The DMAIC Analyze Phase: Root Causes in Six Sigma, ANOVA, Fishbone DIagrams

The DMAIC Analyze Phase: Root Causes in Six Sigma, ANOVA, Fishbone DIagrams
Page content

Objectives of Analyze Phase

The goal of the DMAIC Analyze phase is to identify potential root causes for the process problem being addressed and then confirm actual root causes with data.

Having completed the Measure phase, the project team should have already established a clear problem statement which specifies what the problem is and under what circumstances it occurs. They should have already gathered and analyzed data to establish the baseline performance of the process, relative to the Critical To Quality measures (CTQs) established based on customer input.

The question that the Analyze phase seeks to answer is “Why is this problem occurring?” Another way to ask it is, “What is the cause of the problem?” It is not possible to make improvements to the process until the causal factors are identified.

Potential Root Causes

In many cases, clues to the factors affecting performance are already available based on the work that was done in Define and Measure. Perhaps the team demonstrated that the problem is isolated to one group, and they know that group is using older equipment. Or analysis of the process map may have revealed some fairly obvious sources of inefficiency and delay in the process.

However, this is not sufficient to confirm what is causing the problem for two reasons. One is that, as in all phases of DMAIC, suspicions and hypotheses must be confirmed with data. Not only must the team confirm that these factors are present, they must also confirm that changes in these factors substantially impact the outcome. The other is that the goal of Analyze is to determine root causes, which requires digging deeper than what is apparent on the surface.

Several techniques are employed by Six Sigma project teams to identify potential root causes. One is brainstorming, which is used by team members and, ideally, people involved in performing the process under study, to create a large list of factors which could reasonably affect performance. This list will of course include any factors that were revealed based on the process mapping exercise and the data analysis conducted during Measure.

Another popular exercise is the 5 Whys, which involves repeatedly asking “Why?” until it no longer makes sense to do so. The point is to get past the surface-level answers that are likely to be put forth initially, and to uncover the real underlying issues.

Cause and Effect Diagrams

Once a list of potential root causes has been compiled, the next step is to organize them in a way that makes it easier to prioritize and assess them. Several tools can be used to accomplish this.

The most popular is a fishbone diagram or Ishikawa diagram, which uses a display resembling the bones of a fish to categorize potential causes and illustrate the levels of causation. The main bones are used to reflect high-level categories, such as People, Processes, Technology, and Policies. An example of a fishbone diagram can be downloaded in our Media Gallery.

Another option is a tree diagram which can be used to organize the same information. It is preferable if the amount of information is large and hard to organize in a fishbone, or if the project team wants to create it more quickly and does not have access to a tool for creating a fishbone automatically. And some people just prefer the more straightforward presentation.

Once the diagram has been created, the project team reviews it to determine which seem to be the most likely potential root causes, and to identify any that seem to have consequences in more than one area. These are good candidates for the validation process.

Confirming Root Causes

In some cases, sufficient data is available from the Measure phase to conduct cause-effect analyses during Analyze. Often, however, it is necessary to collect new data so that the relationship between the suspected root causes and the effect under study can be evaluated.

As in the Measure phase, the methods used to analyze the data depend on the type of data collected. Graphical techniques include scatterplots and frequency plots, while statistical techniques include Analysis Of Variance (ANOVA), correlation analysis, and chi-square testing. In Six Sigma, ANOVA is often used when the output measure is continuous (time and money are typical examples) and the input measures or suspected causes are discrete (categorical). Correlation analysis is used when both measures are continuous, chi-square when both are discrete.

Wrapping Up the Analyze Phase

At the end of the Analyze phase, the project team should have at least one confirmed hypothesis regarding the root causes of the problem the project aims to resolve. Once the root cause is known, action can be taken in the Improve phase to counter it.

This post is part of the series: The Six Sigma DMAIC Process

This series provides an introduction to the Six Sigma DMAIC process, and a more in-depth look at each of the individual phases. You’ll learn about the objectives of each phase and common tools to help you reach them.

  1. DMAIC Phase 1: Define
  2. DMAIC Phase 2: Measure
  3. DMAIC Phase 3: Analyze
  4. DMAIC Phase 4: Improve
  5. DMAIC Phase 5: Control