All primers · Mathematics
Conditional probability in 10 minutes
Reading two-way tables, and why a positive test can still mean you are probably fine.
Use counts, not formulas
The fastest way through probability passages is to imagine a concrete group, say 1,000 or 10,000 people, and fill a two-way table with counts. Then every probability is a count divided by another count.
P(A given B) means: among the cases where B happened, what share also have A? So the denominator is always the B column or row, never the whole group.
The base-rate trap
If a condition is rare, even an accurate test gives many false positives, because they come from the large healthy group. Always check how common the condition is before trusting "95% accurate".
Check yourself
Of 200 people who test positive, 50 have the condition. P(condition given positive)?
Show answer
50 / 200 = 25%.
Which number do you divide by for P(A given B)?
Show answer
The number of cases where B happened.