Claim Validationdialectical analysis skilllifestyle

Is Marriage No Longer a Sustainable Institution?

What everyone believes

“Marriage is no longer a sustainable institution.”

The data's verdictmixed

Nearly four in ten Americans — 39% — told Pew in 2010 that marriage is becoming obsolete, up from 28% when Time magazine asked the same question in 1978. The US marriage rate has fallen from 16.4 per 1,000 people at its 1946 peak to 5.1 at its 2020 trough, the lowest ever recorded. Case closed?

Not quite. Because the divorce rate has been falling for forty years too — and the marriages that do form are lasting longer than their parents’ did.

The belief

“Marriage is no longer a sustainable institution.” People marry less, divorce freely, raise children outside wedlock, and the whole arrangement is on its way out. To make that testable, this analysis reads “sustainable” three ways: marriage formation should be collapsing without a floor; the marriages that form should be failing at rising rates; and the institution’s social footprint should be shrinking toward irrelevance. If all three hold, the belief is supported.

Thesis: the decline is real

Start with what the believers get right — the formation numbers are genuinely historic.

US marriages and divorces per 1,000 population, 1886–2022
136 years of national vital statistics. The marriage rate's 2020 trough (5.1) is the lowest ever recorded — below the Great Depression's. But note the second line: divorce has been falling since 1985.

The 2020 figure of 5.1 marriages per 1,000 people isn’t just low — it’s lower than the Great Depression’s worst year (7.9 in 1932), when weddings were postponed for lack of money. And this is not an American quirk.

Marriage rates: 1990 vs. 2022, across 38 countries
Of 38 countries with data in both years, 36 saw marriage rates fall. This is a civilizational trend, not a US anomaly. Source: national statistics offices via OWID.

Every dot below the dashed line is a country that marries less than it did in 1990. That’s 36 of 38. Alongside formation, the institution’s monopoly on family life has broken:

Share of US births outside marriage, 1960–2020
From 5.3% of births in 1960 to 40.5% in 2020 — an eightfold increase. Whatever marriage is now, it is no longer the near-universal container for childbearing. Source: OECD Family Database via OWID.
0%

of Americans said marriage is becoming obsolete when Pew asked in 2010 — up from 28% in 1978. The belief under test is itself widely believed.

Antithesis: the survivor institution

Here’s what the obsolete-marriage story leaves out, and it’s not small.

The divorce rate has been falling for four decades. Look back at the first chart: divorces per 1,000 people peaked at 5.0 in 1985 and have fallen to 2.5 — half the peak, and roughly the level of the late 1960s, before no-fault divorce existed. If marriage were becoming less workable, the marriages that exist should be failing more. The opposite is happening.

The marriages that form are lasting longer. The best long-run public cohort data — England & Wales, tracking each year’s weddings by anniversary — shows it directly:

Share of marriages ending in divorce within 10 and 20 years, by year married (England & Wales)
Couples married in 1994 were the peak: 24.7% divorced within 10 years. Couples married in 2013: 16.8% — a one-third improvement. US cohort research (Rotz 2016; Cohen 2019) finds the same direction. Source: UK ONS via OWID.

People marry later, not never — and later marriages hold. The average US woman now first marries at 28.6, up from 25.1 in 2001. Demographers consistently find later first marriage is one of the strongest predictors of a marriage surviving.

Mean age of US women at first marriage, 2001–2021
From 25.1 to 28.6 in two decades. Marriage is being postponed and selected into — not abandoned. Source: UN World Marriage Data via OWID.

Marriage isn’t dying. It’s consolidating — fewer entries, later entries, and the ones that happen hold together better than their parents’ generation ever managed.

The complication neither side owns

There’s a third reading the data forces, and it belongs to neither the declinists nor the optimists: marriage is becoming a class institution.

Share of US adults married, by education, 1960 vs. 2008
In 1960, college graduates and high-school-or-less adults married at nearly the same rate (76% vs. 72% — a 4-point gap). By 2008 the gap was 16 points (64% vs. 48%) — and Pew's later work shows it widening further. Source: Pew Research Center.

The falling divorce rate and the rising stability are concentrated among the same people still marrying: older, more educated, more financially settled couples. Marriage is thriving as the capstone of already-stable lives, and receding as the foundation of unstable ones. That’s neither “obsolete” nor “sustainable as ever” — it’s the institution narrowing its membership.

Synthesis

Score the belief against its three testable parts:

  • “Formation is collapsing without a floor” — half true. The decline is real, historic, and global (36 of 38 countries). But the US rate rebounded from the 2020 trough of 5.1 to 6.2 by 2022, and roughly two-thirds of American adults still marry in their lifetime. A floor keeps showing up.
  • “The marriages that form increasingly fail” — false, and it’s not close. Divorce has halved since 1985; every recent cohort outlasts the one before it at each anniversary.
  • “The institution’s footprint is collapsing to nothing” — false as stated, but transformed. Marriage lost its monopoly on childbearing (40.5% of births outside it) and its universality across class lines. It survives — selectively.

Verdict: mixed. Marriage as the default life script — universal, early, and mandatory — really is gone, and the believers are right about that. Marriage as an institution is not collapsing: it is contracting into a smaller, later, more durable, and more class-bound form. An institution that keeps a floor under formation, halves its failure rate, and retains two-thirds lifetime participation isn’t unsustainable. It’s exclusive — which is a different problem, and arguably a worse one.

Sources

Methodology

Long-run US marriage and divorce rates (1886-2022) come from Our World in Data's compilation of national statistics office and UN Demographic Yearbook series — for the US, this reflects CDC/NCHS national vital statistics. Cross-country comparison uses OWID's 1990-vs-2022 marriage-rate dataset (38 countries with both years available). The 'marriages that last' evidence uses England & Wales divorce-by-anniversary cohort data (UK ONS via OWID) — the best long-run public cohort series available; US cohort patterns from published demographic research (Rotz 2016, Cohen 2019) show the same direction but lack an equivalent open dataset, which is stated in the text. Popular-belief evidence: Pew Research Center's 2010 survey (39% say marriage is becoming obsolete) and 2023 unpartnered-adults analysis. Every charted number was re-derived from the downloaded source files during the run (the skill's number audit); provenance per figure lives alongside the data files in the analysis folder.

Limitations

Crude marriage/divorce rates are per 1,000 total population — they are sensitive to age structure (an aging population mechanically lowers both), which overstates decline somewhat; rates per unmarried adult would be better but no long-run open series exists. The divorce-by-cohort improvement is demonstrated with England & Wales data, not US data — the direction matches published US research (marriages formed since the 1990s divorce less by each anniversary than 1970s-80s cohorts), but the exact US magnitudes aren't independently derivable from open data here. 'Sustainable' is not a statistical term: this analysis operationalizes it as (a) does marriage formation retain a stable floor, (b) do formed marriages increasingly fail, (c) is the institution's social footprint collapsing to nothing — other operationalizations (e.g., financial sustainability of single-earner households) would be different analyses. Same-sex marriage data, a real part of the institution's modern shape, enters the aggregate rates but isn't separately analyzed here.

~ $ produced with the dialectical analysis skill — thesis, antithesis, synthesis: researched, provenance-audited, and self-judged end to end.