1 Prologue

“Sometimes I wonder,” one mediator said during a seminar break, “if what they call discomfort is actually fear.” The other nodded. “Yes. And when someone says they feel unsafe, I always ask myself: Is this emotional tension, or are we missing signs of violence?”

2 Introduction

The exchange in the prologue, where two mediators asked whether what parents describe as discomfort might in fact be fear, illustrates a recurring challenge in family mediation. Everyday terms such as discomfort and feeling unsafe can refer to very different conditions. They may describe ordinary unease, for example the stress of visiting a public office or speaking about private matters, but they can also signal serious relational risk, including coercive control or postseparation violence. Any intervention must therefore be sensitive to this range. It should avoid conflating mild unease with intimate partner violence, yet it cannot depend on explicit disclosure before risk is taken seriously. The challenge, then, is to create conditions under which these distinctions can be heard.

In family mediation, the most serious risk at that end of the range is postseparation violence. National survey data from Norway indicate that about 8%–10% of women and 2%–3% of men report severe physical partner violence (Thoresen and Hjemdal 2014). Risk often increases around separation, and violence that persists or escalates after a relationship has ended has well-documented consequences for health and quality of life (Seeberg and Bossy 2024). In family mediation, violence is recorded as a theme in roughly 13% of cases between 2019 and 2023, although these figures are based on mediators' recall at case closure and likely underestimate true prevalence (Barne-, ungdoms- og familiedirektoratet [Norwegian Directorate for Children, Youth and Family Affairs] 2024).

2.1 The Norwegian Mediation Context

The institutional context in Norway shapes how these concerns are handled. Mediation is mandatory for all separating parents with children under 16, regardless of whether any dispute is present (Children Act, section 51; Tjersland et al. 2015). This differs from most countries, where mandatory mediation applies only when parents seek access to the courts. In Norway, parents must attend mediation both at the time of separation and before initiating legal proceedings concerning custody, residence, or contact. The model therefore serves a preventive function at the population level, aiming to limit conflict and promote cooperative parenting.

As a result, the mediation population is more heterogeneous than in most international studies, including parents who separate amicably as well as those in active conflict, including where relational risk or violence may be present but less visible. Mediation is offered primarily within family counseling centers, which also provide couple therapy, family therapy, and courses addressing relational difficulties. The service is statutory and free of charge.

Because family mediation is a universal service, it must accommodate situations ranging from cooperative low-conflict cases to cases involving serious relational risk. The system relies on early differentiation so that time, competence, and safety measures can be directed where they are most needed. Cases are classified using an eight-item scored intake tool administered by a nonclinical office administrator (Dittman et al. 2021). Cases are classified as A (low conflict), B (moderate conflict), C (high conflict), or C! (high conflict with risk). C and C! cases require two mediators by regulation (Molden et al. 2019). Where power imbalance is so severe that joint mediation is not viable, parents may be granted separate mediation, a legally defined protective measure distinct from shuttle mediation: parents attend on entirely different days and neither is informed of the other's schedule. This must be decided before the formal mediation invitation is issued. Under the Children Act (§§ 51–52) and the Marriage Act (§ 26), the mediator's purpose is to assist parents in reaching an agreement on parental responsibility, residence, and contact that serves the child's best interests; the mediator facilitates this process but has no authority to determine its outcome. The intake tool was designed primarily as a resource allocation mechanism, not as a clinical screening tool for hidden risk (Eikrem et al. 2023).

Norwegian family mediation covers two principal case types. Separation and divorce cases are opened when parents with children under 16 separate or divorce, as mandatory attendance is required by law. Parental dispute cases are opened when already-separated parents return to the service over ongoing disagreements about residence, contact, or parental responsibility. Although separation cases frequently involve parental dispute, the two types differ in legal basis, the relational stage at which they arise, and, as this study shows, in how risk is re-categorized following the check-in.

2.2 A Structural Gap at Intake

The intake tool functions as intended for its allocative purpose, but its design produces a structural limitation: it depends entirely on what parents choose to disclose in a telephone call conducted by a nonclinical administrator before any clinical relationship has been established. Data protection regulations mean that individual disclosures made during the intake call cannot be passed to the mediator who will conduct the session; only the aggregate category can be shared. The source behind the category is not accessible to the clinician. Under-reporting is well documented: Dittman et al. (2021) found that one third of parents understated their difficulties at the telephone intake compared with an anonymous electronic questionnaire, and that of the parents who reported violence or substance misuse in the questionnaire, only 15% had disclosed this at intake.

