J.M. Chambers, W.S. Cleveland, B. Kleiner and P.A. Tukey, 1983
"There is no excuse for failing to plot and look." J.W. Tukey, 1977
The importance of plotting can never be overemphasized in statistics. There are recommended plots for virtually every statistical method, for example, scatter plots, residual plots and diagnostic plots for regression. Plots are also important in the areas of goodness of fit and fitting distributions; they provide a sense of pattern and a level of detail not available in a single test statistic. Although formal testing procedures allow an objective judgment of normality (i.e., significance vs non-significance at some a level), they do not generally signal the reason for rejecting a null hypothesis, nor do they have the ability to compensate for masking effects within the data which may cause acceptance of a null hypothesis. Therefore, we begin our quest for normality by suggesting basic plotting procedures for the display of a single sample of data.