attraction.lab

In-Laws and Divorce: Half of the Famous 20% Is in the Paper, Half Was Never There

A husband close to her parents: 20% lower divorce odds, published. A wife close to his: that figure is in no paper, and the real result is about race.

Research-backed writing on attraction, dating and relationships. Drafted by AI against a named bibliography, published only when every claim links to its source.
32 min read
Editorial illustration for: In-Laws and Divorce: Half of the Famous 20% Is in the Paper, Half Was Never There

Before the article. If what you are dealing with is not a difficult mother-in-law but a partner or a relative who decides who you may see, checks where you have been, uses money or immigration status as leverage, or makes you afraid of what happens when you say no, that is control rather than family friction, and nothing on this page applies to it. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day; Find A Helpline, in the sources at the bottom, lists free and confidential services by country.

With that said. Two figures have been circulating since December 2012. A husband close to his wife's parents, 20% lower risk of divorce. A wife close to her husband's parents, 20% higher. They appear on law firm blogs, in holiday columns and in wedding-advice pages, always together, almost never with a link to the study.

The study is real, it is good, and reading it changes the figures in four separate ways. The husbands' 20% is in the paper, word for word — and it is per point on a four-point scale, not a comparison between close and not close. The wives' 20% is in no paper: in the model that produced it the odds ratio runs the other way. The result where wives' closeness goes with more divorce exists, but it is an interaction with race, it applies to the White wives in a sample that was 53% Black by design, and no coverage mentions it. And “after 26 years” is sixteen years of marriage, with the last interviews conducted in 2002.

Underneath that there is a quieter literature that nobody quotes, including a follow-up by the same team which found that neither spouse's own account of their in-law relationships predicted divorce — only whether the two of them agreed.

Sixteen sections: where the numbers came from, what the study was, the one-question measure, the half that holds, the half that does not, the race interaction, the date error, what an odds ratio is, the explanation that is labelled as speculation, what the study looked for and did not find, the follow-up, the 72% that is thirty-six couples, the better-designed study that came first, the newest analysis, what has never been counted, and what survives. Our editorial policy explains why every empirical claim here carries a linked source.

#1

The two numbers, and the page they came from

Search for in-laws and divorce and you will meet the same pair of figures, usually in the same sentence. When a husband is close to his wife's parents, the couple's risk of divorce drops by 20%. When a wife is close to her husband's parents, it rises by 20%. Neat, symmetrical, and endlessly quotable.

The earliest version we can find ran on CNN on 13 December 2012, in a holiday column by the sexuality counsellor Ian Kerner. It describes a study by Terri Orbuch following 373 couples who were newlyweds in 1986, and then says: “After 26 years, Orbuch found that when a man reported having a close relationship with his wife's parents, the couple's risk of divorce decreased by 20%. Yet women who said they had a close relationship with their husbands' parents saw their risk of divorce rise by 20%.”

Note the tense in the same paragraph: the study “will be published in a future issue of the journal Family Relations.” The numbers went out four months before the paper did. Women's Health ran its version off the same pre-publication material, with the finding converted into a comparison between groups: “Marriages in which the wife reported having a close relationship with her in-laws had a 20 percent higher risk of divorce than couples where the wife didn't report a close relationship.” The Boston Globe repeated the pair in November 2013. By 2026 the figures live mostly on divorce law firm blogs, where they are reproduced without a link to anything.

The paper exists. It was published in Family Relations in April 2013, by Terri Orbuch, José Bauermeister, Edna Brown and Brandyn-Dior McKinley, and an open copy sits in the University of Michigan's repository. This page is what happens when you read it. One of the two numbers is in there, stated in those words. The other is not in there at all, and the result it was made from is about something the coverage never mentions.

It is the same failure mode as the statistics that circulate about modern dating: a figure survives for a decade because the people repeating it are copying each other rather than the source. The 11%-a-year figure examined in the best age to get married has survived the same way, and hides the same kind of racial interaction. The pattern reaches its limit in the claim that money arguments are the top predictor of divorce, where what circulates is not a mis-cited number but a spoken superlative that no published table supports.

#2

What the study actually was

The Early Years of Marriage Project is a genuinely unusual dataset, and most of what makes it unusual is missing from the coverage.

