There are infinitely many normal curves — one for every pair of μ and σ — and nobody wants a table for each. So we stop measuring in beats, millimeters, or grams and start measuring in standard deviations from the mean. Subtract μ, divide by σ, and every normal curve in the world collapses onto one.
Resting heart rate → z
the area you selectedsimulated sampleμ (and z = 0)μ ± kσ (and z = ±k)the marked value
Reading the curve backwards
Standardizing runs in reverse too, and that is where its value shows. Name a percentile: the standard normal on its own hands you the z that cuts it off, and then x = μ + zσ turns that single z into a cutoff on any scale you please. One curve, one lookup, every distribution.
90.0
The z that cuts it off—
Area left above it—
That one z, translated onto four different scales
Pick a valuedrag the sliders
The variable
Its two parameters, and one observation
Raw value x—
Standardized z—
Percentile % below x—
Raw gap x − μ—
Which area do you want?
The recipe, and its reverse. Standardizing is z = (x − μ) ÷ σ — subtract the center, then divide by the spread. Going back the other way is x = μ + zσ. Standardizing any set of values, normal or not, always yields a mean of exactly 0 and an SD of exactly 1. But when the values were normal to begin with, something stronger happens: the result is not merely centered and rescaled, it is N(0, 1) itself — the standard normal. That is the curve every table, every calculator key, and every software routine is built around, and it is why one of them is enough.
Three things standardizing does not do. It does not make data normal: subtracting and dividing by constants is a linear rescaling, so skew, outliers, and extra peaks all survive untouched. The percentages on this page come from the normal model — the z-score itself is only arithmetic and assumes nothing, but reading a percentile off it requires that the values really do follow this curve. And a z-score is not a grade: negative is not "bad" and positive is not "good," since which direction counts as trouble depends entirely on the variable.
Does a real sample actually standardize?
Each draw generates observations from the current normal distribution, then converts every one of them with the same formula. The gold bars appear on both panels — the identical pile of numbers, read off two different rulers.