Normal Probability Plots

A Normal probability plot provides a simple way to tell whether the numbers in a variable are approximately normally distributed.

Data Desk sorts the values in the variable and then, starting with the smallest value, poses the question: “If this were a sample from the standard normal distribution (that is, the normal distribution with zero mean and unit variance), what would I expect the smallest value to be?” The answer is the first normal score. It depends only on the number of cases in the variable and can be estimated with the NScores function available in derived variables. The probability plot graphs the observed smallest value against the value expected under the assumption of a normal distribution.

The question is repeated for the second smallest value: “If this were a sample from the standard normal distribution what would I expect the second smallest value to be?” The observed second smallest value is plotted against the expected second smallest value.

To make a normal probability plot, select the variable to plot and choose Plot > Normal Prob Plots. Normal probability plots are also available from HyperView menus in several procedures in which the assumption of normality is important.