CDG Quality Control & Process Improvement 3 — Questions and Answers
Question 1: A diamond grader assigns D color to a stone that three senior graders later classify as E. This outcome is best described as:
- An accuracy error with a one-grade deviation (Correct answer)
- A precision error within acceptable repeatability limits
- A systematic bias caused by fluorescence
- A random error with no corrective action needed
Correct answer: An accuracy error with a one-grade deviation
The grade deviates from the consensus correct answer, making it an accuracy (trueness) error, not merely a repeatability issue.
Question 2: Which statistical tool is most appropriate for identifying whether two graders' color grade distributions differ significantly?
- Fishbone diagram
- Two-sample t-test or Mann-Whitney U test (Correct answer)
- Control chart with 3-sigma limits
- Scatter plot of carat weight vs. price
Correct answer: Two-sample t-test or Mann-Whitney U test
Inferential statistical tests (t-test for normal data, Mann-Whitney for non-parametric) formally test whether two graders' grade distributions are statistically different.
Question 3: In diamond QC, 'measurement system analysis' (MSA) evaluates which two primary components?
- Stone value and market demand fluctuation
- Repeatability (same grader, multiple times) and reproducibility (multiple graders) (Correct answer)
- Lab certification level and turnaround time
- Equipment cost and grader salary
Correct answer: Repeatability (same grader, multiple times) and reproducibility (multiple graders)
MSA quantifies repeatability (within-appraiser variation) and reproducibility (between-appraiser variation) to determine if the grading system is reliable.
Question 4: A lab implements a policy requiring all stones above 1.00 ct to receive three independent clarity grades. What quality principle does this reflect?
- Zero defect tolerance for high-value goods
- Increased sample size to improve decision confidence on high-stakes items (Correct answer)
- Regulatory compliance with FTC gemstone standards
- Automated verification replacing manual inspection
Correct answer: Increased sample size to improve decision confidence on high-stakes items
Multiple independent assessments on high-value stones reduce the probability of a single grader error reaching the customer, improving confidence.
Question 5: Which process improvement concept is demonstrated when a grading lab restructures its workflow so graders never handle the same stone twice on the same day?
- Poka-yoke (mistake-proofing)
- Visual management
- Memory-effect mitigation (Correct answer)
- Kanban flow control
Correct answer: Memory-effect mitigation
Preventing a grader from re-examining a stone the same day reduces the memory effect, where recalled prior judgments bias the second grade.
Question 6: When plotting diamond color grades on a frequency histogram, a bimodal distribution (two peaks) most likely indicates:
- A normal grading process with appropriate spread
- Two different grader populations applying different internal grade standards (Correct answer)
- Excessive fluorescence distorting the D-to-Z scale
- An outlier caused by a single misgraded fancy color stone
Correct answer: Two different grader populations applying different internal grade standards
Bimodal distributions in grading data often signal two sub-populations — typically two graders or two grading standards operating simultaneously.
Question 7: A corrective action report (CAR) for a systematic color grading error should include all of the following EXCEPT:
- Root cause identified through analysis
- Containment actions taken on affected stones
- A prediction of future diamond market prices (Correct answer)
- Verification steps to confirm the correction was effective
Correct answer: A prediction of future diamond market prices
CARs address root causes, containment, and verification of fixes — market price forecasting is irrelevant to a process correction.
CDG Quality Control & Process Improvement 3
A diamond grader assigns D color to a stone that three senior graders later classify as E.
This outcome is best described as: