I have a data set of 16000 records of 5 ordinal variables(Customer Satisfaction(response variable), service, quality, knowledge, responsiveness) which are survey responses in 0-10 scale. All the variables are highly left skewed (majority concentrated at 10 scale- approx 80% and rest 20% equally distributed across the others). I need to identify the impact and relation on the response variable from rest 4 independent variables.

Since the data is not normally distributed I am unsure on how to proceed. I thought of doing under-sampling but still unclear on how to proceed after that. Could some one share their knowledge on how to work on such data. Thanks in-advance.

  • 1
    $\begingroup$ Ordinal variable cannot be normally or nonnormally distributed: it is not continuous or interval one. Strictly speaking, even such term as "left skewed" isn't quite appropriate for such data. Rather, the categories are disbalanced by frequency, with greater category tending to have greater n. $\endgroup$ – ttnphns Jul 7 '18 at 16:31
  • $\begingroup$ Got it. but how do I proceed with such an unbalanced data set? $\endgroup$ – Hari Krishna Jul 7 '18 at 17:06
  • $\begingroup$ Welcome to stats.stackexchange.com. you can use multiple regression analysis to meet your goal. $\endgroup$ – Subhash C. Davar Jul 31 '18 at 15:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.