Study guide

4.3 Normal Distribution & Z-Score Calculations

ASQ Certified Quality Technician CQT

Track your progress as you read

Finished sections get a check mark.

1 Chapter 1: Quality Philosophies, Principles, and Management Systems

1.1 Quality Philosophies & Pioneers 1.2 Quality Management Systems, Standards & Documentation 1.3 Customer-Supplier Relationships & Quality Culture

2 Chapter 2: Cost of Quality, Teams, and Professional Ethics

2.1 Cost of Quality (COQ) & Economic Trade-Offs 2.2 Quality Improvement Teams & Meeting Management 2.3 Professional Conduct & ASQ Code of Ethics

3 Chapter 3: Basic Quality Tools & Continuous Improvement

3.1 Process Mapping, Flowcharts, and SIPOC 3.2 Cause-and-Effect Analysis & Check Sheets 3.3 Pareto Analysis & Prioritization 3.4 Histograms & Scatter Diagrams

4 Chapter 4: Descriptive Statistics and Probability Distributions

4.1 Data Types, Sampling Methods, and Populations 4.2 Measures of Central Tendency and Dispersion 4.3 Normal Distribution & Z-Score Calculations 4.4 Discrete Distributions: Binomial, Poisson, and Hypergeometric

5 Chapter 5: Statistical Process Control (SPC) Fundamentals & Variables Charts

5.1 SPC Fundamentals & Process Variation 5.2 Control Limits vs. Specification Limits 5.3 X-bar and R Charts 5.4 X-bar & s Charts and Individual-Moving Range (I-MR) Charts

6 Chapter 6: Attributes Control Charts and Out-of-Control Patterns

6.1 Control Charts for Nonconforming Units: p and np Charts 6.2 Control Charts for Nonconformities: c and u Charts 6.3 Out-of-Control Rules & Process Analysis

7 Chapter 7: Process Capability and Performance Analysis

7.1 Process Stability Prerequisite & Capability Basics 7.2 Potential Capability (Cp) and Actual Capability (Cpk) 7.3 Process Performance (Pp, Ppk) and Long-Term vs. Short-Term Variation 7.4 One-Sided Tolerances and PPM Defect Calculations

8 Chapter 8: Metrology Fundamentals, Standards, and Calibration

8.1 Metrology Foundations & Environmental Controls 8.2 Precision Gage Blocks & Standards Hierarchy 8.3 Calibration Systems & NIST Traceability 8.4 Out-of-Calibration Procedures & Reverse Traceability 8.5 M&TE Identification, Control, and Customer-Supplied Equipment

9 Chapter 9: Measurement Instruments and Measurement System Analysis (Gage R&R)

9.1 Common Dimensional Hand Measuring Instruments 9.2 Fixed Gages, Optical Comparators, and CMM Systems 9.3 Measurement System Analysis: Bias, Linearity, and Stability 9.4 Variable Gage R&R: Repeatability and Reproducibility 9.5 Electronic, Weighing, and Force Measurement Equipment

10 Chapter 10: Technical Drawings, Tolerancing, and GD&T (ASME Y14.5)

10.1 Engineering Drawing Basics & Projection Methods 10.2 Datums and the Datum Reference Frame 10.3 Form and Orientation Controls in GD&T 10.4 Location, Profile, and Runout Controls with Modifiers 10.5 Classification of Product Characteristics and Defects

11 Chapter 11: Inspection Planning, Sampling Plans, and Material Disposition

11.1 Inspection Planning & Inspection Points 11.2 Acceptance Sampling & ANSI/ASQ Z1.4 11.3 Switching Rules & Variables Sampling (Z1.9) 11.4 Nonconforming Material Control & MRB Dispositions 11.5 Types of Measurements, Unit Conversion, and Rounding Rules

12 Chapter 12: Quality Audits

12.1 Quality Auditing Principles & Types 12.2 The Audit Process: Planning, Performance, and Reporting 12.3 Audit Tools and Techniques

13 Chapter 13: Risk Management, Corrective and Preventive Action, and Safety

13.1 Risk Assessment Inputs, Monitoring, and Control Plans 13.2 FMEA, Action Priority, and Technician Workplace Safety 13.3 Root Cause Analysis & Problem-Solving 13.4 Corrective and Preventive Action (CAPA) & Verification 13.5 Preventive Action, Mistake-Proofing, and Read-Across

4.3 Normal Distribution & Z-Score Calculations

Key Takeaways Explain This Topic AI

The Empirical Rule (68-95-99.7) establishes that 68.26% of values fall within μ ± 1σ, 95.44% fall within μ ± 2σ, and 99.73% fall within μ ± 3σ, leaving 0.27% (2,700 PPM) in the bilateral tails outside three-sigma limits.

The standard normal Z-score transformation (Z = (X - μ) / σ) converts any normal distribution into the standard normal distribution Z ~ N(0, 1), expressing distances in units of standard deviations.

To calculate nonconformance PPM, compute Z-scores for Upper and Lower Specification Limits, look up cumulative tail probabilities in standard normal tables, and multiply the sum of tail fractions by 1,000,000.

The Central Limit Theorem (CLT) proves that the distribution of sample averages (x̄) approaches normality with mean μ and standard error σ / √n as sample size increases, regardless of the shape of the parent population.

A confidence interval places a range around a sample estimate at a stated confidence level: x-bar plus or minus z(alpha/2) times sigma over root n when sigma is known, or the t distribution when sigma is estimated from a small sample; the interval narrows with the square root of n and widens as the confidence level rises.

Last updated: September 2026

  • 4.3 Normal Distribution & Z-Score Calculations
  • Foundations of the Normal (Gaussian) Distribution

The normal distribution (often called the Gaussian distribution or the classic bell curve) is the single most important continuous probability distribution in quality engineering. First described mathematically by Abraham de Moivre in 1733 and later expanded by Carl Friedrich Gauss, the normal distribution models natural variation resulting from the combination of many independent, random common causes (such as minor ambient temperature fluctuations, machine micro-vibrations, material alloy inconsistencies, and slight line voltage variations).

Mathematical Probability Density Function (PDF)

The continuous probability density function of a normal random variable X X X with mean μ \mu μ and standard deviation σ \sigma σ is defined by: f ( x ) = 1 σ 2 π e − 1 2 ( x − μ σ ) 2 , − ∞ < x < ∞ f(x) = \frac{1}{\sigma \sqrt{2\pi}} e^{-\frac{1}{2}\left(\frac{x - \mu}{\sigma}\right)^2}, \quad -\infty < x < \infty f ( x ) = σ 2 π ​ 1 ​ e − 2 1 ​ ( σ x − μ ​ ) 2 , − ∞ < x < ∞

where π ≈ 3.14159 \pi \approx 3.14159 π ≈ 3.14159 and e ≈ 2.71828 e \approx 2.71828 e ≈ 2.71828 .

Core Geometric and Metrological Properties

Perfect Symmetry : The distribution is unimodal and perfectly symmetrical around its central axis. The left half is a precise mirror image of the right half.

Coincidence of Central Measures : At the peak center of the distribution, the arithmetic mean, median, and mode are identical: μ = Median = Mode \mu = \text{Median} = \text{Mode} μ = Median = Mode

Points of Inflection : The curve changes concavity (from concave downward to concave upward) at exactly one standard deviation on either side of the mean: μ − σ \mu - \sigma μ − σ and μ + σ \mu + \sigma μ + σ .

Asymptotic Tails : The curve approaches the horizontal axis asymptotically as x x x moves toward ± ∞ \pm \infty ± ∞ , never quite touching zero. Theoretically, any value between − ∞ -\infty − ∞ and + ∞ +\infty + ∞ is possible, though values beyond ± 4 σ \pm 4\sigma ± 4 σ are exceedingly rare.

Unit Area : The total area beneath the probability density curve is exactly equal to 1.0 (representing 100% total probability).

