Hello! I have data from two sets of 10 Likert scale questions, (1-5, strongly disagree to strongly agree). I would like to correlate them with each other using Spearman's correlation. How do I go about this?
Hello Alexa,
If you wish to correlate responses to individual pairs of items, just run an ordinary Spearman correlation between the scores, paired by person/respondent, for each pair of items that is of interest.
If you wish to correlate overall scores across the two sets of items, then first compute a sum of scores for the first set (e.g., Item1 score + Item 2 score ...), and do likewise for the second set of items. Then run a Spearman correlation between the two total scores (again, paired by person/respondent).
Good luck with your work.
Dear Alexa,
there is an issue with the previous answers that needs clarification.
You should consider that Likert scales are typically ordinal scales.
Now, summing scores is illegal for ordinal scales. Nonetheless, it is legal for interval scales, i.e., scales where distances are meaningful.
So, if you want to sum scores, you have to convert your scale to an interval scale.
This is possible if you are able to precisely define the distance between scores.
For instance, suppose you scale is 'very bad', 'bad, 'so-so, 'good', 'very good'. Suppose also that you can make reasonable assumptions concerning distances between scores: for instance, the distance d1 between 'vey bad' and 'bad' is the same as the distance between 'very good' and 'good'; the distance d2 between 'bad' and 'so-so' is the same as between 'so-so' and 'good' and it is d2 = 2/3 d1. In such case, you can create an equivalent interval scale: e.g., 0, d1, 5/3 d1, 7/3 d1, 10/3 d1. By setting d1=3, you get 0, 3, 5, 7, 10. Now you can sum scores.
Please, note that if you cannot define distances between scores, you cannot safely sum scores from a Likert scale.
- Badges
- Science topic
- Similar topics
- Psychology