Dependent events & trees
- Dependent events mean the first thing changes the chance of the second.
- The usual exam version is without replacement, or a tree where the second stage depends on the first.
The tree remembers what already happened
The general rule is . Multiply along a real path, using the probability that belongs to that path.
- With replacement: the bag resets. The second probability stays the same.
- Without replacement: one item has gone. The second denominator drops.
- If a coin, die, or first choice picks the bag, keep that first probability on every path.
Worked example
First rung: two draws without replacement
A bag has 3 red counters and 2 blue counters. Two counters are drawn without replacement. Find the probability of two reds.
- First red: .
- One red has gone, so second red is .
- Multiply along the path:.
Answer
Exam version: one tree, two denominators
A biased coin has . A bag has 4 red and 5 blue marbles. If H occurs, one marble is chosen. If T occurs, two marbles are chosen without replacement. (9709/52 M/J 2024 Q2)
- At least one red after H: .
- At least one red after T: .
- Add the two working paths: .
Answer
For given that on a tree, build the denominator from paths.
- List every path where the condition happens.
- Add those paths. That is the denominator.
- Keep only the paths that also hit the target. That is the numerator.
- Divide numerator by denominator.
In a tree question, is not one branch over one branch. It is target-and-condition paths over all condition paths.
Common mistake
Do not reuse the first denominator after a selection without replacement. From 4 red and 5 blue, if a blue has gone, the next blue chance is , not .
The Bayes shape, without the scary name
- The numerator is one route, or a few routes, that give both the target and the condition.
- The denominator is every route that gives the condition.
- This is why mark schemes reward a correct denominator even before the final answer.