-probit regression model of SPSS data analysis

Probit meaning is the probability unit, and logistic regression, the Probit regression is also used in the case of the dependent variable as categorical variable, usually, the results of the two regression methods are very close, but because the result of probit regression is more abstract and difficult to understand, So the application is not as extensive as logistic regression.

Probit regression is based on the normal distribution theory, and the logistic regression is based on the two-item distribution, which is the difference between the two, when the independent variable is more continuous variable and the normal distribution, you can consider using probit regression, and the independent variable is more categorical variables, You may consider using logistic regression.

In SPSS, there are two processes that can perform probit regression, one is an independent probit process that only handles the case where the dependent variable is classified as two, and the other is the logistic regression process, which changes the connection function to Probit, which does not require the form of the dependent variable.


We still use the two classification logistic regression example to carry out the probit regression analysis of the two classification, and compare the difference between the two.

SPSS default two classification probit regression data is the form of frequency tables, and the data form of this example is a case breakdown, so you need to add a frequency variable count, using the following statement to achieve more convenient

COMPUTE count = 1.
EXECUTE.

After running, a count variable is added to the original data

Analysis-Regression-probit

Let's look at a probit regression example of the frequency data.

Example: To analyze the toxicity of the toxin by studying the relationship between the concentration of a toxin and the lethal amount, the data is composed in the form of frequency tables, as follows

C represents the concentration of toxins, total is the number of mice per group, dead is the number of deaths

Analysis-Regression-probit

-probit regression model of SPSS data analysis