2.3 Approaches to Risk Identification

Internationally, efforts to address this structural condition have produced a range of instruments, positioned along a continuum from psychometric screening to risk-informed inquiry (Figure 1). Spangaro et al. (2009) distinguish routine screening, defined as the systematic application of a standardized test to identify individuals at sufficient risk, from routine inquiry, which draws on clinical judgment rather than test logic. At the screening end, instruments such as the RBRS (Beck et al. 2013), DOVE (Ellis and Stuckless 2006), and FL-DOORS (Booth et al. 2023) are thorough, validated, and behaviorally precise. Their limitation is structural: all are administered before any clinical relationship exists, and depend on a parent's willingness to disclose to an unknown administrator. Closer to the inquiry end, the Mediator's MASIC (Rossi et al. 2022, 2024) places assessment within a confidential clinical interview conducted by the mediator, identifying IPV in 94% of cases where it was reported, compared to 38% with a general, broad-question screen (Rossi et al. 2024). Its limitation is practical: MASIC has been found too time-consuming for routine use under ordinary workload conditions (Rossi et al. 2024). None of the existing instruments address the gap between intake and the first joint session: the moment when a clinical relationship has just been established and disclosure is most possible, but the joint conversation has not yet begun.

2.4 The RIM Check-In

The RIM check-in was developed through a bottom-up collaboration between practitioners and researchers at two family counseling offices in Oslo before expanding to a formal pilot study. The check-in occupies that position on the continuum: the start of the first mediation session, after a clinical relationship has been established but before the joint conversation begins. Each parent is invited to a brief individual conversation while the other waits outside. The mediator asks two questions, scored 0–10: “On a scale from 0 to 10, how uncomfortable is it for you to be in the same room as your ex-partner?” and “On a scale from 0 to 10, how safe do you feel about speaking freely and being in the same room as your ex-partner?” Scores above 5 on either question prompt a follow-up: “What makes you feel uncomfortable or unsafe?” The questions were chosen precisely because they target the two experiential dimensions introduced at the opening of this article: ubehag (discomfort in the mediation situation) and utrygghet (felt unsafety with the other parent). The 5-min format reflects the constraints of mandatory mediation: brief enough to be routine, open enough to bring out what a joint session would suppress. Each parent gives their account independently, before either has heard the other's version. The mediator's responsibility is to pursue what emerges. Spangaro et al. (2011) identify three conditions for disclosure: a direct question, a trusted person asking, and a sense of choice. Intake-based tools satisfy only the first; RIM addresses all three. Because separate mediation must be decided before the formal invitation is issued, risk identified through the check-in cannot trigger a switch to separate mediation, but can inform how the mediator conducts the session and activate statutory reporting duties. Recent legislative changes in Norway have made this gap more pressing.

2.5 Legislative Context

The new Children Act (Stortinget [Norwegian Parliament] 2025a, Prop. 117 L, 2024–2025) establishes shared daily decision-making authority as the default for separating parents regardless of cooperation capacity, prompting concern from several organizations that this assumes cooperation even in cases with histories of violence or coercive control (Barneombudet [Norwegian Children's Ombudsman] 2025; Stine Sofies Foundation et al. 2025; Norsk Psykologforening [Norwegian Psychological Association] 2024). A parallel set of parliamentary recommendations adopted alongside the new Children Act establishes that children who are exposed to, or witness, violence perpetrated by a parent should, as a general principle, not have contact with that parent (Stortinget [Norwegian Parliament] 2025b, Recommendation No. 526 L, 2024–2025).

These resolutions strengthen children's protection but do not extend equivalent provisions to adult parties where histories of violence or coercive control are present. Postseparation conflict spans a wide continuum from temporary disagreements to entrenched hostility (Smyth and Moloney 2019), and terms such as high conflict may conceal significant differences between mutual disagreement and coercive control. The capacity to distinguish between conflict and coercion is therefore central to safe mediation practice.

2.6 Present Study

This article reports findings from an embedded mixed methods study of the risk-informed mediation (RIM) protocol conducted across seven family counseling offices between November 2024 and February 2025. The study was guided by a practice-oriented research approach (Castonguay and Muran 2015), integrating systematic data collection with routine clinical mediation practice.

The study had two primary aims. First, it examined whether the brief check-in protocol was associated with changes in mediators' risk categorization across participating offices. Second, it explored how mediators experienced, interpreted, and integrated the protocol into their day-to-day work. A central analytical focus concerned the clinical significance of dyadic asymmetry—that is, divergent responses between parents to the check-in questions—as a potential predictor of re-categorization. This focus rests on the assumption that asymmetric reporting within a parental dyad may reveal clinically meaningful differences in exposure to insecurity or risk that are not readily identifiable in joint mediation contexts, with direct implications for how risk is identified and addressed in practice. A third question—whether divergent responses between parents predict upward re-categorization, and what that asymmetry reveals about how risk is distributed within a dyad—was identified as a secondary aim following initial data examination.

Four research questions guided the study:

1. To what extent was the check-in protocol associated with changes in risk assessments across the participating offices?
2. How did mediators experience and understand the protocol in day-to-day practice?
3. In what ways did the qualitative material help explain observed quantitative patterns?
4. Do divergent responses between parents predict re-categorization, and if so, what does dyadic asymmetry reveal about how risk may be distributed within a parental dyad?

3 Methods

3.1 Design

The study employed an embedded mixed methods design with a dominant quantitative strand (Creswell and Clark 2017; see Figure 2). Qualitative data from one selected site were integrated within the quantitative strand at the interpretation stage, where findings from Office 2 contextualize patterns observed across all seven offices.

3.2 Participants

Quantitative data were collected from 322 mediation cases across seven Norwegian family counseling offices between November 2024 and February 2025. Cases granted separate mediation at intake were excluded. Offices represented geographic and demographic diversity. Mediators were primarily social workers with family therapy backgrounds or licensed psychologists. One office (Office 2) contributed both quantitative data (n = 66) and participated in a qualitative focus group interview with six mediators.

Under both MAU and RIM, mediators exercise ongoing clinical judgment about risk throughout the mediation process, including decisions about co-mediators, session structure, and whether to hold individual or joint conversations. The formal outcome variable differs between the two periods. In the MAU period, the outcome is the mediator's clinical risk assessment recorded at case closure, as part of routine administrative practice. In the RIM period, the outcome is the mediator's clinical risk categorisation recorded via a brief research questionnaire completed immediately after the check-in at the start of the first mediation session. In both periods, this clinical categorisation is compared with intake category derived from the eight-item intake tool.

3.3 Data Collection

3.3.1 MAU Intake Procedures and Categorization

Before implementation of the RIM protocol, all cases were categorized using a standard phone-based intake procedure administered by a nonclinical office administrator. The tool comprises eight scored questions covering legal basis for mediation, disagreements about residence and contact, co-parenting conflict, duration of difficulties, trust in the other parent as caregiver, prior court proceedings, and problems related to substance use or violence (Dittman et al. 2021).¹ Responses are aggregated into a total score that classifies cases as A, B, C, or C!. The tool was developed from practitioner experience rather than psychometric validation (Dittman et al. 2021).

3.3.2 RIM Check-In Protocol and Implementation

During the first mediation session in the RIM period, each parent was invited to a brief individual check-in using the two structured questions described in Section 1. Scores above 5 prompted a follow-up about the source of discomfort or unsafety. Responses were noted as working notes and not included in the official case file. The conversation typically lasted 5 min per parent. The mediator's revised risk assessment was made immediately after the check-in with both parents, before the joint session commenced. The intake categorization is performed by an office administrator before any clinical relationship exists; the RIM check-in is conducted by the mediator, a trained clinician, within the first session. These are qualitatively different assessment processes carried out by different roles.

3.4 Analytical Strategy: Re-Categorization of Risk as Outcome

We compared mediators' revised risk assessments after the check-in with initial intake categorization, grouping changes as Lower Risk, Unchanged, or Higher Risk. An important structural asymmetry bears on interpretation: in the MAU period the outcome variable is the mediator's assessment at case closure, while in the RIM period it is the mediator's assessment immediately after the check-in at the start of the first session. This limits direct comparison and is acknowledged as a limitation. A further interpretive constraint concerns the B category specifically. When both parents respond at intake and their scores differ, the higher score determines the assigned category. When only one parent responds, any score in the A range is automatically upgraded to B as a precautionary measure (Dittman et al. 2021). The higher-score rule applies to the intake categorization under both MAU and RIM, as intake procedures are identical in both periods. The mediator's subsequent clinical reassessment—at case closure under MAU, and after the check-in under RIM—is a clinical judgment not governed by this algorithmic rule. B-category cases therefore cannot be treated as a uniform group: they may reflect genuine intermediate conflict, dyadic disagreement, or incomplete intake data from a single respondent. A supplementary analysis was therefore conducted restricting the comparison to B-category and above cases (B+). A-category cases more closely resemble family consultation than mediation in the conventional sense (Wing 2009), and including them risks diluting effects that are relevant only to cases where conflict or risk is present at intake.

3.5 Quantitative Analysis

Quantitative data were analyzed using IBM SPSS (version 29.0.2.0). Chi-square tests assessed distributional change after the check-in. Transition matrices, frequency distributions, and cross-tabulations were constructed to identify re-categorization patterns. Effect sizes (Cramér's V) were calculated where relevant. Dyadic analyses examined whether asymmetric responses between parents predicted upward re-categorization.

3.6 Qualitative Analysis

Qualitative data from Office 2 were analyzed using reflexive thematic analysis (Braun and Clarke 2006, 2019, 2024). The focus group was conducted via Microsoft Teams, recorded, and transcribed. The second author led the interview; the first and third authors participated as observers. A follow-up reflection session was also transcribed and included in the analysis. Integration occurred within-site at Office 2, where quantitative and qualitative data were interpreted side-by-side, and interpretively at the regional level, where findings from Office 2 contextualized broader patterns across all seven offices. Joint displays enabled identification of convergent and divergent findings (Creswell and Clark 2017).

3.7 Ethical Considerations

The study was conducted within the statutory Norwegian family counseling service (Bufetat) and was approved by the Data Protection Officer at Bufetat prior to data collection. Under Norwegian research governance, studies conducted as part of routine service development within a statutory body—and which do not involve direct collection of sensitive data from clients—are assessed by the institutional data protection officer rather than a Regional Ethics Committee (REK). This constitutes the appropriate oversight mechanism under applicable Norwegian regulations. The quantitative data were drawn from administrative records generated in the course of normal service operation; no client data were collected as primary research data. Mediators who participated in the focus group interviews were informed about the study's purpose and that their contributions, including direct quotations, might be used in publications; participation was voluntary throughout. All data were anonymized, and data security procedures were followed. The protocol was designed to minimize distress, and mediators received training to respond appropriately to any disclosures.

3.8 Researcher Reflexivity

All three researchers are active in the Norwegian mediation field, positioning them as insider-researchers with access to and accountability within practice. Reflexivity, as used here, refers to an ongoing process in which practitioners and researchers examine their own assumptions, values, and reactions in relation to those they work with (Rothman 2014). Two levels are relevant here. At the practice level, this means attending to how the mediator's own positioning and reactions shape the clinical encounter (Stokkebekk 2024). At the researcher level, it involves critical scrutiny of how embeddedness in the field, prior commitment to the intervention, and insider positioning shape interpretation and reporting. Reflexivity operated at both levels (Braun and Clarke 2019; Whittaker et al. 2023), grounded in critical realism and moderate social constructionism.

4 Findings

4.1 Quantitative Findings

4.1.1 Comparison of MAU and RIM

To evaluate the impact of the RIM protocol, we compared risk re-categorizations across seven offices during the MAU and RIM periods described in Section 2.

Table 1 presents re-categorization by office. Clear shifts are visible particularly at Offices 3, 4, and 7, with bidirectional movement suggesting the protocol revealed both underestimated and overestimated risk.

Table 2 summarizes the statistical tests. Chi-square tests revealed significant differences at Offices 2 (χ² = 15.361, p = 0.001), 3 (χ² = 23.496, p < 0.001), 4 (χ² = 11.470, p = 0.003), and 7 (χ² = 25.462, p < 0.001). Office 6 showed a nominally significant result (χ² = 8.271, p = 0.016) but is excluded from the count given its very small RIM sample (N = 10). The protocol influenced mediators' risk assessments in a measurable way, though with variation across offices.

Effect sizes (Cramér's V) ranged from small (Office 1: V = 0.120; Office 2: V = 0.162) to moderate (Offices 3–4: V = 0.248, 0.219) to strongest at Offices 6 and 7 (V = 0.341, 0.283), suggesting that local implementation context moderates impact. Having established that the protocol had a measurable effect at four offices, we turn to the distribution of re-categorizations across the full sample.

4.1.2 Risk Re-Categorization Patterns Within the RIM Period

We then examined the distribution of re-categorizations within the RIM period. Out of 322 mediation cases, 67.3% (n = 217) assessments remained unchanged following the check-in, 23.2% (n = 74) were re-categorized as lower risk, and 9.5% (n = 31) as higher risk. A chi-square test confirmed a statistically significant overall shift (χ²(2) = 9.96, p = 0.0069; Cramér's V = 0.161), with estimated power of 1.0.

Figure 3 presents a transition matrix for all 322 mediation cases, illustrating how cases moved between intake and post-check-in risk categories. Rows represent the initial risk category at intake (A, B, C, C!), columns represent the post-check-in category, diagonal values (e.g., A → A) indicate unchanged assessments, and off-diagonal values reflect re-categorization, either upward or downward. Six cases initially assessed as high-risk (C!) were reclassified as low-risk (A), suggesting either overestimation during intake or underreporting at check-in. Only one case was re-categorized from A to C!, indicating that the protocol did not produce extreme upward shifts from low to high risk.

4.1.3 Re-Categorization by Mediation Type: B+ Cases

A supplementary analysis restricted to B-category and above cases (B+, n = 202 in the RIM period) examined whether re-categorization patterns differed by mediation type. Among separation and divorce B+ cases, 49.5% were re-categorized downward following the check-in, compared with 26.8% among parental dispute B+ cases, a statistically significant difference (χ² = 11.9, p = 0.003). Upward re-categorization was comparable across types (separation and divorce: 11.0%; parental dispute: 9.3%). When parental dispute B+ cases were compared directly with the MAU B+ reference population (n = 967), no significant difference was found (χ² = 0.7, p = 0.695), indicating that the parental dispute recategorization profile closely matched clinical judgment without the structured check-in. The same comparison for separation and divorce B+ cases yielded a highly significant difference (χ² = 37.8, p < 0.001). These patterns suggest that the aggregate RIM effect is driven primarily by separation and divorce cases. Beyond mediation type, a second question concerns what it is about individual dyads that predicts upward re-categorization.

4.1.4 Dyadic Asymmetry as a Predictor of Risk Re-Categorization

To explore relational dynamics underlying risk assessment, we examined whether dyadic asymmetry, defined as divergent responses between parents on either of the two check-in questions, predicted upward re-categorization following the RIM protocol. In a simple logistic regression model (Table 3), dyadic asymmetry significantly predicted increased risk categorization (B = 0.987, SE = 0.407, z = 2.426, p = 0.015), with an odds ratio of 2.68. A multivariate model yielded a stronger effect (B = 1.356, SE = 0.387, z = 3.50, p < 0.001; OR = 3.88), indicating that dyads with asymmetrical disclosures were nearly four times more likely to be reclassified upward.

Tables 3 and 4 present two components of the same dyadic analysis and are numbered accordingly. A stratified analysis (Table 4) grouped dyads into low (0–1), medium (2–3), and high (4+) risk categories based on a composite of 10 risk indicators. Asymmetry had a significant effect only in low-risk dyads (OR = 7.21, p < 0.001), highlighting the check-in's role in revealing hidden concerns when initial risk appears low. When individual risk indicators (e.g., fear of ex-partner, insecurity in the mediation setting, nonviolent insecurity) were tested separately in multivariate models, none reached statistical significance. It is the discrepancy between parents, not any single risk factor, that prompts concern.

4.1.5 Gendered Patterns in Risk Reporting

To explore how discomfort and insecurity were reported within dyads, we analyzed risk indicators at the dyadic level. Among the 93 dyads in which at least one parent reported risk, three configurations were identified: both parents reported at least one risk factor; only the mother reported risk; only the father reported risk.

Figure 4 illustrates the distribution of symmetric and asymmetric disclosure across each risk indicator, distinguishing cases where both parents, neither parent, or only one parent disclosed that indicator. Discomfort indicators were more frequently reported symmetrically, suggesting shared emotional unease. Unsafety indicators, including partner violence, threats, and substance use, were predominantly reported asymmetrically, pointing to divergent experiences within the same dyad. Among asymmetric cases, mothers more frequently disclosed prior partner violence and care-related mistrust; fathers more often expressed nonviolent insecurity and concerns about the ex-partner's mental health.

The overall proportion reporting at least one risk indicator was similar across genders (19.8% of mothers and 21.8% of fathers), suggesting that both parents report risk at comparable rates but with qualitatively different content. The quantitative strand establishes what the check-in changes; the qualitative strand, drawn from Office 2, addresses how mediators experienced and made sense of those changes.

4.2 Qualitative Findings

Three themes were developed from focus group interviews at Office 2, exploring how the check-in shaped the mediator role. The first, A Door-Opener to the Unspoken, describes how mediators experienced the check-in as creating a moment in which parents could name concerns that intake procedures had not reached, characterized by emotional depth and a sense of individual recognition. The second, Holding Space Without Taking Over, captures the tension between that emotional depth and the nontherapeutic frame of mediation. The third, Balancing Structure and Sensitivity, reflects mediators' practical negotiations between standardized procedure and professional discretion, including challenges with terminology and interpreters. A forthcoming article will present qualitative material from all three focus group interviews.

The three themes reflect the three quantitative patterns. The 9.5% of cases re-categorized upward corresponds to what mediators described in Theme 1, A Door-Opener to the Unspoken: something was revealed that the intake procedure had not reached. The 67.3% of unchanged assessments maps to Theme 2, Holding Space Without Taking Over: the check-in confirmed a safe starting point rather than revealing new concern. The 23.2% of cases re-categorized downward maps to what mediators described as Balancing Structure and Sensitivity: the check-in sometimes confirmed that an intake category reflected precautionary scoring under conditions of uncertainty rather than confirmed risk, and direct conversation with a clinician reduced rather than amplified concern. Together, the quantitative and qualitative strands suggest that the protocol's significance lies not only in the cases it reclassifies, but in the relational conditions it creates: a structured moment in which each parent's account is formed independently, before either has heard the other's version.

5 Discussion

The central finding of this study is that a brief, individually administered risk check-in can reveal concerns that standard intake procedures do not reach, not primarily by reclassifying cases, but by creating a relational moment in which parents can say what they could not, or would not, say in a telephone interview or a joint session. The check-in was, in the mediators' experience, a relational intervention grafted onto an administrative procedure.

5.1 Recalibration or Concealment? Interpreting Downward Re-Categorization

The predominance of downward movement from the intake category to the mediator's post-check-in assessment in the RIM period requires interpretation. Downward movement can reflect four processes: overestimation at intake, where the scoring rule inflates estimates for unsure responses (Dittman et al. 2021); clarification of genuine ambiguity; suppression in the face-to-face encounter; or a containing effect of service contact, where entry into a clinical encounter may itself reduce experienced unsafety. Spangaro et al. (2009) argue that screening itself constitutes an intervention: being asked directly may alter how a parent understands their own situation, which supports the interpretation that unchanged categories are not passive outcomes.

Two structural features of the intake tool help explain the predominance of downward movement. The scoring rule treats an unsure response to the violence and substance use question as equivalent to confirmed risk. A parallel mechanism at B automatically upgrades A-range scores when only one parent has responded (Dittman et al. 2021). Both rules err on the side of caution but produce inflated baselines. The downward pattern is consistent with the protocol confirming genuine low-risk levels that precautionary intake scoring had obscured, particularly in separation and divorce cases where single-respondent upgrades are more common.

At the same time, underreporting at telephone intake is well documented (Dittman et al. 2021), and a structured check-in with a trusted clinician may reveal what the telephone interview did not. The present data cannot separate correction of inflated intake scores from genuine disclosure of suppressed risk, but both processes would produce the pattern of predominantly downward re-categorization observed. The first interpretation—overestimation at intake through the precautionary scoring rules—and the second—clarification of genuine ambiguity—both suggest the protocol is functioning as intended by correcting inflated baselines. The fourth—the containing effect of service contact—is also consistent with intended functioning. The third interpretation—suppression in the face-to-face encounter—cannot be resolved by the present data, as suppression would produce the same observable pattern as correction. Whether the check-in reveals or conceals depends partly on what it asks, and that is the first of three structural dilemmas the findings bring into view.

5.2 Dilemma 1: Scope, Not Too Narrow and Not Too Broad

The check-in protocol faces a scope dilemma that has no clean resolution. If it asks only two broad experiential questions, it identifies that something is present without identifying what, leaving the mediator with signal but not substance. If it extends to behaviorally specific questions about violence, threats, and the child's situation, it risks exceeding the relational conditions that make disclosure possible in the first place, and moves the mediator toward a clinical assessment role the mediation mandate does not support. The RIM check-in asks two questions and follows up when scores exceed five. It does not ask about violence, substance misuse, or the child. What it does is open a space in which parents can raise any of those things, and the qualitative findings show that they do: disclosures covered fear of the other parent's behavior, worry about the child's situation, and parenting capacity, with mothers more likely to raise prior violence and fathers more likely to raise nonviolent insecurity. The FL-DOORS rests on a similar premise: that broad, nondirective questions can yield disclosures that targeted questions suppress (Booth et al. 2023). The cost of this approach is diagnostic imprecision. The protocol signals that something is present without identifying what it is or who is responsible. Jaffe et al. (2008) argue that screening must go further, differentiating the potency, pattern, and primary perpetrator of violence. RIM does not make that differentiation. It creates the opening; further clinical assessment must follow. This scope reflects the purpose of mediation. Katz (2007) draws the contrast with therapy: feelings are acknowledged but not worked through, and the goal is agreement rather than psychological change. Where violence is a factor, therapy involves safety planning and ongoing assessment of whether it is responsible to continue (Snellingen et al. 2024). Risk identification in mediation is not oriented toward such intervention: its purpose is to establish whether safe and equal participation is possible (Wing 2009). What RIM does not address is the information that enters the first session before the check-in begins: the intake category whose source the mediator cannot know.

5.3 Dilemma 2: Neutrality, Asymmetry, and the Child's Best Interests

The neutrality dilemma begins before the first session. When parents arrive, the mediator already holds a risk category derived from a telephone interview conducted by an administrative staff member. A case classified as C! signals that something serious has been reported, but carries no indication of who reported it or whether the other parent is aware. Data protection law means individual disclosures cannot be passed on: the aggregate category can be shared with the mediator; the source behind it cannot. This asymmetry is an unintended structural consequence of applying data protection rules to an administrative intake process not designed with this information flow in mind: the individual disclosure is classified as sensitive personal data and cannot be passed beyond the administrative function that received it, including to the clinician who will subsequently work with the family. The mediator therefore enters the joint session knowing that risk is present without knowing its shape, its direction, or whether both parents are aware that it has been named.

The check-in enters this gap. It does not resolve the information problem, but it gives it a shape. By speaking with each parent individually, the mediator now knows not only that risk is present but something of whose experience it reflects. The findings show this matters clinically: no individual risk indicator predicted upward re-categorization on its own, but dyadic asymmetry did, with an odds ratio of 3.88, rising to 7.21 in low-risk dyads. This finding extends the screening literature, which has predominantly modelled risk as an individual property (Beck et al. 2013; Booth et al. 2023; Jaffe et al. 2008). Where Jaffe et al. (2001) argued that risk remains invisible unless someone asks, the present finding suggests it also remains invisible unless both parents are asked separately and their responses compared. The check-in also opens a limited action space. Where concern is disclosed, the mediator can adjust how the joint session is introduced, consider whether further individual conversations are needed later in the process, or activate statutory reporting duties if the threshold is met. The protocol does not prescribe these responses, but it creates the conditions under which they become possible.

But acting on what has been disclosed is not straightforward. Mediation rests on an assumption of procedural equality, but Wing (2009) argues that reliance on neutrality and self-determination masks how mediation can reproduce structural inequality between parties. Where one parent is afraid of the other's reaction, the precondition for genuine dialogue is not met, regardless of how evenly the mediator distributes speaking time. The individual check-in ensures that each parent's account is formed independently, before either has heard the other's version.

The mediator who has conducted the check-in now holds asymmetric information: one parent may have disclosed fear or prior violence while the other has not. The mediator must lead a joint session from a position formally defined as neutral, while knowing the two parents do not enter the room on equal terms, using information without breaching confidentiality, shaping the session without exposing the disclosing parent, and holding an ethical obligation toward the child that neither question addresses directly. The check-in asks about each parent's discomfort and unsafety, but not what the child experiences in the household where the fear originates (Eikrem and Andenæs 2025). Mediators have a statutory duty under the Family Counseling Act (section 10) to act when there is reason to believe a child is at risk, yet the protocol that reveals that risk provides no structured pathway for acting on it within the first session. The asymmetry revealed by the check-in extends beyond the dyad: it is also an asymmetry between what is known and what can be done. The principles of neutrality, power-balancing, and the child's best interests are simultaneously in play in every mediation, and the check-in brings all three into conflict at once.

The check-in thus creates both a dilemma and a resource: it gives the mediator something to work with that the intake procedure did not, while placing the mediator in a position that the mediation mandate cannot fully support. How well the mediator can hold that position depends both on the conditions under which the protocol is practiced and on the individual mediator's level of skill, confidence, and experience with the questions in use, which is the third dilemma.

5.4 Dilemma 3: Intensity, Not Too Demanding and Not Too Minimal

A protocol that is too demanding exceeds the conditions of everyday practice and is abandoned; one that is too minimal produces no recalibration and adds nothing to what mediators already do. The office-level variation in this study maps directly onto that tension. Four of the seven offices showed significant shifts in risk categorization, while three did not. Office 7, which took part in the prepilot, showed the strongest effect and had the highest number of RIM cases relative to office size, suggesting that accumulated practice with the protocol amplifies its impact. The absence of a significant effect at some offices may reflect a different dynamic: offices where mediators already identify risk consistently at intake may show little movement with RIM precisely because the protocol adds less to what good practice already captures. A further source of variation is the type composition of each office's caseload. Because the re-categorization effect is concentrated in separation and divorce B+ cases, offices with a higher proportion of parental dispute cases would be expected to show smaller aggregate effects, irrespective of implementation quality. This structural factor is not controlled for in the office-level analysis and should be considered in future comparative work. Screening tools are routinely abandoned when they exceed the conditions of everyday practice (Clemants and Gross 2007; Rossi et al. 2024; Sixsmith and Blakey 2025). RIM addresses this through brevity, but brevity does not eliminate interpretive demand. The prepilot at Office 7 documented a recurrent challenge: mediators found it difficult to distinguish between ubehag (discomfort) and utrygghet (unsafety), the two constructs that anchor the protocol. Discomfort with the mediation situation and fear of the other parent can present similarly, and hearing that difference, and responding to it clinically rather than procedurally, requires practice with the questions in use (Eikrem et al. 2023). This may explain why the strongest effect was observed at the office that had spent a full year developing the protocol iteratively: not simply because mediators used the check-in, but because they had learned to use it. This is consistent with Lee et al. (2021) 10-year retrospective on universal risk screening in an Australian mediation and counseling organisation, which similarly found that practitioner familiarity with screening process accumulated over time and influenced quality of risk identification.

This variation is also bound up with how RIM was developed. The protocol emerged from a bottom-up collaboration between mediators and researchers rather than psychometric item testing. A more formally validated tool might offer greater diagnostic precision, but the practice-oriented approach offers something different: tools that practitioners recognize as their own. The cost is reduced precision; the gain is actual use, and it is this tension between precision and feasibility that the three dilemmas, taken together, name.

