What Actually Helps an Anxiously Attached Partner, and What Only Helps Today
Reassurance lowered attachment anxiety on the day and not six months later, in the one study that measured both. Something else predicted the decline.
Before the article. If the anxiety in question is a reasonable response to how somebody is being treated — a partner who monitors a phone or a location, decides who may be seen, controls money or transport, or escalates until there is compliance — then attachment is the wrong lens, and a later section says so at length. 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. This page is for the person on the other side of anxious attachment: the one being asked, again, whether everything is all right. Almost everything written for that person is a list of tips with nothing behind it, and almost everything written about anxious attachment is addressed to the anxiously attached person instead.
There is research, and it is better than the genre lets on. The central result comes from 137 couples followed across five assessments over two years, in which the researchers measured perceived reassurance and attachment anxiety separately and then ran the association twice — once within a wave, once across six-month gaps. Reassurance was strongly associated with lower anxiety in the moment, and had a longitudinal coefficient of exactly zero. What predicted the decline across time was something else entirely, and no page competing for this question mentions it.
Underneath that sit three things nobody puts in the advice. The partner's own report of how much reassurance they provided predicted nothing; only the anxious person's perception did. Expressing insecurity has a measured route by which it makes a partner's reassurance feel less authentic, which is the opposite of what the standard advice predicts. And the standard advice has a price, paid by the person following it, measured in two separate studies.
Twenty-one sections follow: what the ranking pages say, the two-year study, the inversion that preceded it, what reassurance genuinely does inside an argument, the eggshells loop, the cost on both sides, burnout as a boundary condition, the one technique with an experiment behind it, what backing somebody actually means, gratitude, security priming, what happens when the supporter is anxious too, whether this is the hardest attachment style to love, the depression literature hiding behind reassurance seeking, where this stops being about attachment, and the limits. Nothing here is clinical advice, and our editorial policy explains why every figure carries a linked source.
What the pages ranking for this actually say
We ran the search the way a reader would, on 24 October 2026, and wrote down the answer before opening a single paper.
Ten results. Nine of them were therapy practices, coaching sites and attachment-quiz blogs. The first was a numbered list of ten tips — acknowledge the attachment style, be consistent, offer verbal reassurance, watch for triggers, show affection, avoid mixed signals, encourage open communication, build trust, respect their need for reassurance, support their growth — with not one citation anywhere on the page. The others said the same things in a different order.
The only established health brand in the results was Verywell Mind, and it answers a different question: its page is written for the anxiously attached person rather than the partner, and its central instruction is to name your triggers so that your partner can work to reassure you of their commitment in the face of your anxiety. One page told readers that reassurance doesn't work long-term. It did not say how anybody knows that.
The through-line across all ten is a single claim: reassure more, reassure consistently, and never withhold reassurance during an argument. It is intuitive, it is kind, and it is half right in a way that can be stated precisely.
Because the question has been studied. One research group followed 137 couples across two years and measured, separately, how reassured the anxiously attached partner felt and whether their attachment anxiety went down. The two things came apart. No page on that first results page mentions the study, and our editorial policy is the reason every number below arrives with a link attached.
The study that followed 137 couples for two years
Ximena Arriaga, Jami Eller, Madoka Kumashiro, W. Steven Rholes and Jeffry Simpson recruited couples from childbirth preparation classes who were expecting a first child. They measured everyone five times: six weeks before the due date, then at 6, 12, 18 and 24 months after the birth. Of the 192 couples who completed the first wave, 137 complete pairs made it to the end.
Two things were measured at every wave, and the precision matters.
Perceived reassurance came from the Kunce and Shaver Caregiving Scale, and it is exactly what the advice columns describe. Eight items on proximity — When I want or need a hug, my partner is glad to provide it; When I am troubled or upset, my partner moves closer to provide support or comfort. Eight on sensitivity — My partner can always tell when I need comforting, even when I don't ask for it; My partner is very good at recognizing my needs and feelings, even when they're different from his or her own. The two subscales correlated highly and were combined.
Attachment anxiety came from the eighteen-item anxiety subscale of the Experiences in Close Relationships questionnaire, the standard measure. Avoidance was held constant throughout.
Then they ran the same association twice. Once within a wave — how anxious is this person right now, given how reassured they feel right now. And once across waves — does how reassured they felt six months ago predict a change in their anxiety since, holding their earlier anxiety constant.
Within a wave, reassurance worked, and it was one of the strongest results in the paper: b = −0.06, 95% CI [−0.09, −0.04], t = −4.76, p < .001. Not moderated by gender.
Across six months, it did nothing. b = 0.00, t = −0.19, p = .849, with a confidence interval of [−0.03, 0.03]. That is not a small effect or an underpowered one. It is zero with a narrow interval drawn tightly around it, in a dataset large enough to have found the concurrent effect four waves running.
What did predict the decline, and how firmly
Something in the same study did predict attachment anxiety falling over time, and it is not in any of the ten articles.
It was parenting self-efficacy — measured with twelve items about feeling competent in the new role, such as My baby makes me feel more competent and My baby gives me a sense of accomplishment. People who derived more of that sense of competence showed declines in attachment anxiety six months later: b = −0.09, 95% CI [−0.16, −0.01], p = .043.
And here is the symmetry that makes the paper worth reading twice. Self-efficacy did not predict anxiety within a wave (p = .249). Reassurance did not predict it across waves. Each variable was significant in exactly the window where the other was not.
Two honest qualifications belong in the same breath as the number, because the site that leaves them out is selling you something.
First, p = .043 is marginal, and the upper bound of that interval is −0.01. This is a real finding and it is not a large one.
Second, the design cannot establish cause. The authors say so themselves: the mechanism was tested with non-experimental longitudinal data. What they did do is rule out the nearest alternatives. Closeness with the baby did not predict change in anxiety. Parenting satisfaction did not. The partner's parenting self-efficacy did not predict the person's own change. So it is not simply that good things were happening to these couples.
The background trend is worth knowing too: anxiety drifted down a little in everyone across the two years, from 3.34 to 3.03 in the women and 2.74 to 2.50 in the men, while perceived reassurance drifted down rather than up. The decline is the thing being explained, not a thing being promised.
The result that should change how the question is asked
Buried in the same paper is an analysis that reframes the search query itself, and it is reported in two sentences without emphasis.
Both partners answered. So the researchers could compare the anxious person's perception of being reassured against the partner's own report of providing reassurance. The two were correlated, as you would expect: the giving and the receiving are not unrelated.
But the cross-partner association — the partner's report of how much reassurance they provided, predicting the anxious person's attachment anxiety — was not significant in either model. Not within a wave. Not across six months.
