A Framework for Investigating Harm, Establishing Accountability, and Changing Outcomes
By Andrew Paul Klein
Understanding is where we begin. Evidence is how we learn. Accountability establishes what must change. The reduction of harm is how we judge whether we have succeeded.
There is a persistent problem in the way human beings respond to harm. We are often quicker to judge than to understand, quicker to assign blame than to investigate causes, and quicker to announce a solution than to establish whether it works.
We see the consequences of an action and search for someone to hold responsible. Sometimes that is entirely appropriate. People make choices, exercise power, and must answer for the consequences of their conduct.
But responsibility and understanding are not opposing principles. Indeed, accountability becomes more meaningful when we understand what happened, why it happened, who was affected, and what conditions allowed it to occur.
To understand a person is not to excuse their conduct. To establish accountability is not to abandon our responsibility to understand.
This distinction is fundamental to the work proposed here.
Our purpose is not to replace judgement with endless deliberation, nor to suggest that every harmful act can be explained away by circumstances. It is to develop a more effective response to harm: one that examines individual behaviour alongside the social, institutional, economic, and cultural conditions in which that behaviour occurs.
The question is not simply who did something wrong.
It is also how the circumstances arose, what sustained them, why existing safeguards failed, and what can be changed to reduce the likelihood of recurrence.
These are questions that demand investigation rather than assumption.
1. The person behind the behaviour
Human beings are complicated. Their behaviour emerges from the interaction of personal choices, experience, relationships, expectations, opportunities, and constraints.
Sometimes people cause harm deliberately. Sometimes they act through ignorance, fear, prejudice, desperation, or a failure to appreciate the consequences of their actions. Sometimes they operate within institutions that reward conduct they might otherwise question.
These explanations are not interchangeable, and we should not assume that one applies simply because it makes a compelling story.
They are possibilities to investigate.
If we begin by deciding what kind of person we are dealing with, we risk interpreting every subsequent observation to confirm our initial judgement. Evidence becomes a means of defending our opinion rather than testing it.
A more disciplined approach begins with what can be established.
What happened?
What did the people involved know at the time?
What choices were available to them?
What pressures or incentives influenced their decisions?
Who possessed the authority to act differently?
What consequences followed?
And what evidence would help us distinguish between competing explanations?
These questions do not remove moral responsibility. They help us locate it more accurately.
Understanding a person requires us to look beyond the label we have assigned them. Understanding harmful behaviour requires us to examine the circumstances without allowing those circumstances to erase the agency of the person who acted.
We can recognise someone’s humanity while remaining clear about the consequences of their conduct.
That is not weakness. It is a more demanding form of accountability.
2. From individual conduct to repeating patterns
An isolated incident may reveal something about an individual. Repeated incidents may reveal something more.
When similar forms of harm recur across different people, workplaces, institutions, or periods of time, we should ask whether a common set of conditions is contributing to the outcome.
Perhaps warnings are routinely ignored.
Perhaps the people affected have little power to make themselves heard.
Perhaps an organisation rewards short-term financial performance while transferring the costs to workers, families, or the public.
Perhaps the language used to describe a problem makes its consequences less visible.
Perhaps the formal safeguards exist, but nobody is responsible for ensuring that they work.
None of these possibilities should be accepted without evidence. But neither should they be dismissed simply because investigating them would be inconvenient.
A repeating pattern invites a different kind of question.
Instead of asking only why one person behaved badly, we investigate whether the surrounding system makes similar behaviour more likely, easier to conceal, or less costly to those responsible.
This is where the investigation of harm becomes particularly important.
When harm repeatedly occurs under similar conditions, we must investigate the conditions as well as the individuals involved.
A pattern does not prove that every incident has the same cause. It gives us a reason to investigate whether common mechanisms are operating.
That distinction matters. Without it, we risk replacing one form of premature judgement with another.
The objective is not to discover a convenient villain. It is to discover what is happening.
3. The role of institutions
Institutions are not merely collections of individuals. They establish rules, distribute authority, allocate resources, define acceptable conduct, and determine which outcomes are rewarded.
Their influence extends beyond what they formally instruct people to do.
An institution may publicly endorse compassion while rewarding behaviour that leaves vulnerable people without adequate support.
It may declare that safety is paramount while imposing working conditions that encourage unsafe practices.
It may establish complaint procedures while making those who complain fear retaliation.
It may describe a policy as efficient while failing to account for the costs transferred to households, communities, or future generations.
