We Tested Thirty Is Not a Sample Size Rationale

The fastest way to lose credibility in a verification protocol is an unjustified sample size. Thirty units because thirty felt safe.

The fastest way to lose credibility in a verification protocol is an unjustified sample size.

Thirty units because thirty felt safe.

Three because three was what we had.

Neither survives a serious review.

Sample size is a risk decision, not a habit.

Tie the number to the risk of the failure mode. A higher severity hazard demands higher confidence and higher power.

Match the method to the data.

Attribute data, pass or fail, points to binomial approaches.

Variable data points to t-tests, equivalence tests, or tolerance intervals.

State the confidence and the power you are buying, and why that level fits the risk.

Manage both error types.

A Type I error passes a bad design.

A Type II error fails a good one.

Both cost you.

Write the rationale into the protocol before you run it. Not after the data lands.

An auditor does not need your result to find the gap.

An unjustified sample size is a finding on its own.

The number is not the point. The reasoning behind the number is.

rmForge scopes the Hazard Analysis so each risk control ties to a Product Requirement, which is what your sample size rationale should hang from.

#MedicalDevices #RiskManagement #Statistics #ISO14971 #DesignControls #MedTech