The Stories We Tell About Change Are Part of the Change Itself

The hidden problem with trying to explain causality

What if the hardest part of proving change is that every explanation of change already changes the thing being explained?

That sounds like a paradox, but it is closer to a practical truth. In public policy, program evaluation, organizational learning, and even everyday decision making, we often treat causality as if it were a clean line: first an intervention, then an outcome, then a tidy explanation. Yet real life rarely behaves like a laboratory. Effects arrive through winding routes, mixed motives, partial visibility, and human interpretation. The moment we try to describe those routes, we do not merely report reality. We shape it.

This is why a simple search for “what worked” often disappoints. A project may succeed because of a formal strategy, but also because of informal trust, timing, narrative framing, and the expectations created by the very act of evaluation. If you ignore those layers, your causal account may look precise while being shallow. If you include them, the story becomes messier, but also more truthful.

The deeper question is not just, “Did it work?” It is: How do explanations travel through human systems, and what do they do while they are traveling?

Why causality is never just mechanical

A common mistake is to imagine causality as a machine with visible gears. Press button A, outcome B appears. In that view, the evaluator is a mechanic, and the job is to trace the parts. But social systems are not machines. They are partly made of interpretation. People respond not only to events, but to meanings, expectations, and stories about events.

Consider a neighborhood violence prevention program. A report might say the program worked because it created after-school activities, increased mentoring, and reduced idle time. All of that may be true. But the same program may also have worked because it altered the story young people told about the neighborhood, because parents believed the area was finally being taken seriously, or because local leaders began coordinating differently once the program lent them legitimacy.

Those are still causal pathways. They are just not purely material ones. They are pathways of attention, trust, identity, and interpretation. That matters because if you only measure the visible outputs, you may miss the real engine of change.

In human systems, the explanation is often part of the mechanism.

This is one reason linear causality can become misleading. It assumes the world is divided neatly into cause, effect, and observer. But the observer is never outside the system. The act of documenting success can change incentives, which changes behavior, which changes the next round of evidence. The narrative is not decorative. It is operational.

The double life of narrative: method and meaning

Narrative is often treated as something soft, subjective, or supplemental, useful for making data more engaging. That view misses its deeper role. Narrative is both an epistemic tool and a causal force. It helps us know, and it helps produce what we know.

Think about a company trying to turn around a failing product. The leadership team can circulate dashboards, run experiments, and restructure teams. But if the internal story remains, “This company always misses the moment,” people will hesitate, delay, and protect themselves. If the story changes to, “We are the kind of organization that learns fast,” the same people may take different risks, interpret setbacks differently, and coordinate with more confidence.

The numbers do not disappear. But numbers rarely act alone. They become persuasive only when they are embedded in a narrative that explains why they matter, what they mean, and what should happen next. In this sense, narrative is not the enemy of rigor. It is one of the ways rigor becomes socially legible.

This is where many analyses go wrong. They try to separate the “real” causal chain from the “merely” narrative one. But in practice, the narrative often helps assemble the chain. It identifies actors, assigns relevance, and connects scattered events into a sequence people can act on. Without that connective tissue, evidence remains inert.

A useful way to think about this is to distinguish between two levels of causality:

1. Material pathways: resources, timing, infrastructure, incentives, procedures.
2. Interpretive pathways: meaning, legitimacy, confidence, shared expectations, public memory.

Most real-world change requires both. A policy can be well designed in material terms and still fail if the narrative around it creates suspicion. A grassroots initiative can be underfunded but still succeed if it generates a compelling shared account of why participation matters.

A better model: causal pathways as story loops

If causality is partly narrative, then maybe the best way to understand change is not as a straight line, but as a story loop.

A story loop works like this. An action produces an observable result. That result is interpreted through a narrative. The narrative changes behavior, coordination, or belief. That altered behavior then shapes the next result. In other words, the story does not just describe the loop. It becomes one of the loop’s moving parts.

Picture a school introducing a reading intervention. The first test scores improve modestly. If the school tells the story, “This confirms our students can grow quickly when we support them properly,” teachers may invest more energy, students may feel more capable, and families may engage more. If instead the story is, “The district finally gave us an experimental program, but we do not know if it is real,” participation may remain tepid. The causal pathway is not only in the curriculum. It is in the social meaning attached to the curriculum.

This helps explain why some interventions fail to scale even when their pilot results are promising. The pilot worked not only because of its formal components, but because of the local story surrounding it: a charismatic leader, unusually high attention, early novelty, or tight-knit collaboration. When the program is exported, the story loop breaks. The mechanism was never simply the intervention. It was the intervention plus the narrative infrastructure around it.

That phrase, narrative infrastructure, deserves attention. It refers to the stories, symbols, routines, and legitimating frames that allow action to spread. Just as roads and broadband enable physical movement, narrative infrastructure enables confidence, coordination, and adoption. An organization with weak narrative infrastructure may have excellent plans and still struggle to execute them. An organization with strong narrative infrastructure can often make fragile initiatives look more durable than they are.

What looks like persuasion is often coordination. What looks like messaging is often mechanism.

The epistemic risk: when explanations become too tidy

There is a danger, however, in embracing narrative too enthusiastically. Once we recognize that stories matter, we may begin to prefer stories that are elegant over stories that are true. And that is where the epistemic problem begins.

Humans love coherence. We want causes to line up neatly, motives to be intelligible, and timelines to close cleanly. But social change is frequently hybrid, incomplete, and contradictory. A program may work for multiple reasons, some planned and some accidental. An initiative may succeed in one place because of the people delivering it, not because of the model itself. A reform may fail not because the logic was wrong, but because the timing was off or the institution’s memory resisted it.

The temptation is to compress this complexity into a polished account. But when we do, we risk confusing explanatory beauty with explanatory power.

A more disciplined approach is to treat narratives as hypotheses, not trophies. Ask: What did this story make visible that raw data alone did not? What did it obscure? Which actors benefited from this interpretation? Which pathways were strengthened by repeated telling? Which were weakened?

This is especially important in evaluation and reporting. If an organization only tells success stories, it may accidentally train itself to ignore weak signals. If it only tells failure stories, it may paralyze experimentation. The point is not to eliminate narrative. The point is to make narrative accountable to reality while recognizing that reality is partly socially constructed through narrative.

The practical synthesis: evaluate causes, then evaluate the story of the causes

The real breakthrough is to stop treating causal analysis and narrative analysis as separate exercises. They belong in the same room.

A robust inquiry asks two questions in parallel:

1. What happened, through which pathways?
2. What story about what happened changed how people acted afterward?

This second question is often neglected, but it is crucial. A campaign may not only generate an outcome, it may generate a legend about the outcome. That legend can attract support, alter behavior, and influence future decisions more powerfully than the original intervention.

For instance, imagine a city launches a pilot to reduce homelessness through coordinated housing support. The pilot produces moderate but real gains. If the public narrative becomes, “Housing first is the only humane and effective path,” the policy may gain momentum beyond the pilot’s direct effects. If the narrative becomes, “This was a narrow success that depended on exceptional staff,” the same evidence may stall. In both cases, the story changes the future of the intervention.

This suggests a simple but profound principle: every causal explanation has a second life as a social object. People repeat it, contest it, use it to justify resources, and build identity around it. The explanation becomes part of the environment in which the next round of causality unfolds.

That is why serious learning requires more than measurement. It requires narrative literacy. Not the ability to spin a compelling tale, but the ability to detect how explanation itself is operating inside the system.

Key Takeaways

• Do not separate evidence from interpretation too quickly. In human systems, meaning often helps produce the effect you are measuring.
• Look for both material pathways and interpretive pathways. Resources and incentives matter, but so do trust, legitimacy, and shared expectations.
• Treat narratives as causal forces, not just communication tools. A story can change coordination, motivation, and future behavior.
• Test the tidy version of the story against messy reality. Ask what the narrative leaves out, and who benefits from its simplicity.
• When evaluating change, study the loop. The outcome shapes the story, and the story shapes the next outcome.

Conclusion: the explanation is inside the experiment

The deepest shift here is not methodological, but philosophical. We usually imagine that there is a world of events out there, and a separate world of stories in here. First the event happens, then we explain it. First the program works, then we write the case study.

But in social life, that division is unstable. The way we explain change alters the environment in which change continues. The story is not a postscript. It is part of the mechanism. The evaluation is not only a mirror. It is also a participant.

Once you see this, you cannot unsee it. Every report, every theory of change, every success story, every policy brief is doing two things at once. It is trying to describe causality, and it is quietly rearranging the conditions under which causality will happen next.

That is the real frontier of understanding change: not just tracing what happened, but recognizing that our accounts of what happened are already in the room, helping to decide what happens next.