Is the output anova(test='Chisq') and an anova(test='LRT') when comparing logistic regression fits?

Question

I have to run LRT to compare two logistic regression models. when I run the code below I noticed the output seems identical. In this scenario are these identical tests? If so in what scenario would they be different?

input_null <- mtcars[,c("am","cyl","hp")]
null_fit <- glm(formula = am ~ cyl + hp, data = input_null , family = binomial)

input_full <- mtcars[,c("am","cyl","hp", "wt")]
full_fit <- glm(formula = am ~ cyl + hp + wt, data = input_full , family = binomial)


chisq <- anova(null_fit, full_fit, test='Chisq')
lrt <- anova(null_fit, full_fit, test='LRT')

Answer 1

Please read the documentation for ?anova.glm.

You can also choose "LRT" and "Rao" for likelihood ratio tests and Rao's efficient score test. The former is synonymous with "Chisq" (although both have an asymptotic chi-square distribution).

"LRT" and "Chisq" are synonymous; they both trigger the same calculations.