Impacts of Land Use Change on the Regional Climate: A Structural Equation Modeling Study in Southern China
Table 2
The meanings and equations of indices of the goodness-of-fit.
Index
Meaning
Equation
Criterion
/df
is impacted by the difference between sample and model-implied matrices. The small represents the good fitting of model and samples. Degree of freedom (df) is the number of independent variables. Considering that is too sensitive to sample size, /df is selected as a relative chi-square value
—
<5
The probability of error. value and result’s credibility will change in the opposite direction. If , the result is meaningful
—
<0.005
GFI
Goodness-of-fit presents the proportion of covariance in sample explained by implied model. Greater GFI means the good explanation of dependent variables by independent variables
>0.9
CFI
Comparative fit index can avoid the underestimation of fit in small samples. It also tests the difference between the worst (independence) model and model of interest
>0.9
RMSEA
Root mean square error of approximation estimates the discrepancy between model-implied and true population covariance matrix per degree of freedom
<0.08
RMR
Root mean square residual tests one kind of mean of residual
<0.05
NFI
Normed fit index reflects the proportion of worst (independence) model explained by model of interest
>0.9
Notes: is sample’s size, is number of variables, is sample’s relation coefficient, is model-implied relation coefficient, is sample’s covariance matrix, is model-implied covariance matrix, is identity matrix, is chi-square of worst (independence) model, is chi-square of model of interest, df is worst (independence) model’s degree of freedom, and df is model of interest’s degree of freedom.