Normal distribution & z-scores 2.0
About this app
Rebuild 2.0 (preview): bell curve morphing X → Z with the same area, tail and interval probabilities, inverse quantile question, sampling and a men/women comparison.
Standardizing in two steps
Every normal distribution can be reduced to the standard normal N(0, 1). The z-transformation Z = (X − μ) / σ has two steps: subtracting the mean shifts the bell, dividing by σ squeezes or stretches it. Width and height change, but every area stays the same: P(X > x) = P(Z > z) = 1 − Φ(z), where Φ is the cumulative distribution function of the standard normal. The height of the curve is a density, not a probability, and it can exceed 1.
What the app shows
An “X → Z” slider animates the standardization and labels each intermediate step. You choose the event: left, right, between a and b, or outside. The app writes out the calculation with Φ. A quantile panel answers the reverse question, such as the height needed to be in the top 1 %, via x = μ + z·σ with z = Φ⁻¹(p). A sample panel draws 50, 500 or 5,000 values as a histogram. The height presets are based on NHANES data for men and women.
What to look for
Pick “Men” and overlay women for comparison: the same height has a different z-score in each group, because z measures position relative to its own group. Set the threshold to 2.29 m. The model returns a tiny probability even though people of that height exist; the app flags the limits of the normal model far out in the tail. With small samples, compare the observed share with the model probability.
Frequently asked questions
- What does it mean to standardize a normal distribution?
- Standardizing converts each value x into z = (x − μ) / σ. This turns N(μ, σ) into the standard normal N(0, 1). Probabilities are unchanged, so a single table or function Φ covers every normal distribution.
- How do you find a quantile of a normal distribution?
- First find the standard normal quantile z = Φ⁻¹(p), then convert back with x = μ + z·σ. For the top 1 %, z = Φ⁻¹(0.99) ≈ 2.326. With μ = 175 cm and σ = 7 cm the cut-off is about 191 cm.
- Can a probability density be greater than 1?
- Yes. The density f(x) is not a probability but a height measured “per unit”. For a very narrow distribution the curve has to be tall so that the area still equals 1. Probabilities are always areas under the density.
- What is new compared with the first version of the app?
- This version animates standardization from X to Z with the area preserved. It adds interval and outside probabilities, the reverse quantile question, simulated samples as a histogram and a comparison of male and female body heights.
Subject: Data & decisions
More from Data & decisions
- Anchoring effect
- Benford's law
- Berkson's paradox
- Cherry-picking: base period
- Cherry-picking: base period 2.0
- The three-cowboy duel (truel)
- The three-cowboy duel 2.0
- The three means
A VisuApp by heyprof: interactive, free in your browser, no sign-up. What is a VisuApp? · All apps