Percentile Rank Versus Percent Score

The words percentage and percentile sound alike, but answer different questions. A percent score asks, 'What fraction of the available points did this person earn?' A percentile asks, 'Where does this value stand relative to the values in a comparison group?' Confusing them can turn a correct reading of a chart into a false claim about performance.

Percent score measures points earned

A percent score is calculated as points earned divided by points possible, multiplied by 100%. If a student earns 42 of 50 points, the score is 42/50 × 100% = 84%. It describes performance against the assessment's possible points. It does not by itself say how the student performed compared with classmates; that depends on the distribution of their scores.

Percentile rank measures relative position

A percentile rank describes the location of a value after observations are arranged from smallest to largest. In a common interpretation, a score at the 90th percentile is at least as high as about 90% of scores in the reference group, with about 10% at or above it depending on the convention and ties. The precise rule can vary across tests and statistical procedures, so follow the definition stated in the problem or report.

For example, if a reading score is at the 75th percentile in a stated comparison group, the score is higher than or equal to about three-quarters of that group's scores under the report's method. It does not say that 75% of the test questions were answered correctly. A candidate could have a 75% percent score and land at a different percentile depending on how others performed.

Example: same percent, different percentile

Imagine two classes take different quizzes. In Class A, most scores cluster near 90%, so an 85% score might fall below many classmates. In Class B, most scores are near 70%, so the same 85% may rank high. The percent score is 85% in both settings; the percentile rank depends on the comparison group. Never infer a percentile from the percent score alone.

Read the comparison group

A percentile is incomplete without its reference group. A child's height percentile might compare children of the same age and sex in a specified population. A test percentile might use students in the same grade, applicants in one testing window, or a norming sample. Changing the reference group can change the percentile even when the underlying score remains the same.

Look for the population and time period in the wording: 'among these students,' 'relative to the national norm group,' or 'in this sample.' If a chart reports the 60th percentile for a sample, do not generalize it to every person unless the sampling method supports that conclusion. Percentiles describe rank within the defined data, not a cause, a level of mastery, or a universal judgment.

Percentiles and ordered data

To find a percentile from a small data set, first sort the values from least to greatest. The 50th percentile is associated with the median, and the first and third quartiles are commonly associated with the 25th and 75th percentiles. But conventions for locating percentile positions can differ, especially for small samples and when a position falls between two observations. Some methods interpolate; others use a nearest rank or another rule.

If an exam question provides a formula or a specified method, use it consistently. One textbook method locates the kth percentile at position k/100 × (n + 1), where n is the number of ordered observations; if the position is not a whole number, the method explains how to interpolate. Other contexts use another convention. State the method when explaining a calculation, and do not present one convention as universal.

Ties affect the interpretation

Several observations can have the same value. A value tied with others may have a percentile rank that depends on whether the method counts values below it, values at or below it, or averages ranks for ties. This is another reason to follow the convention given by the question. When simply interpreting a report, use its stated definition rather than trying to recalculate a rank using an assumed rule.

Percentile is not percent change

A percentile rank is also different from percent change. If a value rises from 80 to 100, the increase is (100 − 80)/80 × 100% = 25%. A 90th-percentile rank does not mean a 90% increase or a value that is 90% higher. Each term has its own base and purpose: percent score compares points to possible points, percentile compares a value's rank with a distribution, and percent change compares a difference with a starting value.

Common reading errors

The most common mistake is reading '90th percentile' as 'scored 90%.' Another is saying a student scored better than exactly 90% of all people, when the percentile may be based on a particular sample and method. It is also incorrect to treat percentile distances as equal score distances: the gap between the 50th and 60th percentile need not represent the same number of points as the gap between the 90th and 100th.

A percentile tells position, not spread. Two scores may have very different raw-score gaps but similar percentile ranks in one part of a distribution; conversely, a small raw-score change in a dense cluster can move a rank noticeably. To understand how scores are distributed, also consider a histogram, box plot, or other summary rather than relying on one percentile.

A quick interpretation routine

1. Ask whether the number is a percent correct or a percentile rank.
2. Identify the reference group and which direction represents higher values.
3. Read the report's convention for ties or percentile calculation if a precise rank is needed.
4. Describe the percentile as relative standing, not as points earned or a measure of cause.
5. Use a separate calculation for percent change or percent score when that is what the question asks.

Exam takeaway

A percent score measures earned points out of possible points. A percentile rank locates a value within a comparison group. Check the group and the method, and never treat an nth percentile as n percent correct.