The Empirical Rule (The 68-95-99.7 Rule)

For any data set that is approximately normally distributed, the Empirical Rule dictates the fixed proportion of observations falling within integer standard deviation bands around the mean: Empirical Rule Distribution

  • μ
  • /|\
  • / | \
  • / | \
  • --/ | \---
  • ---------------/ | \----------------
  • Tail | 2.14% | 13.59% |34.13%|34.13%| 13.59% | 2.14% | Tail
  • 0.135%| | | | | | |0.135%
  • -------+------------+-----------+------+------+-----------+------------+--------
  • μ-3σ μ-2σ μ-1σ μ μ+1σ μ+2σ μ+3σ
  • | | | |
  • | +--- 68.26% ---+ |
  • | |
  • +---------------------------- 95.44% ----------------------------+
  • | |
  • +--------------------------- 99.73% -----------------------------+
  • Exact Theoretical Percentages

μ ± 1 σ \mu \pm 1\sigma μ ± 1 σ : Contains 68.26% of all parts (approx. 68%). Exactly 34.13 % 34.13\% 34.13% lies between μ \mu μ and μ + 1 σ \mu + 1\sigma μ + 1 σ , and 34.13 % 34.13\% 34.13% lies between μ − 1 σ \mu - 1\sigma μ − 1 σ and μ \mu μ .

μ ± 2 σ \mu \pm 2\sigma μ ± 2 σ : Contains 95.44% of all parts (approx. 95%). Exactly 47.72 % 47.72\% 47.72% lies on each side of the mean.

μ ± 3 σ \mu \pm 3\sigma μ ± 3 σ : Contains 99.73% of all parts (approx. 99.7%). Exactly 49.865 % 49.865\% 49.865% lies on each side of the mean.

Tail Area Probabilities and Scrap PPM

When a process is centered between specification limits located at ± 3 σ \pm 3\sigma ± 3 σ :

The proportion of product falling outside the limits is 1.0000 − 0.9973 = 0.0027 1.0000 - 0.9973 = 0.0027 1.0000 − 0.9973 = 0.0027 (or 0.27% ).

Divided evenly between the two tails: 0.135% falls above the Upper Specification Limit ( + 3 σ +3\sigma + 3 σ ), and 0.135% falls below the Lower Specification Limit ( − 3 σ -3\sigma − 3 σ ).

In parts-per-million: 0.0027 × 1 , 000 , 000 = 2 , 700 PPM 0.0027 \times 1,000,000 = \mathbf{2,700\text{ PPM}} 0.0027 × 1 , 000 , 000 = 2 , 700 PPM total nonconformance (1,350 PPM per tail).

The Standard Normal Distribution ( Z ∼ N ( 0 , 1 ) Z \sim N(0, 1) Z ∼ N ( 0 , 1 ) ) and Z-Scores

Because every manufacturing process has its own unique mean μ \mu μ and standard deviation σ \sigma σ , comparing raw measurements across different processes is cumbersome. The standard normal distribution is a standardized normal curve that has a mean of zero ( μ = 0 \mu = 0 μ = 0 ) and a standard deviation of one ( σ = 1 \sigma = 1 σ = 1 ).

Z-Score Transformation Formula

A Z-score (or standard score) converts any raw measurement X X X into units of standard deviations relative to the process mean:

  • For a Population : Z = X − μ σ Z = \frac{X - \mu}{\sigma} Z = σ X − μ ​
  • For a Sample : Z = X − x ˉ s Z = \frac{X - \bar{x}}{s} Z = s X − x ˉ ​
  • Physical Meaning of Z Z Z

If Z = 0 Z = 0 Z = 0 , the measurement falls exactly on the process mean.

If Z = + 1.50 Z = +1.50 Z = + 1.50 , the measurement lies exactly 1.50 1.50 1.50 standard deviations above the mean.

If Z = − 2.25 Z = -2.25 Z = − 2.25 , the measurement lies exactly 2.25 2.25 2.25 standard deviations below the mean.

Reading Standard Normal Cumulative Probability Tables

Standard normal tables (Z-tables) provide the cumulative area under the curve from − ∞ -\infty − ∞ up to a given Z Z Z value, denoted Φ ( Z ) = P ( Z ≤ z ) \Phi(Z) = P(Z \le z) Φ ( Z ) = P ( Z ≤ z ) .

Cumulative Table Reference Snippet

Z Z Z Φ ( Z ) = P ( Z ≤ z ) \Phi(Z) = P(Z \le z) Φ ( Z ) = P ( Z ≤ z ) Upper Tail: P ( Z > z ) = 1 − Φ ( Z ) P(Z > z) = 1 - \Phi(Z) P ( Z > z ) = 1 − Φ ( Z ) Quality Application Context

  • 0.00 0.50000 0.50000 Process center (50% above, 50% below)
  • 1.00 0.84134 0.15866 Boundary of 1 σ 1\sigma 1 σ band (158,660 PPM tail)
  • 1.28 0.89973 0.10027 90% cumulative confidence threshold
  • 1.645 0.95000 0.05000 95% single-tail confidence limit (50,000 PPM)
  • 1.96 0.97500 0.02500 95% two-tailed confidence threshold (2.5% tail)
  • 2.00 0.97725 0.02275 Boundary of 2 σ 2\sigma 2 σ band (22,750 PPM tail)
  • 2.33 0.99010 0.00990 99% single-tail confidence limit (10,000 PPM)
  • 2.575 0.99499 0.00501 99% two-tailed confidence threshold (0.5% tail)
  • 3.00 0.99865 0.00135 Natural process limit (1,350 PPM tail)
  • 4.00 0.999968 0.000032 31.7 PPM tail
  • Crucial Symmetry Rule for Negative Z Z Z

Because standard normal curves are symmetrical: P ( Z ≤ − z ) = 1 − Φ ( z ) = P ( Z ≥ z ) P(Z \le -z) = 1 - \Phi(z) = P(Z \ge z) P ( Z ≤ − z ) = 1 − Φ ( z ) = P ( Z ≥ z )

For example, P ( Z ≤ − 1.96 ) = 1 − Φ ( 1.96 ) = 1 − 0.9750 = 0.0250 P(Z \le -1.96) = 1 - \Phi(1.96) = 1 - 0.9750 = 0.0250 P ( Z ≤ − 1.96 ) = 1 − Φ ( 1.96 ) = 1 − 0.9750 = 0.0250 .

Step-by-Step Probability Calculations Between Specifications

In quality assurance, customer drawings specify a Lower Specification Limit (LSL) and an Upper Specification Limit (USL) . To compute expected scrap or rework:

Calculate Upper Z-Score ( Z U S L Z_{USL} Z U S L ​ ) : Z U S L = U S L − μ σ Z_{USL} = \frac{USL - \mu}{\sigma} Z U S L ​ = σ U S L − μ ​

Calculate Lower Z-Score ( Z L S L Z_{LSL} Z L S L ​ ) : Z L S L = L S L − μ σ Z_{LSL} = \frac{LSL - \mu}{\sigma} Z L S L ​ = σ L S L − μ ​

Calculate Conforming Proportion ( P good P_{\text{good}} P good ​ ) : P ( L S L ≤ X ≤ U S L ) = Φ ( Z U S L ) − Φ ( Z L S L ) P(LSL \le X \le USL) = \Phi(Z_{USL}) - \Phi(Z_{LSL}) P ( L S L ≤ X ≤ U S L ) = Φ ( Z U S L ​ ) − Φ ( Z L S L ​ )

Calculate Upper Tail Fraction (Scrap above USL) : P ( X > U S L ) = 1 − Φ ( Z U S L ) P(X > USL) = 1 - \Phi(Z_{USL}) P ( X > U S L ) = 1 − Φ ( Z U S L ​ )

Calculate Lower Tail Fraction (Scrap below LSL) : P ( X < L S L ) = Φ ( Z L S L ) P(X < LSL) = \Phi(Z_{LSL}) P ( X < L S L ) = Φ ( Z L S L ​ )

Convert to Parts Per Million (PPM) : PPM = P defect × 1 , 000 , 000 \text{PPM} = P_{\text{defect}} \times 1,000,000 PPM = P defect ​ × 1 , 000 , 000

