Solved

How do I determine what is a statistically significant sample size for my V&V testing?

  • 2 March 2022
  • 4 replies
  • 121 views

Userlevel 5
Badge +1

We’re getting ready to go through Verification & Validation, but we are struggling to understand all the rules around sample size for the various tests we’ll be running. 

How do we determine what is a statistically significant sample size for this testing?

icon

Best answer by Shannon.Hoste 28 March 2022, 02:03

View original

4 replies

Userlevel 2
Badge

The first thing that I’ve found necessary for a solid V&V phase: Having clean, clear, concise requirements (including risk controls) and user needs. This will help you tremendously when you get down to the details of your V&V plan. 

Now, I’m going to add more questions: 

What do you need to prove and to what performance level? Ideally this would be defined back when you worked out your Design Inputs and Product Design & Development Plan.

What form of verification will you utilize (inspection, demonstration, analysis or testing)?

Is there a recognized standard for which conformance can demonstrate verification/validation (e.g. ingress protection ratings)?

For testing, do you have variable or attribute data? Is the method quantative (reliability) or qualitative (human factors)?

I have always loved putting together a good V&V plan, so many pieces to put together. 

 

Userlevel 5
Badge +1

Awesome @Shannon.Hoste - these are great questions to consider.

I tend to try to jump straight to the convergent thinking of a one-size-fits-all solution when, in reality, each one of your questions needs to be applied to each requirement that needs to be verified or validated!

We’ll have to chew on this for a bit - I’ll come back if we have some specific follow-up questions.

Thank you!!

Badge

Greetings all! The sampling plans associated with each input/output tied to those clear and concise requirements should be commiserate to the risk associated with each device or device family. The risk should be evaluated, established, and continually reassessed throughout the lifecycle of the device within the respective devices Risk Management Plan. 

A great resource that I have utilized through the years to establish appropriate sampling plans not only for DV&V, but also incoming inspection, manufacturing acceptance, and audits/effectiveness checks, is the book Statistical Procedures for the Medical Device Industry by Dr. Wayne Taylor. 

Taylor, Wayne (2017). Statistical Procedures for the Medical Device Industry. Taylor Enterprises, Inc., www.variation.com 

Happy to discuss specific sampling plans if anyone has questions!

Best Regards,

Jarod

Userlevel 5
Badge +1

@JMarkley this is so great! I get people asking me about Statistical Procedures every now and then, but I had not run across this book. Super helpful, thank you!

Reply