Is a Z Score a Standard Deviation?
No. A standard deviation is a unit of spread measured in the data's own units — marks, centimetres, seconds. A z score is a count: how many of those units fit between a value and the mean. Confusing the count with the unit is the most persistent misconception in the whole topic.
The distinction in one line: σ is the ruler; z is how many ruler-lengths away the value sits. On a test with σ = 8 marks, a z score of 2 means the value is 2 × 8 = 16 marks above the mean. The z score itself has no units at all.
Why the confusion is so natural
The formula z = (X − μ) / σ divides a distance (measured in marks) by a standard deviation (also measured in marks). The units cancel, leaving a pure number — but the phrase used to read it aloud, "two standard deviations above the mean", keeps the words "standard deviation" inside the sentence, and the ear files z under the wrong noun. A second fuel source: after standardising, the resulting z scores themselves form a distribution whose standard deviation is exactly 1 — so "the standard deviation of z scores is 1" is true, and routinely misheard as "z score is a standard deviation". Both statements are worth keeping, labelled correctly: σ belongs to the original data; the standardised distribution has μ = 0 and σ = 1 by construction. The what is a z score page builds the concept with that hygiene from the start.
Examples that separate them
- Same z, different σ: two classes both place a pupil at z = 1.5. In class A (σ = 6) that is 9 marks above the mean; in class B (σ = 14) it is 21 marks above. The z travelled; the standard deviations did the local work.
- Same gap, different z: a value 10 marks above the mean has z = 2.0 in a tight cohort (σ = 5) and z = 0.5 in a spread-out one (σ = 20). The gap did not change; its meaning did — which is the entire reason z exists.
- Units give it away: "the standard deviation is 8" is an incomplete sentence until "marks" (or cm, or ms) is attached. "The z score is 2" is complete as it stands. Anything that needs units is not a z score; anything unitless is not a standard deviation.
Practically: never report a z where a reader expects a spread, and never plug a z into a formula slot meant for σ (the raw-score reconstruction X = μ + zσ needs both, distinctly). The formula page names every symbol's units as it introduces it — including X — because this confusion is cheapest to prevent at the notation stage.
Also asked as: is z score the same as standard deviation · is a z score measured in standard deviations · z score vs standard deviation.
Related questions
Frequently asked questions
What is the standard deviation of z scores?
Exactly 1, always — standardising rescales the data so the new distribution (the standard normal) has mean 0 and standard deviation 1. That is a property of the transformed set, and a handy check: if your computed z scores do not average 0 with spread 1, an arithmetic error is present.
Can a standard deviation be negative, like some z scores?
No. A standard deviation is a root-mean-square distance and is zero or positive by construction. Negative values in this topic belong to z scores (below-mean positions) — one more behavioural difference between the unit and the count.