Research into intimate-partner homicide has increasingly moved away from trying to identify a single “type” of killerand towards something much more useful: observable patterns of behaviour and circumstances that are associated with increased risk.
This distinction matters. We cannot look at a person and determine that they will become a killer. But research can identify behaviours that deserve serious attention, particularly when several occur together.
A substantial body of research points to previous domestic violence, coercive control, threats to kill or cause harm, stalking, non-fatal strangulation, weapons, sexual violence and separation/estrangement as important risk factors. PubMed
The strongest message from the research: look at behaviour, not personality
There is no single psychological profile that explains intimate-partner homicide.
A systematic review of theories of intimate-partner homicide found that researchers have proposed multiple explanations spanning feminist, evolutionary, sociological and criminological perspectives. The authors emphasised the complexity of the phenomenon rather than one universal explanation. PubMed
That is why statements such as:
“He is narcissistic, therefore he is dangerous.”
or:
“He had childhood trauma, therefore he might become violent.”
are scientifically inadequate.
Risk assessment is much more meaningful when it asks:
What has this person actually done?
1. Previous domestic violence
Perhaps the most consistently important finding is remarkably straightforward:
Previous violence matters.
A major review identified prior domestic violence as the major risk factor for intimate-partner homicide, alongside other factors including guns, estrangement, forced sex, threats to kill and non-fatal strangulation. PubMed
A later meta-analysis examining 17 studies and 148 effect sizes found several particularly strong risk factors, including previous non-fatal strangulation, previous rape, weapon threats, controlling behaviour and threats to harm the victim. PubMed
This is important because homicide often isn’t an entirely isolated event appearing from nowhere.
A global systematic review concluded that intimate-partner homicide commonly represents the culmination of a history of abuse. PubMed
That doesn’t mean that every abusive partner will eventually kill.
It means that a history of violence should not be treated as irrelevant simply because the relationship has subsequently changed or because the person has behaved well for a period of time.
2. Coercive control
Coercive control is particularly important because domestic abuse isn’t necessarily about individual violent incidents.
It can involve a pattern of domination:
- monitoring;
- isolation;
- controlling finances;
- restricting movement;
- controlling social contacts;
- intimidation;
- threats;
- humiliation;
- surveillance;
- stalking;
- controlling communication;
- punishment for asserting independence.
The meta-analysis of intimate-partner homicide risk found controlling behaviours among the strongest identified risk factors. PubMed
This changes the way we understand danger.
The important question isn’t simply:
“How often did he hit her?”
It can also be:
“How much control did he attempt to exercise over her?”
A relationship can therefore be highly dangerous even when physical assaults are relatively infrequent.
3. Threats to harm or kill
Threats should never automatically be dismissed as merely words.
The research identifies previous threats to harm the victim as an important risk factor for intimate-partner homicide. PubMed
That does not mean every threat will be acted upon.
But from a risk-assessment perspective, a documented threat is behavioural information.
There is a significant difference between:
“I’m angry that we’re separating.”
and:
“If you leave me, I will kill you.”
The second provides information about how the person is responding to the loss of the relationship.
4. Non-fatal strangulation
This is one of the most important findings in the research.
Non-fatal strangulation is not simply another form of physical assault.
It is associated with intimate-partner homicide risk and can produce serious physical and psychological consequences even when there are few visible injuries. PubMed
The meta-analysis by Spencer and colleagues identified previous non-fatal strangulation among the strongest risk factors for intimate-partner homicide. PubMed
That is why professionals should not judge the seriousness of strangulation by asking:
“Was there a visible injury?”
The absence of obvious external injury does not mean the event was medically or psychologically insignificant.
5. Weapons
Access to a firearm or other weapon can substantially change the potential consequences of an already violent relationship.
The research has repeatedly identified weapon access and previous weapon threats as important factors in intimate-partner homicide risk. PubMed
Importantly, weapons don’t need to be used for their presence to matter.
A weapon can become part of intimidation and coercive control.
Research examining coercive control, stalking and guns found that separation and stalking could contribute to perceptions of increased risk of potentially fatal gun violence, illustrating why apparently “non-physical” behaviours can be relevant to risk assessment. PubMed
6. Stalking
Stalking is another behaviour that deserves particular attention.
