The Researched Narrative

My Harvard thesis, Good Friday, is a work of researched narrative nonfiction about a crime. Writing it meant holding two jobs at once: the researcher’s job of establishing what happened, and the storyteller’s job of making a reader care. Those jobs pull against each other constantly. These are the principles that kept them working together.

Two clocks

Every narrative runs on two clocks. Story time is the order in which events happened. Discourse time is the order in which the reader learns them. Fiction writers play with the gap between them freely. In true crime, the gap is the whole engine, and it carries risk.

Start with a complete chronology in a spreadsheet, kept separate from the manuscript. Every event gets a date, a source, and a confidence level. Only then decide where the reader enters. You might open with the morning of the crime, with the trial, or with an ordinary day decades earlier. Each choice creates a different kind of suspense, and each one makes a different promise to the reader about what the book is for.

The chronology protects you later. When you move a scene for dramatic effect, you can check in seconds that you haven’t implied a sequence that didn’t happen.

Marking what you know

A researched narrative contains three kinds of statements: documented fact, reasonable inference, and reconstruction. Readers deserve to be able to tell them apart without a footnote on every line.

Fact can be stated plainly. Inference should carry a light signal: “she must have heard,” “the record suggests,” “by his own account.” Reconstruction, such as dialogue or interior thought you couldn’t have witnessed, needs the clearest flag, usually in a note on sources or a sentence in the text explaining what the scene is built from.

The temptation is to smooth all three into one confident voice, because confidence reads well. Resist it. The small hedges are also what make a reader trust the plain statements.

Documents as scenes

Court transcripts, police reports, newspaper archives, and public records requests are often treated as background research to be summarized. They can also be scenes. A transcript has dialogue, conflict, and a setting. A records request that comes back heavily redacted is itself an event, and the redactions can say as much as the text.

When a document carries real drama, consider letting the reader see a short excerpt rather than describing it. The texture of official language next to private grief can do work that no amount of description can.

Who you owe

Crime narratives involve real people who did not ask to be written about: victims, families, witnesses, and sometimes the accused. The writer’s obligations to them don’t end when a fact is accurate.

Some questions I kept beside the draft:

Does this detail serve the reader’s understanding, or only their appetite?
Would I be comfortable if the victim’s family read this sentence?
Have I given the victim a life beyond the crime, or are they only the occasion for the story?
Am I inventing motive where the record shows none?

None of these questions forbids difficult material. They ask you to justify it.

Structure follows the question

The principle I return to most is to name the question the book is trying to answer, in one sentence, and to test every chapter against it. Not “what happened,” which a news article can answer, but something a book is needed for: how a community let it happen, why the case was misread for years, what the people closest to it were carrying. Once the question is clear, chapters that don’t serve it become easier to cut, however hard they were to research.

Try it

Build a chronology of your subject with a source and confidence level for each entry.
Write the same scene twice: once as a confident reconstruction, once marking inference honestly. Read both aloud and notice which one you believe.
Write your book’s question in a single sentence. Tape it above your desk.