In April to June 1986, the researchers took couples who applied for marriage licences in Wayne County, Michigan — Detroit and its surrounds. Couples were eligible if they were the same race, if it was a first marriage for both, and if the wife was under 35. All eligible Black American couples were approached and a random sample of the eligible White American couples. Both partners had to agree; 65% of the Black American couples and 66% of the White American couples did. That produced 373 couples: 199 Black American and 174 White American.

The oversampling was deliberate and it is the design decision that matters most for what follows. The project was built to compare Black and White marriages with enough statistical power to do it properly, which is why 53% of the sample is Black. It is not a cross-section of American marriages and was never meant to be.

Interviews were face to face, in people's homes, with a race-matched interviewer, in the 1st (1986), 3rd (1988), 7th (1993) and 16th (2002) years of marriage. Tracking in years 14 and 16 established who had divorced and when. By Year 16 the researchers had marital status for all but four of the 373 couples, and 172 couples (46.1%) had separated or divorced — 63 of the 174 White couples (36.2%) and 109 of the 199 Black couples (54.8%).

So the study is: couples who married in one Michigan county in 1986, followed for sixteen years, with the last data collected in 2002. Every result below is a result about those people. That is not a complaint — sixteen-year panels with both spouses interviewed are rare and expensive, and this one has produced good work for three decades. It is a boundary, and it is the sort of boundary that goes missing when a finding is turned into advice, in the same way the samples quietly shrink behind the claim that living together before marriage predicts divorce.

#3

The whole measure is one question with four answers

Before any number means anything, you need to know what was measured, and here it is smaller than anyone would guess.

Closeness to in-laws was a single item. Respondents were asked to say how close they felt to their spouse's family, on a scale from 1 (not at all close) to 4 (very close). There was a companion item — how close do you think your spouse feels to your family — asked the same way. Conflict about in-laws was a yes or no question: have you and your spouse ever had any tension or differences about your in-laws. That is the entire instrument.

Mean closeness to in-laws in Year 1 was 3.14 for wives (3.23 White, 3.06 Black) and 3.29 for husbands (3.32 White, 3.20 Black), on that four-point scale. In other words almost everybody said 3 or 4. The variation the analysis is working with is mostly the difference between close and very close.

This changes how every percentage below should be read. The models estimate what happens to the odds of divorce per one point on that scale — moving from a 3 to a 4, say. They do not compare people who are close to their in-laws against people who are not, because hardly anyone in this sample said they were not. Every popular version of the finding makes exactly that substitution. Women's Health's phrasing — wives with a close relationship against “couples where the wife didn't report a close relationship” — describes a comparison the study did not run.

The paper is candid about the limitation. In its own words, the use of single-item measures for closeness to and conflict surrounding both sets of families was not ideal, and the authors say they hope to establish better multi-item indicators in future. We take the same view about instruments elsewhere: the relationship satisfaction test on this site says on its own page what a short scale can and cannot tell you, because a number that comes from four tick-boxes is still a number that came from four tick-boxes.

#4

The husband's 20% is real, and it is in the paper

Start with the half that holds up, because it does hold up.

The analysis is a hierarchical Cox regression — a model for time-to-event data that uses both whether a couple divorced and when. Step one puts in the demographics. Step two adds the family-ties measures. Running it on the husbands' reports, with education, income and their interactions with race already in the model, the result is: a husband's perception of closeness to his in-laws in the first year of marriage significantly predicted the odds of divorce, OR = .79, p < .05.

The paper then states the finding in the sentence that went around the world: “The odds of divorce over time decreased by 20% the closer a husband felt to his in-laws early in the marriage.”

That is the only time the string 20% appears anywhere in the article. It is attached to husbands, and it is per point on the four-point scale.

Two honest qualifications. First, it was the only significant finding among the husbands' family variables — the paper says so directly: no other association between the husbands' family closeness or conflict measures and the odds of divorce reached significance. Second, p < .05 with an odds ratio of .79 in a sample of 373 couples is a real but modest result, the kind that needs replication rather than a headline.

What it does not need is embellishment, and embellishment is what it got. The Boston Globe's version — men close to their in-laws “were 20 percent less likely to get divorced than men who weren't” — turns a per-point slope into a between-group difference. The slope is the finding. The groups are an invention of the retelling.

If you want the general shape of what does predict whether couples last, measured across much bigger samples, the research on couples who stay together is the place to start, and in-laws are not high on that list.

#5

The wife's 20% is not in the paper

Now the other half, which is where the story is.

Run the same model on the wives' reports and the paper reports this: a wife's perception of closeness to her in-laws early in marriage was significantly associated with the odds of divorce over time, OR = 0.82, p < .05.

Read the direction. An odds ratio of 0.82 is below one. In that model, a wife feeling closer to her in-laws went with lower odds of divorce — the same direction as the husbands' result, not the opposite of it. The symmetrical story that every page tells, where the two sexes point in opposite directions by the same 20 points, is not what that model produced.

And the number itself never appears. There is no “20% higher” in the paper. There is no percentage at all attached to wives. The one percentage the authors chose to spell out is the husbands' one, and somebody — a press release, a phone call, a reporter looking for balance — mirrored it.

This matters beyond pedantry, because a mirrored figure is the most persuasive kind of wrong. Two numbers that match look like a pattern. A reader who sees “minus 20 for him, plus 20 for her” assumes someone measured both and found a clean reversal. Nobody did. The same mechanism keeps the five love languages in circulation: a tidy structure travels further than the messy evidence underneath it, and the tidiness is doing the persuading.

There is a result in the paper in which wives' closeness goes with more divorce rather than less. It is in the next model, it is considerably larger than 20%, and it applies to some of the wives in the study and not others.

#6

The finding the coverage removed: it is about race, not about gender

Model 3 adds an interaction between a wife's closeness to her in-laws and race, and that interaction is significant: OR = 0.64, p < .05.

What the authors say it means, in their words: White American couples were at greater odds of divorce the closer wives felt to their in-laws early in the marriage, and this effect was opposite for Black American couples, with the odds of divorce decreasing the closer wives felt to their in-laws.

The discussion puts it as plainly as a journal article ever puts anything. Early family ties were detrimental for marital stability only among White American wives who had close emotional ties to their in-laws. Close emotional ties with in-laws as reported by White American husbands, and by Black American husbands and wives, increased marital stability.

So the real finding is not that wives should keep their in-laws at arm's length. It is that the association ran one way for the White wives in this sample and the other way for the Black wives — and the Black couples were 53% of the study, because the researchers deliberately oversampled them in order to be able to see exactly this kind of difference.

That is the result the coverage deleted. CNN, the Boston Globe and Women's Health all present it as a fact about women. None of the three mentions race at all. The single most-reported finding of a project designed around a race comparison was published with the race comparison removed, and the version without it has outlived the version with it by thirteen years.

It is worth being clear about what the paper does not show. It does not show that this difference is about culture, or about mothers-in-law, or about anything in particular; it shows an interaction term in one sample of Detroit-area couples married in 1986. The authors' own explanation for it is a guess, and the next section is about how that guess became the headline. A finding splitting by group where the popular version reports a single number is a pattern we have run into before, most sharply in whether attachment styles can change and in the survey behind the sleep divorce, where a tick-list of six options became one statistic about couples.

#7

“After 26 years” is sixteen years, and it ends in 2002

A smaller error, but it tells you how the figure travelled.

CNN says “after 26 years.” The Boston Globe says Orbuch has been interviewing the couples “for the past 26 years.” Both are describing the age of the project at the time of writing — 1986 to 2012 — rather than the window the analysis covers.

The analysis covers sixteen years, Year 1 to Year 16 of marriage. The interviews are dated in the paper: 1986, 1988, 1993 and 2002. The divorce outcome is divorce by Year 16. Nothing in the result reaches past 2002.

This is not a trivia point. “26 years” implies a finding that has been watched into the couples' late forties and fifties. It has not. The in-law measure was taken in the first year of marriage, when the average husband was 26 and the average wife 24, and the outcome was counted sixteen years later. It is a statement about how a newlywed's answer to one question relates to whether the marriage survived its first sixteen years.

It is also, inevitably, a statement about a specific cohort. These people married in 1986, before mobile phones, before anyone's parents could text them daily, and in a county with its own history. Any result about how a young couple relates to their parents has a date stamped on it, and the stamp says 1986.

Seventeen years between the last data point and a sentence of advice is a long time for a number to be left unchecked. We have found the same gap behind the rules that circulate with digits in them — the 3-6-9 rule has no origin study at all — and behind figures that look precise because they are specific, which is the trap in what age gaps actually predict.

#8

Odds, risk, hazard: the words are not interchangeable

Everything in this paper is expressed as an odds ratio from a survival model. Everything in the coverage is expressed as “risk.” Those are different objects, and the slippage inflates the finding in the reader's head.