It can include:
- repeatedly appearing where someone is;
- unwanted calls or messages;
- monitoring;
- following;
- unwanted visits;
- surveillance;
- contacting friends or relatives;
- monitoring online activity.
Stalking is not simply evidence that somebody is “having trouble letting go.”
It can represent continuing access-seeking and control after separation.
Research examining intimate-partner violence has identified stalking among behaviours relevant to serious risk assessment, particularly when combined with coercive control and separation. PubMed
7. Separation
One of the most misunderstood findings is that leaving does not necessarily mean the risk immediately disappears.
Estrangement or separation has repeatedly appeared in intimate-partner homicide research as an important factor. PubMed
Why?
Because separation can fundamentally alter the power structure of a coercively controlling relationship.
The controlling partner can no longer exercise the same day-to-day access or authority.
The other person may begin making independent decisions.
They may establish new relationships.
They may obtain their own finances.
They may move.
They may seek legal protection.
For someone whose behaviour has been organised around control, those changes can become highly significant.
The research therefore supports treating separation as a period requiring careful risk assessment in already abusive relationships, rather than assuming that leaving automatically ends the danger. PubMed
8. Sexual violence
Sexual violence is another serious marker.
The major meta-analysis identified previous rape of the victim among the strongest risk factors it examined. PubMed
This reinforces an important principle:
Domestic abuse isn’t only about physical assault.
Sexual coercion, forced sex and reproductive control can be part of a broader pattern of domination.
9. Multiple risk factors matter
Perhaps the most important lesson is that risk assessment should not depend upon one behaviour in isolation.
Consider the difference between:
A relationship with conflict but no history of violence
and:
Previous violence + coercive control + threats + stalking + strangulation + weapon access + separation.
The second pattern contains several independently concerning features identified in the research literature.
That doesn’t mean anyone can calculate with certainty:
“This person will kill.”
Risk assessment doesn’t work like that.
Rather, the accumulation of established risk factors can indicate that greater safeguarding and professional assessment are warranted.
Researchers have therefore developed structured tools such as the Danger Assessment to help professionals systematically examine combinations of risk factors rather than relying solely on intuition. The instrument was validated using cases of intimate-partner femicide and abused women as controls. PubMed
What about psychology and neuroscience?
This is where we need to be careful.
It is tempting to explain intimate-partner homicide entirely through rage, abandonment, narcissism, childhood trauma or an inability to control emotions.
Research does not support such a simple explanation.
A recent systematic review of cognitive factors in intimate-partner and sexual violence examined research on beliefs and cognitions associated with perpetration and victimisation, showing that cognition is an important area of research — but it is one part of a much larger picture. PubMed
The more useful psychological concepts include:
entitlement — “I have a right to control what you do.”
externalisation of responsibility — “You caused my behaviour.”
hostile interpretations — interpreting ordinary independence or boundaries as deliberate provocation.
possessiveness — treating a partner’s autonomy as a threat to the relationship.
coercive control — using multiple behaviours to restrict another person’s freedom.
But these concepts should be used to understand behaviour, not to diagnose an individual from a distance.
The most important conclusion from the research
The evidence does not tell us:
“This type of person becomes a killer.”
It tells us something more useful:
Certain behaviours and circumstances repeatedly appear in cases of severe and lethal intimate-partner violence.
And the strongest signals are often not mysterious.
They are observable.
Previous violence.
Coercive control.
Threats.
Stalking.
Non-fatal strangulation.
Sexual violence.
Weapons or weapon threats.
Separation or estrangement.
When several occur together, they should not be minimised simply because there has not yet been a lethal incident.
The purpose of studying these factors is not to predict someone’s future with certainty.
It is to recognise patterns early enough to take danger seriously.
As the research literature increasingly demonstrates, intimate-partner homicide is often not an inexplicable act appearing from nowhere. It can occur within a much longer history of violence, control, threats and escalating behaviour. PubMed
And that is perhaps the most important message for anyone working with survivors:
Don’t wait for the worst possible outcome before taking the pattern seriously.
If someone has experienced strangulation, threats to kill, stalking, escalating violence or coercive control, those aren’t merely historical details. They are behaviours that the research tells us deserve careful professional risk assessment. PubMed
