The denominator tells you what a probability is about. In a two-way table, a joint probability uses the whole table to describe one cell, a marginal probability uses the whole table to describe a row or column total, and a conditional probability uses only the selected row or column. Read the same table in those three ways and the difference becomes easier to see.
Start with one table and ask what population you are counting
Imagine 200 observations classified by two variables: whether each case is in group B and whether it is in group S. The table below uses the published counts in MacEwan University’s Introduction to Applied Statistics example. The remaining cells follow by subtraction.
As an Amazon Associate I earn from qualifying purchases.
| S | Not S | Total | |
|---|---|---|---|
| B | 10 | 30 | 40 |
| Not B | 20 | 140 | 160 |
| Total | 30 | 170 | 200 |
Each count can answer a different kind of question, depending on the denominator. A cell is a count of cases meeting two conditions. A margin is a total for one variable, ignoring the other. A conditional probability narrows the population to cases meeting a specified condition.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteJoint probability: read one cell
A joint probability asks whether two events happen together: “B and S.” It is written P(B ∩ S) or P(B, S). Find the cell where row B and column S meet, then divide its count by the grand total: 10/200 = 0.05, or 5%.
#1 Best Overall
In general, for events A and B, the cell-over-total calculation is P(A ∩ B) = count(A and B) / total count. The denominator is the full set of observations because the question has not restricted the population to either event.
Marginal probability: read a margin
A marginal probability asks about one variable while disregarding the other. To find the probability of B, use the B row total, 40, over all 200 cases: P(B) = 40/200 = 0.20. To find the probability of S, use its column total: P(S) = 30/200 = 0.15.
These are marginal probabilities because their counts sit on the table’s margins. You can also get a marginal probability by adding the joint probabilities across the values of the variable you are ignoring. For example, P(B) = P(B ∩ S) + P(B ∩ not S). This is why a row or column total is not itself a joint probability: it combines several cells.
Recommended Free Tools
Rank #2
Conditional probability: lock a slice and renormalize
A conditional probability asks about one event within a group already known to satisfy another. The notation P(A | B) means “the probability of A, given B.” Keep only cases in which B is true; the other cases are outside the reference population.
Given B: use the B row total
Among the 40 B cases, 10 are also S. So P(S | B) = 10/40 = 0.25. The denominator is 40, not 200, because the question is about the B subgroup.
Given S: use the S column total
Among the 30 S cases, 10 are also B. So P(B | S) = 10/30 ≈ 0.333. Here the denominator is 30 because the reference population is the S subgroup.
These probabilities differ because they describe different restricted populations. P(S | B) asks what share of B cases are S; P(B | S) asks what share of S cases are B. The shared cell is the same 10 cases, but each percentage compares those cases with a different group.
Why P(A | B) is not P(B | A)
The order around the vertical bar matters. In P(A | B), B identifies the reference population, and A is the outcome being counted within it. In P(B | A), A identifies the population instead. The denominators are P(B) and P(A), respectively, so the answers usually differ.
A useful way to keep the direction straight is to say the notation out loud: “A among B.” The event after the bar is the condition—the group you keep. This also explains why reversing a conditional is not a harmless rearrangement: it changes the question.
Rank #4
Connect the three readings with the product rule and Bayes’ theorem
A conditional probability can be multiplied by the probability of its reference group to recover the joint probability:
P(A ∩ B) = P(A | B)P(B) = P(B | A)P(A)
Both expressions describe the same overlap. The first starts with the B population and takes the share that is also A; the second starts with A and takes the share that is also B.
Rearranging this relationship gives Bayes’ theorem:
Best Value
P(A | B) = P(B | A)P(A) / P(B), provided P(B) > 0.
Bayes’ theorem is useful when the conditional in the direction you need is not the one you know. For instance, you might know the probability of evidence B among cases where A is true, but want the probability of A among cases where B is observed. Identify the prior P(A), the likelihood P(B | A), and the evidence probability P(B); then substitute them into the formula. Penn State’s STAT 414 lesson on Bayes’ theorem describes this as finding a conditional when the reverse conditional is known.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Use the denominator to identify the probability type
Before calculating, ask: “What is my reference population?” The denominator usually settles the question.
- Grand total: a cell count over the total is a joint probability; a row or column total over the total is a marginal probability.
- Conditioning-group total: a cell count over the row or column specified after “given,” “among,” or “of those who” is a conditional probability.
- Changing condition: if the condition changes, the denominator changes. That is why the two conditional directions can have different values.
Colorado State University’s STAT 400 module likewise emphasizes that the denominator determines which probability a percentage represents.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteCheck your reading with a quick practice question
Using the table above, what is the probability that a randomly selected case is not B? The answer is 160/200 = 0.80: it is a marginal probability because it uses the Not B row total and the grand total. What is the probability of not B among S cases? That is 20/30 ≈ 0.667: it is conditional because the denominator is the S column total.
As a final check, within any fixed condition, the conditional probabilities for all possible outcomes of the other variable must add to 1. Among B cases, for example, P(S | B) + P(not S | B) = 10/40 + 30/40 = 1. If they do not, check whether you included every outcome in the slice and used its full total as the denominator.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