Read that against the question people type. How do I reassure an anxious partner is a question about output: what do I say, how often, in what tone. The variable with a measured association in the best longitudinal data on the subject is not output. It is intake — what the other person perceived.
This is not a counsel of despair, and it is not an instruction to try harder. It is a different instruction: the useful information is on their side of the table, and you can only get at it by asking rather than by performing. It also predicts something the genre cannot explain, which is the couple where one person is doing a great deal of visible reassuring and the other does not feel reassured. In that couple, the thing that correlates with anything is the second report, not the first.
The same asymmetry turns up again, independently, in the gratitude research further down this page. And the general question of how far any of this dial actually moves is set out with the stability coefficients in what the longitudinal evidence says about attachment styles changing.
Why none of this is an argument for withholding reassurance
It would be easy to read the last three sections as permission to stop, and that is not what they say. It is worth being explicit, because the one page in our results that told readers reassurance doesn't work long-term was pointing in that direction without the evidence to do it.
The framework behind the two-year study has two halves, and the authors keep them apart on purpose. Arriaga, Kumashiro, Simpson and Nickola Overall set out the Attachment Security Enhancement Model in 2018. One part concerns partner responses that protect a relationship from the erosive effects of immediate insecurity. The other concerns situations that foster more secure mental models underneath. Their own summary is that what allays momentary insecurity may not necessarily reduce attachment anxiety across time, and that the two parts may work in tandem.
Their specific proposal is that attachment anxiety should decline most in situations that foster greater personal confidence and more secure models of the self, while attachment avoidance should decline most in situations that involve positive dependence and more secure models of close others. That is a prediction, and the authors describe the deeper-change half of the model as tentative — which is more honesty than this subject usually gets.
So the measured picture is: reassurance has a strong, repeatedly observed concurrent effect and no detected longitudinal one. Nobody has run the study where a partner deliberately withholds reassurance and the researchers see what happens. That is not a third option with evidence behind it; it is an option with no evidence at all, which is a different thing from a bad option and should not be presented as the finding.
What the evidence does license is a change of expectation. Reassurance is maintenance, not treatment. It is worth doing for what it does.
The same inversion, in the study that found it first
The two-year transition study was not the first time this pattern appeared. It is the second, in a different sample, with different measures, and that is why it is worth trusting more than one result usually is.
In 2014, Arriaga with Kumashiro, Eli Finkel, Laura VanderDrift and Laura Luchies followed 134 romantically committed couples and separated two things a partner supplies. Trust — whether the partner is perceived as available and dependable. Perceived goal validation — whether the partner is perceived as encouraging your own personal pursuits.
Measured at the same moment, the pairings were the intuitive ones. Trust went with lower attachment anxiety. Goal validation went with lower avoidance.
Measured across time, they swapped places. Trust uniquely predicted reduced avoidance later on. Perceived goal validation uniquely predicted reduced anxiety later on.
Two datasets, six years apart, with different operationalisations of the same idea, produce the same shape: the thing that soothes anxiety in the room is not the thing associated with anxiety falling over months. In 2014 the longitudinal predictor was being backed in something of your own. In 2020 it was coming to feel competent in a new role. Those are close relatives.
We have written about the 2014 study before, as part of the question of whether attachment styles change at all. What is new here is the replication and the direction of the advice: that result is usually read as a fact about the insecure person, and it is at least as much a fact about what their partner is in a position to do.
What reassurance does do, measured inside an actual argument
Three behavioural observation studies make the concurrent half concrete, and they are the best argument for carrying on.
SiSi Tran and Jeffry Simpson videotaped married couples discussing important habits each wanted the other to change — a situation built to activate fear of rejection. More anxious individuals felt more negative emotion and showed less coder-rated accommodation. But when their partners were more committed to the relationship, those partners inhibited the urge to retaliate and worked harder at the problem, and the anxious individuals then reported more acceptance and behaved about as constructively as secure individuals did. In that study, wives' commitment had stronger effects on both partners' emotions than husbands' did.
Lorne Campbell, Simpson, Jennifer Boldry and Deborah Kashy ran fourteen days of diaries followed by a videotaped discussion of a problem that had actually come up during the diary period. More anxiously attached people perceived more conflict and reported a tendency for conflicts to escalate; observers rated them as more distressed and as escalating severity. Perceptions of greater daily support had positive effects on their satisfaction and on how they saw the relationship's future.
Jessica Salvatore and colleagues did something rarer. Their 73 target participants had been studied since birth, so childhood attachment was on record rather than recalled. Couples did a ten-minute conflict discussion followed immediately by a four-minute cool-down task, and coders scored how far each person disengaged from the conflict. People with insecure histories recovered less well. But having a partner who recovered better predicted more positive relationship emotions at the time, and, prospectively, the partner's recovery and the target's early security interacted to predict whether the couple was still together two years later — a partner who could let it go buffered the person who could not.
That is a real, repeatedly observed effect, and it is about the minutes after a fight rather than about the disposition. The mechanics of those minutes are the subject of what the repair research actually supports, and if the recurring problem is the shape of the arguments rather than the closeness, our conflict style test is built on the demand-withdraw literature instead.
The loop that makes reassurance stop landing
The single most repeated instruction in this genre is: tell your partner what triggers you, so they can reassure you. There is a paper about what that does, five studies deep, and none of the ten pages mentions it.
Edward Lemay and Margaret Clark called it “Walking on eggshells.” Their model runs like this. When people are insecure about a partner's acceptance, they express emotional vulnerabilities. Expressing them causes them to believe the partner now sees them as highly vulnerable and insecure. That belief causes them to doubt the authenticity of the partner's expressions of positive regard. The doubt predicts perceptions of rejection, which predict derogating the partner and expressing more vulnerability — back to the start.
The studies test it in order. Study 1 established that people's prototypes of interacting with an interpersonally vulnerable person include the partner giving inauthentic expressions of regard. Studies 2A, 2B and 4 found that those who had expressed vulnerabilities doubted their partner's authenticity, and did so because they believed they were seen as vulnerable. Studies 3 and 4 linked those doubts to perceived rejection, and Study 4 followed the chain through to derogation and further expression. Study 5 is the one that makes it causal: experimentally manipulating how vulnerable a person believed they were seen to be increased their doubts about the authenticity of a new acquaintance's expressed emotion.
This is not an argument for saying nothing. It is a specific mechanism with a specific trigger: being seen as fragile, rather than disclosure as such. Reciprocal, gradually deepening disclosure between people who respond well is a different object with a better record, and it is covered in what the research on opening up supports.
