Chapter 3: Measures of Variation

The chapter introduces multiple approaches to quantifying how individual observations deviate from the center of a distribution. The range, defined as the difference between maximum and minimum values, offers computational simplicity but ignores the bulk of the data and remains vulnerable to extreme outliers. Quartiles and percentiles divide ordered datasets into meaningful sections, with the interquartile range measuring the spread of the middle fifty percent of observations and demonstrating superior robustness against anomalous values. Box-and-whisker plots provide visual summaries of five-number data descriptions while enabling quick assessment of distributional symmetry and comparative analysis between datasets. Variance and standard deviation represent more sophisticated dispersion measures by incorporating all data points in their calculations through squared deviations from the arithmetic mean. Because variance produces squared units, standard deviation restores measurements to their original scale while maintaining mathematical tractability. The chapter addresses computational efficiency through computational formulas and emphasizes the critical importance of using exact, unrounded means during calculations. When combining multiple datasets, practitioners must aggregate the underlying sums rather than averaging individual variance values. Finally, the chapter establishes how mathematical transformations affect variation in predictable ways: additive operations preserve spread while multiplicative operations scale variance by the square of the multiplier, providing tools for working with coded or standardized data across different contexts and disciplines.