The sign test is a simple nonparametric test used to evaluate the median of a single sample or to compare two related samples. It looks only at the direction of differences (positive or negative), not their magnitude, making it useful when assumptions of normality are not met. It is often used as a paired test when comparing “before and after” measurements by analyzing the single sample of differences, rather than the two sets of values separately.
The main advantage of the sign test is that it is very easy to use and highly flexible, working even when the data are ordinal or highly skewed. However, it is generally less powerful than other nonparametric median tests, such as the Wilcoxon signed-rank test, because it ignores the size of the differences. Researchers might prefer the sign test when sample sizes are very small, when data are measured only in terms of direction (better/worse), or when the magnitude of changes is unreliable or irrelevant.
For example, a researcher comparing before-and-after test scores for students might use a sign test to see if there is a consistent improvement without assuming that the size of the score change follows a particular distribution.
« Back to Glossary Index