In each case, there is a difference between what the institution says it values and what its arrangements actually encourage.
That difference deserves investigation.
We should examine the formal rules, but we should also examine the incentives, reporting structures, resource allocations, and practical consequences that shape behaviour.
Who benefits from the existing arrangement?
Who bears its costs?
Who has the authority to change it?
Who is expected to explain the outcomes?
And what happens when the evidence contradicts the institution’s preferred account of itself?
These questions are relevant to government, business, education, health care, policing, and the many other settings in which people exercise power over others.
They also help us avoid a common mistake: treating every harmful outcome as the result of a few defective individuals when the organisation itself may be contributing to the problem.
Equally, we must avoid the opposite mistake of blaming an abstract system so completely that individual decisions disappear from view.
Both levels matter.
We need to understand how people act within institutions and how institutions shape the choices available to people.
4. Language, evidence, and the construction of explanations
Language is one of the means by which people understand the world. It is also one of the means by which they can obscure it.
A description may be technically accurate yet leave out the information most necessary for understanding its consequences.
A policy may be presented as reform without a clear account of who will benefit or who will bear the cost.
An organisation may describe a reduction in expenditure as an efficiency improvement without demonstrating whether the underlying service has improved.
A government may announce a successful intervention by measuring the activities undertaken rather than the harm actually prevented.
These are possibilities to test, not conclusions to impose in advance.
The appropriate response is to compare the explanation with the evidence.
What exactly is being claimed?
How has success been defined?
Which outcomes have been measured?
Which have been omitted?
What do the available data establish, and what remains uncertain?
Can the claim be independently examined?
We should be particularly careful when the language of an institution becomes the principal evidence offered in support of its performance.
An organisation’s description of its own success is a claim to be evaluated, not a substitute for evaluation.
This principle applies to our own work as well.
We must not become so attached to an explanation that contrary evidence becomes something to overcome rather than something to learn from.
If our purpose is to reduce harm, our explanations must remain open to revision.
An explanation earns our confidence not because it is persuasive, but because it survives examination against the evidence.
5. Accountability: establishing what must change
Accountability is sometimes reduced to punishment.
Punishment may be justified, particularly where people have deliberately caused harm or abused positions of trust. But punishment alone does not necessarily explain why an event occurred, repair its consequences, or prevent its recurrence.
A meaningful system of accountability must ask several questions.
Who was responsible for the relevant decisions?
What did they know, and what should reasonably have been known?
Were their actions consistent with their duties?
Were the available safeguards adequate?
Did anyone fail to act when action was required?
Were the consequences foreseeable?
And what must now change?
The answers may establish individual responsibility, institutional responsibility, or both.
They may also reveal that a process was poorly designed, that authority was distributed without adequate oversight, or that people were placed in circumstances where foreseeable failures became more likely.
Each finding requires a different response.
An individual who deliberately abuses authority may require disciplinary or legal consequences.
An institution that repeatedly ignores credible warnings may require independent oversight, changes to its governance, or enforceable obligations.
A policy that produces harmful outcomes may need to be redesigned.
A safeguard that exists only on paper must be made effective.
And where people have suffered harm, their needs must not disappear behind the administrative process of deciding who was at fault.
Accountability should therefore do more than identify a past failure. It should establish the responsibilities, corrective actions, and verification procedures necessary to address it.
Without that final step, an investigation may produce a report without producing meaningful change.
6. Finding the points where harm can be interrupted
If harmful patterns are sustained by identifiable conditions, then those conditions may provide opportunities for intervention.
The challenge is to discover where an intervention can make a meaningful difference.
Consider a hypothetical institution in which complaints about mistreatment repeatedly fail to produce action.
An initial investigation might focus on the people accused of misconduct. That is necessary, but it may not be sufficient.
Further investigation might reveal that complaints are assessed by managers whose performance depends on avoiding reputational damage, that reporting procedures are difficult to access, or that employees have little confidence in the independence of the investigation.
These findings would suggest several possible interventions.
Complaints could be reviewed independently. Reporting procedures could be simplified. Retaliation could be monitored. Decision-making responsibilities could be clarified. Outcomes could be published in a form that protects privacy while allowing external scrutiny.
But these are interventions to evaluate, not solutions to assume effective.
We would need to establish whether the changes were implemented, whether people could use the revised system, whether complaints received appropriate consideration, and whether the incidence or severity of harm subsequently changed.
We would also need to consider unintended consequences.
