Animal Cruelty as a Warning Sign: How to Use It Well

The 43 percent figure already has its own piece on this site. It comes from Arluke and Madfis (2014), a study of 23 perpetrators of school massacres, not a study of serial killers. That piece is about how the number got garbled. This one is about a different question: even when the number is right, how should it be used?

That question matters more than it sounds. Behavioural threat assessment is an established procedure in American schools, and animal cruelty is a signal that recurs inside it. There is clear professional consensus on how such a signal should be used and where the use goes wrong. It is rarely written up for a general audience.

The short version

Animal cruelty is a signal worth recording in threat assessment. It is not a test you can screen people with. That distinction is not caution about wording; it is a difference in method. A warning sign exists to trigger assessment and information gathering, not to deliver a conclusion about a person. Used as a screening tool, it fails arithmetically.

The 43 percent figure is accurate, but narrow

The 2014 paper by Arluke and Madfis in Homicide Studies is titled Animal Abuse as a Warning Sign of School Massacres. It examined 23 perpetrators of school massacres between 1988 and 2012.

Of those 23, 43 percent had committed animal cruelty before the attack. The study also found that the cruelty typically targeted anthropomorphised species — dogs and cats — and was usually carried out up close. Both details are meaningful: they point not to someone who dislikes animals, but to a pattern connected to empathy and to a specific victim.

One thing needs stating clearly, and it is the most commonly dropped detail. The authors explicitly discuss the use of their findings as reducing false positive cases. They were not producing a screening checklist. They were warning that misuse of the signal is as real a risk as missing it.

The abstract itself notes that existing research reveals mixed support for this notion and lacks conceptual clarity in the measurement. Even the researchers are saying the field is not clean enough yet to be turned directly into a tool.

The proper use: trigger assessment, not labels

The professional procedure is behavioural threat assessment. It is not one person's judgment. It is a process: gather information from multiple sources, assess it as a team, then provide intervention and support.

Inside that process, animal cruelty sits in a specific place. It is a piece of information worth recording and worth following up. When a school, a parent, or a peer notices a student harming an animal, the correct action is to place it into a fuller picture — what has been happening to this person recently, whether there are other signals, whether anyone is still in contact with them.

Its function is to start the assessment. It authorises no one to move directly from this child harmed an animal to this child will harm people.

Why it fails as a screening tool

There is a concrete statistical point here, and it is worth walking through plainly.

Suppose a signal is genuinely true for only a few people in a hundred. Then even if the signal appears at a high rate among true cases, its share of everyone who shows the signal stays low. This is the base rate problem: the wider the population you screen, the more false positives you generate.

Apply that to animal cruelty. The overwhelming majority of children who harm animals do not become violent offenders — the research is clear on this. If a school treated every student with a history of animal cruelty as a potential threat, it would produce a large number of children treated wrongly, and the few who genuinely needed help would be buried inside that group.

The failure is not moral. It is arithmetic. The signal's predictive validity is not sufficient, and the population base rate is low. That is why professional threat assessment never rests on a single indicator, and requires multiple sources and multiple signals to converge.

Where cruelty sits in threat assessment

In the models promoted at federal and state level, the information to be gathered falls into a few broad categories: what the person has communicated about intent, whether there is planning and preparation, whether there is access to weapons, and prior behavioural history.

Animal cruelty falls into the last category: behavioural history. Its role is to supply context, not to constitute a judgment on its own. Within that category it is usually recorded alongside violence toward people, self-harm, experiences of bullying, and major life disruptions.

Placing it there has a practical benefit. It is not seen in isolation. A single behavioural item standing alone is easily read as a label; inside a picture built from a dozen items it becomes a lead that needs explaining.

What parents can do

If you discover a child in your household harming an animal, the thing not to do is treat it as a small matter that will pass. Nor is it to immediately turn it into a verdict about the child.

What to do is treat it seriously and concretely: establish what happened, what the circumstances were, and how the child describes it. Often the child's account points to something the parents did not know — being bullied, a death in the family, some loss in progress.

If a child is harming animals, the direct step is a professional psychological assessment. That is not labelling the child. It is precisely how you avoid having the child casually labelled by someone else. An assessment produces a fuller picture, which is the exact thing every misuse of this signal lacks.

If your concern is someone else's child, or a classmate, the route is the school's counselling resources or a local child protection line — not public discussion of the individual online. Public identification is the action most likely to cause harm here and the least likely to protect anyone.

What schools and institutions can do

Adopt a written threat assessment process rather than relying on one teacher's instinct. The value of a written process is that it specifies what information is collected, who makes the judgment, and what happens after the judgment.

Write animal cruelty explicitly into the list of behaviors that must be recorded. Many school lists already include self-harm, threats, and bullying, but not this. Adding it costs nothing beyond making its place clear in training.

Training should focus on what to do after noticing, not on how to identify who. The first is a process. The second is a label. That difference determines whether the system helps people or harms them.

Staying consistent with the researchers' own caveat

Return to the paper. The authors use the term warning sign, which in professional usage has a defined meaning: an input that triggers further assessment, not a conclusion.

They also explicitly discuss reducing false positives. That part has been almost entirely lost in popular retelling, in Chinese and English alike, leaving only the 43 percent. Presented alone, 43 percent reads as though nearly half of school shooters abused animals, so nearly half of people who abuse animals will shoot — a conclusion the original study does not support.

How this divides from the other two pieces

Two related articles on this site cover adjacent ground. The violence triangle piece covers the criminological concept, where it came from and where it now stands in the research. The 43 percent piece covers what happened to one number as it circulated. This piece covers application and misapplication — what to actually do when you are holding one of these signals.

Together the three form a chain: where the concept came from, how the number has been used, and what the practical response is. Read any one alone and it is easy to reach a lopsided conclusion.

Things not to do

Do not use a population figure to draw conclusions about a specific person. A rate within a studied group does not transfer to an individual.

Do not publicly identify or discuss a specific minor. Right or wrong, public identification causes irreversible harm and produces almost no protective effect.

Do not assemble animal cruelty and a few other isolated behaviors into a profile. Threat assessment is a holistic judgment made by a team after gathering information. It is not what you get by adding several traits together.

Do not let this material reduce how seriously you take animal cruelty itself. Everything here concerns the boundaries of using the signal, not the signal's importance. The opposite is true: it is worth recording carefully and worth including in assessment. It is simply not worth treating as a conclusion.

How this connects to Wangwang

In Jieyang, the offenders were under 14, were not prosecuted, and were placed in specialist schools for correctional education. That is an educational route, not an assessment route.

The question the research points to applies here too, and more concretely: whether the intervention is real, whether anyone follows up, and whether the signal was recorded anywhere at all. If it was not recorded, it never gets a chance to matter at the next step.

Taking a signal seriously and using it to define a person are two different things. The first requires a system. The second requires only a number. That is the whole of what this piece is saying.