attraction.lab

Splitting Housework and Divorce: The Famous 50% Comes From a Report Almost Nobody Has Read

The report behind "couples who share chores divorce 50% more" is 228 pages in Norwegian. Read in full, the gap is 3.1% against 4.6% over four years.

Research-backed writing on attraction, dating and relationships. Drafted by AI against a named bibliography, published only when every claim links to its source.
44 min read
Editorial illustration for: Splitting Housework and Divorce: The Famous 50% Comes From a Report Almost Nobody Has Read

Before the article, one thing that belongs at the top rather than the bottom. There is a difference between an unfair division of housework and having domestic work used as a means of control. Being monitored or inspected, having your work discounted or mocked, being prevented from working or studying, being made responsible for everything while having no say in anything, or being threatened over it, are patterns of coercive and economic abuse, not disagreements about chores — and in that situation nothing on this page applies. The National Domestic Violence Hotline, linked in the sources below, describes those patterns and is reachable in the United States at 1-800-799-7233 or by texting START to 88788, around the clock. Find A Helpline, also linked below, lists free and confidential services by country. It is worth saying here because in the British national survey quoted later on this page, more women named domestic violence as a reason their relationship ended than named housework.

With that said. Two claims about housework and divorce circulate in parallel and point in opposite directions. The first says that couples who share housework equally divorce about 50 per cent more often, and is credited to "a Norwegian study." The second says that unequal housework leads to breakups, and is credited to "a Swedish study of 1,057 couples." Both are traceable to real documents. Neither is reported accurately, and nobody in the search results puts them side by side.

This page does that, by reading both sources rather than the coverage of them. The Norwegian source turns out to be a 228-page government-commissioned report, written in Norwegian and never translated, which prints the raw numbers the coverage leaves out: the famous 50 per cent is the distance between 3.1 per cent and 4.6 per cent over four years, there is a second and larger figure nobody quotes, the group carrying it contains 313 people, and the division of housework accounted for about one per cent of the variation in who separated. The report's own authors wrote that a direct causal role seems unlikely.

The Swedish source is a real journal article, and its tables say something more interesting than its title. Its separation models run on 803 couples rather than 1,057. And when both partners agree that the wife does most of the housework, there is no effect on dissolution at all. The effect appears when the two partners describe the same household differently — when he says they share and she says she does more.

What follows is eighteen sections: what each document is, what each number means in people rather than in percentages of percentages, why the two studies cannot contradict each other because they did not measure the same variable, why the answer changes between Norway, Sweden, Germany, Britain and the United States, which specific task carries the most signal, the dimension none of them measured, and the most-shared statistic on this subject, which does not appear to exist. Our editorial policy explains why every empirical claim here carries a linked source, and why the pages that cannot be verified say so.

#1

Two sentences about housework circulate, and they point opposite ways

Search for housework and divorce and the results arrive pre-sorted into two camps that never meet.

One camp says sharing is dangerous. Couples who split housework equally had a divorce rate about 50 per cent higher than couples where the woman did most of it. It is usually credited to "a Norwegian study" and it has been reprinted for more than a decade by divorce lawyers, lifestyle magazines and men's-interest sites, almost always without a link.

The other camp says the opposite. Unequal housework leads to breakups. That one is credited to "a Swedish study of 1,057 couples" and it is usually the more carefully written of the two, because it comes from a journal article with a title that asks the question directly.

Both claims are traceable. Neither is reported accurately, and the reason is the same in both cases: the number that travels is the one that fits in a headline, and the numbers that would tell you how much it matters stay in the document.

So this page does the unglamorous thing. It opens the Norwegian report — all 228 pages of it, in Norwegian, because it has never been translated — and reads the chapter the claim comes from. Then it opens the Swedish paper and reads the table the claim comes from. What comes out is not a verdict on whether you should do the dishes. It is a much more useful picture of how small these effects are, how differently the two studies measured the thing they both call "housework," and which of the two questions actually has an answer.

The short version, so nothing depends on you reading to the end. In the Norwegian data, 3.1 per cent of couples where the woman did most of the housework had separated four years later, against 4.6 per cent of couples who split it evenly. That difference is the famous 50 per cent. The report's own authors wrote that it seems unlikely housework division is a direct cause of anything, and the division of housework accounted for roughly one per cent of the variation in who split up.

The rest of this page is about why that is still interesting, and about the one finding in this literature that does replicate across countries — which is not about hours at all.

#2

The Norwegian study is a 228-page report, and chapter 9 is the part being quoted

Start with what the document is, because the genre never says.

It is NOVA Rapport 8/12, titled Likestilling hjemme — "Equality at Home" — edited by Thomas Hansen and Britt Slagsvold and published in 2012 by NOVA, at the time Norway's national institute for research on childhood, welfare and ageing, now part of Oslo Metropolitan University. It runs to 228 pages, carries ISBN 978-82-7894-432-5, and is written in Norwegian bokmål. It is a research report commissioned in the policy context that produced a Norwegian government white paper on gender equality, not an article in a peer-reviewed journal. Nothing in it went through journal peer review, and the report says so by being what it is: a report.

That matters less than it sounds, and more than the coverage allows. Less, because national statistical reports are often better documented than journal articles — this one prints its regression tables and its raw cell counts, which is how the numbers below are available at all. More, because "a Norwegian study found" invites you to picture something that was refereed, and because a 228-page report in Norwegian is a document that English-language writers can cite without ever opening.

The claim comes from chapter 9, written by Slagsvold and Hansen and titled Reduserer likedeling hjemme risikoen for samlivsbrudd? — "Does sharing at home reduce the risk of relationship breakdown?" The data are two linked Norwegian surveys. LOGG (Studien av livsløp, generasjon og kjønn) interviewed about 15,000 people aged 18 to 84 in 2007 and 2008. It was built on top of NorLAG, a panel study with two collection rounds, 2002–2003 and 2007–2008, in which 4,000 people aged 40 to 84 took part in both. Survey answers were then linked to Norwegian population registers, which is how the authors know who separated afterwards without having to ask anyone.

That design is a real strength and worth naming, because it is rarer than it should be in this field. The housework question was answered in 2007–2008, before anyone knew who would separate; the separations were read off administrative records up to the start of 2011. Nobody was asked after the fact to explain their own divorce, which is the weakness of almost every number in the genre — including the ones people give when asked why their marriage ended.

So the study is prospective, register-linked and large. The problem is not the study. The problem is what was done to it on the way out.

#3

The number everyone quotes is one of two, and it is the smaller one

Here is chapter 9's own summary sentence, translated:

"If the man does as much or more housework than the woman, the chance of divorce over the next four years is respectively 49 and 78 per cent higher than when the woman does most."

Two numbers. 49 per cent for couples who split housework evenly, and 78 per cent for couples where the man does most. Elsewhere in the same chapter the first is rounded to "about 50 per cent," which is the version that escaped.

In more than a decade of English-language coverage, the 78 is essentially absent. The pages that repeat this claim give one figure, and it is the lower of the two. That is a strange editorial choice, because the bigger number is the one that makes the folklore interesting: the arrangement with the highest separation rate in these data is not the equal one, it is the one where the man does most.

The underlying model is a logistic regression, printed as Table 9.3, on men and women aged 18 to 79 in LOGG, predicting separation between 2007/2008 and 2011. The reference group is couples where the woman does most of the housework. The odds ratios are:

  • Even split: 1.490 with no controls, 1.565 with controls.
  • Man does most: 1.782 with no controls, 2.150 with controls.

The controls are age, education on a five-point scale, children in the household, married versus cohabiting, and satisfaction with the relationship. The association does not shrink when they are added, which is the finding the report emphasises and which is genuinely worth knowing: whatever this is, it is not simply an artefact of young urban couples being both more egalitarian and more divorce-prone.

The report also notes something the coverage never mentions: there is no statistically significant difference between the even-split couples and the man-does-most couples in this sample. Both groups carry about the same raised chance relative to couples where the woman does most. So the pattern in the data is not a gradient from traditional to egalitarian to reversed. It is traditional on one side, and everything else on the other.

#4

What "50 per cent higher" means in people: 3.1 per cent against 4.6

This is the number the entire genre leaves out, and it is printed in the report as Table 9.2.

Among Norwegian couples followed from 2007/2008 to 2011, the share who had separated or divorced four years later was:

  • Woman did most of the housework: 3.1 per cent (5,989 people)
  • Split evenly: 4.6 per cent (1,833 people)
  • Man did most: 5.4 per cent (313 people)

That is the whole thing. The famous "50 per cent higher divorce rate" is the distance between 3.1 per cent and 4.6 per cent over four years — one and a half percentage points. Put the other way round, 95.4 per cent of the couples who shared housework evenly were still together four years later, against 96.9 per cent of the traditional couples.

Neither number is a relative risk in disguise and neither is being hidden by the researchers. They are in the report, in a table, with the group sizes next to them. They simply did not survive the journey into English, because "a 50 per cent higher divorce rate" is a sentence and "one and a half percentage points over four years" is an argument.

This is the single most common failure in relationship statistics and it is worth learning to spot, because it recurs everywhere. A relative increase on a small base is still small. The same arithmetic turns up in the cohabitation literature, where the famous "living together first raises your divorce risk" effect also shrinks dramatically once you ask what the base rate was and which couples were being compared, and in the research on age gaps, where large-sounding percentage increases sit on top of absolute risks most readers would consider unremarkable.

The habit worth taking from this page: when a page gives you a percentage increase and not a percentage, the percentage is usually the part that would have changed your mind.

#5

The group carrying the biggest number contains 313 couples

The 78 per cent figure — the one that would make the better headline, if anyone used it — rests on a very small group.

Table 9.1 of the report gives the distribution. Of the 8,135 people in the analysis, 73.6 per cent were in couples where the woman did most of the housework, 22.5 per cent split it evenly, and 3.8 per cent were in couples where the man did most. That last group is 313 people. At a four-year separation rate of 5.4 per cent, the entire "78 per cent higher" result is carried by roughly seventeen separations.

It is not nothing — the authors report it as statistically significant, and the direction holds with controls. But a result resting on seventeen events is a result that should be stated with its denominator attached, and a number stated without its denominator is how a tentative finding becomes a law of nature.

The same table gives the comparison from the other direction, which is the honest way to show a weak association. Among the 7,847 people whose relationships were stable, the split was 73.9 / 22.3 / 3.8 per cent. Among the 289 whose relationships ended, it was 64.9 / 29.2 / 5.9. More egalitarian couples among those who separated, fewer traditional ones. The chi-square test is significant at p < .01, and the shift is nine percentage points in the biggest category. That is what a real but modest association looks like when you print it as a table instead of as a ratio.

For completeness, because it bears on how many people this covers: 297 of the survey respondents had separated or divorced by 1 January 2011, and 289 of them — 148 men and 141 women — had answered all the housework questions, which is the group used. The oldest person who divorced was 76 at the time of the interview, a detail that quietly tells you this is a study of all Norwegian couples, not of young marriages.

#6

Housework division accounted for about one per cent of who split up

There is a line in Table 9.3 that settles the question of how much any of this matters, and it has never appeared in English.

The model's Nagelkerke R² is .011 without controls and .198 with them.

R² is a measure of how much of the variation in the outcome a model accounts for. It is not a perfect statistic and for binary outcomes the various versions of it disagree, so treat the exact figure loosely. But the order of magnitude is unambiguous: with housework division as the only predictor, the model accounts for about one per cent of the variation in who separated. Add age, education, children, marital status and relationship satisfaction and it rises to about twenty per cent — meaning almost all of the explanatory work is being done by the other five variables, not by who does the laundry.

This is the number that makes the article honest rather than contrarian. It does not say the association is fake; the association is there, it survives controls, and it points the way the report says it points. It says the association is small, in the specific sense that knowing how a couple divides the housework tells you very little about whether they will still be together in four years. Relationship satisfaction, in the same model, does a great deal more work — which is one of the reasons the site has a satisfaction test that scores the dimensions the literature actually uses rather than printing a verdict.

And the model has one more result in it that goes unquoted. In the same table, higher age, higher education, having children in the household, and being satisfied with the relationship all reduced the chance of later separation. Three of those four are the standard findings of the divorce literature — and the first of them has its own folklore problem, since the famous claim about the best age to get married rests on a model whose turning point is not where the coverage put it. The housework variable is the novelty here, and the novelty is the weakest thing in the model.

#7

The authors' own conclusion is the part the coverage dropped

The report does not claim what the headlines claim it claims. Two sentences from chapter 9, translated:

"It seems unlikely that the division of housework can be a direct cause of later divorce."

"We can, however, conclude that equal sharing of housework does not appear to reduce the chance of divorce, as has sometimes been assumed."

That second sentence is the finding. It is a negative result aimed at a specific hypothesis — the reasonable Norwegian expectation that because sharing is associated with being happier about the division, it should also show up as fewer breakups. It did not. The report says so and then says, in effect, do not read a cause into this.

The authors went further than that, and the next part is the most useful thing in the chapter for anyone who has encountered the folklore version — the idea that men who do housework get left. They tested it. They checked whether who initiated the breakup varied with how the couple had divided housework beforehand, and found no tendency at all, for women or for men.

So within this dataset, the story where the egalitarian husband is dumped for being insufficiently masculine has a direct test, and it fails. What is left is a difference in which couples end up separating, with no difference in who ends it, no plausible direct mechanism, and about one per cent of the variation accounted for.

The explanation Hansen offered publicly is the obvious one and the report's own: these are different couples. Couples who share housework in Norway tend to hold more modern views of marriage, and the women in them tend to have more education and better-paid jobs, so leaving an unhappy relationship is more available to them. That is a selection story, not a causal one, and it is the same logic that explains why people stay in relationships they describe as unhappy — what predicts leaving is not only how bad it is, but what leaving would cost.

#8

The whole English-language version of this comes from one wire interview

There is a reason every English page says the same three things about this report. They are all descended from one article.

On 28 September 2012, The Local's Norway edition published a piece headlined "Couples who share housework divorce more: study," based on an AFP interview with Thomas Hansen. It is short, and it contains the lineage of everything since:

  • The 50 per cent figure, and no mention of the 78.
  • Hansen's quotable line: "The more a man does in the home, the higher the divorce rate."
  • And, in the same piece, the sentence that almost nobody carried forward: "Researchers found no, or very little, cause-and-effect. Rather, they saw in the correlation a sign of 'modern' attitudes."

Hansen's own explanation is there too, in his words: "Modern couples are just that, both in the way they divide up the chores and in their perception of marriage" as less sacred, and "in these modern couples, women also have a high level of education and a well-paid job, which makes them less dependent on their spouse financially. They can manage much easier if they divorce."

He also gave the one speculative mechanism he was willing to offer, hedged: that clear roles may mean fewer arguments, "since you can easily get into squabbles if both have the same roles and one has the feeling that the other is not pulling his or her own weight." That is a hypothesis spoken from a park bench while minding his children, and the coverage turned it into a finding.

One detail from the interview is a useful check that the chain is intact: Hansen said women in Norway still did most of the housework in seven out of ten couples, which matches the report's 73.6 per cent almost exactly. So the wire piece was accurate. It was just thin, as a four-hundred-word wire piece on a 228-page report is bound to be, and it became the only source.

The best page currently on this topic in English is a fact-check by RemindHer, which gets most of it right, correctly calls the 50 per cent figure misleading and correctly notes that the researcher disclaimed causation. Even that page cites the report as "Coverage: The Local, TODAY" — the coverage, not the document. That is not a criticism of them so much as a description of the state of the subject: the primary source has been sitting in an open repository for fourteen years, in Norwegian, and the English-language conversation has been conducted entirely from a wire story about it.

#9

Inside the same report, the other measure runs the other way

Here is the part that makes chapter 9 more interesting than either camp's version of it. The report has two outcome measures, and they disagree.

The register measure — who actually separated — is the one that produced the famous numbers, and it points at egalitarian couples.

The survey measure is different. LOGG asked whether the respondent had considered ending the relationship in the past year, which is the same family of question as divorce ideation. Figure 9.1 breaks the answers down by division of housework, for 4,916 men and 4,766 women. Overall, somewhat more women (8.9 per cent) than men (5.1 per cent) had considered it.

And the pattern by division is this: the chance of having considered a breakup is higher, for both women and men, when they themselves do most of the housework. Women who said they did clearly most were significantly more likely to have thought about ending it than women in evenly-split or slightly-unequal couples. Men who did more than their partner were likewise more likely than men in evenly-split couples — a difference the report reports at p = .061, so on the edge rather than inside.

In the most traditional couples, where the woman did clearly most, it was two to three times as likely that the woman had considered ending the relationship than that the man had, after controlling for age.

So in one chapter, on one dataset: whoever carries the load is more likely to be thinking about leaving, and couples who share evenly are more likely to have actually separated four years later. Those are not contradictory once you notice they are different measures of different things — thoughts of leaving track how the arrangement feels to the person in it, separations track everything including whether leaving is feasible. But they are both in the report, and the genre quotes exactly one of them.

That gap between what people are thinking and what eventually happens is well documented and is a good reason not to read either measure as a forecast. Thinking about ending a relationship is common, often recurrent, and a poor predictor on its own.

Further illustration for: Splitting Housework and Divorce: The Famous 50% Comes From a Report Almost Nobody Has Read
#10

Most people are satisfied with an unequal split, including most of the women doing more

Chapter 7 of the same report measures satisfaction, and its numbers complicate both camps.

