© 2026 JSH Law Ltd. All rights reserved.Seeing the Pattern: What the Major Coercive-Control Frameworks Can — and Cannot — Tell the Family Court
Everyone working seriously with coercive control eventually runs into the same problem: we agree that the pattern matters — but we do not always mean the same thing by “pattern”.
A police risk assessment is looking for something different from a Family Court judge.
A homicide researcher is asking a different question from a sociologist coding years of messages.
A child-protection model is trying to understand something different from an academic theory of entrapment.
And a litigant in person who has 4,000 messages, school records, police disclosure and three years of court orders has a different problem again:
How do I turn all of this into evidence the Family Court can actually understand?
I have been looking closely at a number of the major frameworks used to understand coercive control — including the work of Evan Stark, the Duluth Power and Control Wheel, Michael P. Johnson, Marianne Hester, Emma Katz, Safe & Together, DARA, Jane Monckton Smith, Robert Kaiser and, most recently, Carisa Carlton.
They overlap.
But they are not interchangeable.
And that distinction matters enormously in the Family Court.
The central problem
Coercive control is simultaneously a theory of domination, a pattern of behaviour, a safeguarding concern, a risk indicator, an evidential problem and — in some contexts — a criminal offence.
Different frameworks illuminate different parts of that problem.
The mistake is not using frameworks. The mistake is using a framework outside the purpose for which it was designed — or treating an analytical tool as though it has already proved the facts.
The quick answer: which framework does what?
Evan Stark: What is coercive control and why is it fundamentally about liberty, autonomy and entrapment?
Duluth Power and Control Wheel: What kinds of tactics may be used to maintain power and control?
Michael Johnson: Is this violence occurring within a broader regime of control, or is the context different?
Emma Katz: How does coercive control directly shape children’s lives, agency and relationship with the protective parent?
Safe & Together: What is the perpetrator’s behavioural pattern as a parent, and how does it affect child and family functioning?
DARA / DASH: What is the current safeguarding risk and what protective action may be needed?
Jane Monckton Smith: Does the history resemble a known progression towards intimate-partner homicide?
Robert Kaiser: How is abusive behaviour changing over time — in frequency, intensity and response to restraints?
Carisa Carlton: Can longitudinal records be systematically coded and tested to determine whether the corpus supports a sustained coercive-control pattern?
PD12J and the Family Court: What alleged or established pattern is relevant to welfare, risk and the order the court must make?
Why incident-by-incident thinking fails
The Family Court has already recognised the problem.
In Re H-N and Others [2021] EWCA Civ 448, the Court of Appeal endorsed the principle that coercive and controlling behaviour depends upon understanding a pattern or series of acts, whose impact should generally be considered cumulatively rather than rarely in isolation.
That matters because many acts which form part of coercive control are not inherently unlawful or obviously abusive when viewed alone.
Sending a message is not abusive.
Asking where a child is may not be abusive.
Contacting a school may not be abusive.
Requesting financial information may not be abusive.
Making a court application may not be abusive.
But context can change meaning.
The questions become:
- How often?
- For what purpose?
- Against what history?
- After what boundaries?
- With what effect?
- Directed at whom?
- Across how many areas of life?
- And what happened when the other person attempted to resist?
That is why the January 2026 amendments to PD12J are important.
Where the court uses a schedule or table, it may now expressly include examples of an alleged pattern of coercive and controlling behaviour so the judge can assess the abuse holistically rather than focusing only on individual incidents.
But “holistic” does not mean unstructured.
Quite the opposite.
When the allegation is a pattern, evidential structure becomes more important.
1. Evan Stark: coercive control as entrapment and loss of liberty
Much modern thinking begins with the work of sociologist Evan Stark.
Stark’s contribution was to challenge the idea that domestic abuse can adequately be understood by counting assaults.
His coercive-control framework focuses instead upon domination, restriction of autonomy and entrapment.
Control may operate through:
- isolation;
- intimidation;
- economic restriction;
- micro-regulation of everyday life;
- surveillance;
- degradation;
- violence;
- sexual coercion; and
- constraints upon freedom.
The injury is therefore not simply physical.
It is also the progressive restriction of a person’s ability to live autonomously.
