Define the population, event, or assumptions first
The cases must be mutually exclusive and cover everything. If your categories overlap, such as people who take the bus and people who take the train when some take both, the weighted sum double counts those people and can produce an impossible answer above 1.
What each quantity represents in Law of total probability
Name the experiment or data-generating process, distinguish sample information from population parameters, and state the event or estimate in words. Rebuild missing background in the probability subject guide, then use statistics practice to retrieve the relationship without being told its name.
Interpret a complete Law of total probability calculation
The numerical line is only the middle of the answer; connect it back to the event, data, or uncertainty:
Compute without losing the statistical question
- Name the unknown, and the unit the answer has to come out in.
- Match the symbols to your values. The events $A_i$ form a partition of the sample space.
- Check the conditions before substituting. The $A_i$ must be disjoint, exhaustive, and have positive probability when conditioned upon.
- Substitute, keep exact values to the last line, then test the sign, size, and unit against a rough estimate - The check that catches most probability slips.
Check the range, direction, and interpretation
After using Law of total probability, repeat the calculation with units attached and test the result against a boundary case or a quick numerical estimate.
An inference this formula does not justify
What goes wrong: Adding the conditional probabilities without weighting them by the case probabilities.
Why it fails: With group shares of 0.4 and 0.6 and event rates of 0.2 and 0.5, adding the rates gives 0.7. The weighted total is 0.38, and the unweighted sum can even climb past 1 with enough cases.
Repair: Multiply each conditional probability by the probability of its own case before adding, and check that the case probabilities themselves add to 1.
Questions about interpreting Law of total probability
How do I check that my split is a genuine partition?
Confirm the case probabilities add to exactly 1 and that no outcome belongs to two cases. Both conditions have to hold, since covering everything is not the same as not overlapping.
Can I use more than two cases?
Yes, any number, including infinitely many as long as the total probability still comes to 1. Each case simply adds another weighted term to the sum.
How does this relate to a tree diagram?
Each branch of the tree is one case, multiplying along a branch gives one term, and adding the branches that end in the event gives this formula exactly.
Where does this appear inside Bayes' theorem?
It supplies the denominator. When the overall chance of the evidence is not given directly, this is how you assemble it from the individual cases.
Change one input, predict whether the output should rise, fall, or stay fixed, and only then recalculate.