For example, an increase in reported incidents might indicate that harm had become more common. Alternatively, it might mean that people finally felt safe enough to report what had previously remained hidden.
The numbers alone would not settle the question.
We would need to examine the quality of the evidence, the circumstances surrounding the change, and the experiences of the people affected.
This illustrates why practical interventions require more than good intentions.
We must identify the proposed mechanism of change, establish a baseline, measure the results, and remain willing to revise the intervention when the evidence requires it.
The purpose is not to defend a particular solution. It is to discover which actions make a demonstrable difference.
7. Measuring success by the reduction of harm
Organisations frequently measure what is easiest to count.
They record expenditure, completed activities, processed applications, implemented policies, and compliance with formal procedures.
These measures can be useful. But they do not necessarily tell us whether people’s circumstances have improved.
A service can process more applications while leaving the underlying need unchanged.
A workplace can complete its safety training while continuing to expose workers to avoidable risks.
An agency can meet its administrative targets while failing to provide meaningful assistance to those who need it.
The distinction is between activity and outcome.
If we want to know whether an intervention has succeeded, we must identify the harm it was intended to reduce and establish how that harm can be measured.
Depending on the problem, relevant measures might include:
- The frequency and severity of harmful incidents.
- The number of people exposed to a particular risk.
- The time taken to respond to warnings or complaints.
- The proportion of cases in which corrective action is completed.
- The recurrence of the same failure after intervention.
- The experiences of the people directly affected.
- The distribution of benefits and burdens across different groups.
- The financial and social costs transferred to households or the wider community.
Not every measure will be appropriate to every problem. Some harms are difficult to quantify, and numerical indicators can conceal important differences between individual circumstances.
Qualitative evidence, case studies, interviews, and the testimony of affected people may therefore be essential.
Where possible, findings should be compared with a credible baseline and, when feasible, with a suitable comparison group or other method of assessing what might have happened without the intervention.
We should also distinguish between a change that follows an intervention and a change that the intervention actually caused.
These distinctions help prevent us from claiming success prematurely.
Ultimately, the question is not whether an organisation has completed the activities it promised to undertake.
It is whether the intervention has made a meaningful, demonstrable, and preferably sustained difference to the harm experienced by people.
8. A practical framework for investigation
The principles developed here can be translated into a sequence of investigative steps.
| Stage | Central question | Practical task |
| 1. Understand | What happened, and who was affected? | Establish the circumstances, listen to those affected, and distinguish observation from assumption. |
| 2. Investigate | What explains the outcome? | Examine individual decisions, institutional arrangements, incentives, and available evidence. |
| 3. Identify patterns | Has similar harm occurred before? | Look for recurrence across cases, locations, organisations, or periods. |
| 4. Establish accountability | Who was responsible for what? | Examine authority, duties, knowledge, decisions, omissions, and consequences. |
| 5. Identify intervention points | What could interrupt the pattern? | Develop specific changes linked to an evidence-based explanation of the problem. |
| 6. Evaluate | Did the intervention work? | Compare outcomes with an appropriate baseline, investigate unintended effects, and assess uncertainty. |
| 7. Learn and revise | What must happen next? | Retain effective measures, change ineffective ones, and publish findings that permit scrutiny. |
This is not a rigid formula.
Some investigations will require returning to earlier stages. New evidence may alter the explanation. An intervention may reveal that the original problem was misunderstood. A measure that works in one setting may fail in another.
The framework must accommodate those possibilities.
Its value lies in providing a disciplined way to move from concern to investigation, from investigation to action, and from action to evaluation.
It also creates a record of reasoning that others can examine.
That matters because a framework designed to reduce harm must itself remain accountable.
Its conclusions should be open to challenge. Its evidence should be available for scrutiny wherever privacy, confidentiality, and legitimate legal constraints permit. Its recommendations should be proportionate to what the investigation establishes.
And its authors must be willing to discover that they were wrong.
9. Understanding without surrendering judgement
There is a possible objection to this approach.
If we devote so much attention to understanding the causes of harmful behaviour, do we risk becoming reluctant to condemn it?
I believe the answer depends on what we mean by understanding.
If understanding means inventing excuses for people whose conduct we dislike, then it has little value.
But if understanding means establishing what happened, identifying responsibility, examining the conditions that contributed to the outcome, and determining what must change, then it strengthens rather than weakens accountability.
A person can be held responsible for a decision while we investigate the circumstances in which it was made.
