SUMMARY
A successful DMAIC project needs powerful tools and techniques. Below are the most commonly used methods in the Six Sigma improvement effort.
Statistical Process Control (SPC) involves the measurement and evaluation of variation in a process. It also takes care of the efforts to “control” such variation, or performance. It is an ideal way of monitoring current process performance, predicting future performance, and suggesting the need for corrective action.
The control charts are easy to understand thus can be a very powerful communication tool. In the Six Sigma system, they help teams identify the type and frequency of problems and can even suggest effective corrective action (Measure activities). They also help track results (Piloting activities) and serve as an alarm system, alerting the observer to unusual activities in the process (Response Plan activities). From these activities, you can understand that “Control” means keeping a process operating within a predictable range of variation to maintain the stable, consistently good performance of a process.
If you are using the SPC and Control Charts, you should gather, plot, and review data promptly, choose and prioritize measures carefully, and set or fine-tune your alarms. You need not recalculate control limits too often and assume perfect data.
Tests of Statistical Significance (Chi-Square, t-tests, and ANOVA) look for patterns or tests suspicious data. They confirm, check the validity, and determine the type of distribution among others.
The basics of Statistical Analysis is the Null Hypothesis, that is any variation, change, or difference observed in a population or process is due purely to chance.
You usually use the Chi-Square technique when you have discrete data such as compare defect rates in two locations or check week-to-week changes in consumer product choices.
The t-test is for testing significance when you have two groups or samples of continuous data, e.g. comparing cycle time for a key step or examining customer income levels in two regions.
ANOVA is another test of significance for continuous data. Unlike the t-test, ANOVA can compare more than two groups or samples, e.g. examine customer income levels in four regions.
Some of the important considerations when using tests of statistical significance are that you make sure the data being used is valid and you should perform the right test.
Correlation and Regression analyze the relationships among two or more factors. When you say that two factors are “correlated,” you mean that a change in one will be accompanied by a change in the other. Through statistical calculations you can measure the strength of a possible relationship and will be able draw a number of helpful conclusions.
You use this method only when you have data for tow or more factors that are matched on individual items. The following are the types of Correlation and Regression Analysis:
Design of Experiments (DOE) is for testing and optimizing the performance of a process, product, service, or solution. It gives you the opportunity to plan and control the variables using an experiment, as opposed to just gathering data and observing real-world events known as “empirical observation.”
Some of the advantages of DOE to Six Sigma are that it assesses the VOC systems, the factors that isolate the vital root cause, pilot possible solutions, and evaluate product or service designs.
The basic steps in DOE are:
Failure Modes and Effects Analysis (FMEA) – identifies and prioritizes potential problems (failures) not only in work processes and improvements but also in data-collection activities.
The following are the steps and key concepts:
Two examples of a problem in an on-line catalog
Mistake-Proofing can be thought of as an extension of FMEA, but Mistake Proofing emphasized the detection and correction of mistakes before they become defects delivered to customers. Also known by the Japanese term Poka Yoke, it pays careful attention to every activity in the process. It involves constant, instantaneous feedback.
The basic steps are:
Quality Function Deployment (QFD) prioritizes and translates customer inputs into designs and specifications for a product, service, and/or process. The basics of the QFD involve a special multidimensional matrix called the “House of Quality”. It involves the following:
COMMENTARY
Six Sigma is simply about variation and its control. Its measurement or evaluation is what statistical process control or SPC is all about. Others call it Statistical Quality Control (SQC). The tools employed here is likewise SPC or SQC tools. Without these tools, you cannot go Six Sigma.
The title of the chapter (“Advanced Six Sigma Tools”) however seems to be a misnomer. In truth the so-called “power tools” are a mixture of basic and advanced statistical process control tools used many years ago when the Quality movement circles made their mark in quality improvements. These were the very tools used in problem solving, prioritization, correlations, and design of experiments. You can find these tools in books written by Deming, Juran, and other quality gurus. In fact, control charts were formulated by Shewart way back in the 1920s yet. Let it not be said therefore that these tools were discovered during the Six Sigma era.
As a footnote, as correctly pointed out here, these are improvement tools or techniques, which the entire organization should be familiar with. This makes SPC training a must.