An odds ratio of .79 on a one-point move of a four-point scale is not “your chance of divorce falls by a fifth.” It is a multiplier on the odds, estimated across the whole sample, holding education, income, race and their interactions constant. The couples in this study divorced at 46.1% over sixteen years. A reader who takes “20% lower risk” literally will imagine something like 46% dropping to 26%. That is not what a 0.79 odds ratio per scale point does, and the paper never claims it.

There is a second slippage, more important. These are associations. Nobody assigned anyone to like their in-laws. Everything that might produce both a warm answer in Year 1 and a surviving marriage in Year 16 — how the marriage was already going, how the families behaved, how much money there was, whether a person tends to answer questions warmly — is sitting underneath the estimate. The paper controls for several of those and cannot control for the ones nobody measured.

And a third, which applies to the whole genre: a group average says nothing about a couple. A hazard estimated over 373 marriages does not tell any particular person what their Sunday lunch means.

If what you actually want to know is whether your relationship is in good shape, the answer is not in a number about in-laws at all. The markers that have observational work behind them are about what the two of you do with each other, and the strongest predictors in the biggest analyses are things like commitment and perceived responsiveness — not the family tree.

Further illustration for: In-Laws and Divorce: Half of the Famous 20% Is in the Paper, Half Was Never There
#9

The explanation everyone quotes is labelled as speculation in the paper

The reason always given for the wives' result is the boundary story. Wives who get close to their in-laws find it hard to set limits; closeness invites interference; advice starts to read as criticism. It is in the CNN column, it is in the Boston Globe, it is in Women's Health, and it is quoted from the lead author in all three.

It is also, in the paper itself, explicitly a hypothesis. The authors write that they argue that for White wives, interconnected ties to in-laws early in marriage may interfere with and prevent the formation of a strong bond between spouses, and they trace the idea to Salvador Minuchin's family systems theory and its claim that newlyweds need clear emotional boundaries with their families of origin. On the husbands' side they write that they speculate that close ties connect a husband to his wife's family and so strengthen the bond to his wife.

We argue and we speculate are how a careful author marks the part that is not a finding. The study measured one four-point item and a divorce date. It did not measure boundaries, interference, meddling, or whether anyone took advice as criticism. No variable in it could have.

What happened on the way to publication is that the hedge fell off and extra material was added. The press quotes attribute to the research claims it contains no measure of — that women are constantly analysing their relationships, that they take what in-laws say personally, that men identify as providers first. The CNN column goes further still, adding a scenario in which a wife befriends her husband's parents to recruit them to her side of an argument. None of that is in the paper. It is commentary presented in the voice of a study.

We apply the same rule in the other direction: where a mechanism is a story rather than a measurement, it gets said. The clearest case on this site is the psychology of jealousy, where the popular explanations are confident and the measurements are thin, and the honest page is the one that reports the gap.

#10

What the same study looked for and did not find

A result is easier to judge once you know what else was in the model and came back empty. In this case, quite a lot.

Conflict about in-laws did not predict divorce. The study asked both spouses whether they had ever had tension or differences about the in-laws, and the answers varied in interesting ways — 45% of White husbands against 28% of Black husbands reported in-law tension. Neither that variable nor conflict about one's own family showed a significant association with the odds of divorce. The thing everybody actually complains about was measured and did nothing.

Closeness to one's own family did not predict divorce either, in any of the models.

What did predict divorce, strongly and consistently, were the demographics. Black couples divorced at higher rates in these models — for husbands' models the hazard was 1.88 to 2.12, for wives' 2.20 to 3.11 — and that association is the one the paper spends its discussion trying to contextualise, pointing at income, structural stressors and the sampling design rather than treating it as a fact about marriages. Education was protective on both sides: in the wives' baseline model, the likelihood of divorce decreased by 28% with every additional year of education, which is a larger and better-replicated effect than anything on this page about in-laws, and nobody writes holiday columns about it.

The authors also list what they could not do. They did not examine whether family ties changed over time, only how they stood in Year 1. They did not run dyad-level analyses — husbands and wives were modelled separately. And they flag the single-item measures as not ideal.

That last limitation is the one their own team went back and fixed, and the paper they produced says something different from the one everybody quotes. It is also, as far as we can find, cited by nobody outside academia. The pattern where the correction is quieter than the error is familiar from what the research says about repairing a fight, where the memorable claim and the replicated claim are not the same claim.

