10 Is dating and marriage worth it

In two speed-dating studies, participants filled out more than 100 self-report scales related to partner choice before meeting. Researchers used a random-forest model (an algorithm for finding patterns) to calculate. They could calculate “how easily a person likes others overall”; these questionnaires accounted for 4%–18% of the variance. Variance is the difference between people; the questionnaires explain only a small part of it. They could also calculate “how popular a person is overall,” accounting for 7%–27%. But “whether these two specific people click” could not be calculated; no combination of pre-meeting traits and preferences worked. The same researchers earlier ran another speed-dating experiment. Participants reported their own ideal-type preferences beforehand, such as valuing looks or economic prospects. That preference did not predict their interest in a specific person on the spot.

The Public Security Administration Punishments Law, in force from 1 January 2026, lists entanglement and stalking as a separate item. The text covers “repeatedly sending obscene, insulting, or threatening information, or using harassment, entanglement, stalking, or similar methods to interfere with another person’s normal life.” The act is punishable by detention of up to 5 days, or a fine of up to 1000 yuan. For more serious cases, detention of 5 to 10 days, and a fine of up to 1000 yuan may be added. There is a further layer: with approval from a public-security responsible official, the actor can be ordered not to contact the victim for a set period. Violating that order brings another 5 to 10 days of detention.

Two studies measured how accurately people spot flirting. The first had 52 pairs of opposite-sex strangers talk for 10 to 12 minutes. Afterward each reported whether they had been flirting and whether they thought the other had. The result was that interactions with no flirting were judged more accurately, and interactions with flirting were judged worse. The second used 26 one-minute videos judged by 261 bystanders, with the same conclusion. People can fairly reliably see “the other person was not flirting,” but struggle to see “the other person was flirting.” A third study looked at behaviour. It pooled three speed-dating sessions, 559 people and more than 6600 dates, then followed up. The more distinctive one person’s liking for a given partner at the meeting, the more they later messaged and asked that person out. The average correlation for such links was about 0.15 (a weak association). That shows interest lands in behaviours like initiating contact and suggesting meetings. But this study did not compare behaviour against faces and tone for accuracy.

A machine-learning analysis of 29 labs and 43 longitudinal paired datasets — longitudinal means following the same couples over time — found that the strongest predictors of relationship quality all came from how each person felt about the relationship: sensed commitment from the other, appreciation, sexual satisfaction, sensing the other’s satisfaction, and conflict. Among person-level items the strongest were life satisfaction, negative emotion, depression, avoidant attachment, and anxious attachment. These relationship items explained at most 45% of the variance at the start of the survey. Variance here is the difference in relationship quality between couples; 45% is the part those items can account for. By the end of the study they explained at most 18%. And the predictive power of “items you report yourself” was two to four times that of “items your partner reports.” Adding individual differences and the partner’s reports on top of your own relationship feelings added no further predictive power. Whether the relationship got better or worse during the study was largely unpredictable from any set of self-report measures; the original wording is “largely unpredictable from any combination of self-report variables.”

Pair two strangers and have them take turns answering three sets of progressively deeper questions, 36 in all. Questions range from “Who would you most like to invite to dinner” to “When was the last time you cried in front of someone.” Closeness after 45 minutes was significantly higher than in a control group that made small talk for the same time. It was also higher than the average rating university students in that study gave their closest real-life relationships.

Two studies reach the same conclusion. Long-distance relationships are more stable while apart than same-city ones. The reason is that idealization of the partner is markedly higher in long-distance relationships; the studies measured idealized distortion, romantic love, relationship memories, and perceived agreement. And the less often you meet, the more idealized the picture. But that stability holds only while both sides remain apart; once they become same-city, former long-distance couples are more likely to break up. The study that tracked long-distance to same-city gave these shares: about half of long-distance couples reached moving in together; the other half ended while still apart. Of the half who reunited, one third broke up within 3 months after reunion. Changes participants reported after reunion included: loss of autonomous space, knowing more of both the good and the bad about the other person, hard-to-arrange time, and more conflict and jealousy. Longer gaps between meetings while apart and stronger idealization meant less stability after reunion.