  • Comprehensive Worked Metrology Example: Uncentered Precision Dowel Pins
  • Scenario

A precision Swiss-machining cell manufactures hardened ground steel dowel pins. The engineering drawing specifies outer diameter as: Outer Diameter = 12.000 ± 0.015 mm \text{Outer Diameter} = 12.000 \pm 0.015\text{ mm} Outer Diameter = 12.000 ± 0.015 mm

Lower Specification Limit ( L S L LSL L S L ) = 12.000 − 0.015 = 11.985 mm 12.000 - 0.015 = 11.985\text{ mm} 12.000 − 0.015 = 11.985 mm

Upper Specification Limit ( U S L USL U S L ) = 12.000 + 0.015 = 12.015 mm 12.000 + 0.015 = 12.015\text{ mm} 12.000 + 0.015 = 12.015 mm

A capability study of 100 consecutive pieces demonstrates that the process is in statistical control and normally distributed, with:

Process Mean: μ = 12.003 mm \mu = 12.003\text{ mm} μ = 12.003 mm (process is running slightly oversized by 0.003 mm 0.003\text{ mm} 0.003 mm )

Process Standard Deviation: σ = 0.004 mm \sigma = 0.004\text{ mm} σ = 0.004 mm

The quality technician must calculate the expected scrap fraction, conforming yield, and defect rates in PPM.

Step 1: Compute Upper and Lower Z-Scores Z U S L = U S L − μ σ = 12.015 − 12.003 0.004 = + 0.012 0.004 = + 3.00 Z_{USL} = \frac{USL - \mu}{\sigma} = \frac{12.015 - 12.003}{0.004} = \frac{+0.012}{0.004} = +3.00 Z U S L ​ = σ U S L − μ ​ = 0.004 12.015 − 12.003 ​ = 0.004 + 0.012 ​ = + 3.00 Z L S L = L S L − μ σ = 11.985 − 12.003 0.004 = − 0.018 0.004 = − 4.50 Z_{LSL} = \frac{LSL - \mu}{\sigma} = \frac{11.985 - 12.003}{0.004} = \frac{-0.018}{0.004} = -4.50 Z L S L ​ = σ L S L − μ ​ = 0.004 11.985 − 12.003 ​ = 0.004 − 0.018 ​ = − 4.50

Step 2: Determine Tail Areas from Normal Tables

Upper Tail ( X > U S L X > USL X > U S L ) :

For Z U S L = + 3.00 Z_{USL} = +3.00 Z U S L ​ = + 3.00 , Φ ( 3.00 ) = 0.99865 \Phi(3.00) = 0.99865 Φ ( 3.00 ) = 0.99865 . P ( X > 12.015 ) = 1 − Φ ( 3.00 ) = 1 − 0.99865 = 0.00135 P(X > 12.015) = 1 - \Phi(3.00) = 1 - 0.99865 = 0.00135 P ( X > 12.015 ) = 1 − Φ ( 3.00 ) = 1 − 0.99865 = 0.00135

This corresponds to oversized scrap (or reworkable pins): PPM Upper = 0.00135 × 1 , 000 , 000 = 1 , 350 PPM \text{PPM}_{\text{Upper}} = 0.00135 \times 1,000,000 = \mathbf{1,350\text{ PPM}} PPM Upper ​ = 0.00135 × 1 , 000 , 000 = 1 , 350 PPM

Lower Tail ( X < L S L X < LSL X < L S L ) :

For Z L S L = − 4.50 Z_{LSL} = -4.50 Z L S L ​ = − 4.50 , using the symmetry rule: P ( X < 11.985 ) = Φ ( − 4.50 ) = 1 − Φ ( 4.50 ) P(X < 11.985) = \Phi(-4.50) = 1 - \Phi(4.50) P ( X < 11.985 ) = Φ ( − 4.50 ) = 1 − Φ ( 4.50 )

