1
Q
residual correlation
A
when residual variances contain shared variance
2
Q
identification
A
- scaling the latent variable
option 1:
> mean of factors = 0
> fix 1 factor loading to 1 for each factor
option 2:
> mean of factors = 0
> fix the factor variance to 1 for each factor
- statistical identification
> k should not exceed M
> this can happen if there are too few observed variables
> k does not have a formula in CFA, but needs to be counted
3
Q
global model fit
A
- absolute fit measures
> X2-goodness of fit, RMSEA, SRMR, CFI, TLI - comparative fit measures
> likelihood ratio test (nested models), AIC, BIC, CAIC
4
Q
local model fit
A
modification indices
> indicate how much the X2 will improve (decrease) if that parameter is freed
5
Q
when there is a cross loading
A
add the variable to that factor
6
Q
when there is residual covariance
A
add the covariance to the residual covariance matrix
7
Q
parameters
A
- factor loadings
- residual variances
- factor variances and covariances
