01 / Translation rule
A study can justify a question. It cannot complete the implementation plan.
Responsible translation requires five links: the supported problem, proposed mechanism, lawful authority, safety and due-process safeguards, and an outcome measure capable of disproving the intervention’s value.
02 / Crosswalk
Evidence → intervention → safeguard → measurement.
| Problem | Evidence contribution | Possible intervention | Safeguard | Measure |
|---|---|---|---|---|
| High destructive conflict and child insecurity | Longitudinal studies link conflict, emotional insecurity and later adjustment; constructive conflict may support problem-solving. | Reduce child exposure to hostile conflict; teach bounded, safe adult coordination; provide age-appropriate explanation. | Do not require joint engagement where coercion or violence makes it unsafe. | Child-reported security; school functioning; conflict exposure; adverse events. |
| Loss or weakening of a safe parent–child relationship | Observational studies associate supportive parent relationships with outcomes; contact, support and conflict interact. | Assess relationship quality and barriers; preserve safe involvement; support parenting quality. | No automatic contact presumption; individualized safety and child-experience assessment. | Relationship quality; reliability; child stress; implementation and safety. |
| Service named but not delivered | Effectiveness trials show assignment, attendance and benefit are different stages. | Track referral, provider availability, intake, attendance, fidelity and outcome. | Clinical appropriateness, consent, privacy, provider role and safety. | Time to service; dose received; fidelity; child/family outcome; dropout reason. |
| Family resources depleted by procedure | Natural-experiment and cohort evidence makes material resources relevant to child well-being. | Measure family-borne costs; reduce preventable rework; provide fee, travel and scheduling support. | Do not treat poverty as proof of poor parenting or use cost control to suppress valid claims. | Cash burden; lost work; housing/food stability; participation; abandonment. |
| Process exists but is unusable | Randomized access-to-justice evidence and form research show publication alone does not equal accessibility. | User-test forms, give complete correction instructions, offer bounded assistance, measure completion. | Preserve notice, confidentiality, evidence rules and impartiality. | First-pass completion; correction cycles; time to substantive action; user comprehension. |
| Safety concerns in mediation or agreement | Qualitative survivor research identifies power asymmetry and process harms. | Screen for coercion; offer counsel/support; permit adjudication; avoid pressuring settlement. | Emergency protection and survivor autonomy; no forced joint process. | Safety incidents; informed choice; withdrawal; access to counsel; enforceability. |
03 / Evaluation
Measure implementation before claiming impact.
Reach
Who was eligible, offered the intervention, able to participate and actually enrolled?
Delivery
Was the intended component delivered with the required skill, dose and timing?
Safety
What adverse events, coercion, privacy breaches or unsafe contacts occurred?
Outcome
What changed for the child or family, compared with a baseline or credible comparison?
Equity
Who benefited, who did not, and which language, disability, financial or cultural barriers remained?
Decision rule
What result causes continuation, adaptation, narrowing or termination?
04 / Working aid