RelationWatch Research Brief #01
Escalation patterns in relationship-connected cases. This publication examines patterns documented in a fixed RelationWatch research snapshot; it does not estimate population prevalence.
Brief 01 examines the 169 of 1,000 cases with a recorded separation, divorce, or former-partner condition. The remaining 831 cases are not assumed to be non-separation cases: 754 have another recorded relationship condition, and 77 do not have a usable relationship-status classification. Brief 02 examines that complementary group and its recorded outcomes.
Read Brief #02: relationship conditions beyond separationBrief 01 examines the portion of the fixed 1,000-case snapshot in which separation, divorce, or a former-partner condition was recorded. The analytical denominator for this Brief’s separation findings is therefore 169 cases, not all 1,000 screened cases. It describes documented warning behaviors, relationship contexts, and incident outcomes.
Among the 169 separation-connected cases identified in the fixed 1,000-case snapshot, threats were recorded in 103 (60.9%) and stalking or repeated harassment in 33 (19.5%). Homicide was the recorded outcome in 118 (69.8%) of these separation-connected cases. These are documented associations within the snapshot, not causal findings.
These are the strongest documented intersections within Brief 01’s separation/former-partner subset.
Each row is a documented intersection of relationship condition, warning indicator, and outcome. It does not establish a complete timeline or causal sequence.
The flow begins with a recorded separation-related relationship condition, then shows only behavior-level warning indicators and serious recorded outcomes.
Ranked documented behaviors among the 169 separation-connected cases.
Observed structured co-occurrences, not inferred complete event order.
Distribution of recorded ages in the fixed snapshot. Victims: 313 people; offenders / suspects: 324 people.
Cases and shares of the fixed 1,000-case snapshot; this does not measure national prevalence.
Heatmap cells show how frequently pairs of recorded warning indicators appear together within the snapshot. This is co-occurrence, not correlation or causal proof.
The six cards below are examples selected from the fixed 1,000-case research snapshot. They illustrate the documented patterns; they are not the full dataset on which this Brief’s findings are based. Each pathway uses only the recorded relationship condition, warning indicators, and incident outcome.
Mexico · 2026
Relationship: Separated
Separated → Escalation Pattern / Violence, Divorce Pending / Recent Separation → Homicide
View CaseIndia · 2026
Relationship: Ex-partners
Ex-partners → Escalation Pattern / Violence, Manipulative behavior, Threats → Attempted homicide
View CaseIndia · 2026
Relationship: Separated
Separated → Escalation Pattern / Violence, Threats, Stalking / Repeated Harassment → Homicide
View CaseUnited States · 2026
Relationship: Ex-partners
Ex-partners → Escalation Pattern / Violence, Suspects identified, Threats → Murder-suicide
View CaseUnited States · 2026
Relationship: Ex-partners
Ex-partners → Escalation Pattern / Violence, Suspects identified, Threats → Homicide
View CaseUnited States · 2026
Relationship: Ex-partners
Ex-partners → Divorce Pending / Recent Separation, Mental Health Deterioration / Hopelessness → Murder-suicide
View CaseBrief 01 is deliberately focused on the cases in which a separation, divorce, or former-partner condition was recorded. This accounting keeps its findings separate from the remaining cases rather than treating missing information as a relationship condition.
| Fixed-snapshot group | Cases | Share of 1,000 |
|---|---|---|
| Separation / former-partner condition — covered in Brief 01 | 169 | 16.9% |
| Another recorded relationship condition — covered in Brief 02 | 754 | 75.4% |
| No usable relationship-status classification | 77 | 7.7% |
Dataset: RelationWatch
Research snapshot: 1,000 verified cases
Data source: Publicly available open-source intelligence (OSINT) sources
Analysis: Structured case variables, relationship characteristics, behavioral indicators, event timelines, outcomes, case pathways, and human-reviewed classifications.
Denominators are displayed with each metric where information is incomplete. Missing information is not coded as “No.”
RelationWatch is an OSINT-derived case database and is not a population-based crime registry. Percentages and patterns in this Research Brief describe cases contained in the defined RelationWatch research snapshot and should not be interpreted as national or global prevalence estimates. The absence of a documented behavior may mean it was not documented in available OSINT sources.