The Pareto chart shows the absolute values of the standardized effects from the largest effect to the smallest effect. The standardized effects are t-statistics that test the null hypothesis that the effect is 0. The chart also plots a reference line to indicate which effects are statistically significant.
The reference line for statistical significance depends on the significance level (denoted by α or alpha). Unless you use a stepwise selection method, the significance level is 1 minus the confidence level for the analysis. For more information on how to change the confidence level, go to Specify the options for Analyze Definitive Screening Design. If you use backwards selection or stepwise selection, the significance level is the significance level where Minitab removes a term from the model, known as Alpha to remove. If you use forward selection, the significance level is the significance level where Minitab adds a term to the model, known as Alpha to enter.
Use the Pareto chart to determine the magnitude and the importance of the effects. On the Pareto chart, bars that cross the reference line are statistically significant.
For example, in this Pareto chart, the bars that represent factors A, B, and C cross the reference line. These factors are statistically significant at the 0.05 level with the current model terms.
Because the Pareto chart displays the absolute value of the effects, you can determine which effects are large but you cannot determine which effects increase or decrease the response. Use the normal probability plot of the standardized effects to examine the magnitude and direction of the effects on one plot.
In the first pareto chart, all of the bars are blue because all of the terms are in the model. You can choose to show terms that are not in the model. For more information, go to Select the graphs to display for Analyze Definitive Screening Design. The terms that are not in the model are gray.
For example, in this pareto chart, the blue bars represent the terms that are in the model. The analyst included the terms that are significant at the 0.05 level (A, D, CE, and GG). The analyst also included the terms that make the model hierarchical (C, G ,and E).