Three of the variables in my survey are comprised of 2-item reverse coded scales. One of the reviewers wants me to provide the cronbach alpha values for these scales, whereas by understanding was that we do not need to check the scale reliability of a variable that has only 2 reverse-coded items. Calculating the cronbach alpha value for such scales will be meaningless. Is my thinking correct?
I’d report it, especially if you have reversed-code items. Sometimes people have difficulty understanding these items and their answers are inconsistent. Reporting Alpha will demonstrate if the items were understood as they meant as well as the items homogeneity.
Cronbach's alpha is a measure of internal consistency, 'reliability' is usually used to denote test-retest reliability.
If your variable only consists of two items, then a Cronbach's alpha is still useful as it will tell you how similar responses to them were; but for just two variables you should really run a Pearson's r or Spearman' rho correlation between the (unreversed) variables, as these are statistics that are much easier to interpret (-1.0 to 1.0) and more appropriate in this instance.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Coding