Many phenomena in reality can be divided into two possibilities, or can be attributed to two states, which are represented by 0 and 1 respectively. If multiple factors are used to explain the causal relationship of a phenomenon represented by 0- 1, it may be applied to logistic regression.

1. Substitute all variables into the regression equation and perform stepwise regression analysis. It is necessary to use several techniques to screen variables, while considering the interaction of factors and comprehensive analysis. "

2. It is normal that the results of single factor analysis and multi-factor analysis are different, because single factor analysis often has the influence of mixed factors.

3. It should be understood that before the establishment of multiple regression model, the main function of single factor analysis is to play a screening role. Usually, the factors with p less than 0. 1 or 0.2 are selected to enter the multiple regression model, and the partial regression coefficient in the multiple regression model is meaningful.