biggest change to family life in human history: the number of children people have, and when they have them, has fallen across almost every society, at very different speeds.

Gender, family and sexuality are the parts of life where sociology is most often asked to explain what is "natural". The honest answer from the evidence is layered. Some patterns are very widespread; the forms they take vary a great deal; and who counts as a family, a woman, a man or a couple is itself written differently in different laws and cultures. It is also a field where people arrive with convictions rooted in religion, politics and personal experience. This piece gives the measurements, from several regions and with dates, and keeps the descriptive findings apart from the moral positions, each of which is stated in terms its holders would recognise.

Families: the same jobs, many designs

Every society has to raise children, care for the old and the sick, pass on property or rights and manage sex and marriage. How these are bundled into households differs. Sociologists and anthropologists distinguish the nuclear family (parents and children), the extended family (several generations or siblings together), and many ways of tracing descent: through fathers (patrilineal), mothers (matrilineal) or both.

Some well-known stories need correcting. The belief that industrialisation destroyed a once-universal extended household was challenged by the historian Peter Laslett and the Cambridge Group, who showed that in England the household was already small and nuclear centuries before the factory. The demographer John Hajnal (1965) described a "Western European marriage pattern" of late marriage and many people never marrying, in a region west of a line from St Petersburg to Trieste, found by the 1700s; it is a regional history, not a law of modernisation. Matrilineal societies exist from the Akan of Ghana to the Minangkabau of Sumatra and the Khasi of Meghalaya in India, and show that "patriarchal family" is not the only historical form. In China, Fei Xiaotong's From the Soil (1947) described Chinese social relations as ripples spreading out from the self through kinship, unlike the "bundles of sticks" of Western associations.

Marriage itself varies. In George Murdock's 1967 Ethnographic Atlas, most of the societies recorded permitted polygyny, though in practice most marriages even there were monogamous. Today polygyny is legal in a range of countries, many with Muslim majorities, and illegal in others. Age at first marriage is a clear marker of difference.

Without JavaScript: women's average age at first marriage was 17.2 in Niger (2015), 21.4 in India (2016), 27.5 in the United States (2010) and 32.1 in Norway (2018).

Fertility: the arithmetic of replacement

The total fertility rate (TFR) is the average number of children a woman would have if she lived through today's age-specific birth rates. In low-mortality countries a TFR of about 2.1 keeps each generation the same size (a little over two because not every child is a girl who survives to have children); where child mortality is higher, replacement is higher. The World Bank's series (from UN data) gave a world TFR of about 2.19 in 2024. In 1950 it was close to five.

The tool below runs that arithmetic for three generations. The effect of a rate that looks only a little under 2.1 is slow; the effect of a rate under 1 is dramatic.

Without JavaScript: at a TFR of 0.75 (South Korea, 2024), 100 people are followed by about 36 children, 13 grandchildren and 5 great-grandchildren in a closed population; at 5.93 (Niger), by about 282, 797 and over 2,200.

Why has fertility fallen? Several explanations compete, and most researchers think more than one is right. Gary Becker's economic account says that as wages and the value of education rise, parents choose fewer children and invest more in each (the "quantity-quality" trade-off). Demographers of the second demographic transition (Ron Lesthaeghe, Dirk van de Kaa) point to changing values about autonomy and self-realisation, and to contraception. Peter McDonald's gender-equity argument holds that fertility falls furthest where women have gained equality in schools and workplaces but family life stays traditional, which could help to explain very low rates in South Korea, Japan and Southern Europe. Cost-of-living accounts stress housing and childcare. On policy, reviews of cash and leave incentives (for example, Anne Gauthier's 2007 review) typically find modest effects on completed fertility, mostly on timing. The United Nations' 2024 projections have the world population peaking at around 10.3 billion in the 2080s, then declining.

Gender gaps: it depends what you measure

"The gender gap" is not one quantity. Women are far behind men in some measures, ahead in others, and the gaps do not line up by country. The World Economic Forum's Global Gender Gap Index measures the relative gap between women and men on economics, education, health and politics. Its 2025 edition found that 68.8% of the global gap had been closed, led by Iceland (92.6%, its sixteenth year at the top), and estimated 123 years to parity at the current pace. A well-known caveat is that this index compares women with men inside each country, not women's absolute standing: a poor country with small gaps can score well. Other indices (UNDP's Gender Inequality Index) include absolute health and income terms.

Without JavaScript: women hold 63.8% of seats in Rwanda and 4.2% in Nigeria (2025); women live 7.0 years longer than men in South Africa and 0.6 in Nigeria (2024); the sex ratio at birth is 1.10 boys per girl in China, against about 1.05 worldwide (2024).

Try each measure. Women's labour participation is higher in Niger and Nigeria than in India or Pakistan, because in much of the poorest agriculture nearly everyone works; it says little about pay, ownership or power. Life expectancy runs the other way: women outlive men almost everywhere, by a gap that biology (men's higher mortality from heart disease, accidents and violence, and risk behaviours) and social conditions both shape. The sex ratio at birth shows where preferences for sons have been acted on: the natural ratio is about 1.05 boys per girl, and values well above it, as in China (1.10 in 2024), point to sex-selective abortion. The economist Amartya Sen's 1990 essay "More than 100 million women are missing" started the debate over the cumulative toll of such practices; later estimates differ with method and have been revised.

The wage gap: raw, adjusted and the argument over what to adjust

The raw (or unadjusted) pay gap compares typical pay for all women and all men. It is a fact about the labour market as a whole. The adjusted gap compares women and men who match on hours, occupation, industry and experience. It is closer to a measure of pay differences for "the same job". The two answer different questions, and the argument is not about the arithmetic but about the controls.

Without JavaScript: in this illustration a raw gap of 20 points shrinks to 14 when matched on hours, 8 when also matched on occupation and industry, and 6 when also matched on experience, leaving an unexplained 6.

There is a measurable event behind this. Henrik Kleven, Camille Landais and Jakob Søgaard (2019) followed Danish parents using administrative data: women's and men's earnings tracked each other until the first child, then diverged, leaving a long-run earnings gap of around 20% that is attributable to children. They estimated that child penalties accounted for most of the remaining gender earnings inequality in Denmark by 2013 (about 80%, up from about 40% in 1980). Denmark has generous parental leave and childcare, so the point is not that the gap is a Danish problem but that it shows up even there. Which policies would narrow it (paid leave for fathers, childcare, flexible hours, anti-discrimination law) is a political question, and the evidence on each is itself debated.

Are women and men different? Two honest answers

Psychologists measure differences between two groups with an effect size, d: the gap between the averages divided by the typical spread within groups. A d of 0.2 is called small, 0.5 medium and 0.8 large. In 2005 Janet Hyde reviewed meta-analyses of psychological gender differences and concluded that about 78% of the effect sizes were small or close to zero, summarised as the "gender similarities hypothesis". Some traits do show larger gaps: throwing distance, physical strength and some interests (people versus things) show moderate to large d.

Critics (Marco Del Giudice, Tom Booth and Paul Irwing, 2012) replied that looking at one trait at a time understates the picture. When many personality traits are combined, the overall difference can be much larger, with a multivariate d around 2.7 in their analysis of a personality questionnaire. Defenders of Hyde's view respond that such combined measures depend on the model and the sample, and that overlap in each real-world task is still very large. Both sides use the same data; they disagree on which summary matters. The tool below shows what a given d means in practice.

Without JavaScript: at d = 0.2, the two distributions overlap about 92%, a randomly chosen member of the higher-scoring group beats a member of the other group about 56% of the time, and the higher-scoring group supplies about 59% of the top tenth.

A related dispute: Gijsbert Stoet and David Geary (2018) found that the share of women among science graduates was lower in more gender-equal countries, which they called a "gender-equality paradox" and read as evidence that freer choice leads to more divergent preferences. Critics, such as Richardson and colleagues (2020), argued that the pattern depends on how graduation and equality are measured and that economic necessity pushes women toward STEM careers in poorer countries. Margaret Mead's Sex and Temperament (1935), which reported very different gender roles in three New Guinea societies, is often cited for the argument that roles are made by culture; its fieldwork has been challenged (as Derek Freeman did with her Samoa book in 1983), though the broad finding that roles vary widely across societies stands on much other evidence.

