One learner on Coursera asked me the following question because inconsistent results were found:

# First attempt
f <- levels(iris[,5])
sapply(split(iris[,1],f), mean)
# setosa versicolor virginica
# 5.842  5.822      5.866

# Second attempt
sapply(split(iris$Sepal.Length, iris$Species), mean)
# setosa versicolor virginica
# 5.006  5.936      6.588

Question: Why is it different when he use factor levels from levels?

---

Before answering this question, we can confirm that only the second result is correct by tapply(iris$Sepal.Length, iris$Species, mean).

The problem in the first method is that the length of f is only 3, and that length is shorter than iris$Sepal.Length. According to the help document of split function, “f is recycled as necessary and if the length of x is not a multiple of the length of f a warning is printed,” which means when we split iris$Sepal.Length with f, R regards it as:

Sepal.Length	Species
5.1	setosa
4.9	versicolor
4.7	virginica
4.6	setosa
5.0	versicolor
5.4	virginica

In fact, however, these first six rows above are all setosa. Check it with head(iris). Also, compare the difference between split(iris$Sepal.Length, iris$Species) and split(iris$Sepal.Length, f).

For more information about factor levels, please refer to factor levels and labels.