How to Evaluate Residual Covariances in CFA

Confirmatory facto analysis (CFA) is used by statisticians to determine whether the inferences they've made between multiple pieces of variable data holds true. Computer applications handle the analysis of the complex algorithms, or groups of equations, needed to process the relationships of co-variables in your data. After you've analyzed your data, compare its residuals with your estimated output to find out if your original assumptions about the data were correct.

Instructions

1. Launch your variable modeling application, and then input your CFA model's statements into the program -- enter the model's name, data source, residual covariances and output type.

2. Use your variable modeling application's "Compute" or "Output" command to generate estimates for your data.

3. Locate the list of variables in your CFA model's output and compare them with your estimated output. If a variable has a value of "0," then it has no correlation with the other output variables. If a variable is negative, then it points away from the other variables. Any positive variables you find indicate that the covariances move in the same direction and share a commonality.

References

• UCLA Academic Technology Services: Introduction to Mplus - Featuring Confirmatory Factor Analysis
• Indiana University Center for Statistical and Mathematical Computing: Confirmatory Factor Analysis using Amos, LISREL, Mplus, SAS/STAT CALIS