Satisfaction with the division of housework behaves exactly as you would expect. Men and women are most satisfied when housework is split evenly or when the partner does more — in those arrangements about 95 per cent of both are satisfied. The least satisfied group is the 15 per cent of women who say they do clearly most, and even among them 60 per cent are satisfied with the division.

Satisfaction with the relationship barely moves. Among men who do most and women who do clearly most, 77 and 76 per cent respectively are satisfied with the relationship; in every other arrangement the figures are 81–86 per cent for men and 83–89 per cent for women. The report puts the size of it plainly: the division of housework accounts for 14 per cent of the variation in satisfaction with the division, and 2 per cent of the variation in satisfaction with the relationship — and among men, 1 per cent and 0 per cent.

That is a result worth sitting with, because it cuts against the way the subject is usually written. Housework division is strongly related to how people feel about housework division, and weakly related to how they feel about the relationship as a whole. Even among the women doing clearly most of it, three quarters describe themselves as satisfied with the relationship.

Two warnings go with that, and they are not decoration.

The first is that "satisfied" is not the same as "fair," and this literature has known for decades that people adjust their sense of fairness to what they expect. A high satisfaction figure inside an unequal arrangement is partly a measurement of norms, not only of contentment, and the report itself notes that people who prefer a traditional division tend to find each other.

The second is that a 60 or 76 per cent satisfaction rate means a large minority is not satisfied, and that minority is where everything in this subject actually happens. The useful question is not whether unequal arrangements are generally tolerated — they obviously are. It is what distinguishes the ones that are from the ones that are not. The Swedish paper, which is the other half of this story, has the best available answer, and it is not about hours.

It is worth noticing how often this subject behaves like this. A domestic arrangement that reads as a warning sign from outside turns out, when somebody measures it, to be both common and mostly fine for the people in it — which is exactly what happened with sleeping in separate rooms, where the arrangement the genre calls a sleep divorce is reported by roughly a third of adults and the evidence for it being a symptom of anything is much thinner than the name suggests.

#11

The Swedish paper is a paper, and it does not say what the headlines say

The counter-claim comes from a real journal article: Leah Ruppanner, Maria Brandén and Jani Turunen, "Does Unequal Housework Lead to Divorce? Evidence from Sweden," published in Sociology 52(1), 75–94, submitted in September 2015 and accepted in September 2016.

Its design is genuinely novel and the authors are right to say so. It uses the 2009 wave of the Young Adult Panel Study, a Swedish survey of people born in 1968, 1972, 1976 and 1980, which in that final wave asked participants' partners to fill in a questionnaire too. That gives you both partners' descriptions of the same household — which is the thing almost no study in this field has — and it was then linked to Swedish registers through 2014 to see who split up.

The headline numbers first, since they are what gets quoted. The sample is 1,057 couples. But the breakup analysis does not run on 1,057 couples. Table 2 of the paper gives the figure: actual break-up of the couple between 2009 and December 2014, n = 803 couples, 12.6 per cent. So roughly a hundred separations, and a quarter of the original sample is not in the dissolution models at all.

It is worth noting what the paper is honest about, because it is more candid than its coverage. Of the 1,528 respondents who were married or cohabiting, only 1,074 — 70 per cent — had a participating partner. Among the couples whose partner did not take part, almost 16 per cent ended the union by 2014, against 13 per cent of the analysed sample, and those couples also appear to have shared housework slightly less equally. The authors state the implication themselves: the couples most affected are probably missing, so their estimates are likely too low.

The article is paywalled at the publisher. Semantic Scholar records its open-access status as closed and OpenAlex finds no open repository copy, so the figures below are reported with that in mind: anyone wanting to check them has to go through a library.

#12

Who you ask decides the sign: 2.16 against 0.51

Now the table that the headline version of this paper flattens. These are odds ratios from Table 3, with "we share equally" as the reference group, for the couple-level dissolution outcome.

  • When the woman says I do the most of the housework, the union is 2.16 times as likely to dissolve (p < .01).
  • When the man says my partner does the most, the union is 0.51 times as likely to dissolve (p < .05) — roughly half.

Read those two lines together, because they describe the same household. A couple in which she does more of the housework is twice as likely to split if she is the one reporting it, and half as likely if he is the one reporting it. The arrangement is not what moves the number. Who describes it is.

The satisfaction columns run the same way. When the woman says she does most, both partners' odds of being very satisfied drop sharply — his to 0.55 and hers to 0.40, both at p < .001 — and her odds of having considered breaking up rise to 1.59. When the man says she does most, her odds of being very satisfied rise to 1.64 (p < .01). Same division of labour; opposite readings, depending on whose mouth it comes out of.

And the result that ought to end the "unequal housework causes divorce" summary: when both partners agree that the wife does most, the paper reports that satisfaction is lower for both, but there is no effect on break-up plans and no effect on dissolution. An inequality that both people see and both describe the same way does not show up in the separation data at all.

The mirror case is just as instructive. Where both partners agree the husband does most — the Norwegian report's highest-risk group — his satisfaction is lower (0.43), both partners are substantially more likely to have considered breaking up (3.34 and 2.89), and the dissolution odds ratio is 1.31, not significant. The arrangement that carries the biggest number in Norway carries no reliable separation signal in Sweden.

What the paper does find is about mismatch, and the authors name it with the right word: discrediting. When he reports that they share housework equally but she reports that she does more, both partners are less satisfied, she is more likely to be considering a breakup, and the union is more likely to dissolve. The reverse — when he reports he does more while she reports they share — raises his thoughts of breaking up but does not raise dissolution. And the gentlest finding in the paper is its most practical: when he reports that she does more while she reports that they share — crediting her contribution beyond what she claims herself — her odds of being very satisfied are nearly double (1.97).

That is the real result of the Swedish study, and it is not "unequal housework leads to divorce." It is that not seeing what your partner does is the thing that tracks instability — which is a finding about attention, and sits in the same place as the research on responsiveness and feeling known and on what actually builds attachment over time.

