Speaking Stata: Graphing Agreement and Disagreement

Abstract

Many statistical problems involve comparison and, in particular, the assessment of agreement or disagreement between data measured on identical scales. Some commonly used plots are often ineffective in assessing the fine structure of such data, especially scatterplots of highly correlated variables and plots of values measured “before” and “after” using tilted line segments. Valuable alternatives are available using horizontal reference patterns, changes plotted as parallel lines, and parallel coordinates plots. The quantities of interest (usually differences on some scale) should be shown as directly as possible, and the responses of given individuals should be identified as easily as possible.