How to use (3 steps)
- Pick a mode: from mean & SD, from a score list / frequency table, or reverse from a target percentile/deviation score.
- Enter your score or paste the data. Choose the SD type (population/sample) and tie handling if you use empirical data.
- Results update automatically: deviation score, percentile, top %, and estimated rank when a cohort size is provided. Copy the results or URL to share.
Numbers are processed in your browser only. Decimals and negative values are accepted.
Quick presets
Inputs
Results auto-update as you type. Percentile is based on a normal approximation in mode A/C and empirical counts in mode B.
Deviation score & percentile
Percentile is shown as score ≤ x. Top % = 100 − percentile.
How it’s calculated
- z-score: z = (score − mean) / SD. Deviation score (T-score / hensachi) = 50 + 10 × z.
- Percentile (normal approximation): Φ(z) × 100. Φ uses erf-based approximation with |error| ≤ 1e-6.
- Empirical percentile (data mode): choose ties handling — min (below / N), midrank ((below + 0.5×ties)/N), max ((below + ties)/N).
- Target → score: percentile uses the inverse standard normal CDF, searched in z ∈ [−10, 10]; 0 and 100 are clipped away.
- Estimated rank (optional): floor((1 − percentile/100) × N) + 1, clipped to 1..N.
FAQ
Should I use population or sample standard deviation?
Use population SD when you have the whole cohort. Use sample SD when your list is a sample from a larger group; at least two data points are required for sample SD.
How are ties handled in the percentile?
Choose min (strictly below), midrank (below + half of ties), or max (at or below). Midrank is common and shown by default.
Why can’t I enter 0% or 100% as a target percentile?
0% and 100% would require an infinite z-score in a normal distribution. Use a value slightly above 0 or below 100 (e.g., 0.1% or 99.9%).
Is my score list sent to a server?
No. Calculations run in your browser. Use the Copy URL button only when you want to share the current inputs.
What should I enter first for deviation score or percentile?
Enter the score, mean, and standard deviation first. Confirm whether the distribution assumption is appropriate before interpreting percentile or standardized score outputs.
Comments
Load comments only when you need them.