Top Six Sigma Tools to Use for Big Results: Part 2

In Part 1 of this series, I started reviewing the seven basic quality tools everyone should have in his or her toolbox. You should memorize these Six Sigma tools as shown below:

1. Check sheet
2. Cause and effect diagram
3. Control chart
4. Pareto chart
5. Scatter diagram
6. Stratification
7. Histogram

Having already reviewed the first four Six Sigma tools on this list, I will now discuss the remaining three.

Scatter Diagram

A scatter diagram explores relationships, specifically, to determine if there is a correlation between two variables. If a correlation exists between the variables, when one changes, the other one changes as well. We can use the correlation to predict behavior, especially if one variable is easy to measure and one is not. As an example, if we prove that weight gain in the first trimester of pregnancy correlates well with fetus development, we can use weight gain as a predictor. The alternative to using the scatter diagram in this case would be expensive tests to monitor the actual development of the fetus.¹

To determine if a correlation exists, look at the direction of the points on the scatter diagram. In the diagram shown here, there is a positive relationship because the points are moving in the same direction. As one variable increases, the other increases as well. With a positive relationship, the points move from the lower left corner to the upper right corner.

There can also be a negative relationship. In this chart, the data points move from the upper left to the lower right, demonstrating that as one variable increases, the other decreases.

In addition to the direction of the points, you also need to review the "form." (Translation: what type of pattern do the dots create?) In these examples, the data follows a straight line, suggesting a linear relationship. With linear relationships, we can use correlation and regression to evaluate the data.

Use scatter diagrams when you are trying to determine objectively whether a particular cause and effect are related. Right now, I’m fairly sure there is a correlation between whether you keep reading and whether I move on to the next topic. So, let’s look at stratification.

[eventPDF]

Stratification

Stratification is our friend. Simply put, stratification is the separation of data into categories. It is most frequently used to identity which categories contribute the most to a problem.² You can also use it to impress your friends and coworkers in, say, a discussion at the water cooler. "I just talked to the boss about stratification. What do you think?" Smile and nod as they answer.

Use stratification when data come from several sources or conditions, and when separating the sources or conditions helps to analyze the data.³

Here is a list of common areas for stratification:

• Shifts
• Operators
• Days of the week
• Suppliers
• SKU’s
• Departments
• Error codes
• Defect types

The list could be exhaustive. The best way to stratify is to take the data you have and create a Pareto chart (discussed in Part 1). Then do a Pareto on the Pareto, e.g. find the largest areas of impact and then stratify the data within the top categories.

Histogram

The last of the seven basic quality tools is the histogram. A histogram is a graph used to understand the shape and spread of continuous data. I always (really, I mean always) look at the shape and spread of my continuous data before doing anything else. (Most quality practitioners refer to the histogram as a "frequency distribution.") The shape and spread of the data give good clues about the problem you are trying to solve.

Here are examples of some of the most common patterns:



A bell-shaped curve is one of the most common patterns. (It is also called a "normal distribution.") This shape is natural and is expected as the output of many processes.

An isolated peak shape indicates there may be inefficient processes that are making the data cluster around a certain value.

A double peak (like two camel humps) indicates there are two distinct processes.

Plateau shaped patterns show multiple processes happening within your data set.

Truncated shapes (shapes where both sides appear to have been removed) show that removal or inspection is taking place.⁴

Take a close look at the shape and spread of your data. You will learn a lot in a few short minutes using a histogram.

Wrapping Up the Seven Basic Six Sigma Tools

Well, that’s it for the seven basic tools. My advice is that the simplest Six Sigma tools often are not only the most appropriate, but also often provide the biggest returns. The tool of choice for me, 99 times out of 100, is one of these seven basic quality tools.

References

Example from www.qualityamerica.com, copyright 1995-2008.
Juran Institute, Root Cause Corrective Action Workshop, copyright 2005.
The Quality Toolbox Second Edition, Nancy R. Tague, copyright 2005.
Juran Institute, Quality Improvement Pocket Guide, copyright 1993, 2003.