Toyday Six sigma method is known as an efficient and effective method of solving the problem.The most common methodology in six sigma is DMAIC that includes five steps: Defenition, Measurement, Analysis, Improvement and Control. One of the most important bases of six sigma methodology is the use a lot of statistical techniques and methods for data analysis and identification of the basical cause of defects. On the other hand, one of the prominences of this method comparing to the other improvement methods is the use of statistical methods so that by the use of statistical tools in analysis phase one can measure the effects of potential causes on critical factors. So far, many classifications of statistical techniques and statistical tests have been offered in the analysis phase of six sigma, Which are often divided into two parts: parametric and nonparametric. Of course, there are a lot of difficulties for using these techniques.
The aim of this research in the first sage was to recognize the inadequencies and lack of statistical tests in the analysis phase; then, to present a technique to remove the difficulties by using Multiple Attribute Decision Making in the operation research. The results of this research reveated three classes of major difficulties using the statistical tests:
1- Most of nonparametric statistical tests have less confidence and strenghth than parametric tests.Futher, they do not enjoy sufficient accurancy.
2- When the amount of the considerated data or the sample size is not large enough.
3- In some cases, no statistical tests(either parametric or nonparametric) is proportional to the conditions of six sigma research.
In this research, we reviewed the valid scientific sources available in the field of MADM (Multi Attribute Decisin Making) and the types of classification of these methods to remove defects. At the end, a new method
(algorithm) to apply MADM in the analyse phase of six sigma was developed.
The algorithm pays attention to the Decision Maker, MADM characteristics, the problem characteristics and chracteristics of the obtained solution. Applying of this
new algorithm, in the cases that statistical tests do not have sufficient strength and accurancy, can strongly increase the efficiency of six sigma methodology.