Measures of Dispersion

The Core Idea: Why Does Spread Matter?

Two groups can share the same mean yet have values distributed in very different ways. In one group individuals may cluster tightly around the average, while in another extreme values may be common. This distinction carries critical meaning in contexts ranging from treatment effects to manufacturing quality. Without measures of dispersion, measures of central tendency are misleading; two distributions with identical means but markedly different standard deviations convey fundamentally different information to the researcher. For this reason, centre and spread should always be reported together.

Key Dispersion Statistics and How They Are Computed

Range (R = maximum − minimum) is easy to compute but highly sensitive to outliers because it depends on only two extreme values. The interquartile range (IQR = Q3 − Q1) measures the spread of the middle fifty percent and is resistant to outliers. Variance is the mean of squared deviations from the mean: s² = Σ(xi − x̄)² / (n − 1). Standard deviation returns the result to the original unit by taking the square root: SD = √s². The coefficient of variation expresses SD as a proportion of the mean (CV = SD / x̄), enabling comparisons of spread across variables measured on different scales.

Common Misuses and Misconceptions

The most frequent error is confusing standard deviation with standard error. Standard deviation describes the spread of individuals, whereas standard error expresses the sampling variability of the sample mean: SE = SD / √n. Reporting standard error in small samples to create an impression of precision while implying spread is misleading. Another common mistake is neglecting to use n − 1 in the denominator when computing sample variance; this correction ensures the sample variance is an unbiased estimator of the population variance. Finally, reporting only the range is inadequate because it conveys no information about the shape or internal structure of the distribution.

Importance in Research Practice

Dispersion measures are also a prerequisite for the trustworthiness of statistical tests. Analyses such as the t-test and ANOVA rely on variance estimates; in outlier-rich or skewed distributions, standard deviation can give a misleading impression of stability, and in such cases IQR or the median absolute deviation (MAD) should be preferred. Meta-analyses require standard deviations to compute effect sizes; failure to report SD clearly in publications therefore hampers meta-analytic synthesis. APA publication standards require that standard deviation always be presented alongside the mean; omitting it is a sign of methodological weakness.

Sources