An institution can be required to correct its practices while we recognise the legitimate constraints under which its employees operate.
A policy can be judged by its consequences without assuming that everyone who supported it acted with malicious intent.
And an explanation can be rejected when the evidence does not support it, even when the explanation is emotionally satisfying or politically convenient.
The objective is not to suspend moral judgement indefinitely.
It is to make our judgements better informed, more proportionate, and more likely to lead to effective action.
There will be circumstances in which immediate protection must take precedence over a complete investigation. There will be occasions when the evidence establishes responsibility clearly. There will also be cases in which uncertainty remains.
A responsible framework must recognise these differences.
Understanding is not a reason to delay necessary protection. Nor is uncertainty a reason to abandon investigation.
The work requires both compassion and discipline.
10. From principle to research
The next stage is empirical.
The framework should be tested against actual cases of recurring harm, selected carefully enough to permit detailed investigation.
Rather than attempting to explain every social problem at once, we could begin with a small number of cases in which there is sufficient documentation to reconstruct what happened.
For each case, we would seek to establish:
- The documented harm: What occurred, how frequently, and with what consequences?
- The decisions involved: Who made the relevant decisions, under what authority, and with what information?
- The institutional conditions: What rules, incentives, resources, or organisational arrangements influenced the outcome?
- The available warnings: What evidence of risk existed, and how was it handled?
- The explanatory claims: How did the responsible individuals or institutions describe the problem?
- The accountability response: What investigation, remedy, or corrective action followed?
- The intervention: What changed, and why was that change expected to help?
- The observed outcome: Did the harm decrease, persist, or take a different form?
- The remaining uncertainty: What cannot be established from the available evidence?
The resulting case studies would allow us to examine whether the framework is useful in practice.
They would also help distinguish between different explanations for recurring harm.
Perhaps a problem persists because responsibility is unclear. Perhaps the institution lacks the resources necessary to respond. Perhaps the incentives reward the wrong outcomes. Perhaps the people affected cannot access an effective remedy. Perhaps several mechanisms operate together.
These possibilities should not be collapsed into a single theory before the evidence is examined.
The research should also document unsuccessful interventions. Knowing why an apparently reasonable solution failed may be as valuable as identifying one that worked.
Over time, this process could produce a body of comparative evidence about the conditions under which particular forms of harm arise, the interventions that interrupt them, and the circumstances in which those interventions succeed or fail.
That would move the project beyond an ethical argument and towards a practical research programme.
Its claims would become more precise as the evidence accumulated.
Its methods would become more useful as their limitations were exposed and corrected.
And its value would be judged by the difference it helped people make.
Conclusion: What does success look like?
We cannot change yesterday’s harm.
We can investigate how it happened, refuse to disguise it, and work to ensure that tomorrow does not simply reproduce it.
That requires us to resist two temptations.
The first is the temptation to judge before we understand. It encourages us to settle for explanations that confirm our existing beliefs, even when they leave the underlying problem untouched.
The second is the temptation to understand without acting. It allows investigation to become an end in itself, producing reports, recommendations, and expressions of concern while the same harm continues.
Neither is sufficient.
We need to understand what happened. We need evidence capable of distinguishing between competing explanations. We need accountability that identifies what must change. And we need to evaluate whether those changes have made a difference.
We must also remain willing to revise our conclusions when the evidence contradicts them.
This approach does not promise that every problem can be solved or that every harmful pattern can be eliminated. Some interventions will fail. Some causes will remain uncertain. Some conflicts between legitimate interests will resist simple resolution.
Intellectual honesty requires us to acknowledge these limits.
But uncertainty need not prevent us from making progress. We can identify what is known, investigate what remains unclear, protect people from established risks, and test practical measures while continuing to learn.
The measure of our work is not how confidently we condemn the past, how elegantly we explain it, or how impressive our recommendations appear on paper.
It is whether we have helped prevent avoidable suffering.
The principle with which we began therefore becomes the standard by which the entire project must be judged:
Understanding is where we begin. Evidence is how we learn. Accountability establishes what must change. The reduction of harm is how we judge whether we have succeeded.
That is the beginning of a framework.
Its credibility will depend on the quality of its investigations, its willingness to confront inconvenient evidence, and its capacity to translate understanding into effective action.
Its purpose is simple, even if its application will not be:
To leave people less exposed to harm than we found them.
And to ensure that, when we encounter a repeating pattern of suffering, we do not merely learn to recognise it.
We learn how to change it.