What Stark gives us
A conceptual explanation for why seemingly minor actions can become devastating when they operate as part of a regime.
He moves us from:
“How bad was this particular incident?”
to:
“What freedom did this pattern progressively take away?”
The limitation
Stark gives us a powerful theory.
He does not give a litigant a ready-made evidential protocol for deciding whether 2,000 messages prove coercive control.
The conceptual map still needs translating into source evidence, chronology and legally relevant facts.
2. The Duluth Power and Control Wheel: making tactics visible
The Duluth Power and Control Wheel remains one of the best-known ways of explaining domestic abuse.
It maps recurring tactics including:
- intimidation;
- emotional abuse;
- isolation;
- minimising, denying and blaming;
- using children;
- economic abuse;
- coercion and threats; and
- the use of privilege and entitlement.
Its great strength
It makes invisible behaviour legible.
A person who thinks:
“He never really hit me, so perhaps this wasn’t abuse”
can suddenly see the architecture of control.
Its limitation in court
The wheel is fundamentally an explanatory and educational framework.
It is not an evidential finding.
Recognising that conduct resembles “using children” or “economic abuse” does not establish:
- that the event occurred;
- who was responsible;
- why it happened;
- whether it forms a pattern;
- or what legal consequence should follow.
The wheel tells us what to look for.
It does not determine what has been proved.
3. Michael Johnson: not all partner violence has the same context
Michael P. Johnson’s influential typology makes a different contribution.
He distinguishes between contexts including:
- coercive controlling violence — historically described as “intimate terrorism”;
- situational couple violence;
- violent resistance; and
- mutual violent control.
The key insight is that violence cannot be understood merely by counting who hit whom.
The wider control context matters.
Why this matters
Two people may both have sent angry messages.
Two people may both have shouted.
Two people may both have made allegations.
That does not establish that their behaviour served the same function.
One person’s act may be part of a continuing regime of control.
Another may be defensive, reactive or confined to a particular conflict.
The caution
Typologies are useful analytical devices.
Real cases can be messy.
Labels should not replace detailed evidence about behaviour, context and impact.
Nor is Johnson’s typology itself the legal test applied by the Family Court.
4. Marianne Hester: the same family can look different on different “planets”
Marianne Hester’s Three Planet Model addresses another problem altogether.
It describes the tensions between:
- the domestic-abuse sphere;
- the child-protection sphere; and
- the child-contact / Family Court sphere.
The same behaviour may be interpreted differently depending upon which professional system is looking at it.
A survivor may be told by one agency to separate and protect the child, while another system later expects the same parent to facilitate contact.
What Hester adds
A reminder that pattern recognition is not just an individual cognitive problem.
It is also an institutional problem.
Information gets fragmented because different agencies ask different questions.
That is particularly important now that Child Focused Courts increasingly depend upon effective multi-agency information gathering.
5. Emma Katz: children do not merely “witness” coercive control
Emma Katz’s work is especially important for family justice because it places children inside the coercive-control pattern.
Children are not simply passive witnesses to abuse between adults.
They may experience:
- restrictions upon everyday life;
- fear;
- surveillance;
- control over relationships;
- undermining of a parent;
- post-separation abuse;
- stalking dynamics;
- pressure to carry messages;
- and manipulation of the parent-child relationship.
What Katz changes
Instead of asking only:
“Did the child see the abuse?”
we ask:
“How did the regime of control shape the child’s own life?”
That is a much more useful Family Court question.
6. Safe & Together: perpetrator pattern as parenting behaviour
The Safe & Together Model takes pattern analysis into child-protection practice.
Its focus is deliberately perpetrator-pattern based.
It asks professionals to examine:
- what the perpetrator did;
- how that behaviour affected children;
- how it affected family functioning;
- what the protective parent did to promote safety;
- and what responsibility the perpetrator carries as a parent.
Its strength
It resists a familiar systemic error:
looking closely at the protective parent’s coping while treating the perpetrator’s behaviour as background noise.
Its limitation
Safe & Together is a practice model.
It is not a rule of evidence or a substitute for judicial findings.
A practitioner may use it to improve assessment.
A judge still needs evidence upon which findings can lawfully be made.
7. DARA and DASH: risk assessment is a different exercise again
Police and safeguarding practitioners have another task:
how worried should we be now?
