Greetings,

I am trying to analyse variations in a biological process. I used Anderson-Darling test and found out that the data is not normal. The sampling interval is one day and chances of manipulating data are very rare. The data is presented in the excel file (Count.xls) I attached here. The actual data points are of the form a*10^6.

Since, I initially considered only the multiples, I then checked for lognormality but the same result occured.

Despite that I ran xmr chart and the outliers seem to coincide with the ones that can be analysed with subject matter knowledge. I read from many references that analysing non normal data by standard techniques give significantly faulty results. In this scenario, is it safe for me to go by logic and do normal analysis or should I use any other specific tools?

I am attending BB programme in Pune next month. Do we generally discuss analysis of non normal data in BB Programme or should I go for any specific references?

PS: My excel spreadsheet got saved in GroupFiles Folder

Regards,

Dear RaviShankar,

We will discuss analysis of non-normal data during the BB program. We can of course pick up your queries during the workshop.

Regards,

shantanu kumar

Hi Shantanu,

That would be great. Can you give me some online references so that I can do some homework and come prepared? Secondly, non normality of this parrticular data seems dubious to me as similar data in next process step seems to be normal.

Any suggestions regarding this would be quite helpful.

Regards,

Ravi

Hi Ravi, 

You can read up on Non-Parametric tests and Data Transformation. We will discuss Non-Parametric tests and Data Transformation in detail during the BB workshop. To prepare for BB, I would suggest that you revisit the GB material and Basic Statistics. 

I will suggest that you list down your Six Sigma queries and e-mail me. I will pick up the queries during BB workshop. 

Regards, 

shantanu kumar