I am trying to find the correlation/regression between these two variables:
(both using 5 point likert scales)
I will be testing the correlation/regression between the total scale scores of these variables. Do I need to transform/calibrate these variables so that they all have the same total scale score? Or can I perform the correlation/regression analysis on the variables as they are? If it is the former, how is this done on SPSS?
you don't need to transform/calibrate these variables you can perform the correlation/regression analysis on the variables as they are , all measures depend on 5 likert ,
Hello Karolina,
You may use the two scores without modification (as suggested by Mohamed Ibrahim Mohamed ). In IBM SPSS, analyze/correlate/bivariate will lead you to the proper dialog box (starting from main menu bar).
If you are at all concerned as to whether summated Likert-type items create a score that is ordinal strength, then look at the Spearman (rank-order) correlation. If you're confident that the scores will behave sufficiently well to represent interval strength measures, then the more usual Pearson (product-moment) correlation is the one to use.
Regardless, any linear transformation (adding, subtracting, multiplying, or dividing by a constant) will not change either correlation measure. Nonlinear transformations (e.g., square root) will change the Pearson coefficient but, unless they also modify the order of the cases, the Spearman coefficient will not change.
Good luck with your work!
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Social Media