45 things everyone knows about marriage that aren't true.

"80–90% of couples divorce after a child dies."

~12–16%. Compassionate Friends 2006: ~16%, fewer than half citing the death; 1999 NFO survey (N=14,852): 72% still married, 16% widowed, 12% divorced; Schwab 1998 and a 2003 nursing review found no elevated rate. Origin: Harriet Schiff, The Bereaved Parent (1977), no source cited.

"Men leave their wives when the wife gets sick."

Karraker & Latham 2015 retracted for a one-line coding error (attrition miscoded as divorce): reported 32% divorce was actually ~6%; 35% attrition, not <1%. Corrected: no general gender effect; asymmetry only for heart problems (~6-point higher divorce probability when wives fall ill).

"Couples who wait 3+ years after divorce have 22% lower divorce rates."

Untraceable; appears only on a commercial blog (connectedcouples.app), not in any NCFMR publication. Wolfinger 2007 (N=1,171): no relationship between time to remarriage and remarriage stability.

"60–73% of second marriages fail."

Untraceable. BLS NLSY79 (2024): 39.1% of second marriages had ended by age 55.

"Wait one year before dating" / "one year for every 5–7 years married."

Advice-column folklore, no primary source.

"Rebound relationships are doomed."

Wolfinger 2007 null; Brumbaugh & Fraley 2015 found faster repartnering associated with better well-being (small samples). F as a rule.

"Courtship length and post-divorce waiting are the same thing."

Courtship length before marrying a given partner predicts stability (Francis-Tan & Mialon 2015: 1–2 yrs ≈20% lower hazard; 3+ ≈39–50%); post-divorce waiting has no evidenced effect. Routinely conflated.

"Covenant marriage cuts divorce in half."

Su, Ledermann & Fincham 2023 re-analysis of the Nock data: 10.8% vs 15.8% at five years, not statistically significant; ~5% uptake, self-selected.

"Marital happiness follows a U-curve and recovers in later life."

Cross-sectional artifact; VanLaningham, Johnson & Amato 2001 five-wave panel: decline at all durations, no upturn.

"Online couples are 3× more likely to divorce."

Untraceable. Rosenfeld: no difference on breakup; Cacioppo 2013 (eHarmony-funded): slightly lower.

Precise interracial-divorce multipliers (e.g., "2.3×").

AI-aggregator origin. Real pattern: risk falls between the two same-race groups and is pairing-specific (Zhang & Van Hook 2009; Bratter & King 2008).

Age-gap divorce percentages (3% / 18% / 39% / 95%).

Randal Olson's blog visualization of Francis-Tan & Mialon mTurk data; underlying hazard ≈1.02–1.03 per year of gap; correlational.

"Black divorce rate is 30.8% vs 42% national."

Commercial content (Essence, Gitnux), incompatible denominators; use ACS/BLS.

"Gottman predicts divorce with 90%+ accuracy."

Post-hoc models fitted on the same sample; Heyman & Smith Slep 2001 cross-validation collapses it; positive predictive value ~21–43% at realistic base rates (P1). Gottman's own reliability document concedes the profiles may only map perceptions. Keep contempt and self-soothing as observables; drop the accuracy claim.

"You can tell when someone is lying from body language."

Bond & DePaulo 2006 (24,483 judges): 54% accuracy; gaze aversion, fidgeting, face-touching lack diagnostic value (DePaulo 2003); professionals no better.

"Matching algorithms find compatible partners."

Finkel et al. 2012 (PSPI): no compelling evidence any algorithm works; compatibility is emergent and dyadic.

"Opposites attract."

Dyrenforth 2010 (N=23,250): similarity <0.5% of variance; Montoya 2008: actual similarity stops predicting inside relationships. No support for complementarity.

"Similar personalities are happier together."

Same evidence; personality quality (partner's agreeableness, conscientiousness, low neuroticism) matters, similarity does not.

"Write down what you want and find it."

Stated ideals do not predict in-person desire (Eastwick & Finkel 2008; 2014 meta; 2019). Criteria work only as dealbreaker floors and life-course facts.

Fixed four-box attachment "types" and the anxious/avoidant "trap."

Attachment is two continuous dimensions; 65–80% category stability; enduring change after ~a quarter of major life events (Fraley 2021); anxiety declines with age (Chopik 2019); the trap narrative is a popularization with no established replicated finding.

"The 36 questions make people fall in love."

Aron 1997 produced a temporary closeness state in the lab; no durable-bond claim in the paper; the "one couple married" story is journalism.

"Love languages" as a matching or prediction tool.

Lacks empirical support (flagged across runs).

"Good communication skills" performances (active listening, paraphrasing) mark healthy couples.