From extended standard normal tables, Φ ( 4.50 ) ≈ 0.9999966 \Phi(4.50) \approx 0.9999966 Φ ( 4.50 ) ≈ 0.9999966 . P ( X < 11.985 ) = 1 − 0.9999966 = 0.0000034 P(X < 11.985) = 1 - 0.9999966 = 0.0000034 P ( X < 11.985 ) = 1 − 0.9999966 = 0.0000034

This corresponds to undersized scrap (unrecoverable material loss): PPM Lower = 0.0000034 × 1 , 000 , 000 = 3.4 PPM \text{PPM}_{\text{Lower}} = 0.0000034 \times 1,000,000 = \mathbf{3.4\text{ PPM}} PPM Lower ​ = 0.0000034 × 1 , 000 , 000 = 3.4 PPM

Step 3: Total Nonconformance Rate and Yield Total Nonconformance Proportion ( P def ) = 0.00135 + 0.0000034 ≈ 0.0013534 \text{Total Nonconformance Proportion } (P_{\text{def}}) = 0.00135 + 0.0000034 \approx 0.0013534 Total Nonconformance Proportion ( P def ​ ) = 0.00135 + 0.0000034 ≈ 0.0013534 Total PPM = 1 , 350 + 3.4 = 1 , 353.4 PPM \text{Total PPM} = 1,350 + 3.4 = \mathbf{1,353.4\text{ PPM}} Total PPM = 1 , 350 + 3.4 = 1 , 353.4 PPM Conforming Process Yield = ( 1 − 0.0013534 ) × 100 % = 99.865 % \text{Conforming Process Yield} = (1 - 0.0013534) \times 100\% = \mathbf{99.865\%} Conforming Process Yield = ( 1 − 0.0013534 ) × 100% = 99.865%

Metrology Insight : Because the process mean was shifted high by only 0.003 mm 0.003\text{ mm} 0.003 mm (less than one standard deviation!), over 99.7 % 99.7\% 99.7% of all defective parts occur at the high side ( 1 , 350 PPM 1,350\text{ PPM} 1 , 350 PPM high vs. 3.4 PPM 3.4\text{ PPM} 3.4 PPM low). Centering the tool offset back to 12.000 mm 12.000\text{ mm} 12.000 mm would equalize both tails at Z = ± 3.75 Z = \pm 3.75 Z = ± 3.75 , dropping total scrap from 1 , 353.4 PPM 1,353.4\text{ PPM} 1 , 353.4 PPM to under 0.2 PPM 0.2\text{ PPM} 0.2 PPM !

The Central Limit Theorem (CLT)

The Central Limit Theorem (CLT) is the mathematical engine powering statistical process control (SPC). It explains why control charts work across thousands of diverse manufacturing operations where individual part dimensions are not normally distributed.

Formal Statement of the Central Limit Theorem

Given a population of any arbitrary distributional shape (uniform, exponential, triangular, or skewed) with a finite mean μ \mu μ and finite standard deviation σ \sigma σ :

As the sample size n n n increases, the distribution of the sample means ( x ˉ \bar{x} x ˉ ) approaches a normal distribution .

The mean of the distribution of sample means equals the population mean: μ x ˉ = μ \mu_{\bar{x}} = \mu μ x ˉ ​ = μ

The standard deviation of the distribution of sample means (the standard error of the mean ) equals the population standard deviation divided by the square root of the sample size: σ x ˉ = σ n \sigma_{\bar{x}} = \frac{\sigma}{\sqrt{n}} σ x ˉ ​ = n ​ σ ​

Practical Rules of Thumb for Quality Technicians

Sample Size ( n ≥ 30 n \ge 30 n ≥ 30 ) : If the underlying parent population is heavily skewed or multimodal, a sample size of n ≥ 30 n \ge 30 n ≥ 30 ensures the sample mean distribution is virtually normal.

Moderate Sample Size ( n = 4 n = 4 n = 4 or 5 5 5 ) : If the underlying population is reasonably symmetrical and unimodal, subgroup sizes of just n = 4 n = 4 n = 4 or n = 5 n = 5 n = 5 (the standard rational subgroup size for X ˉ − R \bar{X}-R X ˉ − R charts) produce sample means that conform closely to the normal distribution.