#11

The follow-up nobody quotes: agreeing matters more than being close

In 2021, Katherine Fiori, Amy Rauer, Kira Birditt, Edna Brown and Terri Orbuch published “You Aren't as Close to My Family as You Think” in Research in Human Development, using the same couples — the 355 of the 373 for whom divorce information was available. The question they asked was not how close anyone was. It was whether the two spouses agreed.

Four Year 1 comparisons: whose family the couple spends more time with, whose family they would call on for help, how close the wife is to his family, how close the husband is to hers. Discordance is simply the two answers not matching.

The first finding is that couples disagree about this constantly. Only 58% agreed about which family they spent more time with. Only 42% agreed about which family they would call on for advice or help. Agreement was better on closeness — 86% on the wife's closeness to his family, 84% on the husband's closeness to hers — but just 22% of couples agreed on all four items, and 11% disagreed on three or four of them.

The second finding is the one that reframes the 2013 paper. Neither spouse's own reports of their in-law ties predicted divorce in these models — not time spent, not whom they call on, not closeness. What predicted divorce was discordance about the husband's closeness to the wife's family: HR = 0.54, p = .009, controlling for both spouses' actual reported levels of closeness. Couples who agreed were significantly less likely to divorce over sixteen years than couples who disagreed. Total discordance across all four items pointed the same way without quite reaching significance (HR = 1.20, p = .059).

The authors' summary is worth quoting because it is the opposite of the advice built on their earlier work: whether or not the husband is close to the wife's family seems less important than whether or not the spouses agree that he is or is not.

That is a result about shared reality rather than about mothers-in-law, and it sits with everything else we know about couples noticing the same relationship — which is why the questions that reveal real compatibility are mostly questions that surface a disagreement early rather than questions that score anybody.

#12

The 72% that is thirty-six couples

The 2021 paper contains one striking number, and because it is striking it deserves the same treatment this page has given the others.

The discordance effect was driven almost entirely by one cell. Among couples where the wife thought the husband was not close to her family while he thought he was, 26 of 36 couples had divorced by Year 16 — a 72% divorce rate, against 46% for the sample as a whole. The mirror-image cell, where she thought he was close and he thought he was not, divorced at 48%: indistinguishable from the base rate.

So the signal is specific and asymmetric. It is not disagreement in general. It is the husband believing he has a relationship with her family that she does not think he has.

And it is thirty-six couples. Twenty-six divorces out of thirty-six is a wide confidence interval wearing a sharp number, and the authors present it as an observation about which cell drives the effect rather than as a standalone finding. Treat it accordingly: interesting, consistent with the model, and not a result to build a rule on. The same arithmetic problem — a memorable percentage resting on a handful of couples — is what we found behind the claim that friends-first relationships last longer.

If there is a practical reading, it is modest and it is about noticing rather than fixing. A husband who is sure he gets on with her parents, in a marriage where she does not see it that way, is a couple with two different pictures of the same Sunday. The useful move is finding out that the pictures differ, which costs one conversation. The 2021 authors say as much: even small, incremental changes in shared perceptions — helping couples see eye to eye on a single issue related to in-laws — may be beneficial.

#13

The better-designed study that came first, and measured something else

Before any of this there was a paper that got closer to the thing people mean when they say in-laws are a problem, and it measured discord rather than closeness.

Chalandra Bryant, Rand Conger and Jennifer Meehan published “The Influence of In-Laws on Change in Marital Success” in the Journal of Marriage and Family in 2001, on couples who had been married, on average, for almost two decades. Its design advantage is that it ran the arrow both ways: it tested whether discord with in-laws predicted later marital success and whether marital success predicted later discord with in-laws, which is the obvious alternative explanation and is almost never tested.

What the abstract reports: for wives, discord with mothers- and fathers-in-law predicted their own later perceptions of marital success. Results differed somewhat for husbands' discord with fathers-in-law. And the reverse path — marital success predicting less discord with in-laws — held only for husbands. The authors describe these as among the first prospective longitudinal findings showing that conflicts in extended family relations erode marital stability, satisfaction and commitment over time, even in long marriages.

