Measuring and Modeling Relationships between Variables in Six Sigma

As a Six Sigma team moves into the Analyze stage of the DMAIC process, it looks more closely at the variables and variable interrelationships identified during the Measure stage. As part of the analysis, a scatter diagram of dependent and independent variables is drawn to visualize the form, strength, and direction of their relationships. By determining their correlation coefficient, a linear relationship can be quantified and identified as positive, negative, or neutral. Then, using regression analysis, a model is developed to describe the relationship as a linear equation and then used for predictions and estimations. However, it is essential to analyze the uncertainty in the estimate, to test that the relationship between variables is statistically significant, and that the model is valid. This course discusses two important tools – correlation and regression analysis for measuring and modeling relationships between variables. In terms of correlation, it takes learners through examples of scatter diagrams for two variables, the calculation and interpretation of the correlation coefficient, and the interpretation of its confidence interval. The course also draws learners’ attention to some key considerations in correlation analysis, such as correlation and causation. In terms of regression analysis, the course discusses the simple linear regression model, how to create it using sample data, interpret and use it, and conduct a hypothesis test to check that the relationship between the variables is statistically significant. Finally, the course looks into how residual analysis is used to test the validity of the regression model. This course is aligned with the ASQ Certified Six Sigma Black Belt certification exam and is designed to assist learners as part of their exam preparation. It builds on foundational knowledge that is taught in SkillSoft’s ASQ-aligned Green Belt curriculum.