Formation · Reference Card

Month 08 — Research Craft

Texts

Feynman — "Cargo Cult Science"One sitting, read twice. Mark every specific way scientists fool themselves.

Ioannidis — "Why Most Published Research Findings Are False" (PLoS Med 2005)One sitting; keep as a reference before every analysis decision.

Zinsser — On Writing WellRead the clarity/clutter chapters closely; skim genre chapters.

Core distinction

Form — the visible shape of rigor: stats, citations, hedged language, a submission.

Substance — self-directed honesty about what could make the finding wrong, before anyone else looks.

Small N + many analytic paths + a hot field = false positives are the statistically expected outcome, not the exception (Ioannidis).

Key questions per text

Feynman: Where am I producing the form of research rather than the substance?
Ioannidis: Am I in a high-risk condition right now — and does my language claim more than the design supports?
Zinsser: If I strip every hedge from this paragraph, is there still a real claim left standing?

Bradford Hill (supporting idea): strength · consistency · plausible mechanism · dose-response · temporality — a discipline for resisting overclaiming causation from an observational surgical series, not a checklist to "pass."

Sob medida — the finish line, not the craft

The pattern: 373 registry cases, zero approved manuscripts; manuscripts marked FINAL with PROSPERO still pending. Craft rigor everywhere except the last, checkable step — because shipping is evaluative and building isn't.

The counter-move: a pre-specified, witnessed stopping date for the actual finishing act (registration, submission, response) — not "when it feels ready." Borrow urgency from a deadline and an audience, not the verdict itself.

The tell: built assets keep compounding while the submission count doesn't move — or a scheduled "finishing session" gets spent reorganizing instead of hitting submit.

The month practice

Rewrite one manuscript introduction with brutal clarity — Zinsser's clutter-hunt + Ioannidis's honesty about the claim.

Copy the current introduction into "before."
Circle every hedge, throat-clearing phrase, passive construction.
Rewrite: one sentence on the real gap, one on what the study did, one on the honest claim.
Cut until it survives being read aloud to a skeptical colleague.
Save both to learning-records/writing-craft/YYYY-MM-DD-intro-rewrite.md.

Maxim

Claim only what the design earns, and say it plainly.

Full rationale: SYLLABUS.md · shelf: RESOURCES.md · lesson: 0008-research-craft.html