It also explains the thing partners report most often and understand least — that the reassurance does not seem to stick. Simpson and Overall make the related point in their review: even well-chosen buffering has to be delivered skilfully, because exaggerated or prolonged attempts can be read as insincere and then fail to quell anything.
The cost that is never in the advice
Here is the part that is genuinely missing from every page we read, and it has been in the literature since 2011.
Lemay and Kari Dudley built and tested a model of interpersonal insecurity compensation across three studies and several relationship types. Partners detect a person's chronic insecurities about acceptance. They become vigilant about upsetting them. And they respond with what the authors call affective exaggeration: cautiously inflating positive thoughts and feelings about the person, and concealing negative ones.
The studies show the machinery. In Study 1, people concealed negative sentiments specifically when they believed those sentiments would be observed by the insecure person. In Study 2, different perceivers converged in how much they exaggerated affection towards the same insecure target — it is a response to something real, not a quirk. In Study 3, reported vigilance about upsetting the person predicted heightened cognitive processing of that person's daily insecurity and intensified the tendency to exaggerate affection in response to it.
Then the result with two signs on it. The exaggeration did enhance chronically insecure targets' perceptions of being valued — the advice works. And it also predicted the perceiver's reduced relationship satisfaction, in both of the studies that measured it.
The authors' own summary is the honest one: this is the active, but perhaps dissatisfying, regulation of relationships with chronically insecure partners.
That is not a reason to stop. It is the information needed to budget. Every one of the ten pages recommends this behaviour; none of them says it has a measured price, or who pays it. If you are the person doing it and you have noticed yourself editing what you say before you say it, you have not failed at supporting someone. You are doing the thing the research describes, and the research also describes what it costs.
The same trade, seen from the other side of the table
The companion study is about what the anxiously attached person does, and it is worth reading precisely because it refuses the moralising version.
Overall, Yuthika Girme, Lemay and Matthew Hammond tested whether anxious reactions to relationship threat actually work. Study 1 was a daily diary; Study 2 was observed conflict discussions.
People higher in attachment anxiety felt more hurt on days they faced partner criticism or conflict, and during observed conflicts. Those pronounced hurt feelings triggered exaggerated expressions of hurt that induced guilt in the partner. Partners perceived more anxious individuals' hurt as more intense, and in turn felt more guilt. Objective coders rated more anxious individuals as using more guilt-induction strategies, which led to measured increases in partner guilt.
And the guilt helped them. Anxious individuals whose partners felt more guilty held more stable perceptions of their partner's commitment and more positive evaluations of the relationship. For people low in anxiety, partner guilt was simply detrimental; for anxious people it was stabilising. The authors' conclusion is explicitly against the standard reading: anxious reactions to threat are not uniformly destructive.
The sentence that follows is the one this page is organised around. Those benefits were accompanied by significant declines in the partner's relationship satisfaction.
So two independent studies, from two labs, measuring two different behaviours — the partner's affective exaggeration and the anxious person's expressed hurt — found the same structure. The thing that stabilises one person's felt security is drawn from the other person's satisfaction. That is not a villain and a victim. It is a transfer, and a relationship that runs on it indefinitely is running a deficit somewhere.
Where a feeling and a behaviour have to be told apart carefully, and where the line sits, is the subject of what the research on jealousy actually separates.
What feels like it worked, versus what changed anything
The gap between the immediate reading and the later outcome shows up a third time, in a study that was not about attachment at all, and the repetition is the point.
Overall, Garth Fletcher, Simpson and Chris Sibley videotaped 61 couples while each tried to produce a desired change in the other. They classified the strategies by valence (positive or negative) and by directness (direct or indirect). Then they measured two things: how successful the discussion felt immediately afterwards, and how much the targeted feature had actually changed, rated at three-month intervals across the following year.
Positive-indirect strategies — the soft, oblique, keep-it-pleasant approach — were associated with the highest concurrent perceived success and did not predict any later change.
Direct strategies, positive and negative alike, were initially perceived as relatively unsuccessful, and predicted increased change over the next twelve months as reported by the partner who had been targeted.
That is the same shape as the reassurance result and the trust-versus-goal-validation result, in a third literature: the sensation of a conversation having gone well is a poor instrument for whether it did anything. It is a good instrument for something else — how the evening went — and that is worth having on its own terms.
It also suggests why advice in this genre is so stable and so unfalsified. Every tip on that first results page produces an immediate, visible, positive response. The feedback loop is fast, pleasant and entirely uninformative about the twelve-month question.
Burning out is a boundary condition, not a character failure
Jami Eller and Simpson wrote the paper that asks what limits all of this, and it is open access, which is unusual for the field and useful here. They propose three boundary conditions — motivation, ability and time — on both sides of the exchange, and they are unusually direct about the person doing the supporting, who they note has received much less attention.
Four points from it are worth carrying.
Buffering that is perceived as inauthentic amplifies insecurity rather than reducing it. So does buffering that runs against the norms of that particular relationship. It is not a volume knob.
Support given out of obligation is less effective and more personally costly than support given because you want to give it. Their example is pointed: an agent who buffers because a partner explicitly requested it, or because a therapist prescribed a communication skill, is likely to be less engaged and to come across as less authentic.
Reciprocity is not the norm. In a meta-analysis they cite, partners' communal motivations correlate at about r = 0.26 — related, and nowhere near matched. A partnership where one person consistently does the regulating and the other does not is an ordinary partnership, not a broken one, and the authors are plain that people who are consistently under-benefitted may come to feel underappreciated or burn out.
There is no protocol. They propose that buffering should be scaffolded — more when insecurity is high, less as security builds, with upticks for setbacks — and then state that no established guidelines exist for how to do that. Simpson and Overall add the failure mode from the other direction: when buffering repeatedly fails, partners may give up trying, which amplifies their own dissatisfaction.
They also name the risk on the receiving side, and it is the thing the whole first page of results assumes away: a person can become reliant on a partner as an external emotion regulator and never develop their own regulation, which would maintain the anxiety rather than reduce it. Nobody has measured how often that happens. It is a theoretical worry from the people who study the effect, not a finding, and it is the reason the next three sections exist. If the live question is whether to stay in an arrangement like this, what the research on staying in unhappy relationships has measured is about that directly, and our relationship satisfaction test reports satisfaction, alternatives and investment as separate numbers rather than a verdict.
The one concrete technique with an experiment behind it
Almost everything above is observational. There is one experiment in the neighbourhood that produced a usable, durable effect, and its existence is the best argument for taking the perception finding seriously.
Denise Marigold, John Holmes and Michael Ross started from a problem that will be familiar: people with low self-esteem doubt their value to their partners and resist positive feedback from them, which undermines the relationship and had proved hard to shift in previous research.
