What problem does it solve? Researchers routinely guess at participant counts or dress up tiny qualitative samples as percentages ("60% of users" from 6 people), producing deliverables that collapse under the first "out of how many?" question. This Skill is the canonical reference for how many participants a study needs and whether a given n can carry the claim being made. ## Core Features & Use Cases - Pick-your-number table: Fast defaults per situation — 5–6 for generative discovery, 5 per round for formative usability, ~40 for quant metrics, ~385 for a ±5% survey rate — each with its stop-rule. - Saturation and precision logic: Per-segment saturation stop-rules for qualitative work, and margin-of-error formulas (Adjusted-Wald intervals, effect-size sizing) for quantitative studies. - The hard gate: A four-step check that refuses percentages on qual n, demands confidence intervals on every rate, and offers small-n defenses (counts + severity, or triangulation to analytics) when stakeholders demand numbers. - Use Case: Before reporting "4 of 6 beta testers failed to find reschedule" as "60% of users," run the gate — it rewrites the claim as a count with High severity and routes the rate request to product analytics that actually have the n. ## Quick Start Ask the AI to justify the sample size for your planned usability study and check whether your drafted finding's percentage is defensible at your current n.
