When One Partner Won't Go to Couples Therapy: Who Refuses, and What Working on It Alone Does
In a clinical sample of 913 couples, 40% arrived with mismatched agendas. The one trial of doing the work alone moved the person, not the relationship.
One thing before the article, because it decides whether any of it applies. If the refusal you are dealing with sits alongside fear, threats, having your phone or your movements or your money controlled, being cut off from the people you know, or any physical harm, then the research below is not about your situation: every figure on this page was measured in couples from whom that had been screened out. The section near the end says why, and Find A Helpline, in the sources, lists free and confidential services by country.
With that said. Two things are true about this situation at the same time, and almost no page answering this question holds both.
The first is that you are in the ordinary case. In a clinical dataset of 913 couples arriving for therapy, published in October 2026, 40.3% fell into a mixed-agenda class — out of step on commitment, on being a team and, above all, on hope. In a population sample of 1,487 couples it was about 10%. And in that clinical data, a mismatch in feeling pressured to attend was the indicator that separated the groups least: 53.7% against 46.0%. One of you having to be talked into the appointment is close to the background condition of couple therapy rather than a verdict on the relationship.
The second is that the thing you most want — a way to get them to come — is the thing that has been measured least. What has been measured is what felt pressure costs once they are in the room, what people do instead of going, where a partner's refusal actually ranks among the reasons people do not go, and what happens when one person does the work alone. That last one has a randomised trial, built for exactly this situation, and its result is specific: it moved the person and was not shown to move the relationship.
This page sets all of that out with its sample sizes, including the places where the honest answer is that nobody has looked. For whether the thing being refused works once both people are in the room, what the trials on couples therapy actually report is the companion piece.
Before the rest of this page: the version of this that is not about therapy
Most people typing this are in an ordinary and painful situation. One of them thinks the relationship needs outside help, the other does not want to go, and the question is what to do with that. Nearly everything below is about that situation, and the honest answer has numbers in it.
One thing has to go first, though, because it arrives under the same words and it is not the same problem. If the refusal sits alongside being told who you may see or message, having your phone or your movements or your money controlled, having to account for ordinary hours, anger or punishment or silence arriving as a sanction when you say no, being separated from friends and family a little at a time, or any fear of your partner or any physical harm, then this is not a help-seeking problem and no part of the research below applies to it. Couple therapy trials screen those situations out and measure what is left, which is covered in the section near the end on when a refusal is the signal rather than the obstacle. The useful next step is a person rather than an article. Find A Helpline, in the sources at the bottom, lists free and confidential services by country. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day.
With that said, the question itself contains two questions, and they have different evidence behind them.
The first is how to get someone to come. Very little has been measured about that directly, and what has been measured points somewhere uncomfortable: the one study that looked at felt pressure found it predicts a worse start for both people, not just for the one who was pushed.
The second is what one person can do on their own. That has been tested, once, properly, in a randomised trial built for exactly this situation. It found something specific and worth knowing, and it is not what the first page of search results says.
The short answer, with the numbers that are actually in the papers
Five sentences, each of which the rest of this page unpacks and sources.
You are in the normal case, and there is now a number for it. In a clinical dataset of 913 couples arriving for therapy, a latent class analysis published in October 2026 put 40.3% of them in a mixed-agenda class — partners out of step on commitment, on being a team, and above all on hope. In a population-based sample of 1,487 couples, about 10% looked that way.
The thing you think defines the problem is the weakest signal in that data. Discrepancy in feeling pressured to attend therapy barely separated the two groups: 53.7% in the mixed-agenda class against 46.0% in the aligned one. What separated them was hope, at 83.0% against 0.6%.
Refusal is the third-ranked reason people who considered couple therapy did not go, and it is strongly gendered. Among 231 newlywed couples in high-poverty neighbourhoods of Harris County, Texas, 36% of the women who had considered therapy and not gone named their partner not wanting it, against 14% of the men. Above it for both sexes were cost and not knowing where to go.
Pressure has a measured cost. In 351 couples starting treatment at three training clinics, when one partner reported feeling pressured to attend, the other partner's alliance with the therapist was lower at the fourth session. Alliance is the variable that predicts both dropout and outcome in this literature.
And doing the work alone has been tested once. Ninety distressed individuals — four in five of them women, many of them there because their partner had refused — were randomised to a web programme or a one-month waitlist. Quality of life moved (d = 0.69). Work functioning and perceived health moved. Relationship satisfaction did not (d = 0.23, p = 0.107). The programme helped the person and was not shown to help the relationship.
None of that is a reason to give up on asking. It is a reason to stop treating this as a persuasion problem with a technique behind it, because the measurements do not support one. For what the treatment itself does once both people are in the room, with its recovery and deterioration rates, what the outcome research reports is the companion page; the relationship satisfaction test is the one that reports satisfaction and commitment as two scores rather than one.
This is the ordinary case, and as of October 2026 it has a prevalence
Clinicians have had a name for this for a decade. William Doherty and Steven Harris called them mixed-agenda couples: one partner leaning towards ending the relationship, the other wanting to save it and work on it. The same asymmetry turns up in the literature as polarised couples, as commitment uncertainty, and as leaning-in and leaning-out partners. What it did not have, until this year, was a number from a real dataset.
Yike Tang, Shayne Anderson, Lee Johnson, Steven Harris and Anita Kateregga published that number in the October 2026 issue of the Journal of Marital and Family Therapy, online since August, and the paper is open access. They went at it from two directions.
The population side. Using the National Couples' Health and Time Study, a survey recruited by population-based methods, their analytic sample was 1,487 couples — 2,974 individuals. Classifying by a single item on perceived likelihood of separation, they found:
- 147 couples (9.89%) mixed-agenda: one partner saying separation was unlikely, the other saying it was likely.
- 1,285 couples (86.42%) aligned on staying.
- 55 couples (3.70%) aligned on leaving.
Relationship satisfaction differed across all three groups, F(2, 1468) = 412.30, p < 0.001, and sat exactly where you would expect: highest in the aligned-stay group, lowest in the aligned-leave group, and the mixed-agenda couples in between.
The clinic side. Using the Marriage and Family Therapy Practice Research Network — couples presenting for therapy at university training clinics — 913 couples, 1,826 individuals. A latent class analysis on four indicators returned two classes: an aligned class of 545 couples (59.7%) and a mixed-agenda class of 368 couples (40.3%). Mean satisfaction in the mixed-agenda class was 43.44 against 61.37 in the aligned one, Welch's t(1573.7) = 24.47, p < 0.001.
Two things about the 40%. It is higher than the field's own estimate: Doherty and Harris had put mixed-agenda couples at 25 to 30% of those seeking couple therapy, and the measurement came out above that. And it means that if you are the one who wants to go and your partner is not sure the relationship has a future, you are not an edge case a therapist has never seen. You are something like two in five of the caseload.
The authors are careful about what their numbers are, and so is this page. Both samples are secondary data. The population figure rests on one item about separation likelihood, so it is not the construct as clinicians define it. The clinic indicators were also single items, without published reliability or validity evidence for those specific operationalisations, and the authors say the prevalence should be read as suggestive rather than definitive. They also could not control for the different levels of distress in the two samples, which is the obvious reason a clinical sample would carry more of these couples than the general population. What distress does to a relationship when nobody intervenes is a separate and better-measured question, and why people stay in unhappy relationships sets out the model the literature uses for it.
What separates those couples is not who got dragged in. It is hope
The most useful thing in that 2026 paper is not the headline prevalence. It is the shape of the difference, which runs against the way this question is usually asked.
A latent class analysis reports, for each indicator, the probability that a couple in that class shows mismatch or ambivalence on it. Here are the four indicators, for both classes:
- Hope about the relationship: 83.0% in the mixed-agenda class, 0.6% in the aligned class.
- We-ness, the sense of being a team: 57.0% against 18.5%.
- Commitment to the relationship: 38.1% against 8.4%.
- Feeling pressured to attend therapy: 53.7% against 46.0%.
Read that last line twice. Nearly half the couples in the aligned class — the ones who turned up broadly on the same page about their relationship — had a mismatch in how pressured they felt about being there. The authors say it plainly: pressure to attend differentiated the classes less strongly than any of the other indicators.
That is the single most clarifying finding on this page, and it reframes the question. One of you having to be talked into the appointment is not, by itself, diagnostic of anything. It is close to being the background condition of couple therapy. What marks out the couples whose agendas genuinely differ is whether the two of them still think this can work.