Two things to hold. The full text is behind a paywall and we have not read it; what is above comes from the published abstract and from the description given by Fiori and colleagues, who characterise the sample as a longitudinal study of rural White families and report that the wife's discord with her in-laws was the stronger predictor of both spouses' perceptions of marital success. Second, the asymmetry in the reverse path is the interesting part and the part most easily over-read: in these data, husbands whose marriages were going well later reported less friction with in-laws, while for wives the friction looked more like a cause than a consequence.

Notice what has happened across the three studies. The one that measured conflict found conflict mattered. The one that measured closeness found closeness mattered a little, in opposite directions by race. The one that measured agreement found that neither spouse's own report mattered at all. Three measures, three answers — which is a good reason to distrust any single sentence beginning “research shows that in-laws…”, and the same reason the honest version of do opposites attract is a paragraph about how similarity was measured.

#14

The newest word from the same couples

The Early Years of Marriage data are still producing papers, and the most recent one puts in-laws in proportion.

In 2026, Rosie Shrout, MiKaila Leonard, Rasheedah Adisa, Terri Orbuch, TeKisha Rice Wallace, Melissa Franks and Arthur Aron published an analysis in the Journal of Social and Personal Relationships of how spousal, friend and family relationships relate to marital instability across the same sixteen years. In the author's own plain-language summary for the Society for Personality and Social Psychology, the findings run like this.

Across sixteen years, people's relationships with their spouses and families — but not with their friends — were linked to marital stability. Marriages were most stable in the years when people reported lower marital distress and greater spousal responsiveness: less irritation and resentment, more sense of being understood and cared for. And the key qualifier: even when relationships with their families were unsupportive or tense, marriages were stable as long as spouses felt less distressed and perceived more responsiveness.

The reverse does not hold. When spouses felt more distressed and less responsive than usual, a supportive family helped — but family support did not make up for low responsiveness or high marital distress. Marriages were least stable in the years when both the marriage and the family relationships were strained.

The author's one-line version: a good marriage can protect you from an unsupportive family, but a supportive family cannot fully protect a struggling marriage.

The paper also reports that over the sixteen years, felt support from spouses, friends and family all declined while marital distress and instability rose; that wives generally viewed both their marital and family relationships more negatively than husbands; and that Black wives reported the lowest levels of feeling supported by their spouses — which the author reads in terms of unequal demands rather than as a statement about marriages.

If you want the thing this points at, it is responsiveness: the sense that a partner understands you, values what you said and cares. That construct has carried more weight than almost anything else in relationship research, and it is the subject of what builds attraction over time and of how people actually open up.

#15

Nobody has counted how many marriages end because of in-laws

Several pages on the first screen of search results promise a figure for how many marriages end because of in-laws. As far as we can find, no such figure exists in the research literature, and the nearest thing to it is instructive.

Shelby Scott, Galena Rhoades, Scott Stanley, Elizabeth Allen and Howard Markman interviewed 52 divorced individuals who had received premarital education through the PREP programme, and asked which of a list of reasons were major contributors to their divorce. The list has eleven items. In-laws are not one of them. The nearest item is “lack of support from family,” endorsed by 17.3% of individuals, and it describes an absence of help rather than interference. The items people did endorse were lack of commitment (75.0%), infidelity (59.6%), too much conflict and arguing (57.7%), marrying too young (45.1%), financial problems (36.7%), substance abuse (34.6%) and domestic violence (23.5%).

Fifty-two people is a small, particular sample, and a list cannot capture a reason it does not offer. But the shape of the answer is consistent with the rest of this page: when researchers ask people why their marriages ended, what comes back is about the marriage.

There is also no experiment, and there cannot be one. Nobody has randomly assigned couples to live near their in-laws, to see them more often, or to get on with them. Every number on this page comes from watching what people did and recording what happened afterwards.

And the longitudinal work is thinner than it looks. A 2023 conference abstract from Raquael Joiner, Jacqueline Perez, Thomas Bradbury and Benjamin Karney — working on 375 couples over the first nine years of marriage — opens by noting that empirical studies to date are typically limited to only a couple of measurement occasions, so the stability and impact of changes in these relationships are unclear. Their own result, in a conference abstract rather than a full paper, is that spouses whose marriages dissolved were more likely to fall into the indifferent, ambivalent and difficult profiles of in-law sentiment than into the helpful one.

That is the entire evidence base. It is why the pages promising percentages have none, and why the ones on the first page of results are divorce lawyers: the question gets asked by people who are already a long way down a different road, which is the road covered in thinking about breaking up and in why people stay in unhappy relationships.

#16

What the evidence supports, and where it stops being about in-laws

Collecting what survives all of the above, there are three statements worth keeping and one line that matters more than any of them.

One: agreement beats closeness. The best-designed analysis of this dataset found that neither spouse's own account of their in-law relationships predicted divorce, and that disagreement about one of them did. That points at a conversation, not at a seating plan. Whether he is close to her parents is largely outside anybody's control — distance, work, whether the parents are easy — and the paper says so. Whether the two of you describe the situation the same way is not.

Two: the marriage is the variable. In the most recent analysis of the same couples, a marriage with low distress and high responsiveness stayed stable through tense family relationships, and family support did not rescue a distressed marriage. If both are strained at once, that is the worst position in the data. In practical terms: the in-law problem that is actually a marriage problem is the common case, and it is the one with evidence behind it. Where that is where you are, whether couples therapy works covers what the trials show, and what to do when a partner will not go covers the other half.

Three: conflict with in-laws is a real thing that predicts real outcomes, on the evidence of Bryant and colleagues, and it is not the same variable as closeness. Most advice conflates them. A pattern where one of you keeps being asked to choose sides is a conflict pattern inside the couple, and the conflict style test scores pressing and withdrawing separately because plenty of people do both in the same argument.

And the line. Everything above is about ordinary friction between adults who are stuck with each other. It stops applying the moment someone's behaviour is about control. Being told who you may see, having your movements checked, having money or immigration status used as leverage, being made afraid of what happens when you say no — whether it comes from a partner or from a relative with the partner's cooperation — is not an in-law difficulty and no amount of agreeing with your spouse about it will fix it. That pattern has its own literature, some of which is covered in what trauma bonding does and does not mean, and the services listed at the top of this page and in the sources exist for exactly that. Every empirical claim here carries a linked source, which is the promise made in our editorial policy. More of this under Mindset.

Pulling it together

The short version is that one of the two famous numbers is a real published result about husbands, stated per point on a four-point scale, and the other one does not exist.

The longer version is more useful. The study everybody cites was designed to compare Black and White marriages, found that its most quotable result ran in opposite directions for the two groups, and was reported everywhere with that part cut out. Its authors' explanation — the boundary story about meddling in-laws — is marked in the paper as something they argue rather than something they measured, and the press versions repeat it as a finding while adding claims about men and women that no variable in the study could support.

The same team went back with a better question. Asked whether spouses agree about their in-law relationships rather than how close anyone feels, neither partner's own report predicted divorce at all; disagreement about one of them did. And the most recent analysis of the same sixteen years puts the whole subject in its place: a marriage with low distress and a partner who seems responsive stayed stable through tense family relationships, while family support did not rescue a marriage that was already struggling.

So if there is anything to take from a decade of this figure being quoted, it is not a rule about how close to stand to your in-laws. It is that the couple is the unit that matters, that the cheapest thing available is finding out whether the two of you are describing the same family, and that a number repeated for thirteen years without a link is usually repeated because nobody checked.

Frequently asked questions

Is it true that a wife being close to her in-laws raises the risk of divorce by 20%?

No. That figure appears in no version of the study. The paper — Orbuch, Bauermeister, Brown and McKinley, Family Relations, 2013 — uses the string “20%” exactly once, and it is attached to husbands: the odds of divorce decreased by 20% the closer a husband felt to his in-laws early in marriage, OR = .79, p < .05. For wives, the model reported in the same paper gives OR = 0.82, which runs in the same protective direction rather than the opposite one. There is a result in which wives' closeness goes with higher odds of divorce, but it comes from a model with a race interaction (OR = 0.64, p < .05), and the authors state that it applied to White American couples while the effect ran the opposite way for Black American couples. The “+20%” looks like a mirror of the husbands' figure invented somewhere between the study and the press.

So do good in-law relationships actually protect a marriage?

On the husbands' side of this one study, modestly and in one sample. A husband's self-reported closeness to his wife's family in the first year of marriage predicted lower odds of divorce over sixteen years, per point on a four-point scale, in 373 couples who married in Wayne County, Michigan in 1986. That is a single significant association in a single panel, and it was the only one among all of the husbands' family variables. The same team's 2021 follow-up, using the same couples with a better question, found that neither husbands' nor wives' own reports of connections with in-laws — time spent, whom they call on, closeness — predicted divorce at all. Treat the protective effect as suggestive rather than established.

What did the study actually measure?

Less than the coverage implies. Closeness to in-laws was one question answered on a scale from 1 (not at all close) to 4 (very close): how close do you feel to your spouse's family. Conflict about in-laws was a yes-or-no question about whether the couple had ever had tension or differences about them. Mean closeness was 3.14 for wives and 3.29 for husbands, so nearly everyone answered 3 or 4 and the models are estimating what happens across one point of that narrow range. The paper itself says the use of single-item measures was not ideal and that the authors hope to establish better multi-item indicators. Every popular retelling converts this per-point slope into a comparison between people who are close to their in-laws and people who are not — a comparison the study did not run.

Why does nobody mention race when they quote this study?

We do not know, and it is the most consequential omission in the whole story. The Early Years of Marriage Project deliberately approached all eligible Black American couples and a random sample of eligible White American couples precisely so that Black and White marriages could be compared with adequate power; 199 of the 373 couples were Black. The significant result for wives is an interaction with race, and the paper's discussion says that close in-law ties were detrimental only among White American wives, while for White American husbands and for Black American husbands and wives they went with greater marital stability. CNN in 2012, the Boston Globe in 2013 and Women's Health all report the finding as a fact about women, with no mention of race in any of them.

Was the study really 26 years long?

No. The project began in 1986 and the coverage was written in 2012, which is where 26 comes from. The analysis covers the first sixteen years of marriage: face-to-face interviews in 1986, 1988, 1993 and 2002, with tracking in years 14 and 16 to establish who had divorced and when. Nothing in the result extends past 2002. The in-law question was asked in the first year of marriage, when husbands averaged 26 years old and wives 24, and the outcome was whether the couple had divorced by year sixteen, which 46.1% of them had.

What is the strongest finding anyone has about in-laws and divorce?

Probably the one about agreement rather than closeness. Fiori, Rauer, Birditt, Brown and Orbuch, using 355 of the same couples, compared what each spouse said about four things in year one: whose family they spend more time with, whose family they would call on for help, and how close each spouse is to the other's family. Neither partner's own reports predicted divorce. Discordance about the husband's closeness to the wife's family did — HR = 0.54, p = .009 — controlling for the reported levels of closeness themselves. Couples disagreed far more than you would expect: only 42% agreed about which family they would call on for advice, and just 22% agreed on all four items.

Where does the idea that in-laws interfere with the couple's bond come from?

From family systems theory, and in this paper it is labelled as an argument rather than a finding. The authors write that they argue that for White wives, interconnected ties to in-laws early in marriage may interfere with and prevent the formation of a strong bond between spouses, citing Minuchin's position that newlyweds need clear emotional boundaries with their families of origin. On the husbands' side they write that they speculate about why closeness helps. Nothing in the study measured boundaries, interference, meddling, or whether anyone experienced advice as criticism — the instrument was a four-point closeness item and a divorce date. The explanation may well be right; it is not what was tested.

How many marriages end because of in-laws?

There is no published figure we can find, and the nearest data suggest the question is poorly formed. When Scott, Rhoades, Stanley, Allen and Markman asked 52 divorced individuals which of eleven reasons were major contributors to their divorce, in-laws were not on the list at all; the closest item, “lack of support from family,” was endorsed by 17.3%, and it describes an absence of help rather than interference. What people did endorse was lack of commitment (75.0%), infidelity (59.6%) and too much conflict and arguing (57.7%). It is a small sample and a list cannot capture a reason it does not offer, but every page promising a percentage for in-laws is promising something nobody has measured.

My partner and I fight about their parents constantly. What does the evidence actually suggest?

Three things, none of them a technique. First, that the fight is usually better understood as a disagreement between the two of you than as a problem with the parents: the strongest result in this literature is about spouses not describing the same family the same way, and finding out that your pictures differ costs one conversation. Second, that the marriage is the variable that carried the weight in the most recent analysis of these couples — marriages with low distress and a partner who felt responsive stayed stable through tense family relationships, while family support did not compensate for a distressed marriage. Third, that conflict with in-laws is a different measure from closeness to them, and the study that measured conflict found it did predict later marital success, mainly through the wife's reports. If the pattern is one of you repeatedly being asked to choose sides, that is a conflict dynamic inside the couple. If it involves being told who you may see or being afraid to say no, it is not a conflict dynamic at all, and the helplines at the top of this page are the right place to start.

Sources

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

Keep reading