3

My guess would be discrete because such a variable would only take a countable set of values and "continuous" seems to imply the continuum of the real numbers.

user89964
  • 33
  • 2
  • 1
    It won't count as continuous, since you have points with non-zero probability. Whether it counts as discrete or not depends on exactly whose definition of discrete you use (there are other possibilities than discrete/continuous). By the definition I was taught, the answer is "yes that's discrete", but by another common definition the answer is "no". This doesn't change anything material other than the label we apply of course. – Glen_b Sep 21 '15 at 01:16
  • 1
    A great answer specifying exactly what the pmf of such a random variable might look like is [here](http://stats.stackexchange.com/a/104018/6633) – Dilip Sarwate Sep 21 '15 at 02:19
  • These are both very helpful, clear responses. – user89964 Sep 21 '15 at 03:53

1 Answers1

2

A continuous variable has positive length, area or volume. These terms, respectively, correspond to particular Lebesgue measures. But the Lebesgue measure of $\mathbb{Q}$ is 0; it doesn't have positive length.

Approached from another perspective, it is sufficient to note that $\mathbb{Q}$ is countable, so it's not continuous.

Sycorax
  • 76,417
  • 20
  • 189
  • 313