Friends,

I was wondering which theory I should be using while carrying out the Hypothesis testing? I am doing a Quality Improvement project and the results that have been accrued from the 5 Why's and other tests are basically:

Trainees performance is poor

Training and Quality teams are over occupied

Team is less motivated and a few other findings are there

Which assumption should I take while conducting the hypothesis test??

Thanks in advance for your inputs.

Regards,

Rahul

 

Rahul,

What is your project focused on (scope) with goal and metric so that you will be able to measure if your project is successful or not and if the changes implemented have sustained themselves for the benefit of company's bottom-line?

Each of the statement you have listed can be a hypothesis in itself. Use data to prove or discount if your hypothesis is correct or not.

Hope this helps, but I will be able to better guide you if you can answer my question above.

With 5 why one is able to narrow the potential root causes as there is never one cause that create a quality defect. It has to be combination of two or more with OR / AND qualifier. Just like fire can not be without combination of air (oxygen) AND wood (fuel) AND heat.

This might be off topic however, here is a good read for you for C&E --> http://curriculum.d91.k12.id.us/Curriculum%20Resources/Summer%20School%20Reading/Day%2010%20Cause%20and%20Effect/Cause%20and%20Effect%20Resource.pdf

Thanks for your reply, Ashok. My project is focused on reducing the defect % of the quality in my process. That is how I wish to quantify the pre and post project results. 

About all the fact that a hypothesis can be done on all the factors stated above...is it advisable to conduct the same on all the factors?? Also, if you could please guide me on how to conduct the test since the trial version on Minitab expired on my system? Is there any other option available online for free?

I concur with your point that a 5 Why analysis will narrow down the responsible factors.

Regards,

Rahul

 

Rahul,

Your project is transactional in nature (my assumption based on the factors mentioned in your first question)  Correct?

Can you share with me or identify for me if you have developed a SIPOC (supplier, Input, Process, Output, Customer) diagram and then a SIPOC matrix which helps you prioritize various inputs and related steps in the process quantitatively?

The above step will prepare you to develop surveys/ questionnaire, polls, etc to gather some data on various inputs which then will support or reject specific hypothesis.

If you can share your process map and SIPOC, it will help me better understand what suggestions should I provide you.

If the % defect is not a transaction but a product the above step will lead you in the right direction but then you will not need survey, polls, etc, but understand what attribute of the product is to be inproved and how to measure.

There are other software that one can use and are not very costly to buy. I have used SPC Excel which will help do some of the analysis you wish to do, but is not as powerful as Minitab obviously. I believe if you uninstall Minitab from our computer and clear the registry, you should be able to install Minitab trial version one more time with a different identify (email, name, etc.) You can try it and it may work, as I am not sure how packages are developed for Indian market by Minitab.

Hope it helps.

Ashok

Ashok,

I am attaching the SIPOC with this post. I am arriving at the defect % by taking an average of the all the errors made by the entire team after taking into count all the calls monitored by the quality team.

Thanks for the Minitab suggestion, I thought of it earlier and in case it doesnt work by clearing the registry then I can always use it on a new system with a different emai.

SIPOC Diagram.xls

As always, thanks much for your precious guidance and I look forward to hearing from you again.

Regards,

Rahul

Rahul,

Sorry for delay in responding. I reviewed the SIPOC, but find myself asking a few questions.

I am not sure if you will like me to document detailed thoughts on this forum or if you will like to share with me your email so I can provide you what I have to provide you in terms of examples. Attaching files here sometimes is tedious and very time consuming (I have been trying to upload couple files but have been unsuccessful.)

Here is the start.

Please help me get better understanding on some questions that come to my mind (if you can provide some explanation or make appropriate corrections..... your choice!

1. Why is an output of step "Interview is completed and is ready to be saved as a complete" not an input to the next step "Agent traces the pain areas in the call, if any and listens to them" and so on.....?

2. Supervisor is not an input as indicated in step "Confirms it with the sup and decides if the call can be saved." What do you feel should be appropriate input here... (my suggestion will be identify something in line with "decision tree criteria" or "Departmental/personal goals", etc....

3. Have you used the SIPOC matrix to understand anecdotal importance of specific steps and inputs based on VOC/primary or secondary metrics of your projects?

If you wish I can provide an example for your ready reference.....

If you already have, then share it with me or an example that I can provide can guide you /walk you through the steps.

Purpose is to ensure that ranking you get from the above exercise should align with the result you found with 5Why analysis. If the 2 are different, then one must evaluate which path to follow and why.

Rahul,

Well I realized after my last response that the 2 files I wanted to share with you as example were finally uploaded to folder "Ashok_SIPOC Matrix example". The names of these files are Spliced Coil Project Charter_SIPOC_PF v2.pdf and Root Cause Analysis_Matrix v2.xls.

Hope these help you.

Ashok

Hi

I have a question. The project Y is Continuous. X is discrete and has 3 population. So I would use Anova.

Pre-requisite to run Anova is that the data needs to be normal and should have equal variance.

Now the data is Normal but does not have equal variance. What shall I do now?

 

Dear rahul ,

I want to ask u asimple question u want to apply Hypothesis testing to what u want to find out (assume u r data is variable type) if u want to find out mean of % defect done by ur entire team . but i want to know is ur whole team is working on same process if not than u cant apply Hypothesis testing in that case bcoz if u have more than one process ( as shown in ur SIPOC ) u should find to RTY bcoz Hypothesis testing is use to claim that how much % defect in single process

( and always try to put alternative hypothesis value which our process not want)

Thanks for your post, Amit. Yes, my team is doing the same process. 

Aside to Ashok:

Point 1: Output of step "Interview is completed and is ready to be saved as a complete" is not the input for the next step is cus this steps is in its own entirety. And, the next steps are in continuation

Point 2: Agree with your point, the supervisor should not be the input and I will put the input as work process. Thanks for pointing it out. 

Point 3: I havent used SIPOC to understand the importance of every step w.r.t VOC as yet

- Rahul