It also lines up with the one behaviour that keeps surfacing in the research on couples who last: appreciation that is expressed rather than merely felt. The habits that distinguish couples who stay in love include noticing out loud, and the research on emotional maturity in relationships describes the same thing from the other side — being able to hear that your account of the week is not the only one.

#13

The two studies did not measure the same thing, and one of them used a single question

Before anyone reconciles the Norwegian and Swedish results, there is a prior question that neither camp asks: did these two studies measure the same variable? They did not.

Norway. The division of housework is a four-category index built from the average of six separate questions about specific tasks. The index deliberately excludes maintenance and small repairs — the tasks that are both male-coded and episodic rather than daily. The childcare measure, used in a different chapter, is likewise an average of six items: dressing, bedtime, staying home when a child is ill, play, homework help and transport.

Sweden. The division of housework is one question: "How do you [and your partner] divide housework?" with three options — I do the most, we share equally, my partner does the most.

Those are not variants of one measure. One is a task-level index; the other is a global self-assessment with three bins. And once you notice that, the direction of each study's finding becomes easier to place. A single global question is close to a measure of perception — exactly the right instrument for a paper whose real subject turns out to be whose perception you are using. A six-task index is closer to a description of behaviour, which is the right instrument for asking which arrangements exist and which ones end.

It also means the two studies cannot contradict each other in the way the SERP implies. "Couples who split housework evenly separate somewhat more often in Norway" and "women who say they do most of the housework are in less stable unions in Sweden" are compatible statements about different variables in different countries. The argument between the two camps is an argument between two measures, conducted by people who have read neither.

This is the question worth asking of any relationship statistic once the sources check out: not whether they cited something, but what instrument produced the number. It was the question that mattered for the five love languages, where almost every study uses a forced-choice measure that cannot detect someone who values two things equally, and for the research on how much couples argue about money, where the same national survey gives two different answers depending on which column of the codebook you read.

#14

Both studies threw away the same couples

There is one more thing the two studies share, and it is a limitation rather than a finding.

Both of them get their separation data from population registers, which is the strength of both designs. And both registers have the same hole: they record who is married and who shares a home with a child, but they cannot reliably identify cohabiting couples without children.

The Norwegian report states its exclusion plainly. Breakups among married people are tracked to the start of 2011; breakups among cohabitants with shared children to the start of 2009. Since there is no information about cohabitants without shared children, the authors removed them from the analyses. So the Norwegian sample is married couples plus cohabiting parents.

The Swedish paper does the same thing and quantifies it. Dissolution is measured as the partners no longer living at the same property in 2014; for cohabitants this can only be established if they had at least one child in common in 2009; cohabitants with no shared children are excluded from the breakup analyses, and they are 24 per cent of the original sample.

Two consequences follow, and they run in the same direction.

First, both studies are missing the least stable relationships in their respective countries. The Swedish authors say this outright: cohabiters have less stable relationships than married couples even in Sweden, and housework inequality may well contribute to those separations — their models cannot see it.

Second, in two of the most cohabitation-heavy societies on earth, the samples skew married. That is worth holding onto when either result is generalised to "couples." The relationships most people in their twenties and thirties are actually in are the ones the register cannot follow — the same measurement problem that makes the cohabitation-and-divorce literature so much messier than its headlines, and the reason the available numbers on modern relationships are so much thinner than the confident statistics implies.

#15

The country changes the answer, which is the only way both results can be true

If one finding in this literature deserves to be as famous as the Norwegian 50 per cent, it is this one: the association between who does the housework and whether the marriage lasts is different in different countries.

The cleanest demonstration is Lynn Prince Cooke's 2006 paper in the American Journal of Sociology, which compares West Germany and the United States using event-history analysis. In Germany, moves away from separate gendered spheres — whether in the wife's relative earnings or the husband's relative share of housework — increased the risk of divorce. In the United States, the more stable couples were the ones that adapted by displaying greater gender equity. Same behaviour, opposite association, two countries. Cooke's conclusion is that policy and norms shape what a given arrangement means, and therefore what it does.

The finding has been replicated with different data. Daniela Bellani, Gøsta Esping-Andersen and Léa Pessin, in the Journal of Social and Personal Relationships in 2018, applied event-history analysis to the German Socio-Economic Panel for West Germany and the Panel Study of Income Dynamics for the United States, covering 1986 to 2009. In the US, deviation from an equitable division was destabilising — particularly when the wife was the one under-benefiting, and particularly when both partners worked similar hours. In West Germany, "the degree of inequity has no effect at all," and the male-breadwinner arrangement remained the single most stable one. Their paper is free to read in PubMed Central, which makes it the most checkable thing in this section.

Two more national results fill in the picture, and they are the ones that get least airtime:

  • Britain. Wendy Sigle-Rushton, in Feminist Economics in 2010, used the 1970 British Cohort Study — a period when gendered specialisation was the normative arrangement for families with small children — and found that fathers' contributions to domestic work and childcare stabilised marriages, regardless of whether the mother was employed. That is the opposite of what specialisation-and-trade economics predicted, and the opposite of the Norwegian headline.
  • The United States. Michelle Frisco and Kristi Williams, in the Journal of Family Issues in 2003, looked at perceived fairness rather than hours in a nationally representative sample of dual-earner marriages. Perceived inequity was associated with lower marital happiness for both husbands and wives, and with higher odds of divorce for wives only — and marital happiness did not appear to mediate the link. Their reading was that an unfair-feeling division produces role strain that makes women more likely to end an unsatisfying marriage.

So the full set of national results is: Norway, equal sharing associated with slightly more separation; Sweden, her reporting that she does more associated with more dissolution; West Germany, no effect of inequity at all and traditional most stable; the United States, inequity destabilising and equity protective; Britain in the 1970s, fathers doing more protective.

That is not a literature in disarray. It is a literature saying something coherent and inconvenient: how a couple divides housework does not carry a fixed meaning, so it cannot carry a fixed effect. What a 50/50 split signals in Oslo in 2008 and in West Germany in 1990 are different things about different people, and the register picks up the difference between the people, not the effect of the dishes.

