I have two sets of Likert type items - 10 questions with 1-5 for parameter a, and 30 questions with 1-7 for parameter b.

I want to compare the two results with a Spearman correlation analysis. However, I'm not sure how to justify using the sum, median, or mean of the items for the respective scales.

I've researched the topic extensively and can't find any conclusive answers.

  • $\begingroup$ First, you might consider using Kendall's $\tau$ (tau), which might allow you to view Likert scores strictly as the ordinal data they actually are. I think it's best to rely on statistical software for computing $\tau.$ // Second, in case my opinion on Likert data isn't already clear, I think it is almost always quite wrong to treat Likert data as if numerical. IMHO doing so is responsible for a large fraction of the very large number irreproducible 'findings' that infest journals in psychology and other social sciences. $\endgroup$ – BruceET Jun 21 at 20:13
  • 1
    $\begingroup$ Thank you, I'll use Kendall's tau to look at all the items of both scales, and leave out the likert summations entirely. $\endgroup$ – monkfig Jun 22 at 11:17

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.