There is a practical version of that, and it is a question rather than a tactic. If your partner does not want to go, the thing worth knowing is which of these it is: I do not think therapy is for us, which is a disagreement about a service, or I do not think we are going to make it, which is a disagreement about the relationship. They look identical from the outside and they are not the same conversation. The first is about cost, stigma, scheduling and what a session actually involves — all of which have measured answers, set out two sections from now. The second is the thing the 2026 data says actually sorts these couples, and the named clinical response to it is not couple therapy at all, which is covered further down.
One honest limit, stated by the authors and worth repeating: this study did not look at dropout or at outcomes. It did not measure whether mixed-agenda couples do worse in therapy. The authors connect their classes to earlier findings on alliance and early discontinuance and then call that interpretation speculative, in their own word. Nobody has yet published what happens to the 40%. What does have longitudinal data behind it is the opposite question — which couples hold up over years without anybody intervening at all — and what 137 studies found about couples who stay together is where that sits.
Who tends to be the one not going, and what predicts it
The best-known study of this is more than twenty years old, has a title that says what it found, and almost nothing on the first page of search results mentions it.
Brian Doss, David Atkins and Andrew Christensen published Who's dragging their feet? Husbands and wives seeking marital therapy in the Journal of Marital and Family Therapy in 2003. They surveyed 147 married couples — 294 individuals — who were in the process of seeking marital therapy, which is the hard population to catch: not people remembering it years later, but people mid-decision.
Using multilevel confirmatory factor analysis, they found that getting to therapy is not one step but three relatively independent ones:
- Problem recognition — agreeing with yourself that something is wrong.
- Treatment consideration — thinking of therapy as the thing to do about it.
- Treatment seeking — actually booking it.
Independent is the word to hold on to. Someone can recognise the problem perfectly well and never consider therapy as the answer; someone else can be in favour of therapy in the abstract and never make the call. The refusal you are looking at is at one of these three points, not at all of them, and which one it is changes what is actually being refused.
On average, wives were rated as completing all three steps before their husbands. That is the finding people have in mind when they treat this as a male-reluctance story, and it is real. The predictors were gender-role orientation, demographics, relationship satisfaction, and specific problems — with husbands' dissatisfaction with sex singled out in the paper's own abstract as a notably predictive one.
The same research group went back to the same 147 couples for a second question — not when they came, but why — and published it in Professional Psychology: Research and Practice in 2004. The most common reasons given were problematic communication and lack of emotional affection. The part that matters here is the next sentence of that abstract: within individual couples, spouses showed little agreement on their reasons for being there. Not disagreement about whether to go. Disagreement about what the problem was, among couples who had already agreed to go.
So the asymmetry is not only about willingness, and it does not resolve the moment someone says yes. Two people can walk into the same room with different accounts of why they are in it, which is the ordinary case rather than the troubling one, and is part of why the page on whether couples therapy works is so insistent that the first sessions are mostly spent naming the pattern rather than fixing it.
Both of these papers are paywalled at the publisher, and the figures above are the ones their published abstracts state. No number here is taken from anyone's summary of them.
The men's side of it, measured rather than assumed
There is a small, specific literature on why men do not go, and it is better than the folklore in one particular way: it separates not wanting to from not thinking it would help from not thinking there is a problem.
Douglas Spiker, Joseph Hammer and Kenneth Parnell published the study worth having in the Journal of Social and Personal Relationships in 2018. The sample is the right one: 206 men who reported being unhappy with their romantic relationship. Not men in general, and not men in therapy — men in the situation where the question arises. Four findings from it:
- Conformity to masculine norms accounted for unique variance in help-seeking perceptions beyond neuroticism. That matters because the obvious objection to this whole line of work is that it is personality by another name. Here it was not: masculine norms added something after neuroticism was accounted for.
- Self-reliance and restrictive emotionality behaved differently from each other, with different direct and indirect effects. Valuing handling things yourself and finding it hard to talk about feelings are not interchangeable, and a refusal driven by one is not the same as a refusal driven by the other.
- Perceived need and self-stigma accounted for the most variance in the model. Whether he thinks there is a problem to treat, and what he thinks going would say about him, were the heavy variables — above the masculinity measures.
- And the finding that is genuinely counterintuitive: relationship satisfaction was not associated with attitudes towards couple therapy, but lower satisfaction was associated with feeling more social pressure to go. Being unhappier did not make men think better of therapy. It made them feel more pushed towards it.
That last result deserves to sit next to the pressure finding in the alliance section below, because the two of them together describe a loop that is easy to fall into. The unhappier things get, the more pressure he reports feeling; feeling pressured predicts a worse working relationship with the therapist if he does come; and none of it shifts his view of whether therapy is a reasonable thing to do.
The paper this one builds on, Parnell and Hammer's 2018 study in Psychology of Men and Masculinity, is the origin of the theory-of-planned-behaviour approach to men's relationship help-seeking. It is cited here and no figure is quoted from it: it is closed access and, checked on 9 November 2026, has no publicly readable abstract in Crossref, OpenAlex, Semantic Scholar or Europe PMC. What can be said about it is what Spiker's abstract says about it, which is that it established the framework. That is the whole of it, and this site would rather name a source and say it could not be read than paraphrase a paper from its title.
A note on register, because this is a section about men and this is a site that gets this wrong if it is not careful. None of the above says that reluctance is a character flaw, and the research does not treat it as one. Self-reliance is not stupidity. The measured finding is narrower and more useful: the specific thing being avoided differs between people, and the reasons are enumerable rather than mysterious. Where the pattern is a partner who pulls back from emotional conversations in general rather than from therapy specifically, what the research says about supporting an avoidant partner covers what has and has not been measured about that.
And the women's side, which is not the mirror image
The same lab asked the parallel question about women, and the answer came back more tangled — which is the more interesting result.
Spiker, with Elyssa Berney, Hammer and Katherine Jensen, published it in The Counseling Psychologist in 2020: 302 heterosexual women who were unhappy in their romantic relationships, modelled the same way. Four relational schemas went into it:
- Self-silence — suppressing your own needs.
- Relational-interdependent self-construal — defining yourself through your important relationships.
- Self-reliance — valuing not asking for help.
- Emotional control — preferring not to disclose what you feel.
The authors' own summary of what those four did is the sentence to quote, because paraphrasing it would tidy it up dishonestly: women with those schemas were both less and more likely to intend to seek couple therapy. Direct and indirect paths ran in opposite directions. Their conclusion is that these schemas create conflicting messages for women about seeking couple therapy.
That is worth more than a cleaner finding would be. The cultural story is that women are the ones who want to go and men are the ones who do not, and the help-seeking data does lean that way — wives completed the three steps first in 2003, and 47% of wives against 32% of husbands had considered therapy in the Texas sample below. But the mechanism is not that women simply find it easier. The same orientation towards the relationship that makes someone the one who notices the problem and raises it can also be the thing that keeps her managing it privately instead of taking it to a stranger.
If you recognise yourself in that — the person who has read everything, raised it carefully several times, and is now the one being told they are making too much of it — two other pages here cover the adjacent ground with its own evidence: what has actually been measured about opening up sets out what makes disclosure go well, and what the research says about being too accommodating is about the cost of absorbing a problem on your own.
Both papers in this section and the last are paywalled. The abstracts of the 2018 and 2020 studies are publicly readable and are what the figures above come from; the 2018 abstract was read through OpenAlex, which carries it where the publisher's page does not.
What people actually do instead, and it is probably what you are doing now
There is a population-level answer to this question, and it is the least flattering fact on the page for anybody who writes about relationships.
Tea Trillingsgaard, Trine Sørensen and Hanne Fentz surveyed a random population sample of 1,371 individuals living with a partner in Denmark, and published it in the Journal of Marital and Family Therapy in 2019. Denmark is a high-income, high-divorce country where empirically supported couple interventions have been imported, so the question was what their actual reach is. The answer:
- 3% had ever sought couple education with their partner.
- 7% had ever sought counselling or therapy with their partner.
Then they looked specifically at respondents who had been through a severe relationship crisis, which is the group this page is about. Among them, the most prevalent kinds of help-seeking were individual rather than joint. The top one, for both sexes:
- 38.9% of women and 27.8% of men sought information online.
- 19.4% of women sought individual therapy.
