Study guide
8.3 Variables Sampling with ANSI/ASQ Z1.9
ASQ Certified Quality Inspector CQI
Track your progress as you read
Finished sections get a check mark.
1 Chapter 1: Shop Mathematics & Algebra
1.1 Basic Shop Math & Precision Measurement 1.2 Basic Algebra & Technical Formulas 1.3 Measurement Systems & Unit Conversions
2 Chapter 2: Geometry & Applied Shop Trigonometry
2.1 Applied Shop Geometry 2.2 Right-Triangle Trigonometry 2.3 Shop Trigonometry Applications & Sine Bars
3 Chapter 3: Handheld Variable & Attribute Gages
3.1 Variable Hand Measuring Instruments 3.2 Fixed Attribute & Limit Gages 3.3 Transfer Gages & Precision Accessories 3.4 Special Gages: Electronic, Pneumatic, Force & Environmental Instrumentation
4 Chapter 4: Surface Plate Inspection, Optical Systems & CMMs
4.1 Gage Selection, Care & Environmental Control 4.2 Surface Plate Tools & Layout Inspection 4.3 Optical Comparators & Vision Systems 4.4 Coordinate Measuring Machines (CMM) & 3-2-1 Alignment 4.5 Gage Correlation & Cross-Instrument Agreement
5 Chapter 5: Surface Metrology, Calibration Systems & MSA
5.1 Surface Texture & Finish Measurement 5.2 Calibration Systems & NIST Traceability 5.3 Measurement System Analysis (MSA) & Gage R&R 5.4 Measuring Mass, Form & Profile, and Inclination
6 Chapter 6: Technical Drawing & Blueprint Reading
6.1 Engineering Drawing Structure & Conventions 6.2 Views, Sectioning & Feature Representation 6.3 Dimensions, Tolerancing & Fit Classifications
7 Chapter 7: Geometric Dimensioning & Tolerancing (GD&T)
7.1 GD&T Fundamentals & Datum Reference Frames 7.2 Material Condition Modifiers & Bonus Tolerance 7.3 Form & Orientation Tolerances 7.4 Location, Profile & Runout Tolerances
8 Chapter 8: Sampling Inspection & Acceptance Sampling Plans
8.1 Sampling Fundamentals & Operating Characteristic (OC) Curves 8.2 Attribute Sampling with ANSI/ASQ Z1.4 8.3 Variables Sampling with ANSI/ASQ Z1.9
9 Chapter 9: Inspection Planning, Processes & Material Disposition
9.1 Inspection Planning, Receiving & In-Process Inspection 9.2 First Article Inspection (FAI) & Documentation 9.3 Nonconforming Material Control & MRB 9.4 Inspection Errors & Human Factors 9.5 Product Traceability, Age Control & Shelf Life
10 Chapter 10: Materials Testing & Nondestructive Examination
10.1 Mechanical & Destructive Testing 10.2 Surface & Subsurface NDE Methods 10.3 Volumetric & Electromagnetic NDE Methods 10.4 Functionality, Force & Elastomer Hardness Testing 10.5 Software for Test Equipment & Software Validation
11 Chapter 11: Statistical Process Control & Process Capability
11.1 Statistics Fundamentals & Normal Distribution 11.2 Control Charts for Variables 11.3 Control Charts for Attributes & Run Rules 11.4 Process Capability & Performance Analysis
12 Chapter 12: Quality Systems, Continuous Improvement & Safety
12.1 The 7 Classic Quality Control Tools 12.2 Continuous Improvement & Problem-Solving Methodologies 12.3 Quality Management Systems, Audits & CAPA 12.4 Workplace Safety, Environmental Controls & PPE 12.5 Failure Mode and Effects Analysis (FMEA) & Fault Tree Analysis (FTA)
13 Chapter 13: Quality Fundamentals, Teams & Documentation Systems
13.1 Quality Improvement Concepts, Products & Processes 13.2 Audit Tools, Tracing Techniques & Audit Communication 13.3 Employees as Resources: Empowerment, Team Roles & Team Development 13.4 Quality Documentation, Reference Standards & Document Control
8.3 Variables Sampling with ANSI/ASQ Z1.9
Key Takeaways Explain This Topic AI
ANSI/ASQ Z1.9 (equivalent to MIL-STD-414 and ISO 3951) governs acceptance sampling by variables, evaluating continuous physical measurements (e.g., diameter, hardness, length) rather than attribute pass/fail counts.
Variables sampling achieves equivalent statistical protection (identical OC curve discrimination) with significantly smaller sample sizes (often 50% to 70% smaller) than attribute sampling, making it ideal for costly or destructive testing.
A strict, non-negotiable prerequisite of ANSI/ASQ Z1.9 is that the measured quality characteristic must follow a normal distribution; non-normal data invalidates calculated tail defect estimates.
Acceptance is determined using the Quality Index: Q_U = (USL - X_bar) / s for upper limits and Q_L = (X_bar - LSL) / s for lower limits, measuring how many sample standard deviation units separate the sample mean from tolerance boundaries.
In Form 1 (k-method), the Quality Index Q is compared directly against an acceptability constant k (accept if Q >= k), whereas in Form 2 (M-method), Q is converted to an estimated percent nonconforming p and compared against maximum allowable percent M (accept if p <= M).
Last updated: September 2026
- 8.3 Variables Sampling with ANSI/ASQ Z1.9
- Principles of Variables Acceptance Sampling
ANSI/ASQ Z1.9 ( Sampling Procedures and Tables for Inspection by Variables for Percent Nonconforming ) is the national standard governing acceptance sampling when quality characteristics are evaluated on a continuous quantitative measurement scale . It is the direct civilian successor to military standard MIL-STD-414 and is technically harmonized with international standard ISO 3951 .
The Operational Concept
Unlike attribute sampling (which tallies conforming versus nonconforming parts), variables sampling measures exact quantitative values—such as a shaft outside diameter measured to the nearest 0.0001 in 0.0001\text{ in} 0.0001 in , a titanium tensile specimen pulled to rupture in ksi \text{ksi} ksi , or a hardened bearing raceway tested in HRC. From these continuous sample readings ( X 1 , X 2 , … , X n X_1, X_2, \dots, X_n X 1 , X 2 , … , X n ), the quality inspector calculates two fundamental summary statistics:
Sample Mean ( X ˉ \bar{X} X ˉ ) : Measures process centering relative to specification limits: X ˉ = ∑ i = 1 n X i n \bar{X} = \frac{\sum_{i=1}^n X_i}{n} X ˉ = n ∑ i = 1 n X i
Sample Standard Deviation ( s s s ) : Measures process variability and dispersion: s = ∑ i = 1 n ( X i − X ˉ ) 2 n − 1 s = \sqrt{\frac{\sum_{i=1}^n (X_i - \bar{X})^2}{n - 1}} s = n − 1 ∑ i = 1 n ( X i − X ˉ ) 2
By combining the sample mean, standard deviation, and engineering specification limits, the inspector calculates how far the process distribution sits from tolerance boundaries and mathematically estimates the percent of the lot that falls outside limits.
Variables vs. Attributes Sampling: Technical Comparison
Quality engineers and inspectors select between ANSI/ASQ Z1.4 (attributes) and ANSI/ASQ Z1.9 (variables) by weighing sample size efficiencies against gage costs and mathematical assumptions:
- Parameter Attribute Sampling (ANSI/ASQ Z1.4) Variables Sampling (ANSI/ASQ Z1.9)
- Data Type Qualitative / Binary (Pass/Fail, Go/No-Go) Quantitative / Continuous (Exact numerical dimensions)
Sample Size ( n n n ) Relatively large (e.g., n = 125 n = 125 n = 125 to 315 315 315 ) Substantially smaller (often n = 10 n = 10 n = 10 to 35 35 35 for same protection)
Distribution Requirement Distribution-free (Binomial or Poisson counts) Strict Normality Assumption is mandatory
Multivariate Inspection Multiple characteristics checked simultaneously on one plan Separate sampling plan required for each individual feature
Gaging and Tooling Inexpensive fixed limit gages (plug, ring, snap gages) Precision variable gages (micrometers, bore gages, CMMs)
Process Diagnostic Insight Low; only counts defects, no indication of drift High ; tracks mean shifts ( X ˉ \bar{X} X ˉ ) and dispersion changes ( s s s )
Inspector Math Burden Minimal (counting defectives d d d ) Moderate (calculating X ˉ \bar{X} X ˉ , s s s , Q Q Q , and table interpolation)
The Normality Prerequisite and Verification Techniques
The most critical rule in variables acceptance sampling is that the underlying quality characteristic must follow a normal (Gaussian) distribution .
