We are revising a scale. It has two factors: F1 (3 items: X1, X2, X3)and F2 (3 items: X4, X5, X6).
Results from Mplus showed that the CFA model indicators are acceptable, however, the factor loading is not. What should we do? Is it OK if we keep the items according to the CFA indicators?
- These are the CFA model indicators:
χ2 /df = 13.740/8=1.718
CFI = 0.978
TLI = 0.959
RMSEA = 0.067
- However, the factor loadings of X4, X5, and X6 seem not good. X4 = 2.825 >1, X5 and X6 < 0.3
MPLUS results:
STANDARDIZED MODEL RESULTS
STDYX Standardization
Two-Tailed
Estimate S.E. Est./S.E. P-Value
F1 BY
X1 0.774 0.067 11.581 0.000
X2 0.670 0.067 9.988 0.000
X3 0.614 0.069 8.954 0.000
F2 BY
X4 2.825 6.720 0.420 0.674
X5 0.278 0.663 0.419 0.675
X6 -0.038 0.111 -0.342 0.733
Thanks a lot!
In line with what Samuel Oluwaseun Adeyemo wrote, I suspect that the indicators of F2 (i.e., X4, X5, and X6) may not be highly correlated, leading to an improper solution (negative residual variance for X4). This can happen when indicators are (too) heterogeneous, that is, when their intercorrelations are low and/or very different from one another. If you could post the correlation matrix of the observed variables, that would help.
Thanks a lot! I agree with you Christian Geiser Samuel Oluwaseun Adeyemo
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication