Compare
principle
Model hypothesis: There is no relationship between public factors and between public and special factors (no endogeneity)
Note: The factor loading matrix is not unique, that is, multiple different factors can be selected for explanation.
important elements
The sum of squares of the row elements of A: reflects the dependence of the original variables on the common factors
The sum of the squares of the column elements of A: reflects the contribution of the common factor to x, and measures the importance of the common factor
Parameter Estimation
It is necessary to estimate the factor loading matrix A and the personality variance matrix D
Factor rotation method (SPSS)
Make factors easier to interpret
factor score
Conversely, the common factor can be expressed linearly using x
SPSS
- First test whether the data can be used for factor analysis: KMO test and Bartlett's test of sphericity
- According to the scree plot, there are several factors
- Interpretation of results: common variance (sum of squares of row elements of A), factor loading (making a factor loading scatter plot, that is, the correlation coefficient between variables and common factors), factor score (component score coefficient matrix)
