How to do factor analysis?

Factor analysis (exploratory factor analysis) is used to explore how many factors (variables) the analyzed items (quantitative data) should be divided into, for example, 20 scale items should be divided into how many aspects are more appropriate; the user can set the number of factors by himself, if not, the system will use the root value of the eigenvalues greater than 1 as a criterion for determining the number of factors. In the "Advanced Methods" module, select the "Factor" method, drag and drop the quantitative analysis items into the right analysis box, and click "Start Analysis".

Supplementary note: If there is an expected number of factors you want to extract, you can actively set the number of factors to be output. Checking "Factor Score" and "Composite Score" will generate new variables in the left analysis box, with titles such as CompScore***, *** (Composite Score), *** (Composite Score), *** (Composite Score), and *** (Composite Score). *** (Composite Score), FactorScore ***** (Factor Score). Factor Score can be used for further analysis, such as cluster analysis, regression analysis use, etc.; Composite Score can be used to compare rankings, etc.

The number of factors: in most cases, we have been analyzing with subjective expectations of how we want the questions to be categorized, and at this point you can directly set the corresponding number of factors.

Before the formal analysis of structural validity, the first step is to measure the questionnaire scale through the KMO and Bartlett's test to determine whether it is suitable for factor analysis, the KMO value is used to determine the degree of acceptability of the selected variables in the factor analysis, and to examine the correlation between the variables.

Generally factor analysis requires a kmo value greater than 0.6. Bartlett's test needs attention beyond the treatment. Bartlett's test in principle is to test whether the variables are independent, to determine the correlation of the factors, if the model is significant (the corresponding p-value is less than 0.05) indicates that it is suitable for factor analysis.