The College of Policing’s Domestic Abuse Risk Assessment — DARA — was developed partly because frontline responses had been too incident-focused and insufficiently sensitive to coercive control.
DARA uses structured professional judgment.
It is intended to improve risk identification and safeguarding decisions.
This distinction is essential
Risk assessment is not fact-finding.
A risk tool may properly operate cautiously in the face of uncertainty because safeguarding requires precaution.
A court determining disputed allegations has a different function.
And the evidence base for risk instruments must itself remain open to scrutiny.
A 2026 independent exploratory study of DARA noted that the evidence supporting large-scale use remains comparatively limited and that evaluating tools outside pilot conditions presents significant challenges.
That does not make DARA useless.
It demonstrates why tools should be used for the task they were designed to perform.
8. Jane Monckton Smith: when trajectory becomes a lethality question
Jane Monckton Smith’s Intimate Partner Homicide Timeline takes temporal analysis further.
Her research organises intimate-partner homicide around an eight-stage progression, beginning with histories of control and moving through relationship development, coercive control, triggering events, escalation, changes in thinking, planning and homicide.
Why this work is powerful
It challenges the mythology of the spontaneous “crime of passion”.
It shows why history, control, separation and escalation may matter when assessing very serious risk.
Why caution is essential
Most coercive-control cases do not end in homicide.
It would therefore be irresponsible to take an ordinary Family Court chronology and announce that somebody is “at stage five” of a homicide pathway without an appropriate professional basis.
Monckton Smith’s work is a high-consequence risk lens.
It is not a universal classification system for every difficult co-parenting case.
9. “Paper abuse”, litigation abuse and institutional control
Post-separation control can move into institutions.
Susan Miller and Nicole Smolter used the term “paper abuse” to describe the use of repeated legal proceedings, false reports and procedural mechanisms as a continuation of control.
More recent family-law research has continued examining repeated litigation as a possible method of post-separation coercive control.
This matters because the Family Court itself can become part of the factual pattern.
But this is also an area requiring exceptional care.
Making an application is not abusive merely because the other party dislikes it.
Seeking enforcement is not coercive control merely because it imposes pressure.
Reporting a genuine safeguarding concern is not institutional abuse.
The question is again pattern, function and context.
The court process can be used abusively. But the existence of litigation is not itself proof of litigation abuse.
10. Robert Kaiser: what is the behaviour doing over time?
Robert Kaiser’s emerging Escalation Pattern Analysis adds another useful layer.
Rather than categorising tactics alone, EPA examines change through:
- frequency — is behaviour becoming more frequent?
- intensity — is it becoming more serious, intrusive or harmful?
- inhibitory decline — are boundaries or interventions which previously restrained behaviour becoming less effective?
This asks a different question from Duluth.
Duluth may help identify what kind of tactic is occurring.
Kaiser asks:
what direction is the behaviour moving in?
Its potential
This is particularly useful when a static list of allegations conceals temporal change.
Its present limitation
EPA is still developing.
It should not be presented to the Family Court as though it were an established judicial risk test.
It is best understood, for now, as a promising trajectory lens which can generate better evidential and safeguarding questions.
11. Carisa Carlton: turning “pattern” into something testable
Carisa Carlton’s work interested me because it tackles a problem many frameworks leave unresolved:
how do you operationalise pattern analysis when the evidence is thousands of real-world records?
Carlton is an anthropologist and forensic sociologist whose current Coercive Control Field Check identifies 59 behaviours across 15 behavioural families.
Those families currently include:
- isolation;
- surveillance and monitoring;
- degradation;
- micro-regulation of everyday life;
- economic control and exploitation;
- threats and intimidation;
- gaslighting and reality manipulation;
- sexual coercion and reproductive control;
- children as instruments of control;
- institutional and legal-systems abuse;
- intermittent reward and conditional affection;
- coerced criminality;
- harassment;
- pathologising; and
- violation of protective or restraining orders.
The Field Check itself carries an important warning:
a behavioural match does not establish a legal pattern.
That is the correct starting point.
The forensic method behind it
In Carlton’s published description of her current methodology, a pattern requires four concurrent elements:
- Repetition
- Relatedness
- Continuity
- Asymmetry
The process then goes further.