Standard Error Reduction : If a technician increases subgroup size from n = 4 n = 4 n = 4 to n = 16 n = 16 n = 16 , the standard error of the mean cuts in half ( 16 = 4 \sqrt{16} = 4 16 ​ = 4 vs. 4 = 2 \sqrt{4} = 2 4 ​ = 2 ), tightening control limits and accelerating detection of small process shifts.

  • Real-World Shop Scenarios & Common Exam Traps
  • Common Exam Traps for CQT Candidates

Exam Trap 1: Forgetting to Divide by n \sqrt{n} n ​ for Sample Means : When an exam problem asks for the probability that an individual part exceeds a specification, use Z = X − μ σ Z = \frac{X - \mu}{\sigma} Z = σ X − μ ​ . But when the question asks for the probability that the average of a sample subgroup ( x ˉ \bar{x} x ˉ ) exceeds a limit, you MUST use the standard error: Z = x ˉ − μ σ / n Z = \frac{\bar{x} - \mu}{\sigma / \sqrt{n}} Z = σ / n ​ x ˉ − μ ​ . Using σ \sigma σ instead of σ / n \sigma/\sqrt{n} σ / n ​ for sample averages is the single most common mathematical error on the CQT exam.

Exam Trap 2: Subtracting from 1.0 for Upper Tails : Standard normal tables provide cumulative area from the left ( − ∞ -\infty − ∞ to + Z +Z + Z ). If you need the scrap fraction exceeding an Upper Specification Limit, remember to subtract the table value from 1.0 ( P ( X > U S L ) = 1 − Φ ( Z ) P(X > USL) = 1 - \Phi(Z) P ( X > U S L ) = 1 − Φ ( Z ) ).

Exam Trap 3: Assuming All Physical Features Follow Normal Distributions : Runout, hole position error (true position), surface roughness ( R a R_a R a ​ ), and flatness cannot be negative; their distributions are bounded at zero and naturally follow Chi-square, Rayleigh, or Weibull distributions. Technicians must never blindly apply normal Z-score tables to geometric tolerances without testing for normality.

Confidence Levels and Confidence Limits

Every statistic a technician calculates from a sample — a mean, a standard deviation, a C p k C_{pk} C p k ​ , a gage bias, a percent nonconforming — is an estimate . Repeat the sampling and the number changes. A confidence interval puts a defensible range around that estimate, and the Body of Knowledge expects a technician to explain confidence levels and limits in various situations.

Point Estimate Versus Interval Estimate

A point estimate is a single value: the sample mean x ˉ = 1.2504 " \bar{x} = 1.2504" x ˉ = 1.2504" .

An interval estimate is a range with a stated confidence level : "with 95 percent confidence, the true process mean lies between 1.2501 " 1.2501" 1.2501" and 1.2507 " 1.2507" 1.2507" ."

The endpoints of that range are the confidence limits ; the range between them is the confidence interval .

The confidence level (commonly 90, 95, or 99 percent) is the long-run proportion of such intervals that would contain the true parameter if the sampling were repeated many times. Its complement is α \alpha α : a 95 percent confidence level corresponds to α = 0.05 \alpha = 0.05 α = 0.05 , which is the same α \alpha α used as the significance level in the normality tests of Section 7.1 and the bias t t t -test of Section 9.3.

The Interval for a Process Mean

When the process standard deviation σ \sigma σ is known — the usual situation once a control chart has been running and σ \sigma σ has been estimated from R ˉ / d 2 \bar{R}/d_2 R ˉ / d 2 ​ — the interval uses the standard error of the mean from the Central Limit Theorem: x ˉ ± z α / 2 σ n \bar{x} \pm z_{\alpha/2} \frac{\sigma}{\sqrt{n}} x ˉ ± z α /2 ​ n ​ σ ​

with z α / 2 = 1.645 z_{\alpha/2} = 1.645 z α /2 ​ = 1.645 for 90 percent, 1.96 1.96 1.96 for 95 percent, and 2.575 2.575 2.575 for 99 percent two-sided confidence. When σ \sigma σ is unknown and the sample is small, the sample standard deviation s s s replaces σ \sigma σ and the t t t distribution replaces z z z : x ˉ ± t α / 2 , n − 1 s n \bar{x} \pm t_{\alpha/2,\ n-1} \frac{s}{\sqrt{n}} x ˉ ± t α /2 , n − 1 ​ n ​ s ​

The t t t value is always larger than the corresponding z z z value, and it grows as the sample shrinks — which is simply the arithmetic saying that a small sample buys less certainty.

Worked Example

A technician measures n = 25 n = 25 n = 25 shafts from a stable process and obtains x ˉ = 1.2504 " \bar{x} = 1.2504" x ˉ = 1.2504" with a known process σ = 0.0008 " \sigma = 0.0008" σ = 0.0008" . The 95 percent confidence interval for the process mean is: 1.2504 ± 1.96 × 0.0008 25 = 1.2504 ± 1.96 × 0.00016 = 1.2504 ± 0.00031 1.2504 \pm 1.96 \times \frac{0.0008}{\sqrt{25}} = 1.2504 \pm 1.96 \times 0.00016 = 1.2504 \pm 0.00031 1.2504 ± 1.96 × 25 ​ 0.0008 ​ = 1.2504 ± 1.96 × 0.00016 = 1.2504 ± 0.00031

giving confidence limits of 1.25009 " 1.25009" 1.25009" and 1.25071 " 1.25071" 1.25071" . If the print nominal is 1.2500 " 1.2500" 1.2500" , the interval does not contain the nominal, so the evidence supports a real centering offset rather than sampling noise, and a tool-offset adjustment is justified.

Two-Sided Versus One-Sided Limits

Use two-sided limits when the question is "where is the parameter?" — a process mean, a gage bias.

Use a one-sided confidence limit when only one direction matters: a lower confidence limit on a weld's break strength, an upper confidence limit on a contamination level, or a lower confidence bound on C p k C_{pk} C p k ​ . A one-sided 95 percent limit uses z = 1.645 z = 1.645 z = 1.645 , not 1.96 1.96 1.96 , because the whole 5 % 5\% 5% risk sits in one tail.

What Changes the Width

Change Effect on interval width Why

Increase the sample size n n n Narrower , in proportion to 1 / n 1/\sqrt{n} 1/ n ​ Quadrupling n n n halves the width

  • Increase the confidence level (95 to 99 percent) Wider More certainty demands more room
  • Larger process variation σ \sigma σ Wider Noisier process, less precise estimate

Switch from z z z to t t t (small sample, unknown σ \sigma σ ) Wider Penalty for estimating σ \sigma σ from the data

Interpreting the Interval Correctly

Caution

The interval is random; the parameter is not. "There is a 95 percent probability that the true mean lies in this interval" is the common misstatement. The true process mean is a fixed (if unknown) number. The correct statement is that the procedure produces intervals that capture the true mean 95 percent of the time. On any single interval, either it contains the mean or it does not.

Three further traps: a confidence interval describes the parameter , not individual parts — do not confuse it with the natural process limits of Section 7.1; a wider interval is a statement about measurement and sampling certainty , not about whether the process is capable; and no confidence interval is meaningful unless the process was stable while the data was collected.

Where a Technician Actually Meets Confidence Limits

Situation The interval in use

Capability estimates from small samples A C p k C_{pk} C p k ​ of 1.33 from 30 parts can carry a 95 percent interval from roughly 1.02 to 1.64, which is why 25 subgroups and 100+ parts are required (Section 7.1)

Gage bias studies The bias is significant when its 95 percent confidence interval does not contain zero (Section 9.3)

Calibration certificates Expanded uncertainty U U U is quoted at a 95 percent confidence level with coverage factor k = 2 k = 2 k = 2 (Section 8.3)

Acceptance sampling Producer's and consumer's risk on an OC curve are the α \alpha α and β \beta β that sit behind the same idea (Section 11.2)

Statistical audit sampling A statistically drawn audit sample projects to the population with stated confidence (Section 12.3)

Loading diagram... Standard Normal Z-Distribution and Specification Tail Areas