I am conducting a CFA study of a revised psychometric instrument with a newly hypothesized factor structure. I am not a statistician and gave my cleaned and prepped data to a stats group to run the CFA. They used the marker indicator method and the parameter estimates came back unstandardized.
I am looking for threshold values for unstandardized loadings or some kind of rule like "standardized loadings > x are considered amazing and between x and y are less amazing, and less than y are poor".
Can anybody help me with that?
Thanks,
Heidi
PS Google has been no help at all!
Michael T. Moore, who is on Research Gate, has a publication on confirmatory factor analysis that might help (you need to message him to get it). I tend to use principal components approach, which gives loadings on a range from 0-1 that can be interpreted the same way you view correlations - the loading is essentially the correlation of the item to the underlying factor that has been identified. What is the range of your loadings from the marker indicator method?
Since posting this, I have done additional research and it seems that the standardized parameter estimates are what is used for reporting purposes. I don't think it will be hard to get the standard factor loadings and rework some of my language. I appreciate your help and the M. T. Moore references.
Thanks,
Heidi
- Badges
- Science topic
- Similar topics
- Psychology
- Psychometrics