So the modal response to a serious relationship crisis, in a country with good access to services, is to read something on the internet by yourself. That is what you are doing at this moment, and it is the normal behaviour rather than a failure of nerve. It also sets the bar for a page like this one: if looking it up alone is what most people in a crisis actually do, then the pages they land on had better not be wrong, which is the whole argument of our editorial policy.
A longitudinal American figure sits usefully next to it. Doss, Galena Rhoades, Scott Stanley and Howard Markman followed 213 couples across the first five years of marriage and found help-seeking of some kind was fairly common: 36% of couples sought some form of outside help in that window. The kinds that moved with actual difficulty were relationship books and marital therapy — individual and relationship problems predicted increased use of both in the following year. Attending marriage retreats or workshops, by contrast, was related only to demographic variables: it tracked who you are, not how things are going.
Which is to say the books-and-internet route is not the consolation prize people treat it as. It is the main route, it responds to real distress, and it is almost entirely unstudied as an intervention. It is also the reason the figures that circulate in this genre deserve checking one at a time, which is the exercise in the real numbers on modern dating.
Where a partner's refusal actually ranks among the reasons people do not go
Somebody has measured the reasons, in order, among people who considered couple therapy and did not go. It is the most directly relevant study to this keyword in the literature, and not one page on the first screen of results mentions it.
Hannah Williamson, Benjamin Karney and Thomas Bradbury published it in the Journal of Family Psychology in 2019, and it is free to read in PubMed Central. The sample is 231 newlywed couples living in high-poverty neighbourhoods of Harris County, Texas, identified through marriage licence applications from 2014 and 2015 and interviewed in their homes at baseline, nine months and eighteen months. Median household income was $32,500; about half of each partner group was Hispanic and about a third Black.
First, the base rates:
- 32% of husbands (n = 73) and 47% of wives (n = 109) considered seeking therapy at some point.
- 15% of couples (n = 35) actually received it.
- Of the husbands who considered it, 33% followed through and 67% did not. Of the wives who considered it, 29% followed through and 71% did not.
Then the barriers, among those who considered therapy and did not get it. The top two were the same for both sexes and neither is about the partner:
- Not knowing where to get help or who to see.
- Not being able to afford it.
Third place is where this page lives, and it splits by sex:
- For women, the third barrier was "your spouse or partner did not want to get therapy" — endorsed by 36% of them.
- For men, the third barrier was the belief that individual counselling would be more helpful than couples therapy.
- The sex difference on the partner barrier was significant: 36% of women against 14% of men, two-sided p = 0.008.
Two readings follow, and both are useful. If you are a woman whose partner will not go, you are in a group that is about one in three of the women who got this far and stopped — common enough that it is a named structural barrier in the literature, not a private misfortune. And if you are the one being asked to go, the measured male objection is not usually I refuse to deal with this. It is I think the wrong person is being sent to therapy, which is a claim about format, and the section below on what individual therapy does and does not do is the direct answer to it.
The same paper found two things that moved the needle, and they are oddly concrete. Couples who had received premarital education were more likely to later receive therapy — 25% against 11% — though in a joint model it was the wives' religious attendance that carried the effect, and 87% of the premarital education and 80% of the therapy in this sample happened in religious settings. And knowing someone who had been through couple therapy mattered, but only when both partners did: couples where both knew someone were more likely to go, while couples where only one did were not significantly more likely.
That last result is the closest thing in the literature to a mechanism for changing a partner's mind, and it is not a conversation. It is knowing somebody who went.
The sample is specific — recently married, low-income, Texan, all different-sex couples because of when the licences were issued — and the barrier rankings may not transfer to a couple fifteen years in with a mortgage. The cost and access barriers almost certainly rank differently further up the income distribution.
Does pushing work? The one thing that has been measured says it costs you both
This is the section people come for, and the honest version is narrower than the advice in circulation. Nobody has run a trial of ways to get a reluctant partner into couple therapy. What exists is a study of what felt pressure does once they are in the room, and the result is not comfortable.
Shayne Anderson, Alyssa Banford Witting, Rachel Tambling, Scott Ketring and Lee Johnson published it in the Journal of Marital and Family Therapy in 2020. 351 couples receiving treatment as usual at three family therapy training clinics rated the predictors at intake and the therapeutic alliance at the fourth session. A path analysis gave four results:
- Each partner's dyadic adjustment was directly associated with the quality of their own alliance.
- For male partners, reporting more adverse childhood experiences and more pressure to attend treatment went with lower alliance scores of their own.
- And the crossover effect: when one partner reported feeling pressure to attend therapy, the other partner's alliance decreased.
- For men, there was an indirect path from dyadic adjustment to alliance through pressure to attend.
The third of those is the one to sit with. Pressuring someone into the room does not only cost the pressured person's engagement. It is associated with a worse working relationship with the therapist for the person who did the pressuring — which is to say, for you.
Why alliance rather than some outcome? Because alliance is the variable that has been tied to both of the things anybody cares about here.
It predicts outcome. Myrna Friedlander, Valentín Escudero, Marianne Welmers-van de Poll and Laurie Heatherington meta-analysed the alliance-outcome relation in couple and family therapy in Psychotherapy in 2018: 48 studies, 40 independent samples, a three-level random effects model across 2,568 families, 1,545 couples and 491 effect sizes, giving r = 0.297. They also tested split alliances — where family members' alliances with the therapist are unbalanced — across seven independent samples and 31 effect sizes, and found r = 0.316: the more split the alliance, the poorer the outcome. And a moderator result that belongs in this exact discussion: the alliance-outcome correlations were significantly stronger when families had sought help voluntarily than when they were involuntary or mandated.
It predicts dropout. Anderson, Tambling, Jeremy Yorgason and Erin Rackham followed 994 cases through the first three sessions of individual, couple and family, or high-conflict coparenting therapy. 205 of them — 20.6% — discontinued before the fourth session. Distress and the high-conflict coparenting format were associated with earlier termination, and the effect of both on dropout was mediated by the therapeutic alliance.
Put the three together and the practical conclusion is not never ask. Asking is how anybody ever ends up in a therapist's office, and the alternative to asking is nothing. The conclusion is that an invitation and an ultimatum are not the same instrument, and the difference shows up in a measurable variable at session four, for both of you. If the only way your partner arrives is under duress, the thing you will have bought is the version of therapy with the weaker alliance, the higher dropout risk and the worse expected outcome — and the split-alliance finding says that the imbalance itself is part of the problem, not just the reluctance.
It is worth being clear about what this evidence is. It is correlational, from training clinics, with alliance self-reported at one time point, and it cannot tell you what would have happened to those couples if nobody had pushed. What it does tell you is that felt pressure is not a free move, which is more than the advice pages on this query can say in either direction. Where the sticking point is one specific unresolved rupture rather than a general unwillingness, what the research says about repair after a fight covers the one model that has been tested on exactly that.
What happens if only one of you does the work: the one randomised trial
This is the question underneath the keyword, and there is exactly one trial designed to answer it. It was built for people in precisely this situation, and its result is clean enough to state in one sentence: the programme moved the person and was not shown to move the relationship.
Kathryn Nowlan, McKenzie Roddy and Brian Doss published it in Couple and Family Psychology: Research and Practice in 2017, and it is free to read in PubMed Central. OurRelationship is a web translation of integrative behavioural couple therapy that normally requires both partners; they built a one-person version, OR-I, and tested it.
The sample is the point. Ninety individuals in heterosexual relationships, from 35 states. Some enrolled in the individual programme directly; others had applied to the couple version and been offered the individual one because their partners refused to take part. Of those who chose it directly, some said they did not want to involve their partner, in a few cases because they were weighing whether to end the relationship. 80% were women. Mean age 36.8, 74% married, mean relationship length 10.55 years, and 98% scored in the distressed range on the Couples Satisfaction Index. Everyone was told, by email, that they would probably do better in the couple version; this is the group that still wanted the individual one.
They were randomised to the programme (n = 46) or to a one-month waitlist (n = 44), with no significant pre-treatment differences. 80% completed, in a mean of 25.5 days, with three scripted coach calls totalling about 22 minutes. 97% said they were mostly or very satisfied with what they got.
What changed, against the waitlist:
- Quality of life: d = 0.69, p = 0.002.
- Perceived health: d = 0.49, p = 0.030.
- Work functioning: d = 0.44, p = 0.022.
What did not change:
- Relationship satisfaction: d = 0.23, p = 0.107.
- Negative relationship quality: d = −0.26, p = 0.095.
- Relationship confidence: d = 0.12, p = 0.432.
- Positive relationship quality: d = 0.09, p = 0.523.
