1
$\begingroup$

I have a study in which a well-known previously validated questionnaire (like Beck Depression Index, NEO-FFI, etc.) is used. My question is: do I need to calculate Cronbach's alpha for this questionnaire on my data set? Or, in other words: do I need to demonstrate reliability of a well-known previously validated questionnaire on my data set?

Why I'm asking? I was pretty sure that the answer is "No, this is only required when you create new questionnaire (or validate new language version)". But a reviewer insists on adding this to my article.

Of course, I can calculate alpha on my data set (it turns out to be OK), but still, I wonder if his/her request is justified. Any thoughts (and bibliography ;) ) are welcome!

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

This is standard practice, particularly in clinical psychology/psychiatry journals. The rationale is that while the reliability and validity of the measure has already been established in a particular population, it's worth checking that these results generalise to the population you're studying. If a measure that normally shows high reliability has a lower value in your data, it indicates a potential issue for your study.

By way of an example, imagine a population where many people have sleep problems that are unrelated to depression. You would find lower reliability for your depression measure, since this would weaken the correlation between the sleep item(s) and the rest of the question, and it wouldn't be clear if a change in total scores is due to reduced depression symptoms or from improved sleep.

More generally, you should always at least calculate alpha when using composite measures as a data sanity check.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.