SPSS data analysis-Paired logistic regression model

Lofistic regression model can also be used for pairing data, but its analysis methods and operation methods are different from the previous introduction, the specific performance

In the following areas
1. Each pairing group has the same regression parameter, which means that the covariance function is the same in different paired groups
2. The constant term varies with the pairing group, reflecting the role of non-experimental factors in the pairing group, but we don't care about its size,

Therefore, the conditional likelihood function is substituted for the general likelihood function in the quasi-timely way, and the parameters of reflecting layer are eliminated in fitting.

The process of matching logistic regression model is not directly fitted in SPSS, the data needs to be processed and fitted by other methods, and the fitting method has variable difference fitting and Cox model.

First, the variable difference value fitting

Applicable to 1:1 pairs only, by finding out the difference between the same pair of cases and the control group, the unordered multi-categorical logistic regression model of the difference with no constant term is fitted to achieve the purpose.

Example: A set of data was collected to analyze the relationship between the use of estrogen and endometrial cancer, and in addition to the research factors, additional

Set two variables, data for pairing data, 1 for cases, 0 for control, case for disease, or dependent variable

The variable difference is used to fit, first of all the difference between the variables, you can use the calculation variable process, but the process is only

Can handle a variable, more trouble, we use the syntax editor to program, as follows

When all is checked and then run, the newly generated difference variable will appear sequentially in the original data, and we'll then do unordered multi-categorical logistic regression for these difference variables.

Analysis-regression-Multiple logistic

SPSS data analysis-Paired logistic regression model