A successful DMAIC project needs powerful tools and techniques. Below are the most commonly used methods in the Six Sigma improvement effort.
1. Statistical Process Control and Control Charts identify problems or opportunities.
2. Tests of Statistical Significance (Chi-Square, t-tests, and ANOVA) define problems and analyze their root causes.
3. Correlation and Regression analyze root cause predict results.
4. Design Experiments analyze optimal solutions and validate results.
5. Failure Modes and Effects Analysis prioritize problems and even prevent them.
6. Mistake-Proofing prevent defect and improve process.
7. Quality Function Deployment designs product, service, and process.
Statistical Process Control (SPC) measures and evaluates variation in a process. It also controls such variation or performance. It is an ideal way of monitoring current process performance, predicting future performance, and suggesting the need for corrective action. 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 lim its too often and assume perfect data.
Tests of Statistical Significance, such as 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, which is any variation, change, or difference observed in a population or process is due purely to chance.
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.
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.
Design of Experiments (DOE) test and optimize 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.
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.
Mistake-Proofing emphasizes 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.
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.”
The Six Sigma Way: How GE, Motorola, and Other Top Companies are Honing Their Performance
