Probability assigns a number between 0 and 1 to an event. For a finite uniform sample space, it is the proportion of outcomes belonging to that event. The main challenge is often building the correct sample space rather than simplifying the final fraction.
Uniformity must be justified
The formula requires equally likely individual outcomes in sample space S. A fair die supports that assumption for its six faces; an unequal-sector spinner does not support it for its labeled regions.
If two fair dice are rolled independently, the 36 ordered face pairs are equally likely. Their sums are not equally likely: a sum of 7 occurs in six pairs, while a sum of 2 occurs only in Counting sum labels as though each had equal probability would produce a false model.
The sample space should preserve distinctions that the random process makes. Even if the final event ignores die order, the underlying ordered pairs remain a useful fair basis for counting.
Example: use a complement to simplify counting
Two fair dice are rolled independently. Find the probability that the sum is at least 4. Its complement is a sum less than 4, consisting of and
There are only 3 excluded outcomes, so the desired probability is Directly counting every favorable pair would be longer but must give the same result.
“At least 4” includes 4. Its complement is strictly less than 4, not at most 4. Careful endpoints matter in probability just as they do in inequalities.
Complements require no independence assumption
For every event A, This follows because A and its complement partition the entire sample space. It does not require repeated trials or independent events.
If a device fails with probability 0.03, its probability of not failing is 0.97 under the same specified conditions. But calculating the chance that two devices both avoid failure requires information about their joint behavior. The complement rule by itself does not justify multiplying 0.97 by itself.
Keep single-event complement reasoning separate from additional assumptions about multiple events.
Example: geometric probability
A point is selected uniformly at random from a rectangle 10 units by 6 units. A triangular region inside it has base 4 and perpendicular height 3. Because selection is uniform by area, the probability of landing in the triangle is its area divided by the rectangle's area:
The complement probability is The ratio of perimeters would be irrelevant because the point is chosen from the interior area, not along the boundary.
Uniform selection along a line segment instead uses a length ratio, and uniform selection from a three-dimensional solid uses a volume ratio. Match the measure to the random process.
Expected proportions are not guaranteed counts
If a probability is 0.2, then in 100 comparable trials the expected count is 20. The actual count need not be exactly 20. Probability describes the distribution of possible outcomes, not a promise of exact short-run balance.
Do not confuse an impossible event with an unlikely one. In a finite uniform sample space, zero favorable outcomes means probability zero. In continuous models, individual points can have probability zero even though a point must be selected; interval or region probabilities remain the useful quantities.
Try it yourself
Two independent fair dice are rolled. Find the probability their sum is not 7.
Answer: Six ordered pairs sum to 7: and Thus
What to remember
Build an appropriate uniform sample space, choose counts or geometric measures that match the experiment, and use complements to simplify difficult favorable-event counts.