It uses longitudinal records.
It examines tactic breadth.
It considers temporal distribution.
It looks at both parties rather than coding only one.
It includes a negative-evidence check intended to distinguish coercive control from bilateral hostility or ordinary conflict.
And it examines changes in the recipient’s response across time.
The child-related coding is particularly interesting
Carlton’s current “children as instruments of control” family includes:
- undermining the other parent’s parenting;
- using children as monitors, messengers or leverage;
- custody or child-welfare threats as ongoing control;
- obstructing joint decision-making; and
- gatekeeping access to the child’s providers, records or information.
For private children proceedings, that is immediately recognisable territory.
It moves beyond asking:
“Was the child present when the adults argued?”
and asks:
“Was the child, the child’s information or the parenting relationship itself being used as part of the alleged control?”
That is a much more sophisticated question.
What I think Carlton’s approach gets right — and where caution is needed
There is a lot here that I find valuable.
Strength 1: it tries to operationalise the word “pattern”
Family proceedings use the word constantly.
Yet “there is a pattern” can easily become a conclusion without transparent methodology.
Repetition, relatedness, continuity and asymmetry at least force the analyst to explain why multiple events belong together.
Strength 2: it works at corpus level
One hostile message is not coercive control.
A longitudinal corpus may tell a different story.
This is exactly the problem complex digital Family Court cases now present.
Strength 3: it tests competing explanations
The negative-evidence element is important.
A responsible pattern method should be capable of concluding:
the evidence does not support coercive control.
Otherwise it is not analysis.
It is confirmation.
Strength 4: it understands polite control
This may become increasingly significant in court-monitored communication.
A message can sound reasonable while still functioning to:
- impose an unauthorised obligation;
- extract information;
- control access;
- reopen a settled issue;
- or position the other parent as unreasonable for resisting.
Tone is relevant.
Function may be more important.
Strength 5: children and institutions are inside the model
That reflects the reality of post-separation abuse more effectively than frameworks designed solely around conduct occurring inside an intact intimate relationship.
But there are important limitations
First, this is not an England and Wales Family Court test.
Carlton’s forensic work is currently framed substantially through US legal jurisdictions. Concepts which align with one state’s coercive-control statute cannot simply be imported into PD12J.
Second, the underlying forensic codebook remains proprietary and, in the public material I reviewed, is described as an unpublished research instrument.
The public Field Check is extensive, but it is not the same thing as having the full coding manual, inclusion and exclusion rules and validation material available for independent scrutiny.
Third, external validation matters.
I have not identified in the public material I reviewed a published independent validation study establishing how consistently different trained coders applying the present instrument reach the same result.
That does not establish that the method is unreliable.
It means reliability should not simply be assumed.
Fourth, asymmetry is analytically useful but not itself a legal element of coercive control in England and Wales.
A court still needs to apply the applicable statutory and Family Court framework.
Fifth, written records create selection bias.
A highly strategic person may leave very little explicit evidence.
A traumatised person may write far more.
A litigant who communicates badly may produce a denser paper trail than somebody exercising sophisticated control through apparently reasonable language.
Sixth, thresholds can obscure rare but highly significant conduct.
Some events matter enormously even if they occur once.
A pattern model must not allow recurrence requirements to make a serious one-off act disappear.
Finally, forensic analysis must remain subordinate to the court’s role. An analyst may identify patterns in data. The Family Court determines facts, welfare consequences and orders.
None of those points diminish the value of Carlton’s work.
They identify the questions a serious evidential method should be willing to answer.
Is your coercive-control evidence spread across years of records?
In complex private children cases, the problem is often not lack of evidence.
It is that nobody can see the structure.
JSH Law can provide defined-scope support with:
- domestic-abuse chronologies;
- pattern and evidence mapping;
- source registers;
- schedules of allegations and responses;
- contradiction analysis;
- post-separation abuse timelines;
- digital evidence organisation;
- Cafcass and Child Impact Report analysis;
- witness-statement preparation support;
- position statements; and
- hearing preparation.
The aim is not to label more conduct as abuse. It is to make the evidence capable of being tested.
What does the Family Court actually require?
This is where every framework has to come back to law.
The Family Court is not deciding whether the case satisfies Duluth.
