Sample Size Justification & Statistical Analysis for Performance Qualification (PQ) Studies
Regulatory agencies mandate formal protocols for Performance Qualification (PQ) studies to determine whether products/processes meet the desired requirements. When planning PQ studies, sample size determination is critical to ensure that study results will be meaningful. Methods to determine appropriate sample sizes for various types of problems will be covered.
This 2-day seminar will provide a 12-step process to assist you in writing/reviewing protocols for PQ studies with a focus on sample size justification, acceptance criteria and statistical analysis using Minitab v17. Validation of software will not be covered.
Participants are requested to bring a laptop with Minitab Version 17 software installed.
Upon completing this course participants will be able to:
Determine a Representative Sample for Process Qualification (PQ) study.
Develop product/process acceptance criteria statements
Determine appropriate Sample Size to meet PQ acceptance criteria
Understand how to Analyze PQ data
Understand how to calculate and interpret statistical tolerance limits
Determine and verify the appropriate distribution of the data
Interpret probability plots
Apply normalizing transformations
Handle censored data and mixed failure modes.
DAY 01(8:30 AM - 4:30 PM)
Registration Process - (8:30 am till 8:45 am)
Lecture 1: Introduction
Selecting Representative Sample
Confidence Level Definition
General Flow Chart for 12-step Protocol Development
Lecture 2: Sample Size Determination for Performance Qualification Study
Attribute (Pass/Fail) Response
Censored (Variable) Response
Lecture 3: Non-Censored (Variables) Response
Tolerance Limits Defined
One-Sided vs Two-Sided Limits
Nonparametric Tolerance Limits
DAY 02(8:30 AM - 4:30 PM)
Lecture 4: Analyzing PQ Data using Minitab
Statistical Tolerance Limits
Percentage of Population captured
Lecture 5: Distribution Fitting using Graphical Methods
Lecture 6: Statistical Tests for Distribution Fit
Lecture 7: Normalizing Transformations
Lecture 8: “Non-Normal” Statistical Tolerance Limits
Smallest Extreme Value
Mixed populations or multiple failure modes