Resources What's New Support Download Contact Sales: Arthur-Industrial@outlook.com
Home › Resources › The normal distribution

The normal distribution

The bell curve, the 68-95-99.7 rule, z-scores, and why so many process measurements look normal.

The normal (Gaussian) distribution is a symmetric bell-shaped curve defined by its mean μ and standard deviation . Many process measurements are close to normal because they are the sum of many small, independent effects.

The empirical rule

  • About 68.27% of values fall within of the mean.
  • About 95.45% fall within .
  • About 99.73% fall within , leaving 0.27% (2,700 parts per million) outside.

Z-scores

A z-score says how many standard deviations a value is from the mean. With the standard normal table (or RS-1) you convert a z-score into a probability.

Example: mean 10.000 mm, mm, upper spec 10.012 mm. , so about 0.135% of parts (1,350 ppm) are above the upper spec.

Checking normality

  • Histogram: roughly bell-shaped?
  • Normal probability plot: points should follow a straight line.
  • Anderson-Darling test: a p-value below 0.05 suggests the data are not normal. With very large samples, tiny harmless departures become "significant", so always look at the plot too.

Why it matters

Capability indices and many tests assume normal data. Control charts for averages are much less sensitive to non-normality, because averages of subgroups tend toward normal (the central limit theorem).

Questions and answers

My data fail the normality test. Is my process bad?

Not necessarily. Some characteristics are naturally skewed (flatness, runout, concentricity, cycle times). Use a distribution that fits, a transformation, or non-normal capability. Also check whether the data mix two machines or two shifts.

Do I need normal data for a control chart?

Individuals charts are somewhat sensitive to strong skew. Xbar charts are robust. The chart is still useful for spotting change; just be careful turning its limits into defect rates.

In RS-1: Explore > Normality Test and Probability Plot; the capability analysis checks normality for you.

← All articles