Dear researcher,
Hello
For assessment of factor structure of the A Scale, I performing exploratory and confirmatory factor analysis. Four latent factors were extracted. By CFA, factor structure was assessed. The first point, Factor correlations are very low (for example .06). The second point, factors loading are more than one (for example 1.21). What is your interpretation about factor correlation and factor loading?
What is your interpretation about factor correlation and factor loading when factor correlation is very low and factor loading is more than one (in confirmatory factor analysis)?
Thanking you in anticipation
Hamid
Small correlations among factors is can easily occur, but loadings over 1 are much more problematic. Plus, you don't just have loadings over 1.0, you have them as high as 3 and over, so something is severely wrong with either your data or your specification.
I would suggest you go back to the beginning and look at the frequency distribution of each of your variables (and pay special attention to whether you have accidentally included the scores for missing data). Then I would examine a correlation matrix where the variables are grouped according to their factor assignment (i.e., the variable list would begin with Q30, Q34, Q43, and Q29, then Q35 and so on). If your factor structure makes sense, you will find higher correlations within each factor and lower correlations between the sets of factors.
Are these the standardized loadings? If not, it is not uncommon that the unstandardized loadings exceed 1. So i would suggest to estimate the standardized loadings first.
If these are the standardized loadings, it is uncommon, but they can exceed one to, check this paper of Jöreskog:
http://www.ssicentral.com/lisrel/techdocs/HowLargeCanaStandardizedCoefficientbe.pdf
Best, Manuel
In line with Heinrich, explanation of coefficients is necessary.
Are these scales used before?
In my suggestion go for unidimensionality assumption by means of factor analysis first.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling