Probability Reasoning: Count Outcomes Without Guessing
Probability reasoning starts by defining equally weighted outcomes or explicit weights, then counting only the outcomes that satisfy the event.
How to solve a probability problem
Name the random experiment, the event of interest, and what information is already known. Build a complete sample space with the correct weights. Count favorable outcomes only after checking whether trials are independent, whether outcomes can repeat, and whether order matters. For conditional probability, remove outcomes that conflict with the new information before recalculating.
Probability compares the weight of successful outcomes with all possible outcomes after the event and known information are defined.
Key relationships
- A sample space contains all outcomes allowed by the experiment.
- Replacement can preserve the probability of a later draw.
- Conditional information removes outcomes that no longer fit what is known.
Define the experiment and event
A probability question needs a repeatable experiment and a precise event. Drawing two cards without replacement differs from drawing one card, replacing it, and drawing again. At least one six differs from exactly one six. Rewrite the event before using a formula.
List what is known at the time of the question. Information revealed by the process can change the relevant sample space, while information chosen because of the outcome can introduce selection effects. The wording of how evidence was obtained matters.
Build a sample space with honest weights
A list of outcomes can be counted directly only when its entries are equally likely. Two coin flips have four equally likely ordered outcomes: HH, HT, TH, and TT. The summary outcomes zero heads, one head, and two heads are not equally likely because the middle case has two paths.
Use a tree when events happen in stages, a grid when two choices combine, and combinations when order does not matter. Label branch probabilities so unequal choices remain visible rather than disappearing into a count.
Check independence and replacement
Independent events do not change one another's probabilities. Replacing a drawn object can preserve independence across draws. Without replacement, the contents and denominator change, so the second probability depends on the first result.
Multiply along a path and add across mutually exclusive successful paths. Before multiplying, ask whether the branch probability has changed. Before adding, check that the paths cannot occur together in the same trial.
Use complements for at-least-one events
At least one success often has many overlapping paths. Its complement, no successes, may have one simple path. Calculate the probability of no success and subtract it from one, provided the complement covers every way the target event can fail.
For repeated independent trials with success probability p, no successes in n trials has probability (1-p) raised to n. If the trials are dependent or p changes, calculate the failure probability along the actual branches instead.
Update a conditional probability
When new information is given, restrict the original sample space to outcomes compatible with it and preserve their relative weights. Then ask what share of that reduced space also satisfies the target event. Do not keep impossible outcomes in the denominator.
A useful check is to simulate or enumerate a small case. The check does not replace reasoning, but disagreement between the model and an exact small enumeration often reveals a missing path, duplicated outcome, or false independence assumption.
Count exactly one six without double-counting
Two fair dice produce 36 equally likely ordered outcomes. The event exactly one six has two non-overlapping forms: the first die is six and the second is not, or the second is six and the first is not.
| Event part | Matching outcomes | Count |
|---|---|---|
| First die is 6, second is not | (6,1) through (6,5) | 5 |
| Second die is 6, first is not | (1,6) through (5,6) | 5 |
| Both dice are 6 | (6,6), excluded by exactly one | 0 |
| Total | Ten unique ordered outcomes | 10 of 36 |
The probability is 10/36, which simplifies to 5/18. At least one six includes the same ten outcomes plus (6,6), giving 11/36. The complement method agrees: one minus the probability that neither die is six equals 1 - 25/36 = 11/36.
Audit a probability count
- Confirm that the listed elementary outcomes have equal weight.
- Translate exactly, at least, and at most before counting.
- Check that successful cases do not overlap or get counted twice.
- Solve by a second method when a complement or small enumeration is available.
Practice probability without confusing the score
The reasoning test with answer explanations includes probability as one skill among several and does not claim to estimate IQ.
Use the general logic-solving method to keep facts, deductions, and temporary assumptions separate.