0
$\begingroup$

For example, questionnaire 1 has 20 items scored from 1-5, questionnaire 2 have 10 items, and it is a True/False scale (so the final score is like true =1, false =0, and add them up), can I still do a Pearson correlation between them? If I had to standardize the score, how should I standardize them? just merely averaging everything?

$\endgroup$

1 Answer 1

0
$\begingroup$

The fact some of your items contain different ranges of numeric values (i.e., [1,5] for 18 of your items & [0,1] for your two T/F items) is not problematic. For example, in past work, I would often estimate the correlation between salary and years of experience, which do not have the same range (e.g., [0-50] for years of experience and [30,000, 750,000] for salary). The issue is that your items are categorical. Specifically, your T/F items are dichotomous, and your other 18 items are ordinal. Pearson's correlation is a measure of linear association between two continuous variables. As the number of response options goes to infinity, this will typically be less problematic; however, because two of your items are dichotomous, I would suggest you consider alternative measures of linear association. Given this, the spearman's rank correlation and/or tetrachoric/polychoric correlation may be viable alternatives.

$\endgroup$
2
  • $\begingroup$ Got it!! Thank you so much! :D $\endgroup$ Nov 17, 2022 at 15:47
  • $\begingroup$ No problem @IvesDragonfly! If you found my answer helpful, please accept my answer and give it an upvote! $\endgroup$ Nov 17, 2022 at 16:27

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.