- Depressive symptoms: d = −0.05, and anxiety: d = −0.23, neither significant.
At the three-month follow-up, quality of life and perceived health held. Work functioning significantly deteriorated (d = −0.48, p = 0.040). Relationship measures and mood did not improve over follow-up either, and post-hoc tests found no significant gains from pre-treatment to three months on those. By three months, 10.9% of the couples in the intervention arm had divorced.
The authors' own reading, in their own words, is that the results "provide further evidence that a partner's involvement is needed for more long-lasting change, especially when there are high levels of relationship distress." Their explanation for the pattern is the interesting part: quality of life, work and health are domains that are not specific to the relationship, and they think those improved through coping, resolve, self-efficacy and clarity — including, for some participants, clarity about leaving. Actively working on the problem helped the worker, whether or not it helped the pair.
Now the honesty that the result requires, in both directions. The trial was powered to detect an effect of d = 0.66 for relationship outcomes with 80% power. The relationship satisfaction effect it observed was d = 0.23. A real effect of that size would have been missed by this design, and the authors say so: d = 0.23 is about the same size as the statistically significant effects reported by larger trials of other individual-oriented programmes — PREP Inside and Out at d = 0.25 for men and 0.31 for women, Power of Two at g = 0.24. So "no significant improvement in relationship functioning" is not the same claim as "no improvement", and anyone who tells you this trial proved working alone is useless is overreading it.
Two further caveats. The comparison against a demographically matched subsample of 210 people who did the couple version found no significant differences either, which means this study could not actually show that the couple version beats the individual one — it was not big enough for that question. And the trial comes from the group that developed the programme, which is normal in intervention research and still worth knowing. The couple version of the same programme, with its 300-couple trial and the satisfaction effect the individual version did not get, is covered in the page on what couples therapy does.
The claim on the first page of results that this contradicts
Several pages ranking for this question tell you that going alone works on the relationship. The strongest version of it, on a therapy practice's site, says that research shows couples therapy outcomes improve significantly when even one partner begins, and attributes it to an institute's research without naming a study, a sample or a figure. The same page tells you that starting now, even alone, can prevent further damage.
Set that next to the trial above. The one randomised test of a structured programme delivered to one distressed partner, in the exact population being addressed — 90 people, four in five of them there because their partner would not come — produced no significant change in relationship satisfaction, relationship confidence, or positive or negative relationship quality, and the authors concluded that a partner's involvement is needed. That is not a refutation of the claim, because the trial was underpowered for a small effect. It is a demonstration that nobody has the evidence to make the claim in the confident form it is being made in.
A page from the other direction has the mirror-image problem. A lifestyle site says that you can go to therapy and read all the self-help books, but unless you tackle your problems as a pair, "research shows that it probably won't fix your problems" — again with no study attached. That one is closer to where the evidence actually sits, and it is still an assertion rather than a citation, and it skips the thing the trial did find: measurable gains in the person's own quality of life, health and work functioning. Telling someone whose partner will not come that working on it alone is pointless is both unkind and, on the only data there is, wrong about what the work does.
And one page of the twelve quotes a success figure for a named therapy model — 90% of couples improving through emotionally focused therapy — with nothing attached to it. Where that family of numbers comes from, which meta-analysis it is from, how many couples are under it, and why the biggest figures come from the smallest evidence bases, is set out in where that family of success figures comes from. It is not a number about reluctant partners at all.
Going to individual therapy is not the same as working on the relationship alone
The most common piece of advice on this query is: go on your own, then. Eight of the twelve readable pages raise individual therapy or counselling. It is reasonable advice and it is worth separating into two different things, because they have different evidence.
Going to individual therapy for your own distress is well supported and is not the subject of this page. Going to individual therapy expecting the relationship distress to lift as a result is a specific empirical claim, and it has been tested more directly than you would think.
David Atkins, Sona Dimidjian, Jamie Bedics and Andrew Christensen published the relevant study in the Journal of Consulting and Clinical Psychology in 2009. They looked at two treatment-seeking samples at once: 120 individuals seeking treatment for depression and 134 couples seeking marital therapy. Three findings:
- There was a baseline association between depression and relationship distress in both samples, as expected.
- The asymmetry was striking: relationship distress was far more prevalent among the people being treated for depression than depression was among the couples seeking marital therapy. A lot of people in individual treatment are in distressed relationships.
- Changes in the primary problem were reliably associated with changes in the secondary one — depression improving as relationship distress improved in the couple sample, and the reverse in the individual sample. But the overall effects on secondary problems were quite small.
That is the shape of it. Treating the one thing does move the other thing, in both directions, and it moves it a little. Which is enough to dispose of the two strong positions: individual therapy is not a substitute for couple therapy, and it is not useless for the relationship either.
There is one more measured fact worth having, from the research note that traced where the famous six-year waiting figure came from. Of its 371 participants, 101 had gone to individual therapy specifically for relationship problems — not for depression, for the relationship — and they had waited essentially as long before going as the 270 who went as a couple. The authors had expected the couples to take longer, on the reasoning that two people are harder to get through a door than one. They did not. That whole study, including what it says about how long people actually wait and why the six-year claim does not survive contact with its source, is in the research note that traced the six-year waiting figure.
What has not been measured is the thing people most want to know: whether one partner going to individual therapy makes the other partner more likely to agree to couple therapy later. There is no study of that. Williamson's finding that it helps when both partners know somebody who has been through couple therapy is the nearest adjacent result, and it is about social networks, not about your own course of treatment. What individual work has been shown to move is the individual: whether attachment styles can change is the clearest case of that, and the answer there is a coefficient rather than a yes.
The one place where “they won't go” has randomised evidence, and it is not this
There is a corner of clinical research built entirely around the question how do you get a person who does not want treatment into treatment. It is not in the couples literature. It is in the substance use literature, and because it is the only place the question has been randomised, it is worth knowing — with its boundaries stated clearly.
The approach is called community reinforcement and family training, CRAFT, and the person who turns up for it is the one who is worried, not the one who is unwilling. William Miller, Robert Meyers and J. Scott Tonigan published the trial in the Journal of Consulting and Clinical Psychology in 1999. 130 concerned significant others were randomised to one of three manual-guided approaches, each with 12 hours of contact:
- Al-Anon facilitation, encouraging involvement in the twelve-step programme.
- The Johnson Institute intervention, preparing for a confrontational family meeting.
- CRAFT, teaching behaviour-change skills to use at home.
Follow-up ran for 12 months, with 94% completed. The engagement of initially unmotivated problem drinkers in treatment:
- CRAFT: 64%.
- Johnson Institute: 30%.
- Al-Anon: 13%.
Two further results from the same trial matter more here than the headline. The confrontation arm replicated a known pattern: 70% of the concerned relatives decided not to go through with the family meeting at all, and among those who did, 75% got the drinker into treatment. A high success rate on a thing most people will not do. And: all three approaches were associated with similar improvement in the concerned person's own functioning and relationship quality — regardless of whether the other person ever went. A companion study of CRAFT with drug users, 62 concerned relatives, found 74% engagement over six months and significant reductions in the relatives' own depression, anxiety, anger and physical symptoms, with average scores dropping into the normal range, engaged or not.
Now the boundaries, because this is exactly the kind of finding that gets misapplied. CRAFT is for substance use, where there is a specific treatment the unwilling person is being engaged into. Nobody has tested it, or anything like it, for getting a partner into couple therapy, and the outcome it measures is the other person entering their own treatment, not the pair entering a joint one. The trial also found that engagement rates were higher for parents than for spouses — the role you are in is the harder one. And CRAFT is itself a course of treatment with a manual and twelve hours of contact, not a conversational technique: reading about it is not doing it.
So what is it doing on this page? Two things. It establishes that the question is answerable in principle, which the couples literature has simply never got around to. And the result that transfers best is the least tactical one: in the only randomised comparison that exists, confrontation was the approach most people abandoned, the skills-at-home approach engaged twice as many, and the concerned person's own wellbeing improved under all three. That last clause is the same finding as the OR-I trial reached from the other direction, and two independent literatures agreeing on it is as close to a dependable conclusion as this subject offers: doing something yourself reliably helps you, and whether it moves the other person is a separate question with a worse answer.
Discernment counselling: the named option for exactly this, and the size of its evidence
If the refusal is really about hope rather than about therapy — the distinction the 2026 prevalence data says actually sorts these couples — then there is a clinical approach designed for it, and almost nobody on the first page of results mentions it. One of the twelve pages does.
William Doherty, Steven Harris and Jason Wilde described discernment counselling in the Journal of Marital and Family Therapy in 2016. It is built for the mixed-agenda couple: one partner leaning towards divorce, the other wanting to preserve the relationship and work on it. Three features distinguish it from couple therapy:
- The goal is not improvement. It is clarity and confidence about the next step, whichever step that is, based on a better understanding of what each partner has contributed.
- The sessions are mostly individual. Each partner is seen separately within the session, which is the structural answer to one person being in the room reluctantly.
- The outcomes it is designed to reach include divorce. Making no change, divorcing, and attempting reconciliation through couple therapy are all legitimate endings. Where one partner is reluctant because of an affair, the thing most worth knowing before reading survival percentages is how few couples the best of them were measured on, which is the subject of whether couples stay together after infidelity.
That design is the reason to mention it here. The standard couple therapy format presumes two people with shared goals, which the 2026 paper states as the premise its 40% breaks. Discernment counselling is what the field built for the case where that premise does not hold.
What it is not is well evidenced, and this page is not going to pretend otherwise. The 2016 paper reports an analysis of 100 consecutive cases, in which about half the couples chose to start couples therapy in order to reconcile and most of the rest chose the divorce path, with some longer-term follow-up. That is a case series from the people who developed the approach, not a trial. There is no randomised comparison of discernment counselling against couple therapy, against nothing, or against a waitlist. A later open-access study on what it did for post-divorce coparenting is qualitative and very small — 11 people surveyed and eight interviewed. And the 2026 prevalence paper, written in part by the same group, explicitly lists as future research the question of whether mixed-agenda couples do better starting with discernment counselling than with couple therapy, and calls its own clinical suggestions "preliminary and requir[ing] empirical testing."
So the honest status is: a well-specified approach aimed precisely at this situation, used in practice, described in a peer-reviewed journal, with an evidence base of one case series. That is considerably more than the twelve pages ranking for this question give you, and considerably less than the word "evidence-based" would imply if someone put it on a website. The same distinction — between a treatment that has been described and a treatment that has been tested — is the whole subject of what the trials report about branded therapies, and the clearest example of a branded framework whose popularity ran far ahead of the tests of it is what the research on the five love languages actually measured.
The asymmetry is still there at the courthouse door, and it is measured
One more dataset, because it says something that no advice page says and it changes how final a refusal sounds.
Doherty, with Brian Willoughby and Bruce Peterson, surveyed 2,484 divorcing parents in Family Court Review in 2011, after they had taken a required parenting class. These are not couples in therapy. These are people with a divorce in progress. They were asked whether they believed the marriage could still be saved with hard work, and whether they would be interested in reconciliation services:
- About 1 in 4 individual parents indicated some belief that the marriage could still be saved.
- In about 1 in 9 matched couples, both partners did.
- About 3 in 10 individuals indicated potential interest in reconciliation services.
- In a sub-sample of 329 matched couples, about 1 in 3 couples had one partner interested and the other not, and in about 1 in 10 both were interested.
Findings were consistent across most demographic and marital factors. Only two things strongly predicted interest in reconciliation services: gender, with men more interested than women, and initiator status — far greater interest among those whose partner had initiated the divorce.
Three things follow from that, and they pull in different directions, which is why all three belong here.
The asymmetry does not end. One partner in, one partner out is the characteristic configuration of a couple with a divorce filed, not just of a couple arguing about an appointment. If your situation feels unusually lopsided, the lopsidedness is the pattern rather than the exception.
The gender story is not what the therapy data suggests on its own. In the help-seeking literature women get to therapy first; among divorcing parents it was men who were more interested in reconciliation services. Those are not contradictory findings — different question, different moment, different sample — and together they are a decent argument against any version of this that treats one sex as the reluctant one by nature.
And the initiator effect is the one to hold on to. Interest in saving it tracked who moved first, not how bad things were. Wanting to work on a relationship and not wanting to are not fixed traits of the two people; they move with position, and position changes. What that does not license is waiting: the same literature is clear that distress rarely remits on its own, which is set out with its sources in why people stay in unhappy relationships.
When a refusal is the signal rather than the obstacle
This is near the end for length, not for importance, and it repeats what is at the top because it is the one part of this subject where being wrong has consequences.
Some of what gets described as "my partner won't go to therapy" is not a help-seeking problem. Being told who you may see or message. Having your phone checked, or being required to share your location when it is not mutual and freely chosen. Being made to account for ordinary hours. Money controlled or withheld. Anger, punishment or silence arriving as a sanction when you say no to something. Being separated from friends and family a little at a time. Any fear of your partner, any physical harm, any sense that leaving would not be allowed.
UK statutory guidance on controlling or coercive behaviour names isolating a person from sources of support, monitoring their time and movements, controlling access to money and depriving them of independence as recognised markers of that pattern. Where that is what is happening, the refusal is not an obstacle between you and a solution. It is part of the thing itself, and so is the version where attending is agreed to and then used — turning up and performing cooperation, or using what you said in the session against you afterwards.
The research on this page is structurally silent about those situations, and the silence is informative rather than accidental. The couple therapy trials that produce the effect sizes screened those couples out: intimate partner violence with actual or feared injury was an exclusion criterion in the OR-I trial quoted above, as it is across this literature. Everything measured here was measured in people from whom that had been excluded. There is a specialist literature on conjoint treatment where violence is present, it requires separate screening and separate expertise, and it is not what a general couple therapist's website is describing.
If any of that is recognisable, the useful next step is a person rather than an article. Find A Helpline, in the sources below, lists free and confidential services by country. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day. Where the pattern is jealousy rather than control, the line between the two is worked through in the psychology of jealousy; where the pull is towards a relationship that hurts and is hard to leave, what trauma bonding does and does not mean is a better starting point.
And one distinction that deserves saying out loud, because the advice pages blur it. Ordinary reluctance looks like cost, embarrassment, scheduling, self-reliance, scepticism that talking helps, or fear of being ganged up on — all of which are in the measured barrier lists above, and all of which are negotiable. Control does not appear on those lists, it is not negotiable, and treating it as an incompatibility invites you to keep looking for the better way to ask.
What the pages ranking for this question report, and what they leave out
Thirteen URLs from the first screen of results for this query were collected on 9 November 2026 and twelve were downloaded as plain text and searched; one, GoodTherapy, returned 403 to an automated request and is excluded from the counts. Among the twelve: Psychology Today twice, Healthline, marriage.com, Thriveworks, The List, and six therapy practices or platforms.
Across all twelve:
- Zero mention Doss, Doherty, Nowlan, Trillingsgaard, Williamson, Parnell, Spiker or Anderson — which is to say, zero mention anybody who has measured this.
- Zero contain the phrase effect size.
- Zero give the sample size of a study they describe. The only sample sizes anywhere on the twelve come from one practice's own internal assessment database — 58,221 completed assessments, 6,159 couples where both partners finished — which is product data, not a study.
- One contains the string et al, three times, all of them inside its reference list rather than in the text answering the question. Two further apparent matches turned out to be the letters sitting inside get along and get all, which is a reminder to look at what a count has caught.
- Four mention Gottman. One mentions discernment counselling. One cites a Journal of Consulting and Clinical Psychology finding, about first-session alliance.
- Four suggest going on your own, and eight raise individual therapy or counselling.
- One quotes a success rate for a branded model — 90% of couples improving through emotionally focused therapy — with no source.
Three things in fairness, because a tally like this is easy to use dishonestly.
Healthline's reference list is real. It includes a 2025 systematic review and meta-analysis of digital interventions for relationship satisfaction and a 2024 study of an online Gottman programme. Neither appears as a figure in the part of the page that answers this question, but the sourcing apparatus exists, and that is more than most.
One practice contradicts the stereotype with its own data, and it is worth reporting. Growing Self states that at least 50% of the people who call them for a first free consultation are men wanting to get their wives into counselling, and opens by telling the reader to set aside the idea that this is a male problem. That cuts against the help-seeking literature, and the reason both can be true is the denominator: their 50% is of people who already picked up the phone, which is a long way down the three-step funnel Doss measured in 2003. It is the same denominator problem this site has run into repeatedly, most recently in what has actually been measured about cutting contact with an ex, where every published success rate was computed on the subset for whom it worked.
And one page gives the single best piece of advice on the first screen, which is to not threaten divorce to get someone into therapy. That is the one recommendation in the whole set that the alliance data actively supports, and it is given without the data.