Why Normality Is Mandatory
Variables sampling tables convert the distance between the sample mean and specification limit into an estimated tail area nonconformance percentage. This mathematical translation relies 100% on the theoretical bell curve. If the underlying process is skewed, bimodal, or flat (uniform), the estimated tail area will be wildly inaccurate. An inspector could accept a severely defective lot or reject a perfectly conforming lot because the mathematical model does not match physical reality.
Verification Methods for Inspectors
Before applying ANSI/ASQ Z1.9, quality engineering must verify normality using historical production data or sample datasets:
Normal Probability Plots (Q-Q Plots) : Individual data points are plotted against theoretical normal quantiles. If the distribution is normal, the data points fall tightly along a straight diagonal line. S-shaped curves indicate heavy tails (kurtosis), while bowed curves indicate skewness.
Statistical Goodness-of-Fit Tests : Formal statistical tests, such as the Anderson-Darling test , Shapiro-Wilk test , or Chi-Square goodness-of-fit test , must confirm p p p -values greater than 0.05 0.05 0.05 .
Shop Metrology Warning : Geometric tolerances that are bounded at zero—such as flatness, circularity, perpendicularity, and total runout —naturally exhibit skewed, non-normal distributions (often Chi or Weibull distributions). They cannot be directly evaluated under ANSI/ASQ Z1.9 without a mathematical transformation (e.g., Box-Cox or logarithmic) or falling back to ANSI/ASQ Z1.4 attribute sampling.
- Structure of ANSI/ASQ Z1.9: Inspection Levels and Variability Methods
- Inspection Levels
Similar to Z1.4, ANSI/ASQ Z1.9 utilizes:
General Inspection Levels I, II, III : Level II is the standard default for general variable inspection. Level I provides smaller sample sizes; Level III provides larger sample sizes.
Special Inspection Levels S-3, S-4 : Utilized for destructive or exceptionally costly physical testing.
Variability Estimation Methods
ANSI/ASQ Z1.9 is organized into distinct sections based on how process variability is determined:
Section B: Standard Deviation Method ( s s s -Method) : Process variability is unknown and estimated using the sample standard deviation s s s . This is the most common method on modern inspection benches equipped with digital calipers, CMMs, or digital indicators.
Section C: Range Method ( R R R -Method) : Process variability is estimated using the average sample range R ˉ \bar{R} R ˉ of sub-samples. Historically favored in machine shops before digital calculators because calculating range ( X max − X min X_{\text{max}} - X_{\text{min}} X max − X min ) avoided square root calculations.
Section D: Known Standard Deviation Method ( σ \sigma σ -Method) : Utilized when extensive historical SPC data demonstrates that process variability σ \sigma σ is strictly stable and known. This method yields the absolute smallest sample sizes in the standard.
Single vs. Double Specification Limits and Quality Indices
Engineering blueprints establish either single or double specification limits:
Upper Specification Limit (USL) : Maximum allowable physical dimension (e.g., max diameter 0.5050 in 0.5050\text{ in} 0.5050 in ).
Lower Specification Limit (LSL) : Minimum allowable physical dimension (e.g., min diameter 0.4950 in 0.4950\text{ in} 0.4950 in ).
Calculating the Quality Index ( Q Q Q )
The Quality Index measures the standardized distance (in sample standard deviation units) separating the sample mean from the engineering tolerance limits: Upper Quality Index: Q U = USL − X ˉ s \text{Upper Quality Index: } Q_U = \frac{\text{USL} - \bar{X}}{s} Upper Quality Index: Q U = s USL − X ˉ Lower Quality Index: Q L = X ˉ − LSL s \text{Lower Quality Index: } Q_L = \frac{\bar{X} - \text{LSL}}{s} Lower Quality Index: Q L = s X ˉ − LSL
Notice the symmetry: Q U Q_U Q U and Q L Q_L Q L are always formulated so that a larger positive value of Q Q Q indicates that the sample mean is further inside the specification limit (indicating better quality and lower defect risk).
Acceptance Criteria: Form 1 vs. Form 2
ANSI/ASQ Z1.9 offers two alternative procedural formats for deciding lot disposition:
Form 1: Acceptability Constant Method ( k k k -Method)
Primarily utilized for single specification limits (USL only or LSL only).
