Видео

Yes, unidimensionality is an assumption of omega. Use omega hierarchical for multi-dimensional scales.

Did you check the scoring instructions for any reverse-scored items? Missing this step can cause alpha to be extraordinarily low and cause an error when trying to compute omega.

@@mikekalkbrenner4518 there is no reverse scoring instruction in ways of coping questionnaire, the cronbach alpha is already low therefore I went for Omega, please suggest the way to improve it, item reduction I already know but it is not lifting it up to the threshold level

@@arastupathak3028 I don't recommend deleting items based on the "scale if item deleted" values, as this practice can compromise the content (and construct) validity of the scale. I'm not familiar with the questionnaire that you're using but it sounds like it's lacking reliability evidence of scores with your sample. My recommendation is not to proceed with using the scale, as test scores cannot be valid without being reliable. I know this is very disappointing (I've been there). I would talk with your advisor but one option might be to write about the poor reliability evidence of scores on an existing screening tool with your specific normative sample.

Thank you for your kind words and I'm glad that the video and article were helpful! In my view, the effect size index should be selected based on the data analytic technique (see: doi.org/10.1177/2150137810373613).

In my view, it depends on the nature of the constructs and sample data. If there is a higher-order factor, use Omega hierarchical. However, omega total is probably best if you have three one-dimensional scales and the data are: Consistent with a normal distribution, large sample size, and the scale is comprised of five or more anchor definitions, as ordinal scales tend to closely represent interval level scales when these conditions are met.

@@mikekalkbrenner4518 Thanks so GD much!

Sir, I am using ways of coping questionnaire which is 4 point likert scale, I had very low cronbach alpha in one subscale I have cronbach alpha score of. 04. So I thought to use Mcdonald Omega but it is showing error of assumption violation. Kindly help me out