The failure here is not of expertise. Several of these pages are written by practising therapists and their description of what a first session is actually like is better than anything a literature review can supply — and that description is exactly what a reluctant partner's objections are usually about. The failure is that the research on this specific question exists, is mostly findable, and is not in any of them. It is the same gap this site has measured on one query after another, most cleanly in what micro-cheating actually means, where ten pages described a behaviour and not one of them gave a sample size.
What follows from this if you are the one who wants to go
Five things, and none of them is a script.
Find out which refusal it is. The measured barriers are not a single wall. Cost and not knowing where to go top the list for both sexes; the leading male objection in the Texas data was that individual counselling would be more helpful than couples therapy, which is a claim about format rather than a refusal to engage; and the help-seeking literature separates not recognising the problem, not seeing therapy as the answer, and not getting round to booking, which are three different conversations. A sceptical partner who thinks the wrong person is being sent to therapy and a partner who does not think the relationship has a future look identical and are not.
Treat the hope question as the real one. In the October 2026 data, discrepancy in hope separated mixed-agenda couples from aligned ones at 83.0% against 0.6%, while discrepancy in feeling pressured to attend barely separated them at all. If the answer is that they do not think it can work, the format conversation is not the conversation, and discernment counselling is the approach built for that case — with the honest caveat that its evidence is a case series.
Ask, and do not use leverage. This is the only point here with a direct measurement behind it and the measurement is uncomfortable: when one partner reported feeling pressured to attend, the other partner's alliance with the therapist was lower at session four, and split alliances predict worse outcomes. The one page on the first screen of results that says not to threaten divorce has it right.
Lower the cost of the first step rather than raising the price of refusing. Everything in the barrier data points at logistics as much as at willingness: not knowing where to go was a top-two barrier for both sexes, the couples who had already been through a structured programme were more likely to get therapy later, and knowing somebody who had been through couple therapy mattered when both partners did. A single session, a named therapist already found, a known cost, an online format — these address the barriers that were actually endorsed.
And if they never come, do the work anyway, for the right reason. The one randomised trial of a structured programme for a single distressed partner improved quality of life at d = 0.69, perceived health at 0.49 and work functioning at 0.44, and did not significantly improve any relationship measure. In the CRAFT trials, the concerned person's own functioning improved under all three approaches whether or not the other person ever entered treatment. Two separate literatures, same conclusion: this reliably helps you. Whether it brings the other person round is the question nobody has answered.
If what you want before any of that is a reading of where things currently stand rather than a forecast, the relationship satisfaction test reports satisfaction and commitment as separate scores — which is the distinction the mixed-agenda research turns on — the conflict style test separates pressing for resolution from pulling away, and the attachment style test covers the individual difference the therapy literature actually measures moving. All three say on their own pages what they cannot tell you, and none of them is a diagnosis.
What this evidence cannot tell you
The limits go here rather than in a footnote, because on a page about a treatment decision they are part of the answer.
Nobody has measured whether couples with one reluctant partner do worse in therapy. This is the obvious question and it is open. The 2026 paper that established the 40% figure did not examine dropout or outcome, connects its classes to the alliance findings only by inference, and calls that interpretation speculative in its own text. Everything on this page about what happens next to a mixed-agenda couple is an inference from adjacent studies.
Nothing here is a tested way of changing someone's mind. The pressure-and-alliance result says what felt pressure is associated with; it does not say what to do instead. The CRAFT trials are randomised and they are about substance use, with a different target behaviour, a different relationship role — engagement was better for parents than for spouses — and a twelve-hour manualised treatment rather than a conversation. Reading about it is not having it.
The one trial of working alone was underpowered for the thing people want to know. OR-I could detect d = 0.66 for relationship outcomes with 80% power and the satisfaction effect it saw was 0.23. That is not evidence of no effect; it is an absence of evidence of an effect that larger trials of similar programmes have found at that size. It also came from the team that built the programme, and 90 people is a pilot.
The help-seeking literature is thin, cross-sectional and self-reported. Three of its central papers are paywalled; Parnell and Hammer's 2018 study has no publicly readable abstract anywhere this site could check, so it is named and not quoted. The barrier rankings come from 231 low-income newlywed couples in one Texas county, and the population reach figures from Denmark; neither is a general population of distressed couples. Almost all of it is about different-sex couples, often explicitly so because of how the samples were built.
And the numbers describe groups, not your situation. A 40% prevalence is not a probability about you. A 36% barrier rate is a property of a sample of women in Harris County. The two best studies of who drags their feet disagree in their gender patterns because they were asked at different moments of different people, and the only strong predictor of wanting to reconcile among divorcing parents was who filed first — which is a reminder that these positions move. How much they move has been measured directly: about half of each kind of divorce thinker had shifted category within a year, mostly toward softer thinking.
Every empirical claim on this page carries a linked source, which is the promise made in our editorial policy; claims this site has had to retract are listed on the corrections page. For what the outcome research says about whether any of this works once you are both in the room, what the trials on couples therapy report is the companion piece, and the rest of this section is under Mindset.
Pulling it together
The short version is that the asymmetry is normal, the pressure is not free, and working on it alone is well evidenced for you and not for the pair.
Normal, with a number: 40.3% of 913 couples arriving at training clinics fell into a mixed-agenda class, above the field's own 25-to-30% estimate, and what distinguished them was a gap in hope rather than a gap in willingness to attend. A partner's refusal is the third-ranked barrier among people who considered therapy and did not go — endorsed by 36% of the women in the Texas sample against 14% of the men — behind cost and not knowing where to go, which are both more tractable than they sound.
Not free: in 351 couples, when one partner reported feeling pressured to attend, the other partner's alliance with the therapist was lower at the fourth session, and split alliances predict worse outcomes across 48 studies. Asking is how anyone gets there. Leverage is a different instrument with a measurable cost.
And for you rather than for the pair: the one randomised trial of a structured programme for a single distressed partner improved quality of life, perceived health and work functioning, and moved no relationship measure significantly — while in the only randomised comparison of ways to engage an unwilling person at all, from a different field entirely, the concerned person's own wellbeing improved under every approach regardless of whether the other person ever went.
What nobody has published is what happens next to the 40%. If someone tells you that confidently, ask them which study.
Frequently asked questions
What should I do if my partner refuses couples therapy?
Start by finding out what is being refused, because the measured barriers are not one thing. In 231 newlywed couples in high-poverty neighbourhoods of Harris County, Texas, the top two barriers for both sexes were not knowing where to get help or who to see and not being able to afford it. Third for women was the partner not wanting therapy, at 36%; third for men was the belief that individual counselling would be more helpful than couples therapy — which is an objection to the format, not a refusal to deal with the problem. Separately, Doss, Atkins and Christensen found that getting to therapy is three relatively independent steps: recognising the problem, considering treatment, and seeking it. Someone can be stuck at any one of them. The practical move is to lower the cost of a first step — a named therapist already found, a known price, one session, an online option — rather than to raise the price of saying no. What you would be booking, and what it does, is in what the trials report once both people are in the room.
Can couples therapy work if only one person goes?
One randomised trial has tested a structured programme delivered to a single distressed partner, and the result is worth knowing precisely. Nowlan, Roddy and Doss randomised 90 individuals — 80% women, 98% in the distressed range, many of them there because their partner had refused — to a one-person web adaptation of a couple therapy or to a one-month waitlist. Against the waitlist, quality of life improved at d = 0.69, perceived health at 0.49 and work functioning at 0.44. Relationship satisfaction did not reach significance (d = 0.23, p = 0.107), nor did relationship confidence, positive or negative relationship quality, or mood. The authors concluded that a partner's involvement is needed for lasting change. The honest caveat: the trial could only detect an effect of d = 0.66 for relationship outcomes, so a smaller real effect would have been missed, and d = 0.23 is about the size that larger trials of similar one-person programmes have found. Nobody can tell you confidently that going alone fixes a relationship, and nobody can tell you it does nothing. The couple version of the same programme, tested in 300 couples, did move relationship satisfaction, and that trial is covered in the couple version's own trial.
Is it a red flag if my partner won't go to therapy?
On the data, reluctance by itself is close to ordinary. In the clinical sample of 913 couples analysed in October 2026, a discrepancy in feeling pressured to attend therapy showed up in 46.0% of the couples whose agendas were otherwise aligned, against 53.7% of the mixed-agenda ones — the weakest of the four indicators at telling the groups apart. What did separate them was hope: 83.0% against 0.6%. So the question worth asking is not whether they want to go but whether they think it can work. There is one situation where a refusal is not a reluctance at all, and it is not subtle: if the relationship involves fear, monitoring, financial control, isolation from other people or any physical harm, that is a different problem, no part of the couple therapy evidence applies to it, and Find A Helpline in the sources lists free and confidential services by country.
Should I give my partner an ultimatum about counselling?
The one relevant measurement says it costs you. Anderson and colleagues followed 351 couples entering treatment at three training clinics, rating predictors at intake and the therapeutic alliance at the fourth session. When one partner reported feeling pressure to attend, the other partner's alliance was lower — not only the pressured partner's. Why that matters: across 48 studies and 40 independent samples, alliance correlates with outcome at r = 0.297, and split alliances, where the two partners' alliances are unbalanced, correlate with worse outcome at r = 0.316. In a separate sample of 994 cases, 20.6% dropped out before the fourth session and alliance mediated the effect of distress on dropout. None of that says do not ask. It says that an invitation and an ultimatum are different instruments, and the second one buys the version of therapy with the weaker working relationship.
How common is it for one partner to refuse couples therapy?
Common enough to be a named category in the field. Doherty and Harris estimated that 25 to 30% of couples seeking couple therapy were “mixed-agenda” — one leaning out, one leaning in. In October 2026 Tang and colleagues measured it in two large datasets and found more: in 913 couples presenting at university training clinics, a latent class analysis put 40.3% in a mixed-agenda class; in a population-based sample of 1,487 couples, about 10% had one partner saying separation was likely while the other said it was unlikely. Separately, among 2,484 divorcing parents, about 1 in 3 of 329 matched couples had one partner interested in reconciliation services and the other not. The configuration you are in is the characteristic one, not the exception.
Does individual therapy help a relationship?
A little, and less than people hope. Atkins, Dimidjian, Bedics and Christensen compared two treatment-seeking samples — 120 individuals in treatment for depression and 134 couples in marital therapy — and found that changes in the problem being treated were reliably associated with changes in the other problem, in both directions, but that the overall effects on the secondary problem were quite small. They also found that relationship distress was far more common among people in depression treatment than depression was among couples seeking marital therapy. Worth adding: in the research note that traced the six-year waiting figure, 101 of 371 participants had gone to individual therapy specifically for relationship problems and had waited as long before going as the 270 who went as a couple. What has never been studied is whether one partner's individual therapy makes the other more willing to attend couple therapy later.
What is discernment counselling, and does it work?
It is the approach built for exactly this situation, and its evidence base is smaller than its reputation. Doherty, Harris and Wilde described it in the Journal of Marital and Family Therapy in 2016 for mixed-agenda couples: one partner leaning towards divorce, the other wanting to work on the relationship. It differs from couple therapy in three ways — the goal is clarity about the next step rather than improvement, the sessions are mostly conducted with each partner individually, and divorcing is one of the legitimate outcomes. The published evidence is an analysis of 100 consecutive cases, in which about half the couples chose to start couples therapy to reconcile and most of the rest chose the divorce path. That is a case series from the developers, not a randomised trial, and the 2026 prevalence paper lists testing it against couple therapy for these couples as future work.
Who is usually the one who refuses, men or women?
The help-seeking data leans one way and the divorce data leans the other, which is a good reason not to treat this as a fact about sex. Doss, Atkins and Christensen surveyed 147 couples in the process of seeking marital therapy and found that, on average, wives completed all three steps towards therapy before husbands; in the Texas sample, 47% of wives against 32% of husbands had considered therapy, and 36% of the women who did not go named their partner's unwillingness against 14% of the men. But among 2,484 divorcing parents, the two strong predictors of interest in reconciliation services were gender — with men more interested — and initiator status, with far greater interest among those whose partner had filed. One therapy practice reports that at least half the people calling it for a first consultation are men. Position and timing seem to matter more than sex does, which is also the pattern in what has actually been measured about moving on.
How many couples ever go to couples therapy at all?
Fewer than the amount of advice about it would suggest. In a random population sample of 1,371 people living with a partner in Denmark, 7% had ever sought counselling or therapy with their partner and 3% had sought couple education. Among those who had been through a severe relationship crisis, the most common forms of help-seeking were individual: 38.9% of women and 27.8% of men looked for information online, and 19.4% of women sought individual therapy. In a longitudinal American study of 213 couples across the first five years of marriage, 36% sought outside help of some kind, with books and therapy — rather than retreats — being the kinds that actually tracked distress. Reading a page like this one, on your own, is the normal response to a relationship crisis rather than an evasion of a better one.
Sources
What this article draws on. For the broader bibliography see our sources page.
- Tang, Y., Anderson, S. R., Johnson, L. N., Harris, S. M. & Kateregga, A. (2026) — Portraying mixed-agenda couples: insights from large-scale data sets. Journal of Marital and Family Therapy, 52(4), e70165 — the prevalence figures on this page: 1,487 couples from the National Couples' Health and Time Study, of which 147 (9.89%) mixed-agenda, and 913 couples from the MFT Practice Research Network, of which a latent class analysis put 368 (40.3%) in a mixed-agenda class. Also the source of the conditional probabilities for hope (83.0% against 0.6%) and for pressure to attend (53.7% against 46.0%). Open access; free full text in PubMed Central, PMC13494567, which is the version linked here.
- Doss, B. D., Atkins, D. C. & Christensen, A. (2003) — Who's dragging their feet? Husbands and wives seeking marital therapy. Journal of Marital and Family Therapy, 29(2), 165-177 — 147 married couples (294 individuals) surveyed while in the process of seeking marital therapy; three relatively independent steps (problem recognition, treatment consideration, treatment seeking), with wives on average completing all three before husbands. Paywalled at Wiley; only what the published abstract states is quoted here.
- Doss, B. D., Simpson, L. E. & Christensen, A. (2004) — Why do couples seek marital therapy? Professional Psychology: Research and Practice, 35(6), 608-614 — the same 147 couples, asked for their reasons: most commonly problematic communication and lack of emotional affection, with little agreement between spouses within a couple about why they were there.
- Williamson, H. C., Karney, B. R. & Bradbury, T. N. (2019) — Barriers and facilitators of relationship help-seeking among low-income couples. Journal of Family Psychology, 33(2), 234-239 — 231 newlywed couples in high-poverty neighbourhoods of Harris County, Texas. 32% of husbands and 47% of wives considered therapy; 15% of couples received it. Cost and not knowing where to go were the top two barriers for both; “your spouse or partner did not want to get therapy” was endorsed by 36% of women against 14% of men (p = 0.008). Free full text in PubMed Central, PMC6389366.
- Nowlan, K. M., Roddy, M. K. & Doss, B. D. (2017) — The online OurRelationship program for relationally distressed individuals: a pilot randomized controlled trial. Couple and Family Psychology: Research and Practice, 6(3), 189-204 — the only randomised trial of a structured programme delivered to one partner of a distressed couple. 90 individuals from 35 states, 80% women, 98% distressed, randomised to the programme (n = 46) or a one-month waitlist (n = 44). Quality of life d = 0.69, perceived health d = 0.49, work functioning d = 0.44; relationship satisfaction d = 0.23 (p = 0.107) and no other relationship measure significant. Powered to detect d = 0.66 for relationship outcomes. Free full text in PubMed Central, PMC5786273.
- Anderson, S. R., Banford Witting, A., Tambling, R. R., Ketring, S. A. & Johnson, L. N. (2020) — Pressure to attend therapy, dyadic adjustment, and adverse childhood experiences: direct and indirect effects on the therapeutic alliance in couples therapy. Journal of Marital and Family Therapy, 46(2), 366-380 — 351 couples at three family therapy training clinics, predictors at intake and alliance at the fourth session. When one partner reported feeling pressure to attend, the other partner's alliance decreased. Paywalled; the figures quoted are those in the published abstract.
- Friedlander, M. L., Escudero, V., Welmers-van de Poll, M. J. & Heatherington, L. (2018) — Meta-analysis of the alliance-outcome relation in couple and family therapy. Psychotherapy, 55(4), 356-371 — 48 studies, 40 independent samples, 2,568 families, 1,545 couples and 491 effect sizes; r = 0.297 for alliance and outcome, and r = 0.316 for split alliance and outcome, with correlations significantly stronger where help was sought voluntarily rather than mandated.
- Anderson, S. R., Tambling, R., Yorgason, J. B. & Rackham, E. (2019) — The mediating role of the therapeutic alliance in understanding early discontinuance. Psychotherapy Research, 29(7), 882-893 — 994 cases in individual, couple and family, or high-conflict coparenting therapy; 205 (20.6%) discontinued before the fourth session, and the therapeutic alliance mediated the effect of distress and therapy format on dropout.
- Trillingsgaard, T., Sørensen, T. K. & Fentz, H. N. (2019) — Relationship help-seeking in a Danish population sample. Journal of Marital and Family Therapy, 45(3), 380-394 — a random population sample of 1,371 individuals living with a partner: 3% had ever sought couple education and 7% counselling or therapy with their partner. Among those with a recent severe relationship crisis, 38.9% of women and 27.8% of men had sought information online and 19.4% of women individual therapy.
- Doss, B. D., Rhoades, G. K., Stanley, S. M. & Markman, H. J. (2009) — Marital therapy, retreats, and books: the who, what, when, and why of relationship help-seeking. Journal of Marital and Family Therapy, 35(1), 18-29 — 213 couples followed across the first five years of marriage; 36% sought some form of outside help, with books and marital therapy tracking actual difficulty and retreats tracking only demographics.
- Spiker, D. A., Hammer, J. H. & Parnell, K. J. (2018) — Men in unhappy relationships: perceptions of couple therapy. Journal of Social and Personal Relationships, 36(7), 2015-2035 — 206 men who reported being unhappy in their romantic relationship. Conformity to masculine norms added unique variance beyond neuroticism; perceived need and self-stigma accounted for the most variance; relationship satisfaction was not associated with attitudes towards therapy, but lower satisfaction went with feeling more social pressure to attend. Paywalled at Sage; the abstract was read through OpenAlex and is the only thing quoted.
- Spiker, D. A., Berney, E. C., Hammer, J. H. & Jensen, K. C. (2020) — Maintaining the relationship: relational schemas and women's intent to seek couple therapy. The Counseling Psychologist, 48(6), 801-825 — 302 heterosexual women unhappy in their relationships. Self-silence, relational-interdependent self-construal, self-reliance and emotional control made them both less and more likely to intend to seek couple therapy, which the authors read as conflicting messages rather than a single effect.
- Parnell, K. J. & Hammer, J. H. (2018) — Deciding on couple therapy: the role of masculinity in relationship help-seeking. Psychology of Men and Masculinity, 19(2), 212-222 — the study that established the theory-of-planned-behaviour framework for men's relationship help-seeking. Named here and not quoted: it is closed access and, checked on 9 November 2026, has no publicly readable abstract in Crossref, OpenAlex, Semantic Scholar or Europe PMC. No figure on this page comes from it.
- Atkins, D. C., Dimidjian, S., Bedics, J. D. & Christensen, A. (2009) — Couple discord and depression in couples during couple therapy and in depressed individuals during depression treatment. Journal of Consulting and Clinical Psychology, 77(6), 1089-1099 — 120 individuals seeking treatment for depression and 134 couples seeking marital therapy. Changes in the primary problem were reliably associated with changes in the secondary one, but the overall effects on secondary problems were quite small. Paywalled; the abstract, which carries those figures, is readable in Europe PMC.
- Miller, W. R., Meyers, R. J. & Tonigan, J. S. (1999) — Engaging the unmotivated in treatment for alcohol problems: a comparison of three strategies for intervention through family members. Journal of Consulting and Clinical Psychology, 67(5), 688-697 — the only randomised comparison of ways to get an unwilling person into treatment. 130 concerned significant others, 12 hours of contact each, 12-month follow-up with 94% completion: CRAFT engaged 64%, the Johnson Institute confrontation 30% and Al-Anon facilitation 13%; 70% of those preparing a confrontation never went through with it; all three approaches improved the concerned person's own functioning similarly, and engagement was higher for parents than for spouses. About substance use, not couple therapy.
- Meyers, R. J., Miller, W. R., Hill, D. E. & Tonigan, J. S. (1998) — Community reinforcement and family training (CRAFT): engaging unmotivated drug users in treatment. Journal of Substance Abuse, 10(3), 291-308 — 62 concerned significant others; 74% engaged their resistant relative over six months, and all of them showed significant reductions in depression, anxiety, anger and physical symptoms whether or not engagement happened.
- Doherty, W. J., Harris, S. M. & Wilde, J. L. (2016) — Discernment counseling for “mixed-agenda” couples. Journal of Marital and Family Therapy, 42(2), 246-255 — the description of the approach designed for couples where one partner leans towards divorce and the other wants to work on the relationship, with an analysis of 100 consecutive cases in which about half chose to begin couples therapy and most of the rest the divorce path. A case series, not a trial.
- Doherty, W. J., Willoughby, B. J. & Peterson, B. (2011) — Interest in marital reconciliation among divorcing parents. Family Court Review, 49(2), 313-321 — 2,484 divorcing parents surveyed after required parenting classes. About 1 in 4 individuals believed the marriage could still be saved and about 1 in 9 matched couples both did; in a sub-sample of 329 matched couples, about 1 in 3 had one partner interested in reconciliation services and the other not. The only strong predictors were gender, with men more interested, and initiator status.
- Emerson, A. J., Harris, S. M. & Ahmed, F. A. (2021) — The impact of discernment counseling on individuals who decide to divorce: experiences of post-divorce communication and coparenting. Journal of Marital and Family Therapy, 47(1), 36-51 — 11 people surveyed and eight interviewed, analysed phenomenologically. Cited here as the size of the follow-up evidence on discernment counselling rather than for any quantitative claim. Open access.
- Hubbard, A. K. & Harris, S. M. (2020) — A critical review of help-seeking for couples therapy: clinical implications and next steps. Contemporary Family Therapy, 42(2), 152-162 — the review of this literature, evaluating the methodologies and theories it uses. Paywalled at Springer; its abstract, which is readable, describes a narrative review rather than a pooled estimate, and no figure on this page comes from it.
- Recuento propio del SERP y de las bases bibliográficas, 9 de noviembre de 2026 — Written out so it can be re-run rather than taken on trust. Thirteen URLs from the first screen of results for “my partner won't go to couples therapy” were collected and twelve downloaded as plain text; GoodTherapy returned 403 and is excluded. Across the twelve: zero mention Doss, Doherty, Nowlan, Trillingsgaard, Williamson, Parnell, Spiker or Anderson; zero contain “effect size”; zero give the sample size of a study, the only sample sizes being one practice's own assessment database (58,221 assessments, 6,159 couples); one contains “et al” three times, all inside its reference list, and two apparent matches were the letters inside “get along” and “get all”; four mention Gottman; one mentions discernment counselling; four suggest going alone and eight raise individual therapy. On the same day Google returned the unusual-traffic challenge, Brave an anti-bot check and DuckDuckGo a captcha, so the result set came from the session's own search tool and no “People also ask” block could be collected. PubMed, searched the same day: “couple therapy” with “one partner” and any form of “refuse” returns zero records.
- Find A Helpline — Free, confidential support services listed by country — the referral at the top and near the end of this article, which covers the situation that nothing else on this page addresses.
Keep reading
- MindsetDoes Couples Therapy Work? What the Trials Report, and Where the Six-Year Figure Came FromCouple therapy works, moderately, and fades for about half of couples. The six-year waiting figure is an unpublished thesis on fewer than 20 couples.
- MindsetThinking About Breaking Up: What a National Panel Found Happened NextA quarter of married people had thought about divorce in the past six months. A year later, 31% of them were no longer thinking about it.
- MindsetHow Long Does It Take to Get Over a Breakup? Four Clocks, Four Different AnswersThree of the numbers on the first page of results trace back to one 2007 survey of 155 undergraduates, where the figure was how long ago the breakups were.
- MindsetDoes Living Together Before Marriage Lead to Divorce? Four Things Change the AnswerThe famous 1.31 comes from women under 45 in first marriages. Control for the age they moved in together, not the age they married, and it drops to 1.09.
- MindsetDo Rebound Relationships Work? The Only Experiments Never Put Anybody in OneNot one of the eight readable pages gives a sample size, and the two experiments on the subject never put anybody in a new relationship.
- MindsetDo Couples Stay Together After Infidelity? The Famous 57% Is Eight Couples Out of FourteenThe famous 57% comes from a five-year follow-up of 19 couples: eight of the 14 who disclosed the affair stayed, and one of the five who hid it.