The inspector computes the Quality Index ( Q U Q_U Q U or Q L Q_L Q L ) and compares it directly against the acceptability constant k k k found in Table B-1 or Table B-2 of Z1.9 for the designated Code Letter and AQL:
For an upper specification limit: Accept lot if Q U ≥ k Q_U \ge k Q U ≥ k ; Reject if Q U < k Q_U < k Q U < k .
For a lower specification limit: Accept lot if Q L ≥ k Q_L \ge k Q L ≥ k ; Reject if Q L < k Q_L < k Q L < k .
Form 2: Maximum Allowable Percent Nonconforming Method ( M M M -Method)
Mandatory for double specification limits (both USL and LSL present), and universally applicable to single limits.
Instead of comparing Q Q Q directly to k k k , the inspector uses the calculated Q Q Q values to look up the estimated percent nonconforming from Table B-5 of ANSI/ASQ Z1.9:
Look up Q U Q_U Q U in Table B-5 to determine estimated upper percent nonconforming ( p U p_U p U ).
Look up Q L Q_L Q L in Table B-5 to determine estimated lower percent nonconforming ( p L p_L p L ).
Calculate total estimated percent nonconforming: p = p U + p L p = p_U + p_L p = p U + p L .
Compare individual and total estimated defect percentages against the maximum allowable percent nonconforming ( M M M ) extracted from Master Table B-3:
Accept the lot if : p U ≤ M p_U \le M p U ≤ M , AND p L ≤ M p_L \le M p L ≤ M , AND p = p U + p L ≤ M p = p_U + p_L \le M p = p U + p L ≤ M .
Reject the lot if : p > M p > M p > M , or if either individual tail estimate exceeds M M M .
Step-by-Step Worked Shop Floor Inspection Calculation
To master variables sampling for the ASQ CQI examination, walk through a realistic CNC grinding inspection scenario:
Inspection Problem Setup
Component : Precision ground stainless steel hydraulic spool valve.
Characteristic : Critical spool land diameter.
Engineering Tolerance : 1.0000 in ± 0.0030 in 1.0000\text{ in} \pm 0.0030\text{ in} 1.0000 in ± 0.0030 in .
- USL = 1.0030 in \text{USL} = 1.0030\text{ in} USL = 1.0030 in
- LSL = 0.9970 in \text{LSL} = 0.9970\text{ in} LSL = 0.9970 in
Lot Size ( N N N ) : 1 , 500 1,500 1 , 500 pieces submitted from CNC OD Grinder #4.
Sampling Requirement : Normal Inspection Level II, AQL = 1.5 % \text{AQL} = 1.5\% AQL = 1.5% , Standard Deviation ( s s s ) Method, Form 2 ( M M M -Method).
Step 1: Determine Code Letter, Sample Size, and Acceptance Criteria
Consult ANSI/ASQ Z1.9 Table A-2 (Sample Size Code Letters):
For Lot Size N = 1 , 500 N = 1,500 N = 1 , 500 under General Inspection Level II, the Sample Size Code Letter is J .
Consult Master Table B-3 (Single Sampling Plans for Normal Inspection):
For Code Letter J, the required sample size is n = 35 n = 35 n = 35 .
For AQL = 1.5 % \text{AQL} = 1.5\% AQL = 1.5% , the maximum allowable percent nonconforming is M = 3.32 % M = 3.32\% M = 3.32% .
