This is a post about getting the results after using glm function to generate the logistic regression model.

glm binomial (二值逻辑回归) in car package

coefficient estimate

summary(x)$coefficients[,1] # x is the model name

p-value

summary(x)$coefficients[,4]

odds ratio

exp(coef(x))

confidence intervals

exp(confint(x))

multinom (多值逻辑回归) in nnet package

coefficient estimate

summary(x)$coefficients[,1]

z statistics

z <- summary(x)$coefficients/summary(x)$standard.errors

p-value

(1 - pnorm(abs(z), 0, 1))*2 # z is the z statistics

effect size

多值回归中没有odds ratio的概念

##

references

r documentation - odds ratio

Getting p-values for “multinom” in R (nnet package)