Interpreting standardized vs understandized coefficients for a regression with categorical IV's

Question

I would like to know which stage of the customer journey (my IV with 8 levels) is most important in predicting Net Promoter Score (my DV). I created k-1, or 7, dummy variables, with the base case/reference being one of the stages ("Use the service"). Below is my SPSS output of the multiple regression. What concerns me is:

• The terrible adjusted R square, which seems to indicate only 10% of the variance in my DV explained by the model. Would my dummy coding explain this, or rather just plainly that the stages of the customer journey poorly predict my DV?
• How to interpret the results, more specifically, whether to look at the standardized or unstandardized coefficients, and how to interpret these coefficients in relation to the reference variable?

https://i.sstatic.net/xL5XA.jpg

Answer 1

Since your predictor variable has 8 discrete values you will have only eight distinct predictions (assuming no pair of coefficients is equal). So it is not too surprising that those eight values do not explain much of the variability.

The values of the unstandardised coefficients represent the difference between the predicted value for that group from the reference. So for example people in the stop condition score about 4 points less than people in the use the service condition.

Answer 2

I would agree with this "rather just plainly that the stages of the customer journey poorly predict my DV?" if that's how you interpret 10%. Standardized regression weights don't make much sense for dummy variables. Raw score weights can make sense if the dummy variables are chosen carefully.