After running the McDonald Omega in SPSS 28, I regularly get this statement "Omega cannot be estimated due to negative or zero item covariances. This may be due to items needing to be reverse scored, or to violations of model assumption." Because of the violation assumption, we went from Alpha to Omega. Kindly help!!
Have you checked the correlations among your variables/items? If some of the correlations are negative or zero, this could be the reason for the problem. Perhaps some of the items are negatively worded (reverse scored)? If this is the case, you should recode your items prior to computing omega.
@Christian Geiser yes, I found negative correlation between few items but scale doesn't have any reverse scoring. Kindly help me to over come this
It could simply be that some of the items are bad. If their intercorrelations are low or negative, something must be wrong with either your data or the items. Perhaps some of the items are unreliable or invalid; or the items measure multiple dimensions (factors). Have you examined the factor structure using EFA or CFA?
@Christian Geiser, I did CFA by using Smart PLS for one of the objectives then I found very low Cron bach alpha and low McDonald Omega and AVE, I started removing items from the reflective constructs of each sub domain and in the lower order I remained with minimum 2 items in every sub scales except in planful problem solving and positive reappraisal, the McDonald Omega crossed the threshold of .7 and AVE .5 but again Alpha didn't approve in higher order the 4 sub scales of coping totally removed and remained with 4 subscales.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Probability
- Probability Theory
- Reliability