ABSTRACT
In family mediation, emotional discomfort and perceived unsafety may signal deeper relational risks, including postseparation violence. Standard intake procedures in Norwegian family mediation depend on disclosure to a nonclinical administrator before any therapeutic relationship exists, and research documents systematic under-reporting at this stage. This study evaluates a brief, individually administered check-in protocol developed through the risk-informed mediation (RIM) project, a practice-oriented research collaboration between mediators and researchers in which mediators ask two experiential questions at the start of the first session. Using an embedded mixed methods design, we combined statistical analysis of 322 cases from seven Norwegian family counseling offices with qualitative findings from a mediator focus group at one embedded site. Each office contributed data from both a mediation-as-usual (MAU) period and a RIM period, allowing comparison of risk categorization across the two conditions. Chi-square tests revealed statistically significant shifts at four of seven offices. Within the RIM period, 9.5% of cases were re-categorized upward and 23.2% downward following the check-in. Dyadic asymmetry, defined as divergent responses between parents, predicted upward re-categorization with an odds ratio of 3.88, rising to 7.21 in low-risk dyads. Mediators described the check-in as a relational opening that revealed concerns parents had not raised before; unchanged assessments were not inert but confirmed a safe starting point. The mixed methods findings converge on a single conclusion: the protocol functions not as a classification instrument but as a relational opening that can reach risk intake procedures miss. Effects vary across offices, and the findings surface three structural dilemmas—how broad the inquiry should reach, how a mediator who has learned that one parent fears violence can act on that knowledge within a role that offers no formal protective mechanism beyond reporting duties, and how much interpretive competence the protocol requires of those who use it.