#16

Which task, not how many hours — and the part that is not a task at all

There is one strand of this research that is more useful than any of the divorce findings, and it changes what the question even is.

Daniel Carlson, Amanda Miller and Sharon Sassler, in Socius in 2018, stopped asking how couples divide housework overall and asked which tasks they divide. They compared the second wave of the National Survey of Families and Households, collected in 1992–1994, with the 2006 Marital and Relationship Survey. Two things came out.

The first is that the "stalled revolution" story is overstated for the couples they studied: more recent cohorts of middle- to low-income couples share all the routine tasks more than earlier ones did, with the single exception of shopping, and the biggest changes are in dishwashing and laundry.

The second is the finding worth keeping. Of all the tasks they examined, the division of dishwashing is the one most consequential for relationship quality — especially for women. Equal sharing was more positively related to satisfaction and intimacy and more negatively related to discord in the recent cohort than in the older one. The paper is open access, so this one you can check yourself.

Why dishwashing rather than, say, mowing or home repair? The mechanism the task-level literature points to is that routine housework is daily, unavoidable, time-bound, visible and unrewarding — you cannot do the dishes next weekend, and nobody admires you for it. Episodic male-coded tasks are not like that, which is precisely why the Norwegian index excluded maintenance and repairs from its housework measure.

And there is a dimension that none of the studies above measured, because it is not a task. Allison Daminger, in the American Sociological Review in 2019, interviewed 70 people from 35 couples and argued that what gets waved at as "household management" is a distinct kind of work: cognitive labour, consisting of anticipating needs, identifying options for meeting them, making decisions, and monitoring whether things got done. It is taxing, and it is often invisible to the person doing it as well as to their partner, which is why it is a recurring source of conflict.

Her findings are more precise than the popular "mental load" version, and the precision matters. Women in her sample did more cognitive labour overall, and specifically more of the anticipating and monitoring. But participation in decision-making — the component most closely tied to power and influence — was roughly equal. That is an interview study of 35 couples, not a national sample, and it should be cited as what it is: a well-made argument for a distinction, not an estimate of a quantity.

Put the two together and you get the reason both famous studies may be measuring something shallower than they intend. A couple answering "we share equally" can mean they each do three hours; or that he does the mowing and she does the dishes; or that she decided what needed doing, worked out when, and reminded him, and he then did half of it. Those are three different households with one answer. The Swedish finding that her report and his report of the same household predict opposite outcomes is exactly what you would expect if that were true.

#17

The Harvard 25 per cent does not exist, and there is a real number instead

One more claim has to be dealt with, because it is the most-shared statistic on this topic and it is not a statistic.

It reads: a Harvard Business School study of 3,000 couples found that 25 per cent of divorces are caused by disagreements about housework. It appears on divorce-firm pages, in news write-ups and in advice columns. We could not find a primary source for it, and neither could anyone else who has looked. There is no named survey, no dataset, no HBS publication reporting it, and the accounts do not even agree among themselves — the sample is variously 3,000 divorced couples and 3,000 committed couples, and the finding is variously the top reason, the third reason, and "up to 25 per cent."

The Harvard research the claim is attached to is real and is about something else. Ashley Whillans, Jessica Pow and Joe Gladstone published "Buying (quality) time predicts relationship satisfaction" in the Journal of Personality and Social Psychology in 2025 — a body of work about whether paying to outsource chores improves relationships, not about what causes divorce. A figure quoted in passing inside a working paper is not a finding of that paper, and somewhere along the chain "a Harvard paper mentioned a survey" became "Harvard studied 3,000 couples."

The British equivalent is in the same condition. A "UK study by Lindsey Taylor" finding that 30 per cent of marriages ended over unfairly divided chores circulates on law-firm pages with no author affiliation, no publisher and no date, and does not appear to exist.

This is the most common failure mode in the whole genre and it has a recognisable shape: a number with an institution's name attached and no document behind it. The same pattern turned up in the famous statistic about in-laws and divorce, where half of the quoted pair is in the paper and the other half was manufactured by press coverage, and in the claim that money arguments are the top predictor of divorce, which is a quotation from a university press release rather than a line from a study.

There is a real, representative, properly published number for this exact question, and it is smaller and more interesting. Kirsten Gravningen and colleagues reported it in PLOS ONE in 2017, using Natsal-3, a probability sample survey of 15,162 people aged 16 to 74 in Britain, interviewed in 2010–2012. Among the 706 men and 1,254 women who had experienced the breakdown of a live-in partnership in the previous five years, the reasons given were:

  • Grew apart — men 39.4%, women 35.6%
  • Arguments — 27.0%, 30.3%
  • Unfaithfulness or adultery — 18.0%, 23.6%
  • Lack of respect or appreciation — 17.1%, 24.8%
  • Different interests or nothing in common — 13.3%, 15.7%
  • Money problems — 7.3%, 11.8%
  • Not sharing enough housework — 6.8%, 14.4%
  • Domestic violence — 3.7%, 15.9%
  • Difficulties with sex — 4.9%, 6.3%

So housework is a real reason, cited by about one in seven women and one in fifteen men — nowhere near a quarter, and well behind growing apart, arguments and infidelity. Participants could give more than one reason and most gave one or two, averaging just under two each.

Two details in that table deserve not to be dropped. The first is that for women, domestic violence (15.9%) was cited slightly more often than housework (14.4%), and four times more often than by men. Any honest presentation of this table includes that line. The second is a nuance in the paper's adjusted analysis: women whose most recent ended partnership was a cohabitation were less likely than women whose partnership was a marriage to cite not sharing housework (adjusted odds ratio 0.54). Whatever housework signals about a relationship ending, it signals it more inside marriage.

#18

What this licenses you to do

The question people arrive with is usually practical: does how we split the housework decide whether we last? The research supports a narrower and more useful set of answers.