Sexuality: attraction, behaviour and identity

"How many people are gay?" has no single answer because sexuality has at least three dimensions: who one is attracted to, what one does, and how one describes oneself. They overlap but are not the same, and the labels people use depend on history and language. Alfred Kinsey's 1948 volume on men and the 1953 volume on women introduced a seven-point scale from 0 (exclusively heterosexual) to 6 (exclusively homosexual) to capture the range. His samples were criticised: a 1954 review by Cochran, Mosteller and Tukey for the American Statistical Association found serious problems with his sampling. Modern surveys use probability samples, but results still depend on wording and on how safe people feel answering.

Without JavaScript: in this invented example of 1,000 people, counting only the highest score gives 3% while counting any score above zero gives 12%.

Real figures show the same thing. In the 2021 census of England and Wales, 3.2% of people aged 16 and over (about 1.5 million) identified as gay, lesbian, bisexual or another sexual orientation, and 7.5% did not answer the question. Gallup's 2023 telephone survey in the United States found 7.6% of adults identified as LGBTQ+. The two numbers differ because of method, wording, country and age mix, not because one is wrong.

Where do sexual orientations come from? Three views are common. Constructionists, following Michel Foucault's The History of Sexuality (1976), argue that the categories "homosexual" and "heterosexual" as identities are modern, and that other societies organised same-sex behaviour by age, rank or ritual instead. Essentialist or biological accounts argue that orientation is stable and has biological correlates. A large 2019 genetic study by Andrea Ganna and colleagues (477,522 people, Science) found that many genetic variants of small effect together explain somewhere between 8% and 25% of the variation in whether someone has had a same-sex partner, with no single "gay gene" and no way to predict an individual. Interactionist accounts combine the two: biology shapes dispositions, and society shapes which dispositions become identities. Historians such as John Boswell and David Halperin have argued for a long time about how far the constructionist claim holds.

Societies have recognised more than two genders and various sexualities in their own terms: hijra communities in South Asia (India's Supreme Court recognised a legal third gender in the 2014 NALSA judgment), muxes among the Zapotec of Oaxaca, fa'afafine in Samoa, kathoey in Thailand and two-spirit people, a pan-Indigenous term adopted in North America in 1990. Scholars caution against treating these as one "third gender" category: each has its own history, and members describe themselves in their own words.

What a graduate just sees

• The "traditional family" is not one thing: households, descent and marriage vary, and the small nuclear household is older in some places than industry.
• A TFR is a rate, not a headcount; below about 2.1 a population shrinks without migration, slowly at 1.8 and quickly at 0.8.
• Fertility has fallen nearly everywhere; the reasons are debated, and cash incentives usually have modest effects.
• "The gender gap" must be named: pay, participation, power, health, or the sex ratio at birth. Gaps in one do not predict gaps in another.
• Raw and adjusted pay gaps answer different questions; the dispute is about the controls.
• An effect size d tells you the overlap: even a medium difference leaves two groups heavily overlapping.
• How many people are gay, lesbian or bisexual depends on whether you ask about identity, attraction or behaviour.
• Evidence and moral positions travel separately: a genetic study or a survey count settles neither a marriage law nor a religious teaching.

Where it shows up next

• Race, Ethnicity & Migration (35/79): fertility and family patterns of migrants, and how gender and race categories intersect.
• Public Policy (35/16): parental leave, childcare, pension and anti-discrimination design all rest on these measurements.
• Developmental Psychology (35/24) and Social Psychology (35/25): how children learn gender roles and how stereotypes work.
• In the world: a pension forecast built on a fertility assumption, a pay-transparency law, a census question on sexual orientation, a court case on marriage.

If you want to go deeper

• 📘 Textbook. The free OpenStax Introduction to Sociology (3rd ed.) chapters on gender, sex and sexuality and on marriage and family; and Raewyn Connell, Gender: In World Perspective (Polity, 3rd ed. 2015), which covers several regions.
• 📘 Textbook for the economics. Claudia Goldin, Career and Family: Women's Century-Long Journey toward Equity (Princeton University Press, 2021), the book behind her 2023 Nobel, which tells the story with data.
• 📘 Gentler. Stephanie Coontz, Marriage, a History: How Love Conquered Marriage (Viking, 2005), a readable sweep of how marriage has changed.
• 🌐 Free data. Our World in Data on fertility and the UN Population Division's World Population Prospects.
• 🌐 Free course. Khan Academy: Society and culture has short lessons on gender, family and demographics.