Jump to content

Recommended Posts


My name is Ravi, 

I have started a project for my GB. Project is on Customer satisfaction.

1. I know the VOC

2. I know what are the factors effecting CSAT with the CSAT survey data which we get.

3. I have completed fish bone , XY chart .

4 . I need to start hypothesis tesing, how do i proceed with hypothesis testing.

Need help.

Link to post
Share on other sites

Thanks for the Reply,

I work for a BPO, we work for Bell internet in canada.

I have done a brain stroming sessions with my agents and got few inputs. I have taken these inputs and matched with the real time CSAT comments from customers. Most of them are matching,

Now i want to check whether my agents assumption is correct or not , for that i need to do hypothesis testing.

One thing which is confusing me is, if i have RAW CSAT report ( Includes VOC ) to know why customers are dissatisfied then, do i need to take inputs from the agents regarding the same and do hypothesis testing.

Link to post
Share on other sites

Hello Ravi,

I am Munesh,

If you want to apply hypothesis testing first you complete your analysis.

You have got pssoble causes through brain storming now you do varification of the

possible causes to find out the probable causes.

Do validation of the probable causes to find out the root cause.

Take counter measure to eliminate the root cause.

Now you can apply hypothesis testing to check whether improvement is by chance or it is real.

With regards




Link to post
Share on other sites

Thanks Munesh,

In this case how to do u want me to test hypothesis.

Say if not supported issue is one of the factor causing DSAT, then

Null Hypothesis : Not supported is = DSAT

Alternate hypothesis : Not supported is =/ DSAT.

Is it correct ?

Link to post
Share on other sites

Hello Ravi,

Let me explain you.Suppose I am a supplir of cement and i supply 100 bags of the cement and i say that i am 95% confident that my average weight is 100 kg.Now you want to test whether supplier is right or wrong.Your null hypothesi will be weight = 100 kgs,alternate hypothesis will be weight is not equal to 100 kg.Now you can do t test for mean

In your case i am not clear whether you want to test mean or the variance.

Pls explain.

With regards

Link to post
Share on other sites

Hi Ravi,

With regards to Munesh for helping you out on some crash course with respect to your project, I waould like to support Vishwadeep's comments.

Hypothesis testing is a very critical & effective tool for Six Sigma projects. For effective execution of the project, theoretical knowldge on the subject is a must. Thus I advice you to kindly attend some good training program or go through some good literature on Hyp. testing i.e. Minitab tutorial, Statistics for Managenment by levin & Rubin.


Naveen Pilania



Link to post
Share on other sites


In my case ,

CSAT score for Aug = 75 %

CSAT for Sept = 76 %

CSAT for Oct = 75 %

Mean - 75.3 % this is not equal to my target ( 85 % ) , in this case how do u want me to proceed with hypothesis testing.

Inputs taken from brain stroming are as follows .

1. Not supported issues , Rank - 1

2. Transfer calls , Rank - 2,

3 Connectivity issue 3 ....

in this case do i need to conduct hypothesis testing , if yes how can i test hypothesis.

Link to post
Share on other sites


I have done t test for your data.as given below.P value is less than .05 this mean you are 95% confident that your mean is not equal to target mean.

Your null hypothesis is rejected and alternate hypothesis is accepted.

One-Sample T:mean

Test of mu = 85 vs mu not = 85

Variable N Mean StDev SE Mean

C1 3 75.333 0.577 0.333

Variable 95.0% CI T P

C1 ( 73.899, 76.768) -29.00 0.001

With regards

Link to post
Share on other sites
  • 3 weeks later...


The test done by Munesh is to test the hypothesis whether

- Your 3 month performance is statistically significantly lower compared to your standard of 85% or not.

But from your question I understood that you are trying to find out which of the three identified factors (no-support, transfer, connectiviy) is significantly responsible for the lower customer satisfaction.

For this you have to look at the detailed data regarding how many customers rated these these factors as responsible for their dissatisfaction. I suppose you would have this data. Please do a frequency analysis of this data first.


Link to post
Share on other sites
  • 2 months later...

Hi we may not be able to use t test for this scenario as the x and y is discrete here (percentage cannot be considered as variable). t test can be used only when x is discrete and y is continous.

We need to take a good look at your data and identify whether Y can be programmed to generate continous data. Please go through some literature or attend some training before working on this. Otherwise you might be wasting your time by using the incorrect hypothesis tool

Link to post
Share on other sites
  • 8 months later...

I want to apply hypothesis test for hose manufacturing process.Process is curing.In this hose is kept in steam vessel at 144 degree and 90 minutes.

My null hypothesis is temp is =144 degree

alternate hypothesis is temp <144 degree.

I have little understanding bout this test.I want to clear it with example and exercise.

Link to post
Share on other sites

Dear Anil,

The question that you have posed for curing temperature in steam vessel is incomplete. You need to know more.

  • Have you collected data properly?
  • Are you sure you have taken the right sample size based on alpha, beta and delta (difference you wish to detect)
  • Have you tested normality?
  • Based on the second question above, you may have to choose one sample z or one sample t or one sample sign test.

Best Regards,


Link to post
Share on other sites

Hi Ani ,

For curing process just follow the procedure :

1) Data collection, increase the size of your samples increases the power of your test

so take sample size of 100 as in mafacturing u can easily get these samples also

if the sample size is high no need to check normality as per central limit theorem :Averages of sample taken from non gaussian distribution will tend to follow a gaussian distribution

2) If u know the Standard deviation of the population u can opt for 1 sample z test else by default u choose 1 sample t test

3) Look out for the P value minitab will give u p value in session window By default, all calculations are based on an alpha-level of 0.05.

4) if the pvalue is high null will fly funda<0.05

repeat the test in order to reduce theexperimental error (Human error +Systematic error +Random error)

Hope it clarified



Link to post
Share on other sites
  • 5 years later...

To reduce the NCL we firstly look towards the credit assessment and re engineer the process in order to asses the application and to provide the card to the potential right customer. We require eight pieces of information when applying for a credit card: income, education level, age, length of time at current residence, length of time with current employer, savings, debt, and number of credit cards


We then look forward to reduce and avoid multi-collinearity or when you have too many predictors relative to the number of observations.
Hope it throws some light

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Who's Online (See full list)

    There are no registered users currently online

  • Forum Statistics

    • Total Topics
    • Total Posts
  • Member Statistics

    • Total Members
    • Most Online

    Newest Member
  • Create New...