Direct answer
A normal model is determined by its mean and standard deviation; a z-score expresses signed distance from the mean in standard-deviation units, and tail probabilities come from the model’s cumulative area.
Visual explanation
Z-score as distance and tail area
What this calculation tells you
The normal distribution is a symmetric continuous location-scale family. Standardization converts any member of the family to the standard normal scale while preserving relative position.
Use z-scores to compare values against a defined mean and positive standard deviation. Interpret tail areas only when a normal model or sampling approximation is defensible for the question.
Where it is used
Measurement
Standardize an observation against a documented reference model.
Sampling
Interpret sample-mean probabilities under justified approximations.
Quality
Communicate standardized deviations without replacing process limits.
Education
Connect algebraic standardization to geometric area under a curve.
Common situations
- Converting an observation to a z-score.
- Finding a central or tail probability.
- Comparing values measured on different normal-model scales.
- Separating raw-data normality from sampling-distribution normality.
Start with the statistical question
Use z-scores to compare values against a defined mean and positive standard deviation. Interpret tail areas only when a normal model or sampling approximation is defensible for the question.
A movable observation travels along a bell curve while a lower axis displays its z-coordinate. Separate shading distinguishes left tail, right tail, and central probability.
Worked example
For mean 70, standard deviation 10, and value 85, z=(85−70)/10=1.5. The z-score is a distance statement; the corresponding percentile requires the normal CDF assumption.
Assumptions that carry the result
The standard deviation must be positive. Normality is a model about distribution shape and tails; a sample mean may be approximately normal under central-limit conditions even when raw observations are not.
Interpret the result without overreaching
A large absolute z-score is not proof of error or a universal danger threshold. Multiple comparisons, estimated parameters, selection, and non-normal tails change interpretation.
- Calling density at x the probability of exactly x.
- Using a z-score with a zero or irrelevant standard deviation.
- Applying the 68–95–99.7 rule to every dataset.
Practical questions
Frequently asked questions
Is a z-score a percentile?
No. It is standardized distance; a percentile follows only after applying a distribution CDF.
What is the probability of one exact continuous value?
Under a continuous model it is zero; probabilities belong to intervals, while density describes local concentration.
Does z=2 always mean unusual?
Its model-based tail area may be small, but practical importance and multiplicity depend on context.
Further reading
Authoritative sources
Use these primary and professional resources to check definitions, conventions, or requirements that may extend beyond this guide.