In 2024, 6106000 couples completed marriage registration nationwide under law, down 20.5% from the previous year, with a marriage rate of 4.3‰. In the same year, 3513000 couples completed divorce procedures under law, of which 2622000 were divorce registrations at civil-affairs departments and 891000 were court judgments or mediated divorces. The divorce rate was 2.5‰. The bulletin notes that the marriage (divorce) rate = number of marriage (divorce) couples that year / average total population that year × 1000‰. This is a “crude rate”; the denominator is the whole population, not the married population.

A meta-analysis of older adults pooled multiple studies and recalculated them, covering 53 comparisons and more than 250,000 people. All were cohort studies — follow-up only, with no assigned comparison groups. Married versus non-married all-cause mortality RR 0.88, 95% CI 0.85–0.91 (RR is the risk ratio; CI is the credible range), about 12% lower probability of dying in the same period. In the same analysis, widowed 1.11, divorced or separated 1.16, never married 1.11 — about 11%, 16%, and 11% higher respectively. No difference by sex. The most conservative estimate was RR 0.94, only about 6% lower. Another meta-analysis compared never-married versus married, using 95 papers and more than 500,000,000 people. Result HR 1.24 (1.19–1.30), about 24% higher probability of dying in the same period. And the older the age, the smaller the gap. A further meta-analysis split by sex, 21 studies and more than 7,890,000 people: for both sexes, non-married status co-occurred with total mortality, cancer death, and cardiovascular death. On the male side the gaps for cardiovascular death and total mortality were larger. Never-married men versus never-married women had 31% more stroke deaths and 9% more all-cause deaths.

The 2024 National Time Use Survey tracks unpaid labor, which includes housework and family care. On average, people who perform such tasks spend 2 hours and 45 minutes daily on them; men dedicate 1 hour and 52 minutes, women 3 hours and 29 minutes. The proportion of individuals engaging in unpaid labor each day is 67.5% among men and 83.9% among women. Nationwide, residents spend an average of 1 hour and 17 minutes daily on housework; among those who actually do it, the average rises to 1 hour and 59 minutes — a decrease of 28 minutes compared to 2018. Time spent caring for family averages 30 minutes per day; for those involved, it reaches 1 hour and 46 minutes. In 2018, men spent 45 minutes on housework and 17 minutes on childcare, while women spent 2 hours and 6 minutes on housework and 53 minutes on childcare.

The default rules without a separate agreement are: wages, bonuses, and labor remuneration earned after marriage, returns from production, business, and investment, intellectual-property returns, and inherited or gifted property belong to both spouses jointly (Article 1062). Premarital property, personal-injury compensation, property a will or gift contract clearly gives to one side only, and articles for one person’s exclusive daily use are personal property (Article 1063). The two sides may agree in writing that premarital and postmarital property belongs to each separately, to both jointly, or partly separately and partly jointly. That agreement binds both sides in law (Article 1065). A side that bears more duty because of raising children, caring for elders, or assisting the other side’s work has the right to claim compensation at divorce (Article 1088).

Parents’ contribution toward a home for the couple before marriage is treated as a gift to their own child personally, unless the parents clearly say it is a gift to both. Contributions after marriage follow any agreement. With no agreement or an unclear one, they are handled under joint marital property. From 1 February 2025 there is a further rule. Where during the marriage one side’s parents paid the full amount and the gift contract did not say it was only for their own child, at divorce division the court may award the home to the contributing parents’ child. After that award it weighs several facts to decide whether to compensate the other side and how much: cohabitation and birth of common children, divorce fault, contribution to the family, and the home’s market price at divorce. Where one side’s parents paid part, or both sides’ parents paid, award to one side is based on the source and share of funds, with reasonable compensation (nationwide).