What the evidence will carry:

  • Nobody has shown that splitting housework evenly causes divorce. The study behind that claim is a Norwegian government-commissioned report whose own authors wrote that a direct causal role seems unlikely, whose absolute difference is 3.1 per cent against 4.6 per cent over four years, and in which housework division accounted for roughly one per cent of the variation in who separated.
  • Nobody has shown that splitting it evenly protects you either. That was the hypothesis the Norwegian chapter set out to test, and it was the hypothesis that failed. This is the one place where the folklore and the research agree on the shape of the answer, if not the tone.
  • The strongest and most replicated finding is not about division at all. It is that feeling your contribution is unseen tracks dissatisfaction and instability. In Sweden, her saying she does most predicted dissolution while his saying she does most predicted the opposite, and the single worst pattern was his reporting that they share while she reported that she does more. In the US, perceived unfairness predicted lower happiness for both and higher odds of divorce for wives.
  • Which task may matter more than how many hours. Dishwashing is the task whose division is most strongly tied to relationship quality in the best task-level analysis available, and the daily, visible, unglamorous tasks are the ones that carry the signal.
  • The invisible part is real and badly measured. Anticipating, planning and remembering are work; they are routinely invisible to both partners; and "we split it 50/50" can describe three different households.
  • The country you live in changes the answer, which is the clearest sign that this is about what an arrangement means to the people in it rather than about the labour itself.

What it does not license:

It does not license reading your own arrangement as a prognosis. The effects here are small, selection-driven and context-dependent, and the number of couples who separated in the famous Norwegian cells is in the tens. It does not license the inversion either — the claim that a man should do less around the house to protect his marriage has no causal support anywhere in this literature, and the one direct test of the mechanism behind it, whether the egalitarian partner gets left, found no tendency in either direction.

What it does suggest is that the question worth asking is not how much but whether we are describing the same house. A disagreement about hours is arithmetic. A disagreement about whether the work is being noticed is the thing the Swedish data actually picked up, and it is the version of this argument that tends to be about something else — which is also what the research on recurring arguments and on the couples who last keeps finding.

If the conversation itself is the hard part, that is a conflict-pattern problem rather than a chores problem, and the conflict style test scores the pattern the pair-level literature actually measures. The relationship satisfaction test scores the dimensions used in the research instead of handing out a verdict, and the attachment style test covers what you bring to reading any of it. All three score in your browser and send nothing anywhere. There is more of this under Mindset, including what the research says about partners who will not go to counselling when one person wants the arrangement discussed and the other does not.

And a last word about the two sentences this page started with. Both of them are quotations from documents that say something more careful. One is a 228-page report in Norwegian that has sat in an open repository since 2012. The other is a journal article whose own tables show that the arrangement does not predict dissolution once both partners agree on what it is. Fourteen years of coverage ran on neither.

Pulling it together

The useful finding in this literature was never a divorce rate. It is that the division of housework is strongly related to how people feel about the division of housework, weakly related to how they feel about the relationship, and barely related at all to whether the relationship ends — and that what does track instability is a mismatch between two people's accounts of the same house.

That is why the Swedish result is the one to keep. Her saying she does most predicted dissolution; his saying she does most predicted the opposite; both agreeing she does most predicted neither. The arrangement was not doing the work. The discrepancy was.

The Norwegian result survives as something smaller and still worth knowing: equal sharing does not appear to protect against separation, which is what the researchers set out to test and did not find. Everything built on top of that — that egalitarian men get left, that clear roles save marriages, that sharing chores is a divorce risk — is an inference from a 1.5-percentage-point difference, in a model that explained about one per cent of the outcome, with its causal interpretation disclaimed in the report and in the only interview the author gave about it.

If there is a practical reading, it is unglamorous. The daily, visible, unrewarding tasks carry more signal than the episodic ones. The planning and remembering are work, and they are easy for both partners to miss. And the question that distinguishes the couples in trouble from the couples who are simply unequal is not how the hours divide but whether both people would describe the week the same way.

Frequently asked questions

Do couples who share housework equally really divorce more?

In one Norwegian dataset, slightly. The source is NOVA Rapport 8/12, Likestilling hjemme, a 228-page report edited by Thomas Hansen and Britt Slagsvold and published in 2012, using the LOGG survey of about 15,000 Norwegians aged 18 to 84 interviewed in 2007-2008 and linked to population registers. Over the following four years, 4.6 per cent of couples who split housework evenly had separated, against 3.1 per cent of couples where the woman did most. That 1.5-percentage-point difference is the "about 50 per cent higher" of the headlines. The report also gives a second figure almost nobody quotes: 78 per cent higher for couples where the man did most, a group of 313 people in which roughly seventeen separations occurred. The division of housework accounted for about one per cent of the variation in who separated, and the report's authors wrote that it "seems unlikely that the division of housework can be a direct cause of later divorce."

Did the researchers say sharing housework causes divorce?

No, and they said so in both the report and the press. Chapter 9 of the report states that a direct causal role seems unlikely and frames its finding as a negative result about a specific hypothesis: equal sharing does not appear to reduce the chance of divorce, "as has sometimes been assumed." In the AFP interview that became the only English-language source, The Local recorded that "researchers found no, or very little, cause-and-effect" and that the authors read the correlation as a marker of modern attitudes. Thomas Hansen's own explanation was selection: couples who share housework in Norway tend to hold less traditional views of marriage, and the women in them tend to have more education and better-paid jobs, so leaving is more feasible. The report also tested the folklore version directly and found no tendency for who initiated the breakup to vary with how the couple had divided housework beforehand.

What did the Swedish study actually find?

