Questions: Random Variables: Definition and Classification

Question 1

A programmer writes: 'Let X be a random variable. Since X is random, we can't say anything definite about X = 3.' What is wrong with this statement?

• A. Nothing is wrong — X = 3 is undefined because random variables can't equal specific values
• B. It confuses the function X with a realized value x. P(X = 3) is a perfectly well-defined probability.
• C. It's correct — since X is random, no probability statement about specific values is possible
• D. It's only wrong for discrete random variables; for continuous ones the statement holds

Answer: (B)

A random variable X is a function from outcomes to numbers, not a vague 'unknown.' The notation P(X = 3) means 'the probability that the function X assigns value 3 to the outcome that occurs' — a completely well-defined number. Confusing the function X (the rule) with a realized value x (the output) is the central conceptual error. The statement is wrong because it treats X as fundamentally unknowable, when in fact its probability distribution is precisely defined.

Question 2

X is a continuous random variable representing the exact temperature (in Celsius) at noon tomorrow. A student calculates P(X = 21.5) and gets 0. They conclude their model must be broken. What is the correct explanation?

• A. The student is right — any model that assigns probability 0 to an event is incorrectly specified
• B. P(X = 21.5) = 0 is correct; for continuous RVs, individual points have zero probability because they have zero 'width' under a density curve
• C. The model assigns probability 0 only if 21.5 is outside the support — otherwise it should be positive
• D. Continuous random variables don't have probability at specific values; you must use cumulative distribution functions exclusively

Answer: (B)

For a continuous random variable, probability is spread over regions as a density, not concentrated at individual points. P(X = exactly 21.5) = 0 is correct and expected — a single point has zero length/area under the probability density curve. This does NOT mean the event is impossible in any physical sense; it means you ask P(21 ≤ X ≤ 22) = some positive number instead. The student's model is fine.

Question 3

A random variable is best understood as a function from outcomes in a sample space to real numbers.

• T. True
• F. False

Answer: True

This is the precise mathematical definition. X : Ω → ℝ maps each outcome ω in the sample space Ω to a real number X(ω). Despite the misleading name 'variable,' a random variable is not an unknown algebraic quantity — it is a rule that translates abstract outcomes into numbers, which is what allows calculus and real analysis to enter probability theory.

Question 4

Whether a random variable is discrete or continuous depends on the size of the sample space — discrete random variables come from small sample spaces, and continuous ones from large sample spaces.

• T. True
• F. False

Answer: False

The discrete/continuous distinction is about the *range* of the function (the set of values X can take), not the size of the sample space. A discrete RV takes values in a countable set (even if the sample space is large). A continuous RV takes values in an uncountable interval (like all real numbers in [0, 1]). A coin flip has a tiny sample space, but the number of flips until the first head is discrete with an infinite range.

Question 5

Why does a continuous random variable assign probability 0 to individual values, and why does this not mean those values are 'impossible'?

Model answer: Probability for a continuous RV is spread as density over regions; a single point has zero width under the density curve, so integrating over it gives zero. But 'impossible' means P = 0 in a context where the event can never occur. Here, every individual value has probability 0 yet some value will certainly be realized — the zero probability just means we can't meaningfully distinguish one exact value from its neighbors. You ask about intervals instead.

The resolution is that 'probability zero' and 'impossible' are not synonymous for continuous distributions. Uncountably many values share the unit interval [0,1]; each must receive probability 0 or they couldn't all sum to 1. The right question for a continuous RV is always about ranges: P(a ≤ X ≤ b) = ∫[a to b] f(x)dx for some density f(x) ≥ 0.