It is not deciding whether somebody falls within Johnson’s “intimate terrorism” category.
It is not conducting an EPA assessment.
It is not deciding whether Carlton’s four-prong pattern definition has been met.
It is deciding the issues before it under the law of England and Wales.
PD12J requires the court to identify whether domestic abuse is relevant, and where disputed facts need determination, to focus upon what is necessary for a reliable assessment of risk and welfare.
Current paragraph 19 expressly encourages holistic examination of alleged patterns.
But the Court of Appeal has repeatedly made another point:
the purpose is not to prove domestic abuse in the abstract.
The relevant findings are those which materially bear upon:
- risk;
- the child’s welfare;
- future parenting arrangements;
- contact;
- participation;
- and any protective measures or restrictions required.
This is why evidence architecture matters.
A pattern has to remain connected to the decision the court actually has to make.
The frameworks compared
| Framework | Best question | Strength | Do not use it as |
|---|---|---|---|
| Stark | How does domination restrict liberty and autonomy? | Theory of entrapment | A document-coding protocol |
| Duluth | What tactics are being used? | Accessible pattern recognition | Proof that an allegation occurred |
| Johnson | What context surrounds the violence? | Distinguishes conflict from coercive regimes | A Family Court legal classification |
| Hester | Why do agencies interpret the same family differently? | System fragmentation | An individual abuse assessment |
| Katz | How does control shape the child’s own life? | Child-centred understanding | A risk score |
| Safe & Together | What is the perpetrator’s parenting pattern and its impact? | Child welfare and accountability | Judicial fact-finding |
| DARA / DASH | What safeguarding risk exists now? | Structured professional judgment | Proof of disputed allegations |
| Monckton Smith | Does the trajectory resemble progression towards homicide? | Temporal high-risk analysis | A universal coercive-control classifier |
| Kaiser EPA | How is behaviour changing over time? | Trajectory and escalation | An established Family Court risk test |
| Carlton | Does the longitudinal corpus support a coherent pattern? | Operational coding and competing-explanation testing | A substitute for legal findings or independent validation |
| PD12J | What pattern matters legally to welfare and risk? | The operative Family Court framework | A scientific risk instrument |
What I would take from all of them: a JSH Law pattern-evidence approach
I do not think the answer is to choose one framework.
The better approach is layered.
For practical Family Court evidence organisation, I would ask:
Layer 1 — Legal relevance
What issue does this evidence actually relate to under PD12J and the child’s welfare?
Layer 2 — Source
Where does each event come from?
Layer 3 — Status
Is it an allegation, admission, contemporaneous record, professional opinion or judicial finding?
Layer 4 — Tactic or function
What does the conduct appear to be doing — monitoring, isolating, degrading, restricting, threatening, obstructing, using children or something else?
Layer 5 — Time
What happens when the events are placed chronologically?
Layer 6 — Change
Is behaviour becoming more frequent, intrusive or varied? What happens after boundaries or intervention?
Layer 7 — Alternative explanation
Could the same evidence reasonably reflect ordinary parenting disagreement, legitimate safeguarding action or mutual conflict?
Layer 8 — Child impact
How did the behaviour affect the child’s safety, relationships, emotional wellbeing or everyday life?
Layer 9 — System interaction
Did police, schools, health professionals, Cafcass or the court see different fragments of the same history?
Layer 10 — Court consequence
What finding, direction, safeguard or welfare decision is the evidence actually capable of supporting?
This is not a diagnostic instrument.
It is not a risk score.
It is not an expert opinion.
It is a way to organise evidence so that a court can see both the individual sources and the possible pattern without confusing one with the other.
How the JSH Law Six-Question Check fits
The JSH Law Six-Question Check provides the evidential discipline underneath that analysis.
Where did this information actually come from?
What has actually been proved?
What surrounds the event and where does it sit in the pattern?
Can the other party understand and respond to the allegation?
What does it mean for the child, safety or welfare?
What does the court or another agency actually need to do?
The more sophisticated our pattern analysis becomes, the more important those safeguards become.
Could AI combine all of these approaches?
Technically, AI could help with much of the mechanical work.
