RESEARCH DOSSIER
Couple Therapy and Relationship Interventions
Abstract. Couple therapy and relationship interventions encompass structured clinical and preventive efforts to improve relationship functioning, address dyadic distress, and support relational safety. This critical narrative synthesis traces systemic and behavioral origins through integrative behavioral and emotionally focused approaches, prevention and web-mediated treatment. It evaluates randomized trials, longitudinal studies, meta-analyses, dyadic assessment methods, and selected clinical guidance, distinguishing within-treatment gain from controlled effects and measured satisfaction from safety or relationship continuation. Evidence supports meaningful average improvements in studied populations, while questions of comparative mechanisms, treatment selection, enduring benefit, representation, coercion, and emerging artificial mediation remain open. The manuscript reports no original experiment or pooled quantitative analysis and treats its clinical findings as research evidence rather than individualized treatment advice.
Keywords: couple therapy; relationship interventions; behavioral couple therapy; integrative behavioral couple therapy; emotionally focused couple therapy; relationship education; dyadic interventions; relationship satisfaction; intimate partner violence; clinical efficacy; randomized controlled trials; meta-analysis; web-based relationship interventions; therapeutic mechanisms; relationship safety; Artificial Era
Scientific Scope of Couple Therapy and Relationship Interventions
Couple therapy and relationship interventions comprise structured efforts to change relationship functioning through work with two partners, their patterned interactions, or the shared environment in which their relationship develops. The unit of explanation is dyadic: one partner's behavior changes the contingencies experienced by the other, and their reciprocal responses may sustain support, conflict, withdrawal, or repair. Clinical couple therapy generally addresses established distress or a related disorder, whereas relationship education and prevention may recruit couples before severe distress develops. These categories overlap in technique and differ in recruitment, severity, aims, and clinical responsibility Roddy et al. (2020) Hawkins et al. (2008).
The object of scientific assessment is broader than a single satisfaction score. It includes perceived intimacy, trust, conflict and repair, observable communication, distress in either partner, emotional safety, relationship continuation, and the costs of change. Improved relationship satisfaction is an important endpoint, but continuation can be harmful in coercive relationships, and ending a relationship may sometimes represent a desirable outcome. A defensible effectiveness claim identifies the population, comparison condition, time horizon, unit of randomization, method of outcome measurement, and safety monitoring before interpreting the size or meaning of benefit Karakurt et al. (2016) Alexander & Johnson (2023).
This monograph undertakes a critical narrative synthesis of original randomized studies, longitudinal follow-ups, observational research, meta-analyses, systematic reviews, and formal practice guidance. It does not claim a registered search protocol, exhaustive retrieval, individual participant data, quantitative pooling, or newly collected clinical observations. The research questions concern conceptual boundaries, historical development, explanatory mechanisms, credible comparative outcomes, partner interdependence, prevention, distribution of benefit, safety, and durability. Citations identify the actual studies and authoritative documents on which each assessment rests, and evidence limitations are treated as part of the scientific result Christensen et al. (2010) Barbato et al. (2018).
The study is situated primarily within relationship and family psychology and counseling psychology and psychotherapy, with clinical psychology contributing diagnostic and safety considerations. Its neighbor, research on couple communication and repair, investigates the interaction processes themselves; this dossier instead examines deliberate attempts to alter those processes and the evidence that intervention produces benefits. Parent training, family-based child interventions, and social support outside an intimate dyad remain distinct research objects. The monograph likewise separates the evaluation of a treatment model from general descriptive claims about how relationships operate Karney & Bradbury (1995) Cook & Kenny (2005).
Historical Formation of the Clinical Field
Early marriage counseling, psychoanalytic work with partners, and the family-therapy movement furnished several distinct traditions from which contemporary couple treatment developed. Gurman and Fraenkel's historical review describes successive periods involving marriage counseling, psychoanalytic experimentation, incorporation of systems concepts, and later differentiation and integration of treatment approaches. This history is useful because the object of intervention changed: clinicians moved from advising one spouse or analyzing separate individuals toward examining the ongoing relationship as a target of structured psychological work Gurman & Fraenkel (2002).
Systemic formulations brought recursive interaction, boundaries, and patterns of response into clinical attention. Minuchin's structural family therapy elaborated organization and boundaries within families, while Bowen's family-systems work emphasized intergenerational emotional processes. Those traditions shaped couple practice, although their conceptual influence cannot substitute for controlled demonstrations of therapeutic efficacy. A theoretical claim that dyadic behavior is circular does not by itself identify which intervention component will produce a measurable improvement or whether that improvement outlasts clinical contact Minuchin (1974) Gurman & Fraenkel (2002).
The behavioral tradition strengthened specification of procedures and outcomes through observable exchanges, reinforcement, behavioral rehearsal, communication instruction, and problem solving. Subsequent integrative behavioral couple therapy broadened the treatment target to include acceptance and patterns of polarization around persistent differences. A large randomized comparison of traditional and integrative behavioral treatments thereby made it possible to examine whether an intervention's theoretical distinction predicts a clinically important advantage, rather than resting on an attractive account of what distressed couples need Christensen et al. (2004) Christensen et al. (2006).
Later developments emphasized emotional responsiveness, attachment-related vulnerability, prevention, and remote access. Emotionally focused couple therapy formulated relational distress partly in terms of negative cycles that organize attachment fears and responses; relationship education attempted earlier intervention with lower-intensity services; online programs reduced some geographical and scheduling barriers. Their arrival expanded the set of interventions and outcomes under study. Historical chronological succession should not be mistaken for a linear ranking of scientific validity: each tradition requires its own trials, implementation studies, and independent outcome assessment Rathgeber et al. (2019) Doss et al. (2016) Hawkins et al. (2008).
Defining Relationship Distress and Therapeutic Change
Relationship distress is a description of compromised relationship functioning rather than a unitary psychiatric diagnosis. Partners may report persistent dissatisfaction, hostile or avoidant communication, diminished closeness, sexual difficulties, infidelity, inequity, or disagreement over expectations. The same observed behavior may have different meanings depending on threat, health, finances, caregiving demands, or a partner's history. The measurement task is to distinguish global satisfaction from specific domains and to determine whether both partners recognize the same problem. Their reports are correlated observations, but they are rarely interchangeable Funk & Rogge (2007) Spanier (1976).
An intervention can be considered successful along several dimensions. Symptom reduction concerns within-person distress; relationship improvement concerns the dyad; process change concerns communication or attachment behavior; and functional change concerns the ability to make decisions and manage daily demands. Trial conclusions must distinguish statistically reliable average differences from clinically meaningful change for particular couples. For example, a within-treatment increase may reflect regression to the mean, contextual improvement, or intervention effects; controlled differences and follow-up provide more discriminating evidence Roddy et al. (2020) Christensen et al. (2010).
Relationship stability is especially difficult to interpret. Remaining together is not synonymous with psychological health, and separation following therapy does not necessarily indicate clinical failure. Some programs are intended to support a considered decision about relationship continuation; others pursue improvement among couples who wish to remain together. Retention in treatment, dropout from measurement, and relationship dissolution can become entangled if a study treats absent follow-up data as simple missingness. Every outcome analysis needs a stated estimand and a transparent account of what happened to couples who separated Christensen et al. (2010) Atkins et al. (2010).
The vocabulary of change should therefore remain operational. Satisfaction is a measured evaluation, communication is a pattern of exchange that can be observed or self-reported, intimacy is a multidimensional experience of connection, and safety concerns freedom from intimidation and harm. No one score proves all of them improved. A scientifically useful dossier traces how interventions target these constructs, how studies assess them, and which levels of inference remain warranted when only one type of outcome is reported Baucom et al. (2011) Karakurt et al. (2016).
The Dyad as a Causal and Statistical Unit
A couple supplies two interacting sources of outcome information and one shared treatment context. Partners can respond differently to the same session, and a change in one partner may alter both the other's exposure to stress and the meaning of subsequently reported satisfaction. The actor–partner interdependence model separates associations of a person's characteristics with their own outcomes from associations with the partner's outcomes. It handles nonindependence that invalidates naive analyses treating partners as unrelated individuals, although actor and partner paths alone do not establish causation Cook & Kenny (2005) Cook & Snyder (2005).
Couple trials ordinarily randomize the dyad or the therapeutic service rather than independently assigning each partner to unrelated interventions. Statistical analysis should respect clustering within couples and, in group formats, additional clustering by group or clinician. The choice of estimand matters: average improvement per person can conceal a pattern in which one partner improves and the other deteriorates. Reporting separate partner outcomes, couple-level agreement, and asymmetry of change provides information unavailable from a pooled mean Cook & Snyder (2005) Doss et al. (2020).
Temporal ordering is necessary for evaluating proposed mechanisms. If improved communication and satisfaction are measured at the same visit, their correlation may represent reverse causation or a common change in mood. Repeated assessments, lagged dyadic models, experimentally manipulated modules, and independent behavioral coding permit stronger inference, provided measurement intervals match the hypothesized process. Cook and Kenny explicitly distinguish predictive actor–partner relations from causal effects, an important warning for interpreting seemingly reciprocal pathways in treatment process research Cook & Kenny (2005) Le et al. (2020).
Research design must also consider selection into couple therapy. Couples who volunteer jointly may be more committed, safer, better resourced, or more available than those excluded by recruitment requirements. Therapists can alter inclusion by insisting on both partners' consent, while follow-up samples often overrepresent couples who remain reachable. Randomization strengthens the causal comparison within the enrolled population but cannot erase these limits to generalization. The core scientific challenge is to reconcile internal validity with the breadth of relationships that service systems encounter Roddy et al. (2020) Doss et al. (2016).
Conceptual Models of Relationship Maintenance and Vulnerability
The vulnerability–stress–adaptation framework situates relationship development at the intersection of relatively enduring vulnerabilities, stressful events, and adaptive processes. Karney and Bradbury assembled longitudinal evidence to argue that relationship quality and stability need to be investigated as changing trajectories rather than explained by a single cross-sectional association. For treatment research this framework suggests plausible moderators: a communication procedure may be harder to implement when chronic stress is intense, and apparent relapse may follow a new external demand rather than a simple disappearance of learned skills Karney & Bradbury (1995).
Behavioral accounts emphasize reinforcement, punishment, coercive escalation, and expectancies linked to repeated exchange. Cognitive accounts examine attribution of motives and the interpretations that transform ordinary disagreement into perceived rejection. Attachment-oriented explanations focus on the emotional meaning of availability, threat, and responsiveness; systemic explanations examine mutual regulation and higher-order patterns. These models can converge on a specific observable sequence while disagreeing about the best point of intervention. Comparative trials must therefore test mechanisms, not simply label a treatment with its parent theory Christensen et al. (2004) Spengler et al. (2024).
An important distinction separates relationship risk factors from targets that are readily modifiable. Economic hardship, serious health events, discrimination, and caregiving strain may influence outcomes even when therapy improves particular interactions. Attempts to assign all causality to partners' communication exaggerate their control over structural circumstances. Conversely, treating external stress as determinative would miss evidence that changed patterns of support or problem solving can improve functioning under difficult conditions. Research requires models that incorporate between-couple differences and within-couple change Karney & Bradbury (1995) Doss et al. (2020).
Clinical formulations should be treated as working hypotheses subjected to feedback. A therapist may conceptualize withdrawal as an avoidance strategy, an emotional protection response, an understandable reaction to coercion, or simply fatigue. These interpretations carry different ethical and empirical consequences. The appropriate research question is which interpretation predicts reproducible outcomes under specified conditions and whether treatment benefits each participant. That standard keeps theoretical plurality productive while preserving safety and accountability in relationship intervention science Alexander & Johnson (2023) Christensen et al. (2010).
Traditional Behavioral Couple Therapy and Skills-Based Change
Traditional behavioral couple therapy developed an intervention logic around the functional analysis of exchanges between partners. The clinician identifies patterns that maintain distress, examines reinforcement and avoidance, and encourages behavior that supports cooperation and reduces predictable escalation. Structured communication and problem-solving exercises offer observable targets for sessions and homework. Their appeal as treatment components is methodological as well as clinical: an investigator can code behavior, assess adherence, and compare change to a control condition. However, a procedure's theoretical plausibility does not prove that a particular skill is the necessary mediator of improved satisfaction Christensen et al. (2004) Baucom et al. (2011).
Behavioral techniques commonly address the context in which conflict begins, the sequence of responses it evokes, and the possibilities for constructive exchange. An explicit request may be easier to evaluate than a global accusation; a negotiated task may reduce repetitive blame; positive shared activity can generate experiences against which older expectations are reconsidered. These examples describe hypothesized functional pathways, not guaranteed individual effects. The research problem is to estimate whether skill use increases in daily life and whether measured improvements persist once therapist prompting ceases Christensen et al. (2006) Roddy et al. (2020).
In Christensen and colleagues' randomized investigation, 134 seriously distressed married couples received traditional or integrative behavioral couple therapy. The trial established a demanding comparison between two active programs rather than simply testing treatment against waiting. Differences at particular follow-up points should be read within the full longitudinal record, not taken as permanent superiority. The five-year data eventually showed substantial improvements in both conditions and no significant enduring advantage in satisfaction of one over the other Christensen et al. (2004) Christensen et al. (2010).
One limitation of a skills-only interpretation is that conflict content and stress exposure often remain after competent instruction. Partners can enact a practiced formula while experiencing persistent rejection, fear, or resentment, and observable positivity may be shaped by the artificial conditions of a clinic or laboratory discussion. This does not render behavioral methods unimportant; it requires investigators to examine whether change occurs through skill acquisition, new interpretation, changed exposure to aversive exchanges, or a more complex combination. Comparative mechanism analysis remains essential Baucom et al. (2011) Karney & Bradbury (1995).
Integrative Behavioral Couple Therapy and Acceptance Processes
Integrative behavioral couple therapy reframes recurrent differences as patterns that may be modified by combining behavioral change with acceptance-oriented responses. In this approach, the therapeutic question includes how partners become polarized around durable incompatibilities, attributes, and mutually escalating defenses. Acceptance is operationally relevant when blame, withdrawal, and reactive attempts to change a partner are reduced while communication becomes more flexible. It is not a synonym for tolerating intimidation or relinquishing personal boundaries; safety concerns require a different clinical appraisal Christensen et al. (2004) Karakurt et al. (2016).
Techniques associated with integrative treatment include observing the interaction sequence, identifying contextual triggers, facilitating empathic understanding, and developing a less polarized account of conflict. The hypothesis is that altering the function of behavior within an interaction may produce benefits even when the substantive disagreement remains. This claim is empirically testable through observed emotional responses, changes in hostile exchange, and subsequent satisfaction trajectories. Trials need to distinguish the proposed process from nonspecific factors such as therapeutic alliance, attention, expectancy, and the joint decision to seek help Christensen et al. (2006) Baucom et al. (2011).
The randomized follow-up reported by Christensen and colleagues provides a useful model of longitudinal restraint. At treatment completion, within-group effect sizes for marital satisfaction were 0.90 in integrative therapy and 0.71 in traditional therapy; the difference was not statistically significant. An earlier two-year advantage for the integrative condition subsequently narrowed, and five-year effect sizes relative to pretreatment were 1.03 and 0.92. Clinically significant improvement was observed in 50.0% and 45.9% of the respective couples, indicating both substantial gains and persistent unmet need Christensen et al. (2010).
Predictor analyses from the same cohort examined commitment, marital duration, closeness, and other characteristics as correlates of long-term outcomes. Such analyses are informative about heterogeneity but can become unstable with modest samples, numerous candidate predictors, and subgroup comparisons. Their findings generate hypotheses for stratified future trials rather than rules permitting exact prediction for a particular couple. A serious comparative theory needs independent replication, prospective specification of moderators, and assessment of the potential cost of assigning couples to different modalities Baucom et al. (2015) Christensen et al. (2010).
Emotionally Focused Couple Therapy and Attachment-Oriented Change
Emotionally focused couple therapy conceptualizes a recurring interaction cycle as an organized response to vulnerability, unmet needs, and fears concerning emotional accessibility. Treatment typically encourages recognition of escalation, communication of primary emotional experience, and more responsive exchanges between partners. The attachment vocabulary supplies a theory of why a criticism–withdrawal sequence can acquire meanings beyond the immediate disagreement. These claims concern an intervention model; whether particular emotional enactments are active causal ingredients requires process studies and dismantling designs Spengler et al. (2024) Rathgeber et al. (2019).
The randomized-trial-focused synthesis by Rathgeber and colleagues brought behavioral and emotionally focused treatments into one comparative review. It included 33 suitable primary studies and 2,730 participants, and estimated an overall posttreatment effect of g = 0.60. Its analysis found no statistically significant between-modality difference, notwithstanding numerical differences in some subgroup estimates. The authors cautioned about potential publication bias, and their later follow-up results should not be interpreted as demonstrating that all gains persist indefinitely across treatment schools Rathgeber et al. (2019).
A more recent meta-analysis focused specifically on emotionally focused couple therapy reviewed 20 studies comprising 332 couples. Spengler and colleagues reported a weighted pretest–posttest d of 0.93 and a comparison with viable alternative interventions of d = 0.44, with confidence intervals that differed in precision. Because the included literature mixed randomized and other designs, the magnitude of within-treatment gain cannot be presented as the causal effect attributable to the model. Interpretation also depends on study size, therapist expertise, allegiance, and treatment fidelity Spengler et al. (2024).
Emotional responsiveness may be both a treatment target and a global correlate of perceived satisfaction. When relationship satisfaction improves, partners can retrospectively appraise each other as more responsive, creating a plausible reverse causal pathway. Strong tests therefore require independent behavioral coding, temporal separation of change processes and outcomes, preregistered hypotheses, and adequate representation of relationships beyond the samples on which the original models were tested. The scientific contribution of this school is clearest when its specific mechanisms are measured in a way that alternative accounts can contest Spengler et al. (2024) Baucom et al. (2011).
Systemic, Integrative, and Other Clinical Traditions
Systemic approaches ask how recurrent sequences, implicit rules, and surrounding family networks influence the presenting complaint. Structural accounts highlight patterns of boundaries and organization, while intergenerational models emphasize longstanding emotional processes. These perspectives remain influential for case formulation and training, particularly when couple conflict interacts with caregiving or extended-family obligations. Evidence for any named systemic technique must nevertheless be located in trials of that intervention and population; the historic significance of a theory is a different outcome from its verified comparative effectiveness Minuchin (1974) Gurman & Fraenkel (2002).
Integrative psychotherapy raises a classification problem. A treatment can combine behavioral exercises, emotion work, psychoeducation, individual symptom care, and systemic reframing while retaining the label of an established school. Comparisons across labels are therefore potentially misleading when actual session content converges. Useful research documents treatment manuals, fidelity ratings, therapist training, adherence, competence, dosage, and exposure to concurrent treatments. Such records help identify whether two apparently different interventions received meaningfully different active components Roddy et al. (2020) Rathgeber et al. (2019).
Therapist effects and alliance are important alternative explanations in small trials. Couples may benefit from a trustworthy setting, structured attention, and opportunities to hear one another without interruption independently of a model-specific technique. Therapist allegiance may also shape how well a comparison condition is delivered. Researchers should record prior training, therapeutic orientation, supervision, assignment of clinicians to conditions, and whether assessors were blind to allocation. Without these details, superiority claims can accidentally reward unequal implementation rather than the proposed mechanism Spengler et al. (2024) Christensen et al. (2004).
The proper clinical implication of pluralism is comparative rather than encyclopedic: researchers should distinguish the features that reliably travel across approaches from those that uniquely predict outcome. Common relational opportunities may coexist with specialized effects for particular populations. Factorial designs, process-sensitive measurement, pragmatic trials, and independent replication can refine this question more effectively than allegiance-based rankings of schools. For any claimed advantage, the denominator is the population enrolled and the comparator actually tested, not a general impression that one therapeutic philosophy is more sophisticated Roddy et al. (2020) Christensen et al. (2010).
Mechanisms, Mediators, and Treatment Fidelity
An intervention mechanism is a proposed causal process linking delivery of a component to a meaningful outcome. Communication instruction may alter observed conflict behavior; emotion-focused work may change perceived accessibility; shared activity may change opportunities for reinforcement. Each claim becomes scientific when the process is clearly measured and the causal sequence is challenged by an appropriate design. A variable that changes during therapy and correlates with satisfaction is a candidate mediator, but it is not automatically the pathway by which treatment caused improvement Le et al. (2020) Baucom et al. (2011).
Process measures deserve the same psychometric scrutiny as outcomes. Partner reports of improved communication may partly reflect generalized satisfaction, while observer ratings depend on task selection, coder training, and whether behavior under observation resembles interaction at home. In a follow-up study of traditional and integrative behavioral therapy, observed negativity and withdrawal decreased after treatment, but positivity and problem solving did not follow a uniform improvement pattern. That heterogeneity resists simple claims that all dimensions of communication move together Baucom et al. (2011).
Measurement timing affects mediation analysis. If a mechanism is first measured at the end of treatment alongside satisfaction, temporal precedence has not been established. Repeated sampling during sessions and in everyday life offers a richer account of what changes first, but can increase participant burden and missingness. Multilevel methods are needed to separate stable differences between couples from the within-couple fluctuations relevant to mechanism. Random assignment to intervention components or experimentally varied session procedures can strengthen causal interpretation further Cook & Kenny (2005) Le et al. (2020).
Fidelity is simultaneously a scientific variable and an implementation challenge. It includes delivering the intended components, avoiding contamination by unintended procedures, and providing sufficient competence for a fair test. High fidelity may improve outcomes, yet correlations between fidelity and outcome may also reflect therapist ability or easier cases. A rigorous program examines these possibilities prospectively and remains open to the result that a simpler, scalable intervention produces comparable benefit. Treatment theory ultimately depends on replicable component–process–outcome relationships, not only statistically significant total scores Spengler et al. (2024) Doss et al. (2020).
Relationship Education and Prevention Before Severe Distress
Relationship education targets skills and expectations in couples who may not present with severe clinical distress. Programs often teach communication, conflict management, clarification of expectations, and shared problem solving through workshops, group sessions, or guided materials. Their logic is preventive: modifying relational processes before escalation might reduce later distress. Yet prevention is difficult to evaluate because many enrolled couples would have remained well without intervention, early improvements may fade, and the outcomes most important to policymakers—violence, dissolution, and durable well-being—require longer and more expensive follow-up Hawkins et al. (2008) Blanchard et al. (2009).
Hawkins and colleagues synthesized 86 codable reports yielding 117 studies and more than 500 effect sizes in a meta-analysis of marriage and relationship education. For experimental studies they reported relationship-quality effects around d = 0.30–0.36 and communication-skill effects around d = 0.43–0.45. These are informative summaries of a heterogeneous literature rather than an expectation to be assigned to every program or population. Differences in dosage, assessment type, publication status, and recruitment can explain why average gains do not settle program selection Hawkins et al. (2008).
A second synthesis examined communication outcomes across 97 research reports and 143 distinct evaluations. The findings suggested that education can serve universal and selective prevention, while benefits varied with initial distress and assessment interval. Observational measures tended to yield larger communication effects than self-report, a difference that requires careful interpretation because the measurement methods may capture partially distinct behaviors. Demonstrating communication improvement still leaves unanswered whether later mental health, stability, safety, or parenting outcomes improve because of it Blanchard et al. (2009).
Prevention research must treat resources and social structure as design variables. Couples differ in work schedules, child care, language, access to services, legal security, and exposure to chronic hardship. Group averages from relatively homogeneous samples cannot establish equitable effectiveness for families facing different constraints. Programs should measure reach, attendance, differential dropout, fidelity under ordinary service conditions, and longer-term functional outcomes. The original education meta-analysis expressly identified limited racial, ethnic, and economic diversity and scarce evidence for relationship stability and aggression Hawkins et al. (2008) Karney & Bradbury (1995).
Digital and Guided Online Relationship Programs
Digital delivery separates the physical setting of a relationship intervention from its substantive therapeutic model. Online programs can sequence psychoeducation, structured reflection, partner conversations, and short coaching contacts without requiring regular travel to a clinic. They also change the conditions of exposure: exercises occur in environments that may contain interruptions, coercive monitoring, privacy risks, or uneven technological access. Evaluation should therefore address both therapeutic efficacy and the safety, confidentiality, engagement, and reach of the delivery channel Doss et al. (2016) Doss et al. (2020).
In a randomized trial, Doss and colleagues assigned 300 heterosexual couples to an eight-hour OurRelationship program or a two-month waitlist. The intervention included online activities and brief staff calls. Compared with waiting, the program improved relationship satisfaction with a reported Cohen's d of 0.69 and produced differences in several other relational and individual outcomes. The waitlist design establishes short-term efficacy relative to delayed access, but it cannot by itself establish equivalence to full-length clinician-delivered therapy or resolve how much benefit came from online content versus coaching Doss et al. (2016).
Subsequent follow-up of intervention recipients found maintenance of several outcomes through one year, but a within-treated-group follow-up is weaker evidence about long-term causal superiority than a continued randomized comparison. The relevant question is whether the same trajectories would have occurred without the intervention and whether initial advantage persists under realistic attrition and changing life circumstances. Reporting sustained within-group benefit remains valuable, provided that the absence of a contemporaneous control is acknowledged explicitly Doss et al. (2019) Doss et al. (2016).
Online service models also create opportunities for implementation comparisons: fully self-guided content, supported self-help, structured coaching, synchronous video therapy, and blended clinical care differ in cost, intensity, supervision, and risk. These should be evaluated as distinct packages rather than collapsed under the label of digital treatment. In the Artificial Era, automated dialogue and recommender systems may become parts of such packages; their efficacy and harms require direct testing with both partners and protection against confidential disclosures reaching an unintended party Doss et al. (2020) Karakurt et al. (2016).
Interventions for Couples Facing Economic Constraints
Limited financial resources affect both the prevalence of stressors and the practical possibility of receiving care. Time away from work, transportation, service fees, and childcare responsibilities can reduce participation before therapeutic techniques have an opportunity to help. A program that achieves strong efficacy among highly available volunteers may therefore have low population benefit if access remains unequal. Relationship intervention research must distinguish the magnitude of benefit among participants from the fraction of the eligible population that can feasibly begin and complete an effective service Karney & Bradbury (1995) Doss et al. (2020).
A nationwide randomized trial enrolled 742 low-income couples and compared two supported web programs, OurRelationship and ePREP, with a waitlist. Compared with control couples, those assigned to interventions showed improvement across five reported relationship domains, with a median absolute effect of 0.46 at posttreatment; the advantages were maintained during the subsequent four months. Neither intervention significantly reduced breakup frequency by six-month follow-up. This distinction is central: improved relationship functioning and altered relationship continuation are empirically separate endpoints Doss et al. (2020).
The two programs produced mostly similar results even though one was problem-focused and the other communication-focused. Such convergence is compatible with shared change processes, similar intervention dosage, common coaching factors, or multiple distinct routes to similar endpoints. It does not demonstrate that the programs contain identical active ingredients. Evaluation of negative communication, confidence, problem severity, and emotional support can help specify plausible mediators, but observational associations among improvements remain insufficient for unique causal identification Le et al. (2020) Doss et al. (2020).
Policy translation requires attention to retention and longer-term outcomes. Modest average gains may be important if many couples can access a low-intensity program, but the net effect depends on engagement, adverse events, follow-up duration, and the opportunity cost of alternative services. Structural disadvantage cannot be eliminated through communication training alone. Implementation studies should test whether adaptations in language, schedule, human support, accessibility, and referral pathways preserve benefit and whether participants with high risk receive appropriately differentiated care Doss et al. (2020) Hawkins et al. (2008).
What Preventive Interventions Can and Cannot Demonstrate
The same term improvement is often applied to outcomes with very different causal meaning. A pretest–posttest increase in communication confidence documents change after enrollment. A randomized difference compared with waiting supports a causal effect of access to the tested package during that interval. An active-treatment comparison asks whether a particular program offers incremental benefit over another credible intervention. A long-term follow-up without retained controls examines maintenance among observed participants rather than a persistent counterfactual advantage. These estimands must remain separate in cumulative synthesis Roddy et al. (2020) Doss et al. (2019).
Prevention can also change help-seeking and problem recognition without producing an immediate increase in satisfaction. Structured reflection may reveal previously unrecognized incompatibilities or inequities; among some couples, increased clarity can initially lower an evaluative score. Programs should therefore report multiple outcomes and explicitly state their intended beneficial pathway. The scientific objective is to assess whether participants have greater autonomy, safer interactions, and better functioning, not merely a higher mean score immediately after a workshop Hawkins et al. (2008) Karakurt et al. (2016).
Longer-term prevention outcomes pose substantial threats to interpretation. Relationship breakup changes eligibility for some measures; life transitions alter demand; repeated questionnaires may influence how partners discuss conflict; and selective attrition may remove the most distressed participants. Rigorous trials specify how separation is counted, retain independent follow-up where appropriate, and use sensitivity analyses for missing data. Registries and predeclared outcomes can reduce selective reporting of the most favorable intermediate effect Blanchard et al. (2009) Christensen et al. (2010).
The practical research agenda therefore includes prevention for distinct risk profiles rather than a presumed uniform population. Low-conflict couples, partners entering parenthood, couples coping with a new diagnosis, and people whose relationships involve safety concerns confront different decision problems. A universal skills workshop may benefit some groups while requiring adaptation or exclusion in others. Generalization should be earned through comparative studies with transparent inclusion and exclusion criteria, interpretable comparison conditions, and follow-up long enough to capture meaningful relational trajectories Doss et al. (2020) Karney & Bradbury (1995).
Digital Mediation in the Artificial Era
Digital communication already shapes the environments in which couples negotiate availability, conflict, boundaries, and intimacy. Research on online intervention shows that psychological exercises can be delivered through a technological channel, but that finding does not validate an autonomous artificial clinician or establish that a conversational system can reliably handle intimate risk. The relevant scientific unit is the entire human–technology service arrangement: the partners, the content, human support, consent procedures, data flows, and escalation mechanisms Doss et al. (2016) Doss et al. (2020).
A useful conceptual distinction separates an intervention tool from an additional relational participant. A workbook or scheduling app stores information and organizes tasks; an interactive artificial agent may produce individualized language, remember disclosures, suggest interpretations, and become a third conversational influence. These capacities raise questions about asymmetric disclosure, selective recommendation, emotional attribution, inadvertent alliance with one partner, and how system updates affect continuity. They are investigable hypotheses within human–artificial relations, not clinical benefits that follow automatically from technological novelty Cook & Kenny (2005) Karney & Bradbury (1995).
For automated couple support, an adequate trial would prospectively compare the actual system with appropriate human-supported or noninteractive alternatives, randomize at the dyad level, and record both partners' experiences. Outcome domains should include conflict escalation, perceived manipulation, coercive use of shared data, confidentiality, satisfaction, and agency in addition to engagement. Independent adjudication of adverse events matters because partners may disagree about whether a system's recommendation was helpful or harmful. This design extends existing dyadic methodology to a new intervention substrate Cook & Snyder (2005) Karakurt et al. (2016).
The Artificial Era supplies a horizon for studying new forms of mediated psychological organization while empirical claims remain linked to measured outcomes. A future system's apparent empathy must be evaluated through observable effects and clearly stated operational definitions; user enjoyment alone does not demonstrate safety or therapeutic efficacy. Research should compare system behavior across revisions, detect unequal impacts among partners, and evaluate whether human professional oversight is required for particular risks. Such distinctions preserve the scientific integrity of both relationship psychology and the emerging psychology of artificial systems Doss et al. (2020) Alexander & Johnson (2023).
Couple Therapy in the Context of Depression
Relationship distress and depression can influence one another, but their temporal and causal relationships vary across individuals. Depressive symptoms may impair energy, communication, desire, and participation; chronic conflict can also exacerbate distress and reduce access to support. A couple-focused intervention is therefore sometimes evaluated both for relationship outcomes and for improvement in an identified person's depressive symptoms. Treating those endpoints as separate preserves clarity: a couple may improve its relationship without equivalent change in depression, and individual symptoms can improve while important relational difficulties remain Barbato et al. (2018) NICE (2022).
The Cochrane review by Barbato, D'Avanzo, and Parabiaghi included 14 studies involving 651 participants. Its authors judged evidence for comparable depression outcomes between couple therapy and individual psychotherapy to be low quality, and evidence for some other contrasts to be very low quality. The review suggested potential relationship benefits for distressed couples while stressing the limits of small and heterogeneous trials. It would be scientifically inaccurate to translate these findings into an unconditional claim that conjoint therapy is preferable for depressed patients Barbato et al. (2018).
The United Kingdom's NICE guideline NG222, published in 2022, advises considering behavioral couples therapy for people with depression when relationship difficulties may contribute to depression or partner involvement may assist treatment. Its stated delivery framework refers to behavioral principles and 15–20 sessions over five to six months. A guideline recommendation reflects a health-system judgment using available evidence and service considerations; it is not an original randomized experiment and does not establish the same eligibility criteria for every country or patient NICE (2022).
Clinical assessment must distinguish depression-related risk from relationship distress as such. Severe symptoms, suicidality, coercion, substance use, and other clinical complexities can change the appropriate service setting or sequence of care. This dossier reports research and published guidance rather than providing individualized treatment selection. Future trials should measure both members' mental health, allow for concurrent care, document referral and safety processes, and evaluate whether relational change actually mediates symptom improvement instead of merely occurring alongside it Barbato et al. (2018) Karakurt et al. (2016).
Infidelity, Betrayal, and Relationship Reconstruction
Infidelity introduces questions about trust, disclosure, attachment injury, contested definitions of exclusivity, and the partners' different goals for intervention. Some couples seek reconciliation, others seek clarity concerning separation, and some disagree about whether a relationship can safely continue. These variations affect recruitment and treatment endpoints. A scientifically informed account examines symptom burden and relational functioning separately and considers whether a partner's choice to leave should be counted as treatment failure Atkins et al. (2005) Atkins et al. (2010).
An exploratory analysis of 19 couples reporting infidelity within a randomized trial of 134 distressed couples found greater initial distress among the infidelity subgroup and reported improvement in satisfaction in certain circumstances after disclosure. The small subgroup and post hoc nature of the analysis impose major limits. It was not a dedicated randomized comparison of infidelity-specific interventions, and its patterns cannot support confident predictions about individual decisions or the superiority of a particular repair protocol Atkins et al. (2005) Christensen et al. (2004).
Community-based outcome research also underscores that infidelity is not one homogeneous clinical presentation. The duration of a concealed relationship, discovery circumstances, ongoing contact, co-occurring violence, financial dependence, and conflicting definitions of betrayal create distinct assessment demands. A treatment model can appear effective by recruiting only couples who have already chosen reconciliation. More informative studies would report recruitment denominators, readiness for conjoint work, early discontinuation, and both partners' goals at intake Atkins et al. (2010) Atkins et al. (2005).
Research on repair should operationalize the mechanisms it invokes: restored trust, behavioral consistency, responsive acknowledgment, negotiated boundaries, and reduction in intrusive distress. Their order and interaction are empirical questions. Self-reported forgiveness or relationship continuation alone cannot demonstrate recovery, particularly if a partner experiences pressure to forgive. An ethical research agenda must permit a range of informed outcomes and evaluate emotional and physical safety independently from the ideal of preserving a partnership Atkins et al. (2010) Karakurt et al. (2016).
Intimate Partner Violence, Coercion, and Safety
Intimate partner violence is a critical boundary condition for any relationship intervention. Physical or sexual violence, coercive control, stalking, threats, and intimidation can make equal participation or honest disclosure impossible. A therapist's apparent neutrality may have asymmetric effects when one partner possesses substantially greater power to punish the other after a session. Consequently, research cannot treat ordinary conflict, mutual escalation, and coercive domination as one interchangeable category for recruitment or treatment recommendation Karakurt et al. (2016) Alexander & Johnson (2023).
Karakurt and colleagues' 2016 systematic review and meta-analysis identified six studies evaluating couple treatment in contexts of violence. Its conclusions concerned selected circumstances and were explicitly preliminary; they do not establish safety or efficacy of conjoint therapy across all violent relationships. The limited evidence base and possibility of retaliation require careful screening and credible procedures for responding to risk. A positive average symptom or violence measure in a selected trial cannot license an unscreened intervention in a setting with coercion Karakurt et al. (2016) Stith et al. (2012).
An additional methodological problem concerns typology. Some frameworks distinguish situational violence from coercive control, but a 2023 systematic review of empirical clustering studies found variable solutions and mixed support for influential classification models. Rigid assignment to an assumed subtype may obscure dynamic risk, discrepancies in partner reports, and context-dependent behavior. Investigators should describe the actual criteria and assessment methods used, document risk changes, and test whether subgroup distinctions reliably predict treatment safety across settings Alexander & Johnson (2023).
Research ethics require protected opportunities for disclosure, ongoing assessment of coercion, independent informed consent, and adverse-event reporting that can detect retaliation outside the session. The methodological principle is stronger than a generic assertion of caution: the intervention itself may change exposure to danger, so harms must be measured as outcomes. Specialist safety processes and qualified clinical judgment remain indispensable. A published research monograph can explain these evidentiary requirements without presenting conjoint treatment as an individual recommendation for people experiencing abuse Karakurt et al. (2016) Stith et al. (2012).
Clinical Selection, Consent, and Adverse Outcomes
Participant selection is part of the intervention being evaluated. Trial eligibility often depends on both partners' willingness to attend, relationship status, symptom profile, language, access to a particular clinician, and absence of predefined high-risk conditions. Selection criteria determine the population for which causal inference from randomization is strongest. A sample of mutually consenting couples with manageable conflict cannot be used to estimate benefit for partnerships in which one member cannot speak freely or safely decline participation Roddy et al. (2020) Karakurt et al. (2016).
Consent in conjoint care is necessarily relational while remaining individually grounded. Each partner needs an understandable account of the intervention, confidentiality arrangements, potential disclosure pressures, and available alternatives. Researchers should predefine how disclosures from individual contacts are handled and how outcomes will be collected if partners separate or one withdraws. This matters statistically: a partner may remain in an outcome study after ending therapy, and avoiding such participants can systematically bias estimates toward apparently successful relationships Christensen et al. (2010) Doss et al. (2020).
Adverse outcomes include more than immediate symptom deterioration. Escalation of conflict, increased surveillance, disclosure-related retaliation, pressure to remain together, privacy breaches, and unequal burden can be consequential even when an average satisfaction measure rises. Harm assessment should combine independent partner reporting with clear escalation procedures and adequate follow-up. Small studies are poorly powered to detect uncommon severe harms, making the absence of recorded events particularly weak evidence of universal safety Karakurt et al. (2016) Alexander & Johnson (2023).
For clinical claims, the evidentiary hierarchy must reflect the question at issue. Randomized treatment comparisons answer average efficacy questions within selected groups; independent clinical reviews synthesize the broader record; formal guidelines articulate practice recommendations under their jurisdiction; and expert safety assessment addresses circumstances that individual trials exclude. This monograph treats its clinical discussion as research synthesis. It records no examination of patients, issues no prescription, and does not assign a medical or clinical reviewer's credentials that have not been verified NICE (2022) Barbato et al. (2018).
What Meta-Analyses Establish About Efficacy
Meta-analysis is useful because individual couple-therapy studies vary in size and comparison condition, but synthesis depends on whether apparently separate reports share participants. Roddy and colleagues analyzed 58 studies representing 40 unique samples and 2,092 couples. The distinction between reports and independent samples matters: treating multiple publications from the same trial as independent can overweight one research program. The analysis reported large within-group improvements in relationship satisfaction among treatment recipients and much smaller changes in waitlist groups, while also documenting changes in communication, intimacy, and partner behavior Roddy et al. (2020).
The reported mean within-treatment gain in relationship satisfaction was Hedges' g of 1.12 with 95% confidence interval 0.92–1.31, compared with g of 0.12 and interval −0.04 to 0.29 among waitlisted couples. These statistics describe distinct trajectories; subtraction of two pooled within-group estimates is not automatically equivalent to a directly estimated randomized treatment effect. Their evidence supports the conclusion that evaluated couple interventions often produce meaningful improvement in the studied populations, while uncertainty remains about which components, delivery conditions, and follow-up contexts determine that improvement Roddy et al. (2020).
Rathgeber and colleagues restricted their comparison of behavioral and emotionally focused therapies to randomized trials, obtaining an overall posttest estimate of g = 0.60 across their selected literature. The estimate and the larger pre–post value in Roddy's broader synthesis should not be viewed as contradictory measurements of one parameter. They differ in eligible studies, comparison structure, outcome definitions, and statistical estimands. Similarly, Spengler and colleagues reported a favorable emotionally focused therapy synthesis, but their 332-couple sample and mixture of study designs limit the certainty of a universal comparative ranking Rathgeber et al. (2019) Spengler et al. (2024).
Credible cumulative evaluation asks whether outcomes persist in independent service settings, whether the treatment comparator was active and competent, how attrition was handled, and whether negative trials remain unpublished. Confidence intervals and sensitivity analyses communicate uncertainty in pooled estimates but do not repair bias caused by systematic differences in who enrolls, how treatments are delivered, or which outcomes are reported. Stronger claims arise when several designs converge on a stable functional effect while maintaining transparent distinctions among their populations and purposes Roddy et al. (2020) Barbato et al. (2018).
Measurement, Observer Coding, and Clinical Significance
Research has used the Dyadic Adjustment Scale and later Couples Satisfaction Index to operationalize relationship functioning. Spanier's 1976 work differentiated domains including satisfaction, consensus, cohesion, and affectional expression. Funk and Rogge's later work used item response theory to improve precision of relationship-satisfaction measurement. Scale scores remain reports about experienced relationships rather than direct observations of the entire dyadic system. Their validity depends on language, population, measurement invariance, and how closely the items correspond to the intervention's target Spanier (1976) Funk & Rogge (2007).
Multi-method assessment offers important safeguards. Self-report captures private evaluations and perceptions unavailable to observers, whereas coded interaction tasks can assess expressed negativity, withdrawal, cooperation, and problem solving under standardized conditions. Ecological momentary reports and diary methods can record conflict closer to its occurrence. Each method has limitations: partners can share a response bias, coders may see behavior constrained by the laboratory, and repeated sampling can change the very interaction it measures. Convergence among methods is informative precisely because their error structures differ Baucom et al. (2011) Cook & Snyder (2005).
Change should be evaluated at the individual and couple levels. A dyadic mean may conceal widening disparity between partners, and one partner's satisfaction may increase as the other's options narrow. Clinically significant change classifications can distinguish recovery-like scores, partial improvement, unchanged distress, and deterioration, but their thresholds depend on instrument norms and decisions about reliability. Christensen's follow-up illustrates the value of reporting the proportion of couples with meaningful improvement alongside group-level effect sizes and separation status Christensen et al. (2010).
The statistical model should explicitly handle within-dyad dependence, repeated visits, and missing assessments. Cook and Snyder applied actor–partner reasoning to nonindependent couple outcomes, and later work used intensive longitudinal measurement to explore candidate relational mechanisms. Analysts must predefine whether their primary result is average individual change, change in the couple's shared functioning, the probability both partners benefit, or the probability of a particular clinical transition. Transparent reporting of these estimands would make evidence across therapeutic schools much more comparable Cook & Snyder (2005) Le et al. (2020).
Durability, Separation, and Long-Term Recovery
Maintained treatment gains are a central question because relationship functioning changes under recurring life stress. A single posttreatment score provides a narrow observation during a period of intensive support. Following couples for years requires explicit attention to newly occurring medical events, births, caregiving obligations, financial stress, and relationship transitions. Apparent stability of an aggregate average may conceal strong individual divergence. Longitudinal evidence is therefore indispensable for claims about durable therapeutic effect and still rarely permits simple causal attribution many years after randomization Karney & Bradbury (1995) Christensen et al. (2010).
The five-year comparison between integrative and traditional behavioral couple therapy followed the original 134 randomized distressed married couples. At the final follow-up, 25.7% of integrative-treatment couples and 27.9% of traditional-treatment couples were separated or divorced. Relationship outcomes for those who stayed together were substantial, but excluding separated couples from a satisfaction analysis would change the target population. The trial's later similarity between active treatments also shows why isolated two-year significance should not be rewritten as a durable hierarchy of therapies Christensen et al. (2010).
Analyses of predictors at five years reported relationships involving commitment, duration, and desired closeness, but prediction is a distinct task from estimating a treatment's causal effect. Predictor associations may vary by baseline distress and intervention; the large number of plausible variables raises overfitting concerns. Their clinical usefulness requires performance testing outside the development cohort and assessment of calibration, fairness, and consequences of decisions based on the prediction. Randomized treatment assignment alone does not make a post hoc predictive model externally valid Baucom et al. (2015).
For online interventions, follow-up evidence indicates that some initially observed gains can persist through a year, but these studies need to be interpreted with their available comparator structure and attrition. For prevention programs, the temporal challenge is even greater: short-term communication skills may be meaningful without proving reductions in later dissolution, depression, or violence. The scientific synthesis accordingly separates maintained score improvement, comparative treatment advantage, relationship continuation, and safety over time as four analytically different questions Doss et al. (2019) Hawkins et al. (2008).
Representation, Transportability, and the Next Generation of Trials
Much of the historical couple-therapy evidence was developed among married, different-sex couples who consented jointly to research. Roddy and colleagues expressly noted that the samples in their synthesis comprised opposite-sex couples, limiting claims about other relationship configurations. Similarity of relational processes across groups is a hypothesis that can guide research, but it does not validate automatic transfer of effect sizes. Future trials should recruit diverse sexual orientations, gender identities, relationship structures, cultures, and age groups, and then assess measurement invariance and differential treatment response Roddy et al. (2020).
Treatment accessibility and social position are integral determinants of transportability. Language, insurance coverage, transportation, immigration-related fear, economic strain, disability access, and digital literacy affect who enters a study and how a structured treatment fits ordinary life. Pragmatic trials can estimate effects within real services, while explanatory trials can clarify mechanism under controlled conditions. Both are required. Differential attrition and harms should be disaggregated where sample size permits, rather than concealing heterogeneity behind a reassuring overall effect Doss et al. (2020) Hawkins et al. (2008).
A cumulative program should combine prespecified primary outcomes, independent registration, active comparators, longer follow-up, therapist-level clustering, validated partner-specific measures, and explicit adverse-event procedures. Dismantling and factorial designs can test whether acceptance, emotion, communication instruction, human coaching, or other components are necessary for specific outcomes. Micro-randomized interventions and repeated daily measurement can clarify short-term sequences, while long-range cohorts establish whether those sequences predict enduring functioning. Each method answers a different part of the mechanism question Le et al. (2020) Cook & Kenny (2005).
Future psychological interventions involving artificial systems add version control and data governance to traditional fidelity assessment. An intervention whose dialogue policy changes during a trial is not necessarily the same treatment for all participants. Research should preserve the tested system configuration, audit privacy and discriminatory impacts, record adverse responses from both partners, and compare automated support with appropriate human and nonautomated conditions. Ethical evaluations should also consider whether one partner can use a system to monitor or pressure the other. These are concrete design problems to solve before broad claims of clinical equivalence are warranted Karakurt et al. (2016) Doss et al. (2016).
Conclusion
Couple therapy and relationship interventions form an established but heterogeneous scientific field organized around deliberate attempts to improve relational functioning, reduce clinically relevant distress, and support safer and more effective ways of interacting. Its strongest aggregate evidence documents meaningful average gains among the couples studied in controlled and longitudinal research. Behavioral, integrative behavioral, emotionally focused, preventive, and web-delivered interventions all contribute testable models, although current evidence does not justify a universal hierarchy across modalities or a promise of benefit for every relationship Roddy et al. (2020) Rathgeber et al. (2019) Christensen et al. (2010).
The central explanatory task is to connect components to mechanisms and mechanisms to outcomes across time. Relationship satisfaction, communication, intimacy, symptom change, safety, and stability are related but independent dimensions of evaluation. Trials improve causal inference when they respect dyadic interdependence, use suitable comparison conditions, monitor attrition and adverse events, and measure both partners rather than allowing one voice to represent the couple. Meta-analyses sharpen these conclusions when they preserve the distinction between treatment-group gains and between-group effects Cook & Snyder (2005) Roddy et al. (2020).
Clinical boundaries are essential to the meaning of an intervention. Evidence relating to depression, substance-related disorders, infidelity, and violence arises from different selected populations and study designs. When coercion or risk of retaliation exists, an average positive relationship outcome cannot demonstrate safety. Specialist assessment, informed individual consent, confidentiality arrangements, and clinically appropriate alternatives remain crucial; the present synthesis does not issue individual treatment recommendations. Prevention and online delivery expand access while introducing fresh questions about follow-up, equitable reach, and protection of private information Barbato et al. (2018) Karakurt et al. (2016) Doss et al. (2020).
The continuing research program is to establish durable effects, identify reproducible mechanisms, represent diverse relationships, and develop interventions whose benefits and harms can be independently measured. In the Artificial Era, new tools may become influential participants in the organization of relationship communication, which makes controlled evaluation and accountability more important. The scientific framework remains clear: define the dyad and research object; specify the intervention; measure every participant's experience and safety; test a credible counterfactual; and follow outcomes long enough to distinguish immediate change from durable psychological improvement Karney & Bradbury (1995) Doss et al. (2016).
References
- Alexander, E. F., & Johnson, M. D. (2023). On categorizing intimate partner violence: A systematic review of exploratory clustering and classification studies. Journal of Family Psychology, 37(5), 743–752. https://doi.org/10.1037/fam0001075
- Atkins, D. C., Eldridge, K. A., Baucom, D. H., & Christensen, A. (2005). Infidelity and behavioral couple therapy: Optimism in the face of betrayal. Journal of Consulting and Clinical Psychology, 73(1), 144–150. https://doi.org/10.1037/0022-006X.73.1.144
- Atkins, D. C., Marín, R. A., Lo, T. T. Y., Klann, N., & Hahlweg, K. (2010). Outcomes of couples with infidelity in a community-based sample of couple therapy. Journal of Family Psychology, 24(2), 212–216. https://doi.org/10.1037/a0018789
- Barbato, A., D’Avanzo, B., & Parabiaghi, A. (2018). Couple therapy for depression. Cochrane Database of Systematic Reviews, 2018(6), CD004188. https://doi.org/10.1002/14651858.CD004188.pub3
- Baucom, B. R., Atkins, D. C., Rowe, L. S., Doss, B. D., & Christensen, A. (2015). Prediction of treatment response at 5-year follow-up in a randomized clinical trial of behaviorally based couple therapies. Journal of Consulting and Clinical Psychology, 83(1), 103–114. https://doi.org/10.1037/a0038005
- Baucom, K. J. W., Sevier, M., Eldridge, K. A., Doss, B. D., & Christensen, A. (2011). Observed communication in couples two years after integrative and traditional behavioral couple therapy: Outcome and link with five-year follow-up. Journal of Consulting and Clinical Psychology, 79(5), 565–576. https://doi.org/10.1037/a0025121
- Blanchard, V. L., Hawkins, A. J., Baldwin, S. A., & Fawcett, E. B. (2009). Investigating the effects of marriage and relationship education on couples’ communication skills: A meta-analytic study. Journal of Family Psychology, 23(2), 203–214. https://doi.org/10.1037/a0015211
- Christensen, A., Atkins, D. C., Baucom, B., & Yi, J. (2010). Marital status and satisfaction five years following a randomized clinical trial comparing traditional versus integrative behavioral couple therapy. Journal of Consulting and Clinical Psychology, 78(2), 225–235. https://doi.org/10.1037/a0018132
- Christensen, A., Atkins, D. C., Berns, S., Wheeler, J., Baucom, D. H., & Simpson, L. E. (2004). Traditional versus integrative behavioral couple therapy for significantly and chronically distressed married couples. Journal of Consulting and Clinical Psychology, 72(2), 176–191. https://doi.org/10.1037/0022-006X.72.2.176
- Christensen, A., Atkins, D. C., Yi, J., Baucom, D. H., & George, W. H. (2006). Couple and individual adjustment for 2 years following a randomized clinical trial comparing traditional versus integrative behavioral couple therapy. Journal of Consulting and Clinical Psychology, 74(6), 1180–1191. https://doi.org/10.1037/0022-006X.74.6.1180
- Cook, W. L., & Kenny, D. A. (2005). The actor–partner interdependence model: A model of bidirectional effects in developmental studies. International Journal of Behavioral Development, 29(2), 101–109. https://doi.org/10.1080/01650250444000405
- Cook, W. L., & Snyder, D. K. (2005). Analyzing nonindependent outcomes in couple therapy using the actor-partner interdependence model. Journal of Family Psychology, 19(1), 133–141. https://doi.org/10.1037/0893-3200.19.1.133
- Doss, B. D., Cicila, L. N., Georgia, E. J., Roddy, M. K., Nowlan, K. M., Benson, L. A., & Christensen, A. (2016). A randomized controlled trial of the web-based OurRelationship program: Effects on relationship and individual functioning. Journal of Consulting and Clinical Psychology, 84(4), 285–296. https://doi.org/10.1037/ccp0000063
- Doss, B. D., Knopp, K., Roddy, M. K., Rothman, K., Hatch, S. G., & Rhoades, G. K. (2020). Online programs improve relationship functioning for distressed low-income couples: Results from a nationwide randomized controlled trial. Journal of Consulting and Clinical Psychology, 88(4), 283–294. https://doi.org/10.1037/ccp0000479
- Doss, B. D., Roddy, M. K., Nowlan, K. M., Rothman, K., & Christensen, A. (2019). Maintenance of gains in relationship and individual functioning following the online OurRelationship program. Behavior Therapy, 50(1), 73–86. https://doi.org/10.1016/j.beth.2018.03.011
- Funk, J. L., & Rogge, R. D. (2007). Testing the ruler with item response theory: Increasing precision of measurement for relationship satisfaction with the Couples Satisfaction Index. Journal of Family Psychology, 21(4), 572–583. https://doi.org/10.1037/0893-3200.21.4.572
- Gurman, A. S., & Fraenkel, P. (2002). The history of couple therapy: A millennial review. Family Process, 41(2), 199–260. https://doi.org/10.1111/j.1545-5300.2002.41204.x
- Hawkins, A. J., Blanchard, V. L., Baldwin, S. A., & Fawcett, E. B. (2008). Does marriage and relationship education work? A meta-analytic study. Journal of Consulting and Clinical Psychology, 76(5), 723–734. https://doi.org/10.1037/a0012584
- Karakurt, G., Whiting, K., van Esch, C., Bolen, S. D., & Calabrese, J. R. (2016). Couples therapy for intimate partner violence: A systematic review and meta-analysis. Journal of Marital and Family Therapy, 42(4), 567–583. https://doi.org/10.1111/jmft.12178
- Karney, B. R., & Bradbury, T. N. (1995). The longitudinal course of marital quality and stability: A review of theory, method, and research. Psychological Bulletin, 118(1), 3–34. https://doi.org/10.1037/0033-2909.118.1.3
- Le, Y., Roddy, M. K., Hatch, S. G., & Doss, B. D. (2020). Mechanisms of improvements and maintenance in online relationship programs for distressed low-income couples. Journal of Consulting and Clinical Psychology, 88(12), 1091–1104. https://doi.org/10.1037/ccp0000616
- Minuchin, S. (1974). Families and family therapy. Harvard University Press. https://doi.org/10.4159/9780674041127
- National Institute for Health and Care Excellence. (2022). Depression in adults: Treatment and management (NICE guideline NG222). National Institute for Health and Care Excellence. https://www.nice.org.uk/guidance/ng222/chapter/recommendations
- O’Farrell, T. J., Schumm, J. A., Murphy, M. M., & Muchowski, P. M. (2017). A randomized clinical trial of behavioral couples therapy versus individually-based treatment for drug-abusing women. Journal of Consulting and Clinical Psychology, 85(4), 309–322. https://doi.org/10.1037/ccp0000185
- Rathgeber, M., Bürkner, P.-C., Schiller, E.-M., & Holling, H. (2019). The efficacy of emotionally focused couples therapy and behavioral couples therapy: A meta-analysis. Journal of Marital and Family Therapy, 45(3), 447–463. https://doi.org/10.1111/jmft.12336
- Roddy, M. K., Walsh, L. M., Rothman, K., Hatch, S. G., & Doss, B. D. (2020). Meta-analysis of couple therapy: Effects across outcomes, designs, timeframes, and other moderators. Journal of Consulting and Clinical Psychology, 88(7), 583–596. https://doi.org/10.1037/ccp0000514
- Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and the Family, 38(1), 15–28. https://doi.org/10.2307/350547
- Spengler, P. M., Lee, N. A., Wiebe, S. A., & Wittenborn, A. K. (2024). A comprehensive meta-analysis on the efficacy of emotionally focused couple therapy. Couple and Family Psychology: Research and Practice, 13(2), 81–99. https://doi.org/10.1037/cfp0000233
- Stith, S. M., McCollum, E. E., Amanor-Boadu, Y., & Smith, D. (2012). Systemic perspectives on intimate partner violence treatment. Journal of Marital and Family Therapy, 38(1), 220–240. https://doi.org/10.1111/j.1752-0606.2011.00245.x