Step 2: Perform Measurements and Calculate Sample Statistics
The inspector randomly draws n = 35 n = 35 n = 35 spools from the lot and measures each land diameter on an optical micrometer calibrated to 0.00005 in 0.00005\text{ in} 0.00005 in . The computed sample statistics are:
∑ X = 35.0630 in ⟹ X ˉ = 35.0630 35 = 1.0018 in \sum X = 35.0630\text{ in} \implies \bar{X} = \frac{35.0630}{35} = 1.0018\text{ in} ∑ X = 35.0630 in ⟹ X ˉ = 35 35.0630 = 1.0018 in
∑ ( X − X ˉ ) 2 = 0.00001224 ⟹ s = 0.00001224 34 = 0.00060 in \sum (X - \bar{X})^2 = 0.00001224 \implies s = \sqrt{\frac{0.00001224}{34}} = 0.00060\text{ in} ∑ ( X − X ˉ ) 2 = 0.00001224 ⟹ s = 34 0.00001224 = 0.00060 in
Step 3: Compute Upper and Lower Quality Indices ( Q U Q_U Q U and Q L Q_L Q L ) Upper Quality Index: Q U = USL − X ˉ s = 1.0030 − 1.0018 0.00060 = 0.0012 0.00060 = 2.00 \text{Upper Quality Index: } Q_U = \frac{\text{USL} - \bar{X}}{s} = \frac{1.0030 - 1.0018}{0.00060} = \frac{0.0012}{0.00060} = 2.00 Upper Quality Index: Q U = s USL − X ˉ = 0.00060 1.0030 − 1.0018 = 0.00060 0.0012 = 2.00 Lower Quality Index: Q L = X ˉ − LSL s = 1.0018 − 0.9970 0.00060 = 0.0048 0.00060 = 8.00 \text{Lower Quality Index: } Q_L = \frac{\bar{X} - \text{LSL}}{s} = \frac{1.0018 - 0.9970}{0.00060} = \frac{0.0048}{0.00060} = 8.00 Lower Quality Index: Q L = s X ˉ − LSL = 0.00060 1.0018 − 0.9970 = 0.00060 0.0048 = 8.00
Step 4: Look Up Estimated Percent Nonconforming in Table B-5
Consult Table B-5 ( Table for Estimating the Percent Nonconforming Using Standard Deviation Method ) for n = 35 n = 35 n = 35 :
For Q U = 2.00 Q_U = 2.00 Q U = 2.00 : The table yields an estimated upper tail nonconformance of p U = 1.62 % p_U = 1.62\% p U = 1.62% .
For Q L = 8.00 Q_L = 8.00 Q L = 8.00 : Because the mean is 8 standard deviations above LSL, the lower tail nonconformance is essentially zero: p L = 0.00 % p_L = 0.00\% p L = 0.00% .
Calculate total estimated percent nonconforming: p = p U + p L = 1.62 % + 0.00 % = 1.62 % p = p_U + p_L = 1.62\% + 0.00\% = 1.62\% p = p U + p L = 1.62% + 0.00% = 1.62%
Step 5: Adjudicate Lot Disposition
Compare estimated defect percentages against M = 3.32 % M = 3.32\% M = 3.32% :
- p U = 1.62 % ≤ 3.32 % p_U = 1.62\% \le 3.32\% p U = 1.62% ≤ 3.32% (Upper tail satisfies requirement)
- p L = 0.00 % ≤ 3.32 % p_L = 0.00\% \le 3.32\% p L = 0.00% ≤ 3.32% (Lower tail satisfies requirement)
p total = 1.62 % ≤ 3.32 % p_{\text{total}} = 1.62\% \le 3.32\% p total = 1.62% ≤ 3.32% (Total satisfies requirement)
Disposition Decision : ACCEPT THE LOT . Even though the sample mean was shifted upward toward USL ( 1.0018 in 1.0018\text{ in} 1.0018 in vs nominal 1.0000 in 1.0000\text{ in} 1.0000 in ), the tight process standard deviation ( 0.00060 in 0.00060\text{ in} 0.00060 in ) kept total estimated nonconformance ( 1.62 % 1.62\% 1.62% ) well below the permissible M M M limit ( 3.32 % 3.32\% 3.32% ).
Real Shop Inspection Scenarios & Common Exam Traps
Exam Trap: Confusing Form 1 and Form 2 Decision Directions :
In Form 1 ( k k k -method), larger Q Q Q is better: the lot is accepted if Q ≥ k Q \ge k Q ≥ k . In Form 2 ( M M M -method), smaller p p p is better: the lot is accepted if p ≤ M p \le M p ≤ M . Novices frequently invert these inequalities on certification exams.
Exam Trap: Assuming All Physical Features Can Use Z1.9 :
A candidate may assume that any feature dimensioned on a print can use variables sampling. Remember: Z1.9 requires a normal distribution and continuous variables. Visual defect counts, thread go/no-go checks, solder bridge inspections, and un-transformed geometric tolerances (runout, flatness) cannot use Z1.9.
Exam Trap: Failing to Check Individual Tails in Double Limits :
Under Form 2 for double limits, satisfying p total ≤ M p_{\text{total}} \le M p total ≤ M is not enough! Both individual tails must also satisfy p U ≤ M p_U \le M p U ≤ M and p L ≤ M p_L \le M p L ≤ M . While mathematically p U + p L ≤ M p_U + p_L \le M p U + p L ≤ M usually implies each is ≤ M \le M ≤ M , specific table adjustments for asymmetric tolerances require verifying individual tails.