Percentile rank formulas
Count B, the values below your value, and E, the values equal to it. N is the total number of values.
Strictly below = B ÷ N × 100 · at or below = (B + E) ÷ N × 100
The midpoint version splits ties in half, so a value in the middle of the data ranks near 50. It's the one used most often in education and testing, and it's the main answer here. The other two are shown because some textbooks and software use them. With no ties they differ by 100 ÷ N.
Worked example
Ten test scores: 55, 61, 67, 70, 70, 72, 78, 81, 85, 90. For a score of 70, three scores are below it (B = 3) and two equal it (E = 2):
So 70 has a percentile rank of 40. By the other definitions it's 30 (strictly below) or 50 (at or below). A score of 80 isn't in the list, but 7 scores are below it, so it ranks 70.
Percentile rank vs. percentile
They go in opposite directions. A percentile rank starts with a value and tells you what percent of the data is below it. A percentile starts with a percent and tells you the value: "the 90th percentile is 85.5". To find percentiles and quartiles, use the statistics calculator.
Limitations
- With a small data set, each value moves the rank a lot. With 10 values, the rank jumps in steps of 5 or 10.
- Official test percentiles, like SAT or ACT, compare you with a large national group, not the scores you type in.
- The calculator gives the exact rank within your data. It doesn't assume a bell curve. If your data is roughly normal, the normal distribution calculator can estimate a percentile from the mean and SD instead.