Marginal ANOVA test in R

Question

I am trying to run a marginal (also called type-III) ANOVA using the following code. Unfortunately, I get "Error: $ operator is invalid for atomic vectors". A google search turned up another person getting the same error (see here), but unfortunately no solutions.

rm(list = ls())

data(iris)

iris.lm <- lm(formula = Sepal.Length ~ Sepal.Width + Petal.Length , data = iris)
print(anova(object = iris.lm))
print(anova(object = iris.lm , type = "marginal"))

Answer 1

For future reference, the function Anova from the package car accepts a type argument (such as "III" or 3). This function is not to be confused with the function anova from the standard R library, which does not accept a type argument.

Answer 2

Using a traceback you can localize the error:

print(anova(object = iris.lm , type = "marginal"))
Error: $ operator is invalid for atomic vectors
> traceback()
7: deparse(x$terms[[2L]])
6: FUN(X[[2L]], ...)
5: lapply(objects, function(x) deparse(x$terms[[2L]]))
4: anova.lmlist(object, ...)
3: anova.lm(object = iris.lm, type = "marginal")
2: anova(object = iris.lm, type = "marginal")
1: print(anova(object = iris.lm, type = "marginal"))

So you get the error because you try to call terms on an object that not support it. This reproduce the error:

 lapply(list(iris.lm,type='marginal'),terms)
 Error: $ operator is invalid for atomic vectors

Or just :

terms('marginal')

But Why do you expect that this works? Maybe I miss something, but I think it is not mentioned that anova can take type as an argument.