Something more specific than its title suggests. Ruppanner, Brandén and Turunen, in Sociology 52(1) in 2018, used the 2009 Young Adult Panel Study — the first wave to survey both partners — linked to Swedish registers through 2014. Of 1,057 couples, the dissolution analysis runs on 803, of which 12.6 per cent separated. The key results are about whose report you use. When the woman said she did most of the housework, the union was 2.16 times as likely to dissolve. When the man said his partner did most, it was 0.51 times as likely — roughly half. When both partners agreed the wife did most, satisfaction was lower for both but there was no effect on break-up plans or on dissolution. The pattern that carried the most risk was what the authors call discrediting: he reports that they share, she reports that she does more. So the finding is about mismatched accounts rather than about the division itself.

Is it true that 25% of divorces are caused by arguments over chores?

There is no traceable source for it. The claim is usually attributed to a Harvard Business School study of 3,000 couples, and we could find no named survey, dataset or HBS publication reporting it; the versions in circulation do not even agree on whether the 3,000 were divorced or still together, or on whether housework was the first or third reason. The Harvard research the figure is attached to is about something else: Ashley Whillans, Jessica Pow and Joe Gladstone's "Buying (quality) time predicts relationship satisfaction," published in the Journal of Personality and Social Psychology in 2025, which concerns outsourcing chores and relationship satisfaction. A real figure for the same question exists and is smaller: in Britain's Natsal-3 survey, among 706 men and 1,254 women who had been through the breakdown of a live-in partnership in the previous five years, 6.8 per cent of men and 14.4 per cent of women cited not sharing enough housework, well behind growing apart, arguments and unfaithfulness.

Does unequal housework predict divorce in the United States?

The American evidence points the opposite way from the Norwegian headline, and it is about perceived fairness rather than hours. Frisco and Williams, in the Journal of Family Issues in 2003, used a nationally representative sample of dual-earner marriages and found perceived inequity in housework associated with lower marital happiness for both husbands and wives, and with higher odds of divorce for wives only; marital happiness did not appear to mediate that link. Bellani, Esping-Andersen and Pessin, in the Journal of Social and Personal Relationships in 2018, analysed the Panel Study of Income Dynamics alongside German panel data from 1986 to 2009 and found deviation from an equitable division destabilising in the US — especially when the wife under-benefited — while in West Germany "the degree of inequity has no effect at all" and the male-breadwinner arrangement remained the most stable. Lynn Prince Cooke's 2006 paper in the American Journal of Sociology reached the same cross-national conclusion a decade earlier.

Why do the Norwegian and Swedish studies disagree?

Largely because they did not measure the same variable. The Norwegian report built a four-category index from the average of six questions about specific tasks, deliberately excluding maintenance and small repairs. The Swedish paper used a single question — "How do you divide housework?" — with three options. One is closer to a description of behaviour, the other closer to a measure of perception, which is fitting given that the Swedish paper's real finding is about whose perception you use. There is also the country difference, which is documented rather than speculative: the association between a couple's division of labour and their stability runs in opposite directions in West Germany and the United States, in two separate studies using different data. A division of housework does not carry a fixed meaning, so it does not carry a fixed effect.

Which chores matter most for a relationship?

The best task-level analysis available points at dishwashing. Carlson, Miller and Sassler, in Socius in 2018, compared the 1992-1994 wave of the National Survey of Families and Households with the 2006 Marital and Relationship Survey and found that of all the tasks examined, the division of dishwashing was the most consequential for relationship quality, especially for women. They also found that recent cohorts of middle- to low-income couples share all routine tasks more than earlier ones, except shopping, with the largest changes in dishwashing and laundry. The mechanism the literature points to is that routine housework is daily, unavoidable, time-bound, visible and unrewarded, which is not true of episodic, traditionally male-coded tasks — and is why the Norwegian index left maintenance and repairs out of its housework measure in the first place.

Does the mental load count as housework in these studies?

Mostly not, and that is a real limitation of all of them. Allison Daminger, in the American Sociological Review in 2019, interviewed 70 people from 35 couples and argued that household management is a distinct kind of work — cognitive labour — made up of anticipating needs, identifying options, making decisions and monitoring progress. It is taxing and often invisible to both the person doing it and their partner, which makes it a recurring source of conflict. Her findings are more precise than the popular version: women did more cognitive labour overall and specifically more of the anticipating and monitoring, but participation in decision-making, the component most tied to power, was roughly equal. That is an interview study of 35 couples rather than a national sample, so it is best cited as an argument for a distinction rather than an estimate of a quantity. The practical consequence is that "we share equally" can describe three quite different households, which may be why the Swedish paper found opposite results depending on which partner was answering.

Are most people unhappy about an unequal split?

No, and the Norwegian report is the best source on this because it reports the figures by arrangement. Satisfaction with the division of housework is around 95 per cent among both men and women when it is split evenly or when the partner does more. The least satisfied group is the 15 per cent of women who say they do clearly most — and even among them, 60 per cent describe themselves as satisfied with the division. Satisfaction with the relationship barely moves at all: 76 per cent among women who do clearly most, against 83 to 89 per cent in other arrangements. In the same analysis, the division of housework accounted for 14 per cent of the variation in satisfaction with the division and 2 per cent of the variation in satisfaction with the relationship. Two cautions go with those numbers: "satisfied" is partly a measurement of what people expect rather than of fairness, and a 60 per cent satisfaction rate still leaves a large minority for whom the arrangement is a live problem.

Where can I read the original studies?

The Norwegian report is free. NOVA Rapport 8/12, Likestilling hjemme, is catalogued in Norway's national research repository with a permanent handle, and the PDF of all 228 pages is available through the Internet Archive because the original university address no longer resolves; both links are in the sources below. It is in Norwegian. The Swedish paper is paywalled: Semantic Scholar records its open-access status as closed and OpenAlex finds no open repository copy, so a library is the route. Of the surrounding literature, the openly readable pieces are the most useful anyway — Bellani, Esping-Andersen and Pessin's German-American comparison is free in PubMed Central, Carlson, Miller and Sassler's task-level paper is open access in Socius, and Gravningen and colleagues' Natsal-3 analysis of why relationships ended is free in PLOS ONE.

Sources

What this article draws on. For the broader bibliography see our sources page.

Keep reading