Why does R coerce a one dimensional data frame result to a vector? (and how can I stop it?)

Question

> age <- c(1,2,3)
> weight <- c(4,5,6)
> p <- data.frame(age=age, height=height)
> p
  age height
1   1      4
2   2      5
3   3      6

> p[p$age > 1,]
  age height
2   2      5
3   3      6

> age
[1] 1 2 3

> age.df <- data.frame(age=age)

> age.df
  age
1   1
2   2
3   3

> age.df[age.df$age > 1,]
[1] 2 3

I want the subset of the dataframe that I asked for, not a vector, in which the rownames and column names are lost. Why does R do this? Can it be prevented?

Answer 1

Use drop=FALSE:

> age.df[age.df$age > 1, , drop=FALSE]
  age
2   2
3   3

A useful reference is the R Inferno (http://www.burns-stat.com/pages/Tutor/R_inferno.pdf); see section 8.1.44.