Debt in the following situations is joint marital debt. First, both spouses signed. Second, one side later acknowledged it — both accept the debt. Third, one side during the marriage borrowed in their own name for ordinary family daily needs. Money one side borrowed in their own name that clearly exceeds ordinary household spending is not joint marital debt. The exception is when the creditor (the party seeking repayment) can prove the money was used for the couple’s joint life, joint production or business, or was acknowledged by both. Home-purchase money clearly exceeds ordinary daily family needs; the burden of proof is on the side asserting the claim — the creditor must produce evidence. But if that money really became a home the couple live in together or own jointly, the creditor can often prove the “used for joint life” layer. So what actually helps is the step in item 11 (when parents contribute, write whether it is a loan or a gift) (nationwide, from 2021).

The new marriage-registration regulation, in force from 10 May 2025, changed the materials inland residents must produce to marry. The text says only two things need be produced: “(1) the person’s resident identity card; (2) a signed statement that the person has no spouse and has no lineal blood relationship and no collateral blood relationship within three generations with the other party.” The household register is no longer required. The registration authority is “the civil-affairs department of the county-level people’s government, or a township (town) people’s government designated on the principle of convenience by the people’s government of the province, autonomous region, or municipality directly under the central government.” The old regulation’s limit of going “to the place of permanent household registration of one of the parties” has been removed. The regulation also says “the marriage-registration authority shall not charge a fee for handling marriage registration,” and “where the parties meet the conditions for marriage, registration shall be made on the spot and a marriage certificate issued.” The situations for refusing registration are four: “(1) not yet of statutory marriage age; (2) not the complete voluntary will of both the man and the woman; (3) one or both already have a spouse; (4) lineal blood relatives or collateral blood relatives within three generations.”

The legal consequence of concealment is clear. The new marriage-registration regulation provides that “where one party has a serious disease, that party shall truthfully inform the other party before marriage registration; where that is not done truthfully, the other party may, under Article 1053 of the Civil Code, ask a people’s court to revoke the marriage.” The state treats the premarital check as the first gate in preventing birth defects. The current targets are that “the premarital medical examination rate and the coverage rate of the target population for pre-pregnancy eugenic health checks remain above 70% and 80% respectively.” The document also requires “promoting premarital medical examination institutions and marriage-registration places being sited near each other for public convenience,” and promoting “one-stop” service for “marriage registration, premarital medical examination, and fertility guidance.” The Civil Code provides that a party with a serious disease shall truthfully inform before marriage registration; if not, the other party may ask a people’s court to revoke the marriage. A request to revoke must be made within one year of the day the person knew or should have known the ground for revocation. The premarital check is not a precondition for registering. A 2022 reply by the Tai’an Municipal Health Commission to a CPPCC proposal stated: “While the 2003 Marriage Registration Regulation cancelled ‘mandatory premarital checks,’ it also made clear that marriage-registration authorities must not attach other duties when handling registration.” Article 8 of the current regulation’s list of marriage-registration materials also has no premarital-check certificate.

Negotiated divorce requires a written divorce agreement and a personal application at the marriage-registration authority (Article 1076). Within 30 days from the day the registration authority receives the application, either side may withdraw. Within 30 days after that period ends, both must personally apply for the divorce certificate; not applying counts as withdrawal (Article 1077). When one side disagrees, go to litigation; the court should mediate first. With bigamy or cohabitation with another person, domestic violence or abuse or abandonment, gambling or drug use that is repeatedly unreformed, living apart for two full years because of emotional breakdown, and similar situations, if mediation fails divorce shall be granted. If divorce was refused and they then live apart for one full year, a new suit shall be granted (Article 1079). In 2024 there were 3513000 divorces nationwide, of which 891000 went through court judgment or mediation.

The elders’ benefit is real. But it does not automatically convert into your health benefit (item 8) or relationship quality (item 4 — whether a relationship is good depends on how you feel). The time account (item 9), the money account (items 10 to 12), and the cost of leaving (item 15) are all borne by the parties. After you keep the accounts separate, whether “task-style marriage” has high or low cost-effectiveness is something you can see yourself.

