A Practical Guide to Statistical Verification. Issue 1 (13th September 2016)

• Record Type: Journal Article
• Authors: Kolozs, James; Black, Carissa
• Is Part Of: INCOSE International Symposium. Volume 26:Issue 1(2016)
• Page Start: 862
• Page End: 876
• Publication Date: 2016-09-13
• Subjects: Systems engineering -- Congresses; Systems engineering -- Periodicals; 620.0011
• DOI: 10.1002/j.2334-5837.2016.00198.x
• Languages: English
• ISSNs: 2334-5837

Abstract

Abstract: Statistical techniques are required for medical devices and can be used to make verification more meaningful in any industry. Sometimes, a company needs to predict how their total population of devices will perform. Other times, teams simply need stronger justifications than 'that's how many we had' or 'we always test 5' for test plans and sample sizes. However, there are an overwhelming number of statistical methods and references available. We have identified three tests that can be used to verify many types of device requirements: pass/fail (binomial), tests on the population (tolerance interval), and tests on the mean (t‐tests). This paper provides the information needed to choose a test, pick the right statistical claim and related parameters, determine the sample size needed to pass, and write acceptance criteria that makes it easy to determine a pass or fail.