iSixSigma

Attribute Data

Six Sigma – iSixSigma Forums Old Forums General Attribute Data

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #49637

    Dillon
    Participant

    Hello:
     
    I have a question about attribute data and using it to perform correlation analysis.   I have a large data set, but much of the data are attribute data, such as issued/not issued.  Cash/nocash, paper/plastic, etc.
     
    Would it be statisitcally correct of me to convert these attribute values to a discrete value, such as 1 for cash, and 2 for nocash.  By converting to a number, I can run a quasi correlation test to see which variables most impact the desired result.  chi-Square tests are ok, but i prefer a specific correlation score?
    Thank you  
     
    Thanks!

    0
    #169922

    Robert Butler
    Participant

    You will need to tell us more about the measured response before anyone can offer anything meaningful. Specifically, what kind of a measure is it?

    0
    #169927

    Sloan
    Participant

    Mr. Butler is correct. However just on the surface of your question, it looks like you might be headed down the path of logistic regression rather than linear regression. Logistic regression can be quite a bit trickier than linear regression so you would need some experienced help with it if you have never attempted it. 

    0
    #170069

    Vikas
    Member

    Sore information is required on what all data you have and of what type and above  that what is your perspective of analyzing the data, as in what do you want to prove / disprove from that data.
    But in general, correlation analysis should be done on continuous data only and not on such discrete data, as you have. For discrete data toold like proportion testing and Chi Square testing can be used.

    0
Viewing 4 posts - 1 through 4 (of 4 total)

The forum ‘General’ is closed to new topics and replies.