A probability table pairs every possible outcome of an event with its likelihood of occurring. Building one takes three steps: list all possible outcomes, calculate the probability of each, and organize the pairs into rows or columns that sum to exactly 1 (or 100%). The format works for everything from coin flips to survey data, and once the table is built, you can use it to answer questions about expected values and ranges of outcomes.
The Two Rules Every Probability Table Must Follow
Before building anything, you need to know what makes a probability table valid. There are only two rules:
- Every individual probability must be between 0 and 1. No outcome can have a negative probability or a probability greater than 1.
- All the probabilities must add up to exactly 1. This confirms you’ve accounted for every possible outcome. If your column sums to 0.98 or 1.03, something is off.
These rules apply whether you’re working with two outcomes or two hundred. They’re also the quickest way to check your finished table for errors.
Building a Table From Equally Likely Outcomes
The simplest probability tables come from situations where you can count every possible outcome. Here’s the process, using coin flips as an example. Suppose you flip two coins and want to track how many land on heads.
First, write out the full sample space. Two coins give you four equally likely results: HH, HT, TH, and TT. Next, define your variable. Here, X equals the number of heads, so X can be 0, 1, or 2. Now map each value to its outcomes: X = 0 matches TT (one outcome), X = 1 matches HT and TH (two outcomes), and X = 2 matches HH (one outcome). Finally, divide each count by the total number of outcomes (four) to get the probabilities.
The finished table looks like this:
- X = 0: P(X) = 0.25
- X = 1: P(X) = 0.50
- X = 2: P(X) = 0.25
Check: 0.25 + 0.50 + 0.25 = 1.00. The table is valid.
A Fuller Example: Sum of Two Dice
Rolling two dice produces 36 equally likely combinations (6 sides × 6 sides), but the sums range from 2 to 12. Some sums are far more common than others because multiple combinations produce them. For instance, only one combination makes a 2 (both dice show 1), but six combinations make a 7 (1+6, 2+5, 3+4, 4+3, 5+2, 6+1).
The complete probability table:
- Sum = 2: 1/36
- Sum = 3: 2/36
- Sum = 4: 3/36
- Sum = 5: 4/36
- Sum = 6: 5/36
- Sum = 7: 6/36
- Sum = 8: 5/36
- Sum = 9: 4/36
- Sum = 10: 3/36
- Sum = 11: 2/36
- Sum = 12: 1/36
Add all the numerators: 1+2+3+4+5+6+5+4+3+2+1 = 36. Divided by 36, that gives 1. The symmetry here is worth noticing. The probabilities mirror around the center value of 7, which is the most likely outcome.
Building a Table From Real-World Data
When outcomes aren’t equally likely, you’ll typically start with frequency data, meaning counts of how often each outcome appeared in a dataset. The U.S. Census, for example, recorded household sizes across the country. To turn those raw counts into a probability table, you divide each category’s count by the total number of observations.
Here’s what the resulting table looks like for household size based on 2010 Census data:
- 1 person: 0.267
- 2 people: 0.336
- 3 people: 0.158
- 4 people: 0.137
- 5 people: 0.063
- 6 people: 0.024
- 7+ people: 0.015
The steps are the same every time. Identify all categories, count how many observations fall into each one, then divide each count by the grand total. The result is a proportion that represents the probability of randomly selecting that outcome.
Adding a Cumulative Probability Column
A cumulative probability column answers the question “what’s the probability of getting this value or less?” You create it by adding a running total as you move down the table.
Using the household size example: the cumulative probability for 1-person households is just 0.267. For 2-person households, it’s 0.267 + 0.336 = 0.603. For 3-person households, 0.603 + 0.158 = 0.761. You continue this way until the final row, which should always equal 1.000.
This column is useful when you want to answer range-based questