Summary of the six-sigma ANALYZE stage

In the Analyze phase, the measurements collected in the Measure phase are analyzed so that hypotheses about the root causes of variations in the measurements can be generated and the hypothesis subsequently validated. It is at this stage that practical business problems are turned into statistical problems and analyzed as statistical problems. This includes:

• generate hypotheses about possible root causes of variation and potential critical Inputs (X’s);
• identify the vital few root causes and critical inputs that have the most significant impact; and
• validate these hypotheses by performing Multivariate analysis.

Key questions:

• Which Inputs actually affect our CTQ’s most (based on actual data)?
• By how much?
• Do combinations of variables affect outputs?
• If an input is changed, does the output really change in the desired way?
• How many observations are required to draw conclusions?
• What is the level of confidence?

The Analyze phase offers specific statistical methods and tools to isolate the key factors that are critical for a comprehensive understanding of the causes of defects:

• Five Why’s - use this tool to understand the root causes of defects in a process or product, and to penetrate through incorrect assumptions about causes.
• Tests for normality (Descriptive Statistics, Histograms) – this is used to determine if the collected data is normal or abnormal so as to be properly analyzed by other tools.
• Correlation/Regression Analysis - to identify the relationship between process inputs and outputs or the correlation between two different sets of variables.
• Analysis of Variances (ANOVA) - this is an inferential statistical technique designed to test for significance of the differences among two or more sample means.
• FMEA (Failure Mode and Effect Analysis) - applying this tool on current processes enables identification of sufficient improvement actions to prevent defects from occurring.
• Hypothesis testing methods - these are series of tests in order to identify sources of variability using historical or current data and to provide objective solutions to questions which are traditionally answered subjectively.