It could:
- extract dates;
- group communications by behavioural category;
- identify repetition;
- calculate intervals;
- compare behaviour before and after court orders;
- identify source duplication;
- map records from different agencies;
- flag possible contradictions;
- and build a provisional chronology.
That could make complex Family Court files substantially easier to analyse.
But it creates an equally serious danger.
Once you tell AI to find coercive control, it may start finding coercive control everywhere.
Confirmation bias can be automated.
An AI may:
- misclassify legitimate parental communication;
- treat multiple reports of one event as independent corroboration;
- infer motive from tone;
- mistake temporal sequence for causation;
- ignore exculpatory material;
- or quietly convert an allegation into a finding.
AI can help find a possible pattern.
It cannot be allowed to decide that the pattern is true.
A responsible workflow would therefore be:
AI extraction → human verification → source checking → alternative explanation → legal relevance → court-ready evidence
not:
Upload messages → ask “Is this coercive control?” → file the answer.
I think my case involves coercive control. What should I actually do?
1. Do not start with a label
Start with events, dates and sources.
2. Preserve the underlying records
Messages, orders, reports and emails matter more than a later summary of them.
3. Mark the evidential status
Separate allegation, admission, professional record and judicial finding.
4. Build the chronology
Patterns are difficult to assess when events remain scattered across folders.
5. Identify possible functions
Was the behaviour monitoring, isolating, obstructing, threatening, financially controlling, using children or something else?
6. Test your own theory
What is the strongest innocent or alternative explanation for the same evidence?
7. Look for change over time
Do not simply count incidents. Ask whether the character or frequency changed.
8. Identify what happened after boundaries
Court orders, police intervention, separation or communication restrictions may provide useful context.
9. Put the child back into the analysis
How did this affect the child’s lived experience and welfare?
10. Identify institutional fragments
Police, school, GP, local authority and Cafcass may each hold different pieces.
11. Avoid overclaiming
Do not turn a research framework into an expert diagnosis.
12. Ask what the court actually needs to decide
A pattern only becomes useful Family Court evidence when it is connected to a material welfare or risk issue.
The future of coercive-control evidence is not another checklist
What strikes me most after looking across these frameworks is that none of them is wrong simply because another exists.
They are solving different problems.
Stark gives us liberty.
Duluth gives us tactics.
Johnson gives us context.
Hester gives us systems.
Katz gives us the child’s lived experience.
Safe & Together gives us perpetrator-pattern parenting analysis.
DARA gives us structured safeguarding judgment.
Monckton Smith gives us high-risk temporal progression.
Kaiser gives us behavioural trajectory.
Carlton gives us an attempt at systematic longitudinal coding.
And the Family Court gives us the legal question:
what has been proved, what matters to welfare and what does the court need to do?
The future, in my view, is not choosing one framework and declaring it the answer.
It is building better bridges between them.
See the tactic. Reconstruct the trajectory. Test the alternative explanation. Identify the child impact. Preserve the source. Then let the court decide the facts.
That is a much safer route than either extreme:
reducing coercive control to isolated incidents, or seeing a pattern simply because we expected to find one.
Need help making a complex evidence pattern intelligible?
JSH Law provides defined-scope support for litigants in person in private children proceedings in England and Wales.
Support can include:
- domestic-abuse and post-separation abuse chronologies;
- evidence-source registers;
- pattern and behaviour timelines;
- contradiction and evidential-gap analysis;
- schedules of allegations and responses;
- C1A preparation support;
- witness-statement preparation support;
- Cafcass and Child Impact Report analysis;
- position statements;
- digital evidence organisation;
- appeal evidence analysis;
- bundle preparation support; and
- hearing preparation and McKenzie Friend support where appropriate.
The objective is not to create a label. It is to make the evidence clear enough to be tested.
Related JSH Law analysis
Selected research and official sources
- Carisa Carlton — Coercive Control Field Check / Coercive Control Observatory
- Robert Kaiser — Women’s Safety Institute / Escalation Pattern Analysis
- Evan Stark — Coercive Control: How Men Entrap Women in Personal Life
- Jane Monckton Smith — Intimate Partner Femicide: An Eight-Stage Progression to Homicide
- Practice Direction 12J — Child Arrangements and Domestic Abuse and Harm
- Home Office — Controlling or Coercive Behaviour Statutory Guidance
