Answer: (1) Causality refers to the relationship that changes in some variables cause changes in other variables. Causality can be direct or indirect (that is, there may be intermediate variables); It can be full of one cause and many fruits, or it can be multiple causes and one fruit. The causal relationship between variables must meet the following conditions: there is an interpretable correlation; There is a certain time sequence; Can't be a virtual relationship (that is, after one relationship is replaced by another, it proves that the original relationship is untenable); The direction of causality cannot be changed and so on.

(2) How to determine the causal relationship.

① induction

A. The method of seeking common ground, also known as the method of emphasizing, means that the phenomenon studied appears in different cases, and only one antecedent of each case is the same, and the others are different. This same antecedent may be the reason for this phenomenon.

B. Difference method, also called difference method, means that the phenomenon studied appears in one situation but not in another, and only one antecedent of these two situations is different, and the others are the same, so this different antecedent is the reason for this phenomenon.

C. the combination of seeking common ground and seeking differences is the combination of seeking common ground and seeking differences, that is, there is only one same antecedent when the phenomenon under study appears, but there is no such antecedent when it does not appear, so this antecedent is the reason for this phenomenon.

D.* * * Reform means that the research object changes with the change of other antecedents under the same but only one different condition, so this antecedent is the reason for this phenomenon.

E. residual method is suitable for causal analysis of compound phenomena, and there are many factors affecting compound phenomena. Except for the part with known causality, there must be causality between the rest of the research object and other influencing factors.

② experimental design method

Through experimental design, irrelevant variables are effectively controlled, so as to ensure that the changes of dependent variables are really caused by independent variables, and thus determine the causal relationship.

③ Statistical analysis method

For example, use structural equation model to discuss causality.

(3) Regression analysis is a statistical method to find the functional relationship between one or more door variables and a dependent variable through observation, so there is not necessarily a causal relationship between the independent variable X and the dependent variable Y in the regression equation.

The basic idea of regression analysis is to calculate regression coefficient according to many observed values, establish regression equation and test the significance of regression coefficient. Homologous analysis is a mathematical way to express the relationship between variables. Y can be predicted by regression equation according to X, but regression analysis cannot establish causality between variables.