Purpose
The purpose of this course is to provide the participants with the methods and techniques necessary to determine whether measurement systems yielding discrete (rather than continuous) data are in a state of control and capability.
This course introduces the concepts and implications of measurement control for all discrete data systems. The methods stressed within this course correspond to systems which generate nominal data. Examples of these systems include go/no-go gauging systems, categorical (e.g., bitter/sweet/sour) sensory scales, and low resolution continuous data systems.
After completing this seminar, the participant will be capable of assessing the precision and accuracy of all nominal data measurement systems, regardless of the number of scale categories or the number of judges and inspectors.
Time Requirement
5 days
Number of Participants
25 participants maximum
Prerequisites
An Introduction to Statistical Process Control and Capability, Advanced Statistical Process Control and Capability, Guidelines for a Practical Approach to Gauge Capability Analysis, Advanced Gauge Control and Capability Analysis, Experimental Design and Industrial Statistics — Levels I, II, and III
Primary Resource Materials
Guidelines for a Practical Approach to the Assessment of Discrete Data Measurement Systems — Volume I: Nominal Data Applications
Content Outline
- Introduction to the Issue of Discrete Data Measurement Analysis
- The Measurement of Agreement (Precision): An Overview of Statistical Indices of Agreement for Nominal Data Scales
- Guidelines and Procedures for Measurement Processes Yielding Nominal Data
- General Guidelines and Observations
- A recommended Procedure for a Short-Term Agreement Study
- The Statistical Analysis of Agreement-Two Inspectors/Judges
- Two Inspectors-Two Categories
- Two Inspectors-More Than Two Categories
- Testing Hypotheses Associated with Kappa
- Testing the Hypothesis that K'=0
- Testing the Hypothesis that K' Equals a Value Other than 0.
- Statistically Comparing Two Independent Kappa Values
- The Statistical Analysis of Validity (Accuracy) for Nominal Data Systems
- The Statistical Analysis of Agreement: Light's Extension to More than Two Inspectors/Judges
- Procedures for Two Scale Categories
- Procedures for Scales with More Than Two Categories
- Testing Hypotheses Associated with K'
- The Statistical Analysis of Agreement: Fleiss Extension to More Than 3 Inspectors
- Procedures for Two Scale Categories
- Procedures for Scales with More Than Two Categories
- Testing Applicable Hypotheses
- Assessing Control and Capability of Measurement Systems Yielding Nominal Data