MSA

(Measurement System Analysis)

Handbook

BY: MUHAMMAD IRFAN HAIDER

 What is MSA?
Measurement System Analysis (MSA) is a critical tool in quality control that helps ensure the
accuracy and reliability of data collected from measurement systems. This module provides an
overview of the fundamental aspects of MSA, explaining its significance, key categories, and
essential terminologies.
σobs2(Observed Process Variance) =
σactual2(Actual Process Variance) + σmsa2
(Measurement System Variance)

Instrument:
Definition: A device or system used to measure a physical quantity with
higher accuracy, resolution, and sometimes digital processing.
Examples: Digital Vernier Caliper, Coordinate Measuring Machine
(CMM) , Oscilloscope, Spectrophotometer, Micrometer with digital
readout
Gauge:
Definition: A tool or device designed to check, compare, or verify whether
a dimension or characteristic falls within a specified tolerance range.
Examples: Plug Gauge (Go/No-Go), Snap Gauge, Thread Gauge, Dial
Gauge, Bore Gauge

Key terms used in MSA:

• Accuracy – The closeness of a measured value to the true value.
• Precision – The degree to which repeated measurements under unchanged conditions show
the same results.
• Bias – A consistent deviation from the true value.
• Stability (Drift over Time)– The ability of a measurement system to produce consistent
results over time.
• Linearity (Non-Uniform Bias Across Range)– The consistency of bias across the
measurement range.
• Repeatability – The variation in measurements taken under identical conditions by the same
operator.
• Reproducibility – The variation in measurements when different operators use the same
equipment.
• GR&R Study (Gauge Repeatability & Reproducibility) – A statistical method to analyze
the variability in measurement systems.
• Resolution – The smallest measurement increment that an instrument can detect.
• Tolerance % in MSA: % of measurement variation compared to specification limits
• True Value: The actual value of a measured characteristic in an ideal condition. It is often
unknown because all measurement systems have some level of imperfection.
• Reference Value: Accepted Value. A value determined from a reliable source, such as a
calibrated standard or master gauge, which serves as a benchmark for measurement accuracy.

BY: MUHAMMAD IRFAN HAIDER

 Two Categories of MSA
1. Variable MSA (Continuous Data Analysis)
When to Use: When measurements are numerical (e.g., length, weight, voltage, pressure). Common in
manufacturing, machining, and lab testing.
Key Analysis Method Gauge Repeatability & Reproducibility (GR&R) :
• Repeatability → Variation when the same person measures the same part with the same tool multiple
times.
• Reproducibility → Variation when different people measure the same part using the same tool.
• Bias, Stability, Linearity → Other factors affecting measurement accuracy over time.
2. Attribute MSA (Discrete Data Analysis)
When to Use: When measurements are categorical (e.g., Pass/Fail, Go/ No-Go, Accept/Reject).
Common in visual inspections, defect grading, and quality audits.
Key Analysis Methods:
• Kappa Studies → Evaluates agreement among inspectors beyond chance (e.g., Cohen’s Kappa).
• Attribute Agreement Analysis (AAA) → Checks consistency in inspectors' judgments.
• Signal Detection Analysis → Determines if inspectors correctly identify defects vs. false alarms.
Types of Error:
Type I Error (Producer’s Risk): A good part is incorrectly rejected, leading to unnecessary
scrap, rework, or increased costs.
Type II Error (Consumer’s Risk): A defective part is mistakenly accepted, causing field
failures, customer complaints, and warranty claims.
Precision-to-Tolerance Ratio (PTR):
It is a metric used to assess whether a measurement system is precise enough relative to the
required product tolerances.
Formula for PTR: PTR=6σMS/Tolerance
Where: σMS = Measurement system standard deviation, Tolerance = The allowable range of variation
in the product specification
Interpreting PTR Values: PTR < 10% → Excellent measurement system (high precision). ,
PTR between 10-30% → Acceptable but may need improvement.
PTR > 30% → Poor measurement system (high variability), requiring improvement or replacement.

Type 1 Gage Study:

It is a quick and simple method used to evaluate the accuracy and repeatability of a
measurement system. It involves: A single operator measuring a single part multiple times
using the same gage. Comparing the measurements to the reference value to determine bias.
Gage Capability Index, Cg & Cgk:
Capability analysis determines if the measurement system is suitable for process control by
evaluating the ratio of measurement variation to product tolerances.
Gage Capability Ratio (Cg): Measures repeatability of a gage.
Cg=Tolerance/6σ

Cg > 1.33 → Acceptable, Cg < 1.33 → Needs improvement.

Gage Performance Index (Cgk): Considers both bias and repeatability.
Cgk=Tolerance−2∣Bias∣ / 6σ

Cgk > 1.33 means the system is accurate and precise.

BY: MUHAMMAD IRFAN HAIDER

Expanded Uncertainty (U): U=k × σ Where: k = Coverage factor (typically 2 for 95%
confidence). σ = Standard deviation of measurement data.
NDC:
It represents the number of distinct categories that a measurement system can differentiate within the
variation of the process.
It is calculated using the formula: NDC=(1.41×PV)/GRR
Where:
• PV (Process Variation) = Part-to-Part Variation (Standard deviation of the parts)
• GRR (Gage Repeatability and Reproducibility) = Measurement System Variation
Interpretation of NDC:
• NDC < 5 → Poor Measurement System (Not capable of distinguishing parts reliably)
• NDC ≥ 5 → Acceptable for process monitoring
• NDC ≥ 10 → Ideal for effective process control
Why is NDC Important?
Helps determine if the measurement system can differentiate between different part variations.
A low NDC means the measurement system is too noisy and not suitable for control or improvement
purposes. Used in Gage R&R studies to assess measurement capability.
Analysis of Variance (ANOVA) in MSA

The Analysis of Variance (ANOVA) method is used in Gage R&R analysis to statistically determine the
sources of variation in a measurement system. It helps in quantifying and isolating measurement system
errors such as repeatability, reproducibility, and interaction effects.
Key Components of the ANOVA Table
The ANOVA table consists of six main columns that provide a structured way to analyze variation:
Source – Identifies the cause of variation (e.g., parts, appraisers, interaction, repeatability).
DF (Degrees of Freedom) – Represents the number of independent values available for estimating
variation.
SS (Sum of Squares) – Measures total variation within each source by calculating the squared deviations
from the mean.
MS (Mean Square) – Found by dividing SS by DF; represents the variance estimate for each source.
EMS (Expected Mean Square) – Defines the expected variance components for each MS term in the
table.
F-ratio – Used for statistical significance testing. If the F-ratio is high, the interaction effect is
significant, meaning the measurement system may be influenced by operator-part interaction.
Decomposition of Variation in ANOVA
The ANOVA method breaks down the total variation into four key components:
1. Parts (Product Variation) – Reflects natural differences in part measurements.
2. Appraisers (Operator Variation or Reproducibility) – Accounts for variation due to different
operators measuring the same part.
3. Interaction (Parts × Appraisers) – Identifies if some appraisers measure certain parts differently,
indicating an operator bias issue.
4. Repeatability (Gage/Equipment Error) – Measures variation when the same appraiser repeatedly
measures the same part.

BY: MUHAMMAD IRFAN HAIDER

 Gage Repeatability and Reproducibility Data Collection Sheet
Part No. & Name: 8775, Shaft Gage Name: Shaft Gage Date: 1-1-25
Characteristics: Outside Diameter Gage Number: 9166 Performed by: SSM
Specifications: 25 plus or minus 5 Gage Type: Indicator

Upper Spec 11 # of Trials = 3 K 1 = 0.5908 Xbar diff = 0.447 D 4 = 2.58

Lower Spec 9 # of appraisers = 3 K 2 = 0.5231 Rbarbar = 1.033 R p = 2.3611111
Total Tol 2 # of parts = 10 K 3 = 0.3146

Part
Appraiser/Trial # Average
1 2 3 4 5 6 7 8 9 10
A 1 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 10.244
2 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 9.345
3 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 10.236
Average 10.2 9.946667 6.893333 10.36667 10.33333 10.75667 10.53333 10.4 10.06667 9.92 Xbar a = 9.941667
Range 0.2 0.12 9.08 0.8 0.3 0.27 0.1 0.4 0.1 0.08 Rbar a = 1.145
B 1 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 10.236
2 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 10.52
3 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 9.345
Average 10.43333 10.18333 7.1 10.43333 10.3 10.72333 10.43333 10.33333 10.2 10.19667 Xbar b = 10.03367
Range 0.8 0.83 9.7 0.8 0.4 0.27 0.4 0.2 0.5 0.91 Rbar b = 1.481
C 1 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 10.52
2 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 10.249
3 ####### ####### ####### ####### ####### ####### ####### ####### ####### ####### 10.397
Average 10.46667 10.7 10.53333 10.36667 9.99 10.13 10.56667 10.36667 10.4 10.36667 Xbar c = 10.38867
Range 0.7 0.2 0.1 0.4 0.23 0.9 0.7 0.1 0.5 0.9 Rbar c = 0.473
Xbarbar = 10.12133
Part Average 10.36667 10.27667 8.175556 10.38889 10.20778 10.53667 10.51111 10.36667 10.22222 10.16111
Rp = 2.361111
[(Rbar a = 1.145 + Rbar b = 1.481 + Rbar c = 0.473) / # of appraisers = 3] = Rbarbar Rbarbar = 1.033
(Max Xbar = 10.3886666666667) - Min Xbar = 9.94166666666667) = Xbar diff Xbar diff = 0.447
(Rbarbar = 1.033) x (D4 = 2.58) = UCL R UCL R = 2.66514

Gage Repeatability and Reproducibility Report - % VARIATION

Part No. & Name: 8775, Shaft Gage Name: Shaft Gage Date: 01-01-2025
Characteristics: Outside Diameter Gage Number: 9166 Performed by: SSM
Specifications: 25 plus or minus 5 Gage Type: Indicator

Upper Spec 11 # of Trials = 3 K1 = 0.5908 Xbar diff = 0.447 D4 = 2.58

Lower Spec 9 # of appraisers = 3 K2 = 0.5231 Rbarbar = 1.033 Rp = 2.3611111
Total Tol 2 # of parts = 10 K3 = 0.3146

Measurement Unit Analysis % Total Variation

Repeatability - Equipment Variation (EV) Percent Equipment Variation
EV = (Rbarbar) x (K1) %EV = 100 [ EV / TV ]
= (1.033) x (0.5908) Trials K1 = 100 [0.6103 / 0.9831 ]
= 0.6103 2 0.8862 = 62.079 %
3 0.5908
Reproducibility - Appraiser Variation (AV) Percent Appraiser Variation
AV = SQRT{(Xbar diff x K 2)^2 - (EV/[(# parts) x (# trials)])} %AV = 100 [ AV / TV ]
= SQRT(0.0547 - 0.0124) = 100 [0.2056 / 0.9831 ]
= 0.2056 Appraisers 2 3 = 20.913 %
K2 0.7071 0.5231
Repeatability & Reproducibility (GRR) Percent Gage Repeatability & Reproducibility Variation
GRR = SQRT[ (EV^2) + (AV^2) ] %GRR = 100 [ GRR / TV ]
= SQRT [ 0.3725 + 0.0423 ] = 100 [0.644 / 0.9831 ]
= 0.644 = 65.507 %

Part Variation (PV) Parts K3 Percent Part Variation

PV = Rp x K3 2 0.7071 %PV = 100 [ PV / TV ]
= 0.7428 3 0.5231 = 100 [0.7428 / 0.9831 ]
4 0.4467 = 75.557 %
Total Variation (TV) 5 0.403
TV = SQRT[ (GRR^2) + (PV^2) ] 6 0.3742 Number of Distinct Categories that can be Distinguished
= SQRT [0.4147 + 0.5518 ] 7 0.3534 ndc = 1.41( PV / GRR )
= 0.9831 8 0.3375 = 1.41 (0.7428 / 0.644 )
9 0.3249 = 1.626 or approximately: 1
10 0.3146

Gage Repeatability and Reproducibility Report - % TOLERANCE (5.15 Sigma)

Part No. & Name: 8775, Shaft Gage Name: Shaft Gage Date: 01-01-2025
Characteristics: Outside Diameter Gage Number: 9166 Performed by: SSM
Specifications: 25 plus or minus 5 Gage Type: Indicator

Upper Spec 11 # of Trials = 3 K1 = 0.5908 Xbar diff = 0.447 D4 = 2.58

Lower Spec 9 # of appraisers = 3 K2 = 0.5231 Rbarbar = 1.033 Rp = 2.3611111
Total Tol 2 # of parts = 10 K3 = 0.3146

Measurement Unit Analysis % Total Tolerance

Repeatability - Equipment Variation (EV) Percent Equipment Variation
EV = (Rbarbar) x (K1) x (5.1 Sigma) %EV = 100 [ EV / Total Tol ]
= (1.033) x (0.5908) x (5.1) Trials K1 = 100 [3.1125 / 2 ]
= 3.1125 2 0.8862 = 155.625 %
0.59083
Reproducibility - Appraiser Variation (AV) Percent Appraiser Variation
AV = (5.1 Sigma) x SQRT{(Xbar diff x K 2)^2 - (EV/[(# parts) x (# trials)])} %AV = 100 [ AV / Total Tol ]
= (5.1) x SQRT(0.0547 - 0.0124) = 100 [1.0484 / 2 ]
= 1.0484 Appraisers 2 3 = 52.420 %
K2 0.7071 0.5231
Repeatability & Reproducibility (GRR) Percent Gage Repeatability & Reproducibility Variation
GRR = SQRT[ (EV^2) + (AV^2) ] %GRR = 100 [ GRR / Total Tol ]
= SQRT [ 9.6877 + 1.0991 ] = 100 [3.2843 / 2 ]
= 3.2843 = 164.215 %

