[Total: 0    Average: 0/5]
The idea of this technique is to compare the results by departing and then going up the elements in a very specific order.


The idea of this technique is to compare the results by departing and then going up the elements in a very specific order. This method will reduce the search for causes in 1 to 3 source components of variability.

The technique is applicable in the assembly operations where there are good and bad products. The output variable must be quantitatively measurable and reproducible. One must also be able to disassemble and reassemble the products without significant modification of the initial performance.

1-Determine the quantitative parameter of the defect

We identify the parameter that we use to measure the defect. It must be quantitative: a diameter, a depth, a temperature…

2-Select a good and a bad product

With regard to the chosen parameter, we select 2 Extreme components, the best of the best (BoB – Best of the best) and the worst of the worst (WoW-worst of the worst). These 2 components must sweep the whole of the defect variability.

3 – Disassemble and reassemble

In this exploratory phase, one:

  1. Measure our setting.
  2. Dismantle the two pieces.
  3. Pull up the two pieces.
  4. Measure our parameter again.

If the difference D between the good and bad exceeds the difference D (between the 2 measurements of the same component) of a ratio of 5, one concludes a significant and reproducible difference between the good and the wrong Part. We can continue the experiments without worrying about non-reproducibility.

Otherwise, it is most likely that the variability brought about by successive fixtures and Disassemblys may be the cause of misinterpretation. This technique will not be able to be used. We will turn to a multi-vari analysis or a paired comparison.

4-classifying components in an assumed order of influence

Based on judgement and knowledge at a priori, the components are classified in order of decreasing importance (of the most influential at least influential) of the perceived problem.

5 – Swap the fixtures

For each component, the fixtures are switches, starting with the most influential components. Then we measure the variable of the defect.

  • No change: The best is the best and the worst is the worst. Component A is therefore irrelevant.
  • Partial change: A is not the only important component. A could be a cause of the Pink X family.
  • Complete reversal: The best becomes the worst and the worst becomes the best. A is a priori the Red X.

6-Reassemble the parts

Some of the results are obtained, the parts are reassembled to return to the initial situation.

7 – Repeat steps 5 and 6

The tests are continued by continuing this permutation of components with the other components of the list and following the descending order defined in step 4.

If in step 5 we had a complete reversal of the situation, we identified the Red X. However, experiments will continue to confirm this hypothesis.

If at this point there was no specific change in the results, it is probably that a component is not the source of the variation, but the interaction between 2 components can be the source of the problem. We resume in step 4 to determine the components for which we suspect that the interaction is paramount. Then we will continue a cycle between steps 5 and 6.

8-Analyzing the results

Once all the tests are done, the results are analyzed and the cause or causes of the problem are determined.


We want to understand which component is failing in a cylinder, which speeds up the output speed. At first, we dismantle and assemble the 2 cylinders (the best and the worst), and we get for the 2 the same result as at the initial.

  • BoB: Output Speed of 1m/min
  • WoW: Output speed of 5M/min

We have in our case d = 5 and D = 0, we believe that we have a significant and reproducible difference.

We estimate that 3 components can cause the problem, in order of importance:

  1. The piston
  2. The seal
  3. The stem

We do a permutation of the different fixtures and we get the following table:



Assembly 1

Piston WoW

3.5 m/min

Piston BoB

1.5 m/min

Mounting 2

Seal WoW

1 m/min

Seal BoB

5 m/min

Assembly 3

Stem WoW

4.5 m/min

Stem BoB

2 m/min


In terms of results, we conclude that the seal is not at issue, the results being identical to the initial.

The other two components are the cause of the problem. The stem is the Red x and the piston the Pink x.


A. K. Bhote, K. R. Bhote (2000) – World class quality

S. Amster, K. L. Tsui (1993) – Counterexamples for the component search procedure

S. H. Steiner, R. J. MacKay (2005) – Statistical Engineering: an ALGORITM for reducing variation in manufacturing processes

K. S. Vinay, G. Praveena, H. Hamakrishna (2014) – Industrial scrap reduction using Shainin technique

Share This