Their intervention was not about how the compliment was delivered. It was about what the recipient did with it afterwards. Across three studies, participants were asked to describe a recent compliment from their partner either concretely — what was said, what happened — or abstractly, in terms of its meaning and significance: what it says about them, what it says about the relationship.
In the abstract condition, people with low self-esteem felt more positively about the compliment, about themselves and about the relationship — as positively as people with high self-esteem felt. And the effect was still evident two weeks later.
The caveat has to travel with it, so here it is. This was measured on self-esteem, not attachment anxiety. The two are correlated and they are not the same measurement, and nobody has run this manipulation with the attachment anxiety subscale as the moderator. We are not going to describe it as an attachment intervention, because it has not been tested as one.
What it supports is narrower and still worth knowing: when reassurance does not land, there is at least one demonstrated route by which the same words can be made to land, and it runs through the receiver rather than the giver. It also means the honest version of the usual advice is not say it better. It is closer to ask what they made of it.
Backing their thing is not a figure of speech
If two independent longitudinal datasets point at the same family of predictors, it is worth being precise about what is in that family, because the vague version of this advice is everywhere and the measured version is narrower.
In the 2014 study the longitudinal predictor of falling anxiety was perceived goal validation: the partner perceived as encouraging your own personal pursuits. In the 2020 study it was self-efficacy in a new and important role, measured as feeling more competent and getting a sense of accomplishment from it.
Brooke Feeney's secure-base work puts a qualifier on both that the advice columns drop. Across an observational phase, in which couples were videotaped discussing personal goals, and an experimental phase in which support behaviour was manipulated, what predicted the recipient's happiness, self-esteem and perceived likelihood of achieving a specific goal was support of their goal strivings that was responsive and nonintrusive. Taking over is not backing. Advising is not backing. Improving their plan is not backing.
And two limits that belong here rather than in a footnote.
The 2020 authors themselves write that future research should compare parenting versus work self-efficacy, which could have divergent results. Nobody has run that comparison. So “support their career and their attachment anxiety will fall” is an extrapolation from one life domain to another, offered by us as a reasonable guess rather than a result.
And the alternatives they ruled out tell you what the finding is not. Closeness with the baby did not predict change in anxiety. Satisfaction with parenting did not. The partner's own competence did not. What predicted it was the anxious person's own sense of being good at something that mattered.
The behaviours that tend to accompany that kind of support are the subject of what has and has not been measured about emotional maturity, and what accumulates over months rather than evenings is in what actually builds attraction over time.
Gratitude was measured, and it mostly helped the other insecurity
One specific behaviour in this area has five studies behind it, and the result is more interesting than the headline because it does not go where you would expect.
Yoobin Park, Emily Impett, Geoff MacDonald and Lemay examined whether receiving expressions of gratitude from a partner buffers insecurely attached people against low satisfaction and commitment.
Study 1 found the negative associations between attachment avoidance and both satisfaction and commitment were weaker among people who perceived their partner expressed gratitude more often — and found the same pattern for anxiety and satisfaction. Study 2 traced it through beliefs about the partner's communal strength. Study 5 followed people for three months: perceiving high levels of a partner's gratitude enhanced avoidantly attached people's feelings of being cared for three months later, which went with greater satisfaction and commitment.
Two details are the reason this section exists.
First, their internal meta-analysis across the five studies concluded that the benefits of perceiving a partner's gratitude may be specific to buffering the negative effects of attachment avoidance on satisfaction. The thing that worked best worked best for the other dimension. Advice that hands this to everyone is over-generalising a paper that checked.
Second, in Studies 3 to 5 it was perceived gratitude, not the partner's self-reported gratitude, that buffered daily dissatisfaction. That is the Arriaga asymmetry again, in a different lab with a different behaviour: what the giver reports giving did not do the work.
This is also a reminder that these two dimensions are not interchangeable and the tailoring is not cosmetic. Overall, Simpson and Helena Struthers videotaped 180 couples and found that avoidant people's anger and withdrawal eased specifically when their partner softened influence attempts and respected their autonomy. Simpson, Heike Winterheld, Rholes and Minda Oriña found that avoidant people were calmed by instrumental caregiving — concrete suggestions — while emotional caregiving made them worse, the reverse of the secure pattern. Girme, Overall, Simpson and Fletcher, pooling four studies and 298 couples, found the effect of practical support on avoidant recipients was curvilinear: low-to-moderate support made things worse, and only high levels improved them. Those three studies, the thirty-second coding inside them and what they imply for the opposite advice — give them space — are the subject of what actually helps an avoidant partner.
Security priming: where a trait number did move, and the asterisk
There is one experimental literature in which a trait-level attachment number has moved, and it is worth looking at squarely, both because it is encouraging and because it is routinely oversold.
Security priming means experimentally activating mental representations of a supportive attachment figure — asking someone to visualise or write about a time they felt cared for — and measuring what shifts. Omri Gillath, Gery Karantzas, Daniel Romano and Katherine Karantzas meta-analysed 120 studies across 97 articles, N = 18,949, and found an overall effect of d = .51, largest for affect. Their own bias assessment found mixed evidence for possible asymmetry, which is the authors being careful. Crucially, nearly all of those outcomes are states measured shortly after the prime.
Two studies pushed at the trait question.
Katherine Carnelley and Angela Rowe primed participants with attachment security on three occasions across three days, then measured them two days later with no prime beforehand. Those repeatedly primed reported more positive relationship expectations, more positive self-views and lower attachment anxiety at the end than at baseline, while the neutral group did not change. The published abstract does not report a sample size and the full text is paywalled; we could not read the method.
Carnelley, with Mona-Maria Bejinaru, Lorna Otway, David Baldwin and Rowe, then took it into a clinical sample: 48 outpatients with depressive disorders, primed in the lab and then with three daily primes by text message, followed up a day later. Felt security was higher than control at every time point. Symptoms of anxiety and depression fell more than in the control condition, but the difference was only significant at one time point, and the authors name the moderate sample size as limiting their power.
What this is: evidence that the attachment system is responsive, and the only place where a dispositional measure has moved in a controlled design. What it is not: a partner. Nobody has shown that a person functions as a repeated prime, and the follow-up windows here are days. The broader question of what does and does not shift a baseline, including why the self-help version has never been trialled, is set out here with the numbers.
If the person doing the reassuring is anxious too
Nearly all the advice assumes the supporting partner is secure, available and steady. One study asked what happens when they are not, and it is the most practically important limit on this page.
Shanuki Jayamaha, Girme and Overall ran two studies in which couples had two video-recorded discussions, each about the other's most important personal goal. The person whose goal was being discussed reported how distressed they felt. The person in the supporting position reported how valued and appreciated they felt. Independent observers coded how far the supporter was critical and derogating versus warm and understanding.
The chain held across both studies. Supporters higher in attachment anxiety felt less valued and appreciated when the person they were supporting was more distressed. And lower feelings of being valued went with those supporters showing more negative support behaviour. In Study 2 the effects emerged only for male providers, which the authors report rather than smooth over.
The consequence is not confined to the conversation: consistent with prior work, greater negative behaviour by support providers predicted declines in the recipient's relationship quality over time.
Eller and Simpson reach the same place from theory. An anxious supporter is dealing with their own escalated emotion and may not register the other person's distress at all; an avoidant one is likely to be disengaged and to miss the signals. Both then face the separate problem of responding in a tailored way.
So “just be consistently warm” is advice with a prerequisite. If reading your partner's distress as evidence that you are failing them is the thing that happens first, you are the person in that study, and the useful move is probably not a better script. Two insecure people can of course make this work; what the literature says is that nobody should be surprised when the standard advice does not fit, and what the evidence says about telling love from anxious attachment covers how much of this is shared surface.
Is this the hardest attachment style to love?
It is one of the questions Google suggests alongside this one, so it deserves an answer rather than a dodge, and the answer starts by refusing the premise.
There are no styles to rank. The measured object is two continuous dimensions — anxiety and avoidance — and when Fraley and colleagues ran taxometric analyses on two samples, testing both general attachment representations and relationship-specific ones, the structure came out dimensional rather than categorical in every case they tested. “Which style is hardest” is a question about boxes that nobody has been able to find, which is set out with the analyses in what actually moves, and how slowly.
What has been pooled is the association between each dimension and relationship outcomes. Benjamin Hadden, C. Veronica Smith and Gregory Webster meta-analysed those associations and tested something nobody had: whether they change with how long the relationship has lasted. Both insecure dimensions were associated with lower satisfaction and lower commitment, and for both, the negative associations were more negative in samples with longer relationship durations. Their caution belongs with the result — those comparisons are across cross-sectional samples, so this is a pattern across relationships of different lengths rather than a measured trajectory within couples.
Neither dimension emerges as the hard one. What emerges is a slope, and a reason to take the twelve-month question seriously rather than the evening one.
And the dyadic literature makes the ranking question the wrong level of question anyway. Its founding observation, repeated at the top of both review papers, is that insecurely attached people have less happy and less stable relationships but that the quality of those relationships depends on how their partners respond. That is not a property of a style. It is a property of a pair.
Reassurance seeking has its own literature, and it is about depression
When the frequency of the asking is the thing in the room — the same question several times a night, the answer never holding past the hour — there is a measured literature on exactly that, and it is filed somewhere people searching for attachment advice rarely look.
Thomas Joiner and Gerald Metalsky ran six studies on excessive reassurance seeking. Studies 1 and 2 established it as a cohesive, replicable construct, distinguishable from related interpersonal variables. Study 3 found it showed diagnostic specificity to depression in a clinically diagnosed sample, where other interpersonal variables did not. Study 4 is the one that matters most: among people who were symptom-free at baseline, those who went on to develop depressive symptoms had elevated reassurance-seeking scores while still symptom-free, and were not elevated on the other interpersonal measures. Studies 5 and 6 found it predicted depressive reactions to stress.
Two things follow, and neither is a diagnosis.
The first is that persistent reassurance seeking is not only an attachment phenomenon, and the literature that has followed it prospectively points at depression rather than at working models. If that is the shape of what you are living with, reading more about attachment styles is unlikely to be the highest-value next step, and a professional is.
The second is a caution about this page and every page like it. A research scale is not an assessment, and nothing here can tell you what is going on with a specific person. Our attachment style test returns two scores and says on its own page what a score like that cannot tell you — which is most of what people want from it.
When this stops being a question about attachment
One thing on this page can be said with more confidence than anything measured on it, so it goes here where it will be read rather than buried at the bottom.
The vocabulary of anxious attachment gets used to explain two very different situations, and in both of them it is the wrong lens.
When the fear is accurate. If someone is anxious because a partner monitors their phone or location, decides who they may see, controls money or transport, escalates when they say no, or punishes them with silence until they comply, that is not an attachment pattern to be reassured away. It is a reading of their circumstances. Attachment language is actively harmful here, because it hands people a vocabulary in which the problem is their nervous system rather than someone else's conduct, and that framing keeps people in these relationships for years.
When the label is being used against someone. A score on a questionnaire is not a diagnosis, anxious attachment appears in no diagnostic manual, and no page is in a position to tell anyone that their account of their own relationship is a symptom. If a term like this has been handed to you as the reason your perception should not be trusted, the term is not what is happening.
And it runs the other way too, which is worth naming plainly rather than leaving as an implication. Yuji Kanemasa, Yuki Miyagawa and Takashi Arai followed people across four waves in a year and analysed the 356 whose partners left them. Attachment anxiety at the first wave predicted stalking behaviours after the breakup, and the path ran through psychological intimate partner violence perpetrated during the relationship, as well as through post-breakup anger and rumination. The honest qualifier is in the same abstract: the direct correlation between attachment anxiety and later stalking was not significant, and these are indirect effects in a structural model, so what the study supports is a route rather than a prediction. Checking a partner's phone, demanding their location, requiring an account of their time: those are controlling behaviours whatever is driving them, and insecurity is an explanation rather than a permission. The distinction between a feeling and a behaviour, and where the line sits, is worked through at length in the psychology of jealousy, and the research behind isolation as a measured dimension of maltreatment is in what the evidence on hidden relationships shows.
If any of this is live rather than interesting, help lines come first. 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, around the clock.
What none of this can tell you
The limits belong on the page, because the likeliest use of a page like this is to get an answer it cannot give.
It cannot tell you that reassurance is pointless. The concurrent effect is the best-established finding here — it held across five waves in 137 couples, and it is visible in behavioural observation studies of couples in real conflict. What has not been found is a longitudinal one, which is a different claim.
It cannot promise the self-efficacy route. That result sits at p = .043, in one sample, during one life transition, measured by one twelve-item scale, with a design its own authors say cannot establish cause. It is the most useful thing on this page and it is one finding.
No experiment has manipulated reassurance over time. Nobody has assigned partners to reassure more or less and measured attachment anxiety months later. Everything in the “does it help in the long run” column is observational, and the one causal literature — security priming — tests a laboratory procedure rather than a person.
Nobody has shown that a secure partner makes an insecure person secure. The ingredients have been measured piecemeal. The study has not been run. We said so on the page about whether attachment styles change before this page existed, and nothing published since has changed it.
The samples are narrow. The two-year study recruited from childbirth classes, was 82% white, 95% married, and its data collection began in 2002. The 2014 study is 134 committed couples. The observation studies are from New Zealand and the United States, heavily student and community samples, and mostly mixed-sex. The one durable experimental effect was demonstrated on self-esteem rather than attachment.
And it cannot tell you about your partner. Every measure here is self-report, and the one direct comparison available found that a partner's own report of providing reassurance predicted nothing about the other person's anxiety.
What survives is small, specific and more use than a list of ten tips. Reassurance reliably helps in the moment and has not been shown to change the underlying pattern. The things associated with the pattern changing are being backed in something of your own and coming to feel competent at something that matters. What the giver reports giving is not the variable; what the receiver perceives is. The support has a measured cost to the person supplying it, and so does the hurt that induces it, and a relationship can run on that transfer for a long time before anyone names it. More of this under Mindset, and the claims this site has had to retract are listed on the corrections page.
Pulling it together
The short version is smaller than twenty sections and more use than ten tips.
Reassurance works in the moment, reliably, and it has never been shown to lower attachment anxiety over months. Those are two findings, not a contradiction, and the people who established the second are explicit that the first still matters. What has been associated with the pattern itself shifting is narrower and less obvious: being backed in your own pursuits, and coming to feel competent at something that matters to you — the second of those at a marginal p-value, in one sample, during one life transition.
The finding that should change the question is the quietest one. A partner's own report of how much reassurance they provided predicted nothing about the other person's anxiety, concurrently or six months later. Only the anxious person's perception did. That is not a reason to try harder; it is a reason to ask rather than to perform.
And the cost is real on both sides. Partners of chronically insecure people conceal irritation and inflate affection, which does make the insecure person feel valued and does predict the partner's own satisfaction falling. Anxious people's expressions of hurt induce guilt, which does stabilise their sense of the relationship and does predict their partner's satisfaction falling. Two labs, two behaviours, the same transfer. Knowing that it is a transfer rather than a character flaw is most of what this research has to offer, because it is the only version of the situation in which both people are doing something describable rather than something wrong.
Frequently asked questions
How do you reassure an anxious attachment partner?
The behaviours that have been measured are proximity and sensitivity: being glad to give a hug when one is wanted, moving closer rather than away when the other person is upset, and noticing that comforting is needed before being asked. Those are the items on the scale Arriaga and colleagues used with 137 couples, and perceiving more of them was strongly associated with lower attachment anxiety at the same moment, b = -0.06, p < .001. Two things complicate the instruction. The same study found that the partner's own report of providing reassurance predicted nothing about the other person's anxiety — only the anxious person's perception of receiving it did. And the behavioural observation work suggests what lands is less a script than a demonstration of commitment: in Tran and Simpson's study of married couples in conflict, anxious individuals whose partners were more committed, and therefore inhibited retaliation and worked at the problem, felt more accepted and behaved about as constructively as secure individuals did.
Does reassurance make anxious attachment worse?
Nobody has shown that, and the claim circulates without evidence. What has been shown is narrower. Over five waves and two years, perceived reassurance was strongly associated with lower attachment anxiety within a wave and had a longitudinal coefficient of 0.00, 95% CI [-0.03, 0.03], p = .849 — no effect in either direction. Separately, Lemay and Clark's five studies describe a route by which expressing insecurity leads people to believe they are seen as fragile, which leads them to doubt the authenticity of a partner's expressions of regard; and Simpson and Overall note that exaggerated or prolonged reassurance can be read as insincere and then fail to quell anything. So reassurance can stop landing, for reasons that have been mapped. Deliberately withholding it has not been tested by anyone, and it should not be presented as the finding.
What actually reduces attachment anxiety over time?
In the only study that tested both things in the same sample, it was self-efficacy. Across 137 couples and two years, people who derived more personal competence from their new parenting role showed declines in attachment anxiety six months later, b = -0.09, 95% CI [-0.16, -0.01], p = .043, while perceived partner reassurance predicted nothing across the same lags. Arriaga's earlier study of 134 committed couples points the same way from a different angle: perceived goal validation — feeling that a partner encourages your own pursuits — uniquely predicted reduced attachment anxiety over time, while trust predicted reduced avoidance. Both results are observational and the authors say plainly that they cannot establish cause. The broader question of how far an attachment pattern moves at all, and over what timescale, has stability coefficients attached to it and they are set out on our page about whether attachment styles change.
What is the hardest attachment style to love?
The question assumes four categories, and taxometric analyses of attachment data have come out dimensional rather than categorical every time they have been run. What exists is two continuous dimensions, anxiety and avoidance, and everyone has a position on both. On the nearest answerable question, Hadden, Smith and Webster meta-analysed the links between each insecure dimension and relationship satisfaction and commitment, and found both negative — and more negative in samples of longer relationships, a comparison across cross-sectional samples rather than a trajectory within couples. Neither dimension comes out as the hard one. The dyadic research makes the ranking question the wrong level anyway: its founding observation is that insecurely attached people have less happy relationships but that the quality of those relationships depends on how their partners respond.
Can a secure partner heal an anxious partner?
Nobody has demonstrated it. It is a plausible mechanism and the ingredients have been measured piecemeal — partners can demonstrably down-regulate an anxious person's distress in the moment, and committed partners' accommodation during conflict produces measurably better behaviour — but the study showing that having a secure partner makes an insecure person chronically secure has not been run. The closest longitudinal evidence points somewhere adjacent: what predicted declines in attachment anxiety was the anxious person's own sense of competence in a new role, not the partner's reassurance and not the partner's own self-efficacy, which was tested and did not predict it. Treating a relationship as a treatment also has a known failure mode, described by Eller and Simpson: a person can come to rely on a partner as an external emotion regulator without developing their own regulation.
Is it exhausting to support a partner with anxious attachment?
There is a measurement of this, and the answer is that the exhaustion is documented rather than imagined. Lemay and Dudley found across three studies that people who detect a partner's chronic insecurity become vigilant about upsetting them and respond with affective exaggeration — inflating positive feelings and concealing negative ones. It worked on the recipient, who felt more valued. It also predicted the exaggerator's own relationship satisfaction declining, in both studies that measured it. Overall and colleagues found the mirror image: anxious people's expressions of hurt induced guilt in partners, which stabilised the anxious person's view of the relationship and was accompanied by significant declines in the partner's satisfaction. Eller and Simpson add that partners who consistently give without reciprocity may come to feel underappreciated or burn out, and that no guidelines exist for how to pace any of it.
What should you not say to an anxiously attached partner?