Gottman found paraphrasing exchanges in 4.4% of happy-couple interactions and predicted nothing.

"How he treats the waiter tells you how he'll treat you."

Plausible heuristic; no dedicated study links it to partner treatment. The inconsistency version (warmth switched by status) is the defensible signal.

"Once a cheater, always a cheater."

Directionally supported but probabilistic: ~3× odds (Knopp 2017, N=484, re-partnered sample); models explain 17–21% of variance.

"You must love yourself before someone can love you."

No evidence as a prerequisite (Jansen et al. 2025/26, N=460: partial relationship only); low self-esteem is a risk factor, not a gate.

"Do the work on yourself first, then you'll attract the right person."

Joel 2020: individual traits ~5% of partner satisfaction; individual therapy not established as improving relationship outcomes (Carr 2025); transfer happens only through specific trainable behaviors and relationship-loaded variables.

"High-value man" / hypergamy as a law of female behavior.

Stated resource preference is real (Buss), small, culturally moderated (Kasser & Sharma 1999), and absent in actual choice (Eastwick & Finkel; Todd 2007).

Money-script "types" (avoidance, worship, status, vigilance) as established categories.

Four-factor structure failed in a diverse sample of 2,686 (Reiter 2025).

"Choice overload" as an iron law of dating apps.

Replication failures (Scheibehenne); real but fragile.

"1 in 3 women aged 35–39 won't conceive within a year."

Traces (Twenge 2013) to a 2004 Human Reproduction paper built on French birth records from 1670–1830. Modern prospective data: Dunson 2004 (N=782) — 18% at 35–39 do not conceive within a year, i.e., ~82% do (vs 86% at 27–34), many more within two years; Steiner & Jukic 2016 — gradual decline, no difference 30–31 vs 32–33.

"Men's fertility doesn't age."

Paternal age adds ~2 de novo mutations per year (doubling ~16.5 years); ~2× schizophrenia/ASD risk at 40 vs 20 from low bases; male fecundability declines from late 30s.

"90% of bipolar marriages end in divorce" / "60–75%" / "2–3×."

Untraceable to peer-reviewed sources (commercial blog origin); a 2–3× multiplier on a ~40–50% base is arithmetically implausible. MUSUBI (N=1,071): 2.8% divorced over 2 years.

"ADHD raises divorce risk 30–50%."

Not traceable to a peer-reviewed effect size; small clinical samples (Eakin N=33) show poorer adjustment tied to untreated ADHD.

"It takes ~9 years to seek help after recognizing a problem."

N=19 qualitative study; illustrative only. NESARC's ~8–10-year treatment gap is the defensible figure.

"One problem gambler affects ≥7 other people."

Clinical consensus, not a peer-reviewed effect size.

Financial infidelity as one number.

Range 13–43% depending on wording (TD Bank vs Bankrate/NEFE); cite as a range. Overstated when given as a single figure.

"Absent Black fathers."

CDC NHSR 71 (2013): Black fathers highest on daily care (70% vs 60% white, 45% Hispanic) and homework help (41% vs 28%/29%); nonresident Black fathers more involved than other nonresident fathers; involvement declines less on repartnering (Fragile Families, N=2,447).

"The Black marriage decline is caused by X" (culture, or economics, or incarceration alone).

Each is real and partial: economics 24.4–46.3% of the decline (Sawhill & Venator), incarceration 2.4 points per 1-point rise (Charles & Luoh), culture ~12%; no monocausal account survives (Ruggles 2022).

"Culture of poverty" / Moynihan pathology framing.

Superseded by strengths scholarship (Billingsley, Hill, Nobles) and by structural evidence; kin and congregational support are measurable protective factors.

"Wait 6 months / 1 year before introducing a partner to your child."

Clinical consensus, not data; the evidenced variable is the number of transitions the child experiences (Fomby & Cherlin; Cavanagh & Huston).

"Marriage makes you happier and healthier" as a large causal effect.

Cross-sectional selection; longitudinal bump only; never-married wellbeing often comparable (Greitemeyer 2009; Sarkisian & Gerstel 2016); the male mortality gap is largely isolation and selection, buffered by non-spousal support.

"Serial monogamy is fine for the kids as long as each relationship is good."

Child outcomes worsen with each transition regardless (Fomby & Cherlin 2007; Cavanagh & Huston 2006/2008). F [RC].

"A shared baby will bond the stepfamily."

"His, hers, and ours" findings mixed; shared child does not reliably protect against dissolution. Overstated [RC].

"Success sequence" (education → work → marriage → children) as causal guarantee.

Wilcox/IFS framing contested as confounding privilege with causation [AA].