What holds the three dilemmas together is a structural feature of the check-in that none of them resolves: speaking with each parent individually makes visible what joint sessions suppress. Mothers and fathers disclosed risk at comparable rates but with qualitatively different content, and the gap between their accounts, the dyadic asymmetry, was the only consistent predictor of upward re-categorization. The check-in does not eliminate that inequality; it names it. Office 2 allows the two strands of evidence to be read together directly, and the joint display below shows concretely what that looks like in practice.

5.4.1 Mixed Methods Integration

Table 5 presents a joint display integrating quantitative and qualitative findings from Office 2, where both strands were collected. All quantitative figures refer to Office 2 cases (n = 66).

Taken together, the two strands support a conclusion neither could reach alone: the check-in functions not as a classification instrument but as a relational opening. The 67% of unchanged cases were not passive: mediators described confirmation of a safe starting point as itself clinically meaningful.

6 Limitations

The design is naturalistic, with no control over confounding factors and no systematic assessment of implementation fidelity. The intake tool against which RIM is compared has not been psychometrically validated, and observed recategorization reflects the combined effect of RIM and the properties of that tool, which cannot be disentangled. The dataset does not record whether one or both parents responded at intake, limiting interpretation of B-category patterns specifically. The MAU and RIM periods reflect different assessment logics and time points, which limits direct comparison. Generalisability is bounded by the Norwegian context of mandatory, universal mediation, which differs structurally from jurisdictions where mediation is sought only in disputed cases. Interview order was not standardized, separation initiation was not recorded, and mediator-level variation was not controlled for. The RIM check-in is administered once, at the start of the first session; the number of sessions per family varies by category, and a single assessment point cannot capture how disclosure readiness and perceived risk may shift across a longer mediation trajectory.

7 Conclusion and Implications

This study evaluated a brief risk-informed check-in in Norwegian family mediation. Effects vary across offices and the protocol's scope is constrained by existing regulations. This article reports the first phase of a three-part practice-oriented research program. The protocol is already being scaled to additional offices, and the present findings are feeding directly into that development work. These implications rest on the central pattern: the check-in opened a space for disclosure rather than classifying risk, dyadic asymmetry was the only consistent predictor of upward re-categorisation, and mediators described it as a door-opener whose unchanged outcomes confirmed a safe starting point.

The findings indicate that RIM primarily enables inquiry. It opens a space for disclosure rather than directly classifying risk. It is positioned as a supplement to existing intake procedures, not a replacement for them, and the two are complementary rather than one compensating for the other. Risk-informed inquiry reflects a clinical orientation that holds attention to relational risk alongside the`mediator's broader responsibilities, navigating the child's best interests, neutrality and power balance (Stokkebekk 2024). Screening and risk-informed inquiry may serve overlapping but distinct disclosure motivations. Some parents disclose more readily at a distance, where anonymity lowers the threshold; others disclose only once a clinical relationship and a sense of trust have been established. These are competing accounts of what supports disclosure, and the present study does not test them directly.

Within RIM, the two experiential questions function as a primer rather than a complete assessment. They lower the threshold for disclosure, establish the legitimacy of the topic, and create the relational conditions in which parents can articulate concerns at the check-in. A structured extension, summarized in Table 6, would follow a different logic. Where the experiential questions ask how the parent feels, behaviorally specific questions ask what has happened. Rossi et al. (2022, 2024) demonstrate that this distinction matters clinically: general distress questions and behavior-specific questions tap different dimensions of risk and complement rather than substitute for each other. An extended protocol would therefore preserve Q1 and Q2 as the relational entry point and add behaviorally specific questions conditionally, only when scores above 5 signal that something is present. The three domains where under-reporting is most common in the Norwegian intake data are fear of the other parent's reaction, threats and violence, and concern for the child's situation. Whether they can be administered within 5 min without disrupting relational quality is an empirical question the development project is positioned to address. Extending the protocol raises practical questions about follow-up and system readiness: where violence is identified, the service system must be equipped to respond.

Several aspects of the present findings extend beyond what the project can directly address. RIM is not a screening instrument and cannot be evaluated as one; it is an inquiry tool that opens a clinical conversation. As the Limitations note, the Norwegian intake instrument has not been psychometrically validated, and the present findings should be read in that context. Returning to the screening-inquiry continuum in Figure 1, the findings suggest that the intake and the inquiry functions may need to be considered together. RIM adds the inquiry functions the system has lacked. It does not address the screening function. A second and third article will address cross-case qualitative synthesis and full mixed methods integration. 

The dilemmas of scope, neutrality, and intensity set out in the discussion are not failures of the protocol. They are structural features of the mediator role that the check-in makes visible precisely because it works.

8 Epilogue

Discomfort and unsafety are not simply emotional states. They are signals with relational, legal, and ethical implications. The RIM protocol offers a structured way for practitioners to hear those signals, but acting on them requires institutional conditions that tools alone cannot provide. Better mediation practice begins with a willingness to ask the difficult question and to take the answer seriously.