Sign Test for Median of Single Population

Use the sign test when data does not follow a known distribution, sample sizes are small, or parametric test assumptions cannot be met. The sign test serves as a nonparametric method to test hypotheses about the median of a single population without requiring normality assumptions. It provides a robust alternative for skewed distributions or when traditional statistical methods are inappropriate.

Data points greater than the hypothesized median receive positive signs, while those smaller than the median receive negative signs. Data points that exactly match the median are ignored because they do not contribute evidence toward either hypothesis. The test then evaluates the balance between positive and negative signs to determine if the sample contradicts the null hypothesis.

The null hypothesis typically states that the population median equals a specific value. The alternative hypothesis could be that the median is not equal to, less than, or greater than the tested value, depending on the research question. In the example with student test scores, the null hypothesis posits that the median equals 45.

When the total number of positive and negative signs exceeds 25, the test statistic is calculated using the z statistic rather than exact binomial methods. For a one-tailed test at a 0.05 confidence level, the critical z-value is -1.645. If the test statistic falls below the critical value, you reject the null hypothesis.

A significant imbalance between positive and negative signs suggests the sample data contradicts the null hypothesis, indicating an alternative median value for the population. If most scores fall below the hypothesized median, the abundance of negative signs provides evidence that the true median differs from the proposed value.

The single-population sign test assesses the central tendency within one group relative to a fixed value, focusing solely on whether a sample median deviates from a standard. Unlike sign tests for matched pairs or paired samples used in before-and-after studies, it does not compare two groups or conditions, making it ideal for evaluating a single sample against a hypothesized median.

When the test statistic falls below the critical value, you reject the null hypothesis. This rejection suggests that the dataset's median significantly differs from the hypothesized value. In the student test score example, rejecting the null hypothesis indicates the median score is less than 45, supporting the alternative hypothesis.