I have a set of data including a 10 point rating scale (overall satisfaction) and about 20 service rating questions (measured using a 5 point satisfaction scale).
What I would really like to do is to run a regression analysis to measure how well correlated each of the satisfaction statements are with the overall rating question (and ideally to be able to order them by ‘importance’ / key drivers).
Are there any suggestions out there on how I should prepare the data and use stats IQ / Regression to best achieve this. I have statsIQ and have tried running with the full data (using the 10 pt and 5pt scale data) but not confident in the results.
Any thoughts would be appreciated!
Userlevel 5
+15
Hi @Ashards I don't know anything about statsIQ, but I can speak to your point on comparing your statements in terms of importance. To do that, you will want to standardize your regression coefficients (dividing by standard deviation). I'm not sure whether statsIQ can do this for you, but you can also calculate them manually. This will make the coefficients comparable.
Userlevel 6
+51
Hello @Ashards You are doing it right! you have to confirm that you are taking your variables as numerical, after you select everything and run the regression you will see something in blue "Show regression parameters summary" click there and you will see a table and all your variables would have an importance weight where you can see which one is more important for your model and for your key variable (overall satisfaction) those percentages are like the coeficients of the regression equation
Hope this works!
Hope this works!
Leave a Reply
Enter your username or e-mail address. We'll send you an e-mail with instructions to reset your password.