# Pareto

Viewing 6 posts - 1 through 6 (of 6 total)
• Author
Posts
• #41747

marie
Participant

I have a Pareto chart of data that shows reasons or root causes why my defect is being created.  What Hypothesis test should I do to prove the root causes are significant?

0
#131255

Anirvan Sen
Participant

When you trying to do establish a correlation, you can run a correlation or a regression test and use the p-value or the r-square values (in correlation, you can use the Pearson’s correlation) to see if indeed there is a strong relation or not.
Cheers,
A.

0
#131272

Clancy
Participant

What type of data do you have?  Attribute… Variable…?

0
#131273

Jeff Korfhage
Participant

If you have identified the root cause from your Pareto, you should implement some type of corrective action to prevent or reduce this cause from occurring. After this action has been implemented, start collecting data on the occurence of the defect. Then you could run a hypothesis test on the before and after data to see if there is a significant difference between the two data sets. This would validate your corrective action as the reason for the improvement of your defect rate.

0
#131920

R.M.Parkhi
Participant

Dear Marie,
I suggest you should carry out multi-vari study of the process to identify the variation due to-
1.within component,
2.component to component,
3. time related i.e. temporal
Attack the the first two of the above. Mostly, you should be in position to analyse & fix the problem with engg. analysis.
This is well explained in the book ” World Class Quality & How to Make It Happen ” by Mr. Keki R. Bhote: Publishers -American Management Association.
Pl. feel to ask if you have a query.
With regards,
R.M.Parkhi

0
#131963

Mazzino
Participant

I think it depends on the nature of the problem and the kind of data involved. You have several tools to do that, such as DOE, correlation analysis, means tests, variance tests, etc. but that will strongly depend on the type of data (discrete, continuous, etc.), the distribution (z test, t test, nonparametric tests) the physical relation between variables (linnear, cuadratic, etc)