Power Analysis in Covariance Structure Modeling Using GFI and AGFI.

Area of Science:

• Statistics
• Psychometrics
• Structural Equation Modeling

Background:

• Extends previous research on power analysis for covariance structure models.
• Focuses on tests of overall model fit using GFI and AGFI fit indexes.

Purpose of the Study:

• To develop procedures for power analysis when null and alternative model fit levels are specified using GFI or AGFI.
• To evaluate the behavior of power as a function of degrees of freedom for these indexes.

Main Methods:

• Conducting power analyses for covariance structure models.
• Specifying null and alternative model fit levels using GFI (Goodness of Fit Index) and AGFI (Adjusted Goodness of Fit Index).

Main Results:

• For GFI, power decreases as degrees of freedom increase, a counter-intuitive finding.
• For AGFI, power increases as degrees of freedom increase.
• Establishing appropriate null and alternative hypothesis values for GFI and AGFI is problematic.

Conclusions:

• The behavior of GFI and AGFI in power analyses presents challenges for model evaluation.
• The RMSEA (Root Mean Square Error of Approximation) index is recommended as a preferable basis for power analysis and model evaluation.