There is no tested list, and anyone offering one has made it up. Two things do have evidence behind them. Overall, Fletcher, Simpson and Sibley found with 61 couples that soft, indirect, keep-it-pleasant approaches to a problem felt the most successful immediately and predicted no change at all over the following twelve months, while direct approaches felt unsuccessful and predicted measured change — so the most pleasant thing to say is not reliably the most useful. And Jayamaha, Girme and Overall found that support providers who felt unvalued during a discussion behaved more critically, and that critical support behaviour predicted declines in the recipient's relationship quality over time. If anything generalises from that, it is less about phrasing than about not running the conversation while feeling accused.
Why doesn't reassurance seem to stick?
One mechanism has been mapped in detail. Lemay and Clark's five studies describe a loop: people who are insecure about acceptance express their vulnerabilities, which leads them to believe the partner sees them as highly vulnerable, which leads them to doubt whether the partner's expressions of positive regard are authentic. Those doubts predicted perceived rejection, which predicted derogating the partner and expressing more vulnerability. The final study made it causal by manipulating how vulnerable a person believed they were seen to be, which increased their doubts about a new acquaintance's expressed emotion. The practical implication is uncomfortable for the standard advice, which is to name your triggers so your partner can reassure you: the measured problem is being seen as fragile rather than disclosure itself. There is also one experiment on the receiving end that worked — Marigold, Holmes and Ross got people with low self-esteem to describe the meaning and significance of a partner's compliment rather than its details, and they felt as positively about it as high-self-esteem people did, two weeks later. That was self-esteem, not attachment anxiety, and nobody has repeated it with the attachment measure.
Is constantly needing reassurance a sign of something else?
It can be, and the relevant literature is not the attachment one. Joiner and Metalsky ran six studies on excessive reassurance seeking and found it to be a coherent construct showing diagnostic specificity to depression in a clinically diagnosed sample. In their prospective study, people who were symptom-free at baseline but later developed depressive symptoms had elevated reassurance-seeking scores while still symptom-free, and were not elevated on other interpersonal measures. Two later studies found it predicted depressive reactions to stress. None of that is a diagnosis, and a research scale is not an assessment. What it supports is a practical point: if the frequency of asking is the thing dominating a relationship, more reading about attachment styles is unlikely to be the highest-value next step, and a qualified person is.
When is anxious attachment not the right explanation?
When the fear is accurate, and when the label is being used as an argument. If someone is anxious because a partner monitors their phone or location, decides who they may see, controls money or transport, or escalates when they say no, that is coercive control and attachment language makes it worse by relocating the problem into the anxious person's nervous system. And a questionnaire score is not a diagnosis: anxious attachment appears in no diagnostic manual, and if the term has been handed to someone as the reason their account of their own relationship should not be trusted, the term is not what is happening. It runs in the other direction too. Kanemasa, Miyagawa and Arai followed 356 people whose partners left them across a year and found that attachment anxiety predicted post-breakup stalking indirectly, through psychological intimate partner violence perpetrated during the relationship and through post-breakup anger and rumination — the direct correlation was not significant, so what they describe is a route rather than a prediction. Insecurity explains controlling behaviour; it does not license it.
Sources
What this article draws on. For the broader bibliography see our sources page.
- Arriaga, X. B., Eller, J., Kumashiro, M., Rholes, W. S. & Simpson, J. A. (2021) — Self-efficacy and declines over time in attachment anxiety during the transition to parenthood. Social Psychological and Personality Science, 12(5), 658-666 — 137 couples across five waves and two years; the concurrent and lagged coefficients, the cross-partner null and the ruled-out alternatives are read from the authors' open preprint
- Arriaga, X. B., Kumashiro, M., Finkel, E. J., VanderDrift, L. E. & Luchies, L. B. (2014) — Filling the void: Bolstering attachment security in committed relationships. Social Psychological and Personality Science, 5(4), 398-406 — 134 committed couples; trust and goal validation swap places between the concurrent and longitudinal analyses
- Arriaga, X. B., Kumashiro, M., Simpson, J. A. & Overall, N. C. (2018) — Revising working models across time: Relationship situations that enhance attachment security. Personality and Social Psychology Review, 22(1), 71-96 — the Attachment Security Enhancement Model, which separates buffering momentary insecurity from revising the underlying model
- Simpson, J. A. & Overall, N. C. (2014) — Partner buffering of attachment insecurity. Current Directions in Psychological Science, 23(1), 54-59 — the Dyadic Regulation Model of Insecurity Buffering; open access, and the source for the caution that exaggerated or prolonged buffering reads as insincere
- Overall, N. C. & Simpson, J. A. (2015) — Attachment and dyadic regulation processes. Current Opinion in Psychology, 1, 61-66 — open access review of which partner responses assuage which insecurity, and of the transition-to-parenthood findings
- Eller, J. & Simpson, J. A. (2020) — Theoretical boundary conditions of partner buffering in romantic relationships. International Journal of Environmental Research and Public Health, 17(18), 6880 — open access; motivation, ability and time as limits, the burnout and over-reliance risks, and the absence of any scaffolding guidelines
- Lemay, E. P. & Dudley, K. L. (2011) — Caution: fragile! Regulating the interpersonal security of chronically insecure partners. Journal of Personality and Social Psychology, 100(4), 681-702 — three studies on affective exaggeration; it raised targets' sense of being valued and predicted perceivers' reduced satisfaction
- Lemay, E. P. & Clark, M. S. (2008) — “Walking on eggshells”: How expressing relationship insecurities perpetuates them. Journal of Personality and Social Psychology, 95(2), 420-441 — five studies, the last experimental; expressing vulnerability predicts doubting the authenticity of a partner's regard
- Overall, N. C., Girme, Y. U., Lemay, E. P. & Hammond, M. D. (2014) — Attachment anxiety and reactions to relationship threat: The benefits and costs of inducing guilt in romantic partners. Journal of Personality and Social Psychology, 106(2), 235-256 — a daily diary and an observed-conflict study; partner guilt stabilised anxious individuals and was accompanied by significant declines in partners' satisfaction
- Overall, N. C., Fletcher, G. J. O., Simpson, J. A. & Sibley, C. G. (2009) — Regulating partners in intimate relationships: The costs and benefits of different communication strategies. Journal of Personality and Social Psychology, 96(3), 620-639 — 61 couples, change rated at three-month intervals for a year; positive-indirect strategies felt successful and predicted nothing