Part Variation (PV) Parts K3 Percent Part Variation

PV = (Rp) x (K3) x (5.1 Sigma) 2 0.7071 %PV = 100 [ PV / Total Tol ]
= (2.36111111111111) x (0.3146) x (5.1) 3 0.5231 = 100 [3.78828 / 2 ]
= 3.78828 4 0.4467 = 189.414 %
5 0.403
Total Tolerance (Total Tol) 6 0.3742 Number of Distinct Categories that can be Distinguished
TV = Upper Spec - Lower Spec 7 0.3534 ndc = 1.41( PV / GRR )
= 11 - 9 8 0.3375 = 1.41 (3.78828 / 3.2843 )
= 2 9 0.3249 = 1.626 or approximately: 1
10 0.3146

BY: MUHAMMAD IRFAN HAIDER

The Kappa statistic (κ) is a measure of inter-rater agreement that quantifies how well operators
agree beyond what is expected by random chance. It is commonly used in attribute agreement
analysis (AAA) when inspectors classify parts into categories (e.g., Pass/Fail, Defective/Non-
Defective).
Kappa (κ) Statistic Formula in MSA
The Kappa (κ) statistic is used in Measurement System Analysis (MSA) to assess the agreement
between inspectors (appraisers) beyond what would be expected by random chance. It is
particularly useful in Attribute Agreement Analysis (AAA) when evaluating pass/fail or
categorical decisions.
Formula: κ=(Po−Pe) / (1−Pe)
Where: Po(Observed Agreement): The actual percentage of times inspectors agree on
classifications. Pe(Expected Agreement): The probability that agreement occurred by chance.
Kappa (κ) Value Interpretation
< 0.00 Poor agreement (worse than random)
0.00 - 0.20 Slight agreement
0.21 - 0.40 Fair agreement
0.41 - 0.60 Moderate agreement
0.61 - 0.80 Substantial agreement
0.81 - 1.00 Almost perfect agreement

Kendall’s Coefficient of Concordance (W):

Kendall’s W is used in Measurement System Analysis (MSA) when multiple raters (more than
two) rank or score a set of items. It helps evaluate the consistency of subjective assessments,
commonly used in visual inspections, defect grading, and quality evaluations.
W=(12∑Ri2−3N2(k+1)) / (k2(N3−N))
Where: N = Number of items being ranked, k = Number of raters (appraisers)
Ri= Sum of ranks for item i
Kendall’s W Value Level of Agreement
1.00 Perfect Agreement
0.80 – 0.99 Strong Agreement
0.60 – 0.79 Moderate Agreement
0.40 – 0.59 Weak Agreement
< 0.40 Poor Agreement (high variability)

• Kappa (κ) Amongst Inspectors: Measures agreement when different operators classify parts.
• Kendall’s W Amongst Inspectors: Evaluates if inspectors rank items in a similar order.
Attribute Gage R and R

Known Population Operator #1 Operator #2 Operator #3 Agree?

Sample # Attribute Trial #1 Trial #2 Trial #1 Trial #2 Trial #1 Trial #2 TRUE/FALSE
1 NG G G G G G G FALSE
2 G G G G G G G TRUE
3 G G G G G G G TRUE
4 NG NG NG NG NG NG NG TRUE
5 G G G G G G G TRUE
6 G G G G G G G TRUE
7 NG NG NG NG NG NG NG TRUE

Attribute GRR 8
9
10
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
TRUE
TRUE
TRUE
11 G G G G NG G G FALSE
12 G G G G G G G TRUE
13 G G G G G G G TRUE
14 G G G G G NG NG FALSE
15 NG G G G G G G FALSE
16 G G G G G G G TRUE
17 G G G G G G G TRUE
18 G G G G G G G TRUE
19 G G G G G G G TRUE
20 G G G G G G G TRUE
Appraiser Score 95% 93% 93% 80%
% Effective
AIAG Guidelines
Acceptable for the appraiser >90%
Marginallay acceptable for the appraiser - may need improvement >80%
Unacceptable for the appraiser - needs improvement <80%

BY: MUHAMMAD IRFAN HAIDER

BY: MUHAMMAD IRFAN HAIDER
https://www.linkedin.com/in/irfan-haider-730379154