Pooling 126 studies and more than 72,000 people (a meta-analysis). Higher marital quality went with better physical health; the correlation r was between .07 and .21. A correlation of 0 is no relationship; 1 is perfect lockstep. The correlation with lower death risk was r = .11. Smaller cardiovascular reaction when arguing was r = -.13. The authors said these effects are all small — about as large as the link between habits such as diet and health. In another crossover experiment, 42 healthy couples each stayed in hospital twice for 24 hours. Once they had a mutually supportive talk; once they discussed a disagreement. After arguing, experimentally made blister wounds healed more slowly. Couples who showed more hostility in both sessions had wound healing at only 60% the speed of low-hostility couples. On the morning after arguing, their blood inflammation markers (IL-6, tumor necrosis factor) also rose more.

Article 1045 of the Civil Code lists near relatives as spouses, parents, children, siblings, grandparents, maternal grandparents, grandchildren, and maternal grandchildren. Article 1127’s statutory heirs are first spouses, children, and parents, then siblings, grandparents, and maternal grandparents. A partner is in neither list. Article 1219 covers surgery, special examinations, and special treatment. When the patient cannot or should not be told, medical staff must explain to the patient’s “near relatives” and obtain consent. Three paths can fill the gap. First, Article 33 allows an adult and “another individual willing to serve as guardian” to agree in writing in advance who the guardian will be. After the person loses or partly loses civil capacity, that person becomes the guardian, and the guardian is the statutory agent. Second, Article 165 provides that a written authorization of agency must state the agent’s name, the matter, the scope, and the term, signed or sealed by the principal. When the principal dies or loses civil capacity, the agency ends. Third, Article 1133 allows a will that gifts personal property to an individual outside the statutory heirs. An article by the Yichang Municipal Justice Bureau on the Hubei Provincial Justice Department website states that same-sex partners can notarize a voluntary-guardianship agreement naming the other as future guardian (nationwide).

Article 1051 of the Civil Code provides that a marriage is invalid only in three situations: bigamy, a prohibited kinship for marriage, and not yet of statutory marriage age. Articles 1052 and 1053 provide that revocation is available only in two situations: marriage under coercion, and concealing a serious disease before marriage. Article 17 of Interpretation (I) of the Marriage and Family Book provides that a request to confirm invalidity on grounds outside those three is rejected by judgment. So the only exit is divorce. A registered marriage of convenience is also a lawful marriage. Article 1062 provides that wages, bonuses, and labor remuneration after marriage are joint marital property. Article 1064 provides that debts both signed and debts one side borrowed for ordinary family daily life are joint marital debt. Divorce damages are in Article 1091. Situations that support damages are limited to bigamy, cohabitation with another person, domestic violence, abuse or abandonment of a family member, and “other major fault.” Article 2 of Interpretation (I) limits “cohabitation with another person” to cohabitation with an “opposite-sex person outside the marriage.” So a same-sex relationship can only be argued under “other major fault.” Damages include mental-harm damages; the deceived side must raise them together when suing for divorce; if divorce is refused they are not supported. After the Civil Code took effect, no published court judgments of this kind were found. Under the old law, a 2013 article on the Jiangsu Higher People’s Court website discussed a case of a wife suing for divorce because the husband hid homosexuality. The article listed two views. One held that homosexuality was not among the statutory faults then, so neither a larger property share nor mental compensation should be supported. The author favored the other: treat the deceived side as the “fault-free party,” care for her in dividing property, and also award mental consolation money (nationwide).

A 2020 meta-analysis pooled 58 studies and 2092 couples. After couples therapy, the before–after change in relationship satisfaction was g = 1.12 (credible range 0.92–1.31); the authors called it a large effect. g is an effect size; see the README glossary. For couples assigned to wait, over the same period g = 0.12 (−0.04 to 0.29), with no clear improvement. Couples who started worse improved more. A 2012 review wrote that couples therapy brought improvement for 70% of treated couples. Another randomized trial took 134 couples in long-term, serious discord through about 8 months of two kinds of behavioral couples therapy. At completion, satisfaction change was d = 0.90 and 0.71. By year 5, 50.0% and 45.9% of couples in the two approaches still had clear improvement. The authors call that clear improvement “clinically meaningful improvement.” At the same time, 25.7% and 27.9% had already separated or divorced.

The Civil Code provides that “upon completion of marriage registration, the marriage relationship is established. Where marriage registration has not been handled, it shall be made up.” The Supreme People’s Court’s interpretation splits people who have not registered but live in the name of spouses into two kinds. Those who already met the conditions for marriage before 1 February 1994 are treated as a de facto marriage. After that date, the court “shall inform them to make up the marriage registration.” If they do not make it up, it is handled as cohabitation. If one side dies and the other wants to inherit as a spouse, the same principle applies. A spouse is a first-order heir, so the side that never registered is not in that line. For the cohabitation itself, the court does not accept a request to “dissolve” it. It only accepts disputes over dividing property and raising children. After the registration is made up, the marriage takes effect “from the time both parties met the substantive requirements for marriage provided in the Civil Code.” Children are not affected. “Children” in the Civil Code inheritance book expressly includes children born outside marriage, who are first-order heirs the same as children born in marriage.

The Civil Code provides that spouses have “equal disposal rights” over joint property. Where during the marriage one side hides, transfers, sells, damages, or squanders joint property, the other side may ask a court to divide joint property without divorcing first. When dividing property at divorce, the side that did so may get less or nothing. If discovered only after divorce, you can sue again for another division. Interpretation (II), in force from 1 February 2025, spells out two situations. Article 6 covers tipping. Where one side, without the other’s consent, tips on a livestream platform with joint property in an amount clearly beyond ordinary household spending, that counts as “squandering.” If the other side asks for division during the marriage, or at divorce for the tipping side to get less or nothing, the court shall support it. Article 7 covers gifts outside the marriage. Where one side, for bigamy, cohabitation with another person, or another purpose that breaches the duty of fidelity, gifts joint property to someone else, and the other side asserts the gift is void, the court shall support it. Once void, the recipient must return the money. You can also on that basis ask for division during the marriage, or at divorce for the other side to get less or nothing. In a typical case released with the Supreme People’s Court’s interpretation, one side during the marriage transferred 730,000 yuan over more than 4 years to a person in an improper relationship; that person transferred 170,000 yuan back. The court ordered return of the 560,000 yuan actually received; the claim that “part of the money had been spent” was not supported (nationwide).

The Civil Code provides that how much support is paid and for how long, and the form and times of visitation, are first agreed by both sides; if agreement fails the court decides. A child under two years old in principle lives with the mother. From age two, judgment follows “the principle most favorable to the minor child.” From age eight, the child’s own wishes are to be respected. From February 2025, where one side has domestic violence, gambling or drug use, an affair or cohabitation, or has seized or hidden the child, the court preferentially considers letting the other side raise the child. The support amount looks at the child’s actual needs, the parents’ means, and local living standards. With fixed income, it is generally 20% to 30% of total monthly income. With two or more children the share may rise, but generally not above 50% of total monthly income. Without fixed income, refer to that year’s total income or the industry average at that share. For example, monthly income of 8000 yuan at that share is 1600 to 2400 yuan a month (self-calculated). Support generally runs to age 18. After an agreement or judgment sets the amount, the child may still when necessary claim reasonable costs above the original amount — for example illness or schooling that exceeds it. An agreement that one side pays no support is also valid. But if the custodial side’s living standard falls markedly, or the child’s necessary costs rise markedly, the child may still ask the other side to pay. Support not paid as agreed after divorce can be sued for as arrears. After a judgment or mediation document takes effect, if the other side does not perform, you can apply for compulsory enforcement. A person who refuses to assist visitation may be detained or fined, but “compulsory enforcement cannot be carried out against the child’s person or the visitation act” (nationwide).