- Tran, S. & Simpson, J. A. (2009) — Prorelationship maintenance behaviors: The joint roles of attachment and commitment. Journal of Personality and Social Psychology, 97(4), 685-698 — videotaped married couples; partner commitment buffered anxious individuals' reactions to threat
- Campbell, L., Simpson, J. A., Boldry, J. & Kashy, D. A. (2005) — Perceptions of conflict and support in romantic relationships: The role of attachment anxiety. Journal of Personality and Social Psychology, 88(3), 510-531 — fourteen days of diaries followed by a videotaped discussion of a real problem
- Salvatore, J. E., Kuo, S. I.-C., Steele, R. D., Simpson, J. A. & Collins, W. A. (2011) — Recovering from conflict in romantic relationships: A developmental perspective. Psychological Science, 22(3), 376-383 — 73 targets studied since birth; a partner who recovered better buffered insecure targets and predicted stability two years on
- Jayamaha, S. D., Girme, Y. U. & Overall, N. C. (2017) — When attachment anxiety impedes support provision: The role of feeling unvalued and unappreciated. Journal of Family Psychology, 31(2), 181-191 — two observational studies; the predicted effects emerged only for male providers in Study 2
- Marigold, D. C., Holmes, J. G. & Ross, M. (2007) — More than words: Reframing compliments from romantic partners fosters security in low self-esteem individuals. Journal of Personality and Social Psychology, 92(2), 232-248 — three studies; the abstract-meaning manipulation was still evident two weeks later, in low self-esteem rather than attachment anxiety
- Park, Y., Impett, E. A., MacDonald, G. & Lemay, E. P. (2019) — Saying “thank you”: Partners' expressions of gratitude protect relationship satisfaction and commitment from the harmful effects of attachment insecurity. Journal of Personality and Social Psychology, 117(4), 773-806 — five studies; their internal meta-analysis suggests the benefit may be specific to avoidance, and perceived rather than reported gratitude did the work
- Overall, N. C., Simpson, J. A. & Struthers, H. (2013) — Buffering attachment-related avoidance: Softening emotional and behavioral defenses during conflict discussions. Journal of Personality and Social Psychology, 104(5), 854-871 — 180 couples, coded every 30 seconds; softening attenuated avoidant anger and withdrawal
- Simpson, J. A., Winterheld, H. A., Rholes, W. S. & Oriña, M. M. (2007) — Working models of attachment and reactions to different forms of caregiving from romantic partners. Journal of Personality and Social Psychology, 93(3), 466-477 — avoidant partners were calmed by instrumental rather than emotional caregiving
- Girme, Y. U., Overall, N. C., Simpson, J. A. & Fletcher, G. J. O. (2015) — “All or nothing”: Attachment avoidance and the curvilinear effects of partner support. Journal of Personality and Social Psychology, 108(3), 450-475 — four studies pooled, N = 298 couples; low-to-moderate practical support made avoidant recipients worse and only high levels helped
- Feeney, B. C. (2004) — A secure base: Responsive support of goal strivings and exploration in adult intimate relationships. Journal of Personality and Social Psychology, 87(5), 631-648 — observational and experimental; the support that predicted recipients' outcomes was responsive and nonintrusive
- Gillath, O., Karantzas, G. C., Romano, D. & Karantzas, K. M. (2022) — Attachment security priming: A meta-analysis. Personality and Social Psychology Review, 26(3), 183-241 — 120 studies, N = 18,949, d = .51; the outcomes are states measured shortly after the prime
- Carnelley, K. B. & Rowe, A. C. (2007) — Repeated priming of attachment security influences later views of self and relationships. Personal Relationships, 14(2), 307-320 — primes on three consecutive days, trait measures two days later; the published abstract reports no sample size and the full text is paywalled
- Carnelley, K. B., Bejinaru, M.-M., Otway, L., Baldwin, D. S. & Rowe, A. C. (2018) — Effects of repeated attachment security priming in outpatients with primary depressive disorders. Journal of Affective Disorders, 234, 201-206 — 48 participants; symptom differences significant at only one time point, with power named as a limitation
- Hadden, B. W., Smith, C. V. & Webster, G. D. (2014) — Relationship duration moderates associations between attachment and relationship quality. Personality and Social Psychology Review, 18(1), 42-58 — meta-analyses of both insecure dimensions against satisfaction and commitment, with duration as the moderator across cross-sectional samples
- Joiner, T. E. & Metalsky, G. I. (2001) — Excessive reassurance seeking: Delineating a risk factor involved in the development of depressive symptoms. Psychological Science, 12(5), 371-378 — six studies, including a prospective test in initially symptom-free participants
- Rholes, W. S., Eller, J., Simpson, J. A. & Arriaga, X. B. (2021) — Support processes predict declines in attachment avoidance across the transition to parenthood. Personality and Social Psychology Bulletin, 47(5), 810-825 — the avoidance half of the same 137-couple dataset; note that this article appears twice in the journal's own record, also at 47(7), 1119-1134
- Kanemasa, Y., Miyagawa, Y. & Arai, T. (2024) — Attachment anxiety and the dark triad increase stalking after breakups of romantic relationships through psychological maltreatment of romantic partners and reactions to breakups. Aggressive Behavior, 50(1) — four waves across a year; the final sample was the 356 participants whose partners left them
- Brennan, K. A., Clark, C. L. & Shaver, P. R. (1998) — Self-report measurement of adult attachment: An integrative overview. In J. A. Simpson & W. S. Rholes (Eds.), Attachment Theory and Close Relationships (pp. 46-76). Guilford — the Experiences in Close Relationships questionnaire and the two-dimension structure used throughout the studies above
- Find A Helpline — Free, confidential helplines by country, for anyone who needs to talk to someone now
- National Domestic Violence Hotline (United States) — 1-800-799-7233, or text START to 88788 — including their list of warning signs of abuse, which covers monitoring, isolation and control of money or transport
Keep reading
- MindsetHow to Reconnect After a Fight: What the Repair Research ShowsResearch says little about how to fight well and a great deal about the hours afterwards. Most advice for those hours is untested or points the wrong way.
- MindsetThe Psychology of Jealousy: The Feeling, the Behavior, and the Line Between ThemThe feeling is ordinary and almost everyone has it. What a minority do with it is a different subject, and the research is clear about where the line falls.
- MindsetLove or Anxious Attachment? How to Tell the DifferenceFrom the inside they feel the same. They come apart in your body, in how long reassurance lasts, and in whether the intensity needs uncertainty to survive.
- SignsFlirting or Just Being Friendly? What the Research Can Actually Tell YouThe behaviours are the same. What differs is who they are aimed at, why, and what the person reading them already wants to see.
- SignsSigns A New Relationship Has Real PotentialSeparating the predictors with longitudinal evidence behind them from the ones that are just folklore.
- DatingDoes Distance Make You Like Someone More? What the Evidence SaysAbsence does raise how much you like someone, by withholding what would revise the estimate. At least a third of long-distance couples split after reuniting.