A story can make scattered evidence intelligible.
It can put events into sequence, hold a person inside a context, carry fragments across time and give a reader something coherent enough to think with.
That does not make it proof.
Clyde Street kept returning to both sides of that problem.
Keith repeatedly returned to narrative while thinking about performance, coaching, history and analytical work. He also kept placing story beside sources, models, metrics, partiality and other people’s accounts.
The question is therefore not whether stories matter.
It is what a story can let us know — and what it can never establish merely by becoming persuasive.
Hearing other narratives
In February 2010, Keith wrote Speaking About Performance: Hearing Other Narratives. He returned to Donald Polkinghorne's work on narrative knowing alongside Miller Mair and John van Maanen, naming these as influences reaching back to qualitative research he had done in the 1980s.
Keith's claim was enrichment rather than replacement. Discussion of sporting performance, he wrote, could be enriched by openness to other narratives, and he described the post itself as part of an ongoing story about coming to know performance.
That gives "way of knowing" a bounded meaning here. Narrative can connect actions and events so that relations become easier to see. It does not become evidence merely by making those relations intelligible.
Structure can do analytical work
Two years later, Keith encountered that possibility in a form he had not written himself.
In 9.79, he reflected on the documentary The Race That Shocked The World, about the 1988 Olympic 100 metres final and its aftermath.
The documentary brought together archive footage, contemporary interviews and athletes looking back twenty-four years later. Keith called it an exceptional resource for qualitative enquiry in high-performance sport and singled out the power of its narrative structure.
The structure matters.
Historical images and later testimony do not become the same kind of evidence simply because a documentary places them beside one another. But the juxtaposition can bring past images and later recollection into the same inquiry.
Narrative is doing work here by arranging evidence.
It is not certifying the evidence.
Keith’s praise of its narrative structure does not, by itself, verify the recollections or historical claims assembled within the documentary. The power of the story does not remove the need to ask where its materials came from and what they can support.
A story can contain a model
A particularly revealing Clyde Street return came in 2017.
In An Introduction to Analytic Narratives for Coaches and Students, Keith encountered a historical study of cooperation between human whalers and orcas in Twofold Bay. The surviving evidence was fragmentary. The researchers nevertheless tried to build a simple fuzzy rule-based model of a complicated social system.
Crucially, the authors did not present the model as a clean solution. They described its limitations and treated it as something that could be revised as more evidence and understanding became available.
Keith followed that trail back to Polkinghorne and to Philippe Mongin’s work on analytic narratives.
This was not story against model.
It was story with model.
Keith became interested in what happens when historical sequence, incomplete evidence, formal representation and interpretation are allowed to sit in the same account. He described his purpose as opening conversations about evidence and models and extending the “epistemic reach” of sport analytics with coaches and students.
But he placed a restraint inside the same post.
He quoted Polkinghorne’s requirement that historical narrative remain answerable to traces of past events. Keith then added his own emphasis: actionable insight depended on rigour in how data were collected and analysed, whether those data were comprehensive or fragmentary.
That makes this source unusually important.
Narrative can help assemble an intelligible account from pieces that do not arrive already joined. A formal model can become one of those pieces. Neither gains validity simply because the resulting story hangs together.
Stories are shared objects
By 2018, Keith was approaching story as something that happened between people as well as inside an account. In Sharing Stories, he moved in one day between a data-science conversation and a school public-speaking event in which Year 5 pupils told family stories.
For analysts, his concern was what happened when the work stopped at delivering data. He imagined another role: being an audience for stories, sharing stories and co-creating stories with coaches.
That is a proposition about the analytical relationship, not evidence that co-created stories produced better decisions. Its narrower significance is that a story can become a shared object around which people locate experience, ask questions and make connections.
Coherence is not a validity test
That distinction becomes especially important when the story starts with data.
In Data and coherent narratives, Keith followed a discussion of the need to embed data in coherent narrative. But the rest of the post is just as important as that starting point.
The post moved through open data and code, the distinction between data-informed and data-driven decisions, organisational data cultures, the qualities of robust metrics and the warning carried by Goodhart’s law.
The surrounding material prevents an easy conclusion.
A coherent narrative does not make the underlying measure reliable.
A neat story does not turn a poor proxy into a good one.
Keith’s post places narrative inside a larger ecology of data, audience, checking and judgement. The story can help make the data meaningful to someone. It cannot confer validity on them.
Every telling has a cut
In July 2019, Keith returned directly to Miller Mair in Poetry, prose and poetics.
Mair’s argument was his own. Keith made that ownership visible while describing how deeply it had influenced him.
The part Keith returned to concerned acts of telling. A telling is composed. It carries intention. It is partial.
That need not be treated as a confession that stories are useless.
It is a warning about what a story is.
To call a telling composed and partial is to make selection visible: where the account begins, what is placed beside what, what is left out, and which words carry it.
Keith connected Mair’s poetics to imaginative attention: hearing differently, seeing differently and using language to pursue understanding.
The evidential limit must stay beside that possibility.
Making partiality visible gives the reader something important to judge. It cannot make factual looseness harmless.
Later that month, in Performance and Analytic Narratives, Keith returned to formal analytic narrative through Margaret Levi. Levi’s framework kept players, goals, rules, constraints, sequence and temporality in the same textured account.
Keith placed that beside Elo displays and observations about performance environments.
Again, the interesting object is the conjunction.
The indicators can help organise a puzzle. The narrative can situate the indicators in time and context. But three selected teams and their ratings cannot, by themselves, prove a culture, a cause or an explanation.
A story can help us see what we are asking.
It cannot answer more than the evidence can carry.
Stories beside data
By 2019, Keith was still placing narrative beside data rather than against it. In Sharing history he described qualitative and quantitative knowledge as dimensions on a continuum in which neither excluded the other.
The deeper sociology-of-knowledge inheritance that Keith elsewhere made explicit belongs in The Sociology Keith Carried With Him. Here the narrower point is enough: a data-rich analytical practice could still make room for story, context and human meaning without treating narrative as a substitute for verification.
What a story can carry
Clyde Street does not give us a single theory of narrative. Across these encounters, story is repeatedly asked to do bounded intellectual work: arrange sequence, place measures in context, connect testimony with an earlier record, hold fragmentary evidence together provisionally, and make selection and partiality more visible.
None of that allows a story to certify itself. Narrative cannot repair invalid data, supply a missing witness, turn a selected proxy into a cause, make contested history final or take ownership of another person’s experience.
So "way of knowing" has to keep its limit in view. Story can function that way when it makes relations, sequence, context and human meaning available for inquiry - while coherence is not asked to do the work of evidence.
Question to carry
When a story makes something easier to understand, what would we need to see before treating that understanding as something the evidence can carry?
Sources followed
Keith Lyons, “Speaking About Performance: Hearing Other Narratives” — 10/02/26
Keith Lyons, “9 79” — 12/07/11
Keith Lyons, “An Introduction To Analytic Narratives For Coaches And Students” — 17/02/03
Keith Lyons, “Sharing Stories” — 18/08/03
Keith Lyons, “Data And Coherent Narratives” — 18/10/14
Keith Lyons, “Poetry Prose And Poetics” — 19/07/17
Keith Lyons, “Performance And Analytic Narratives” — 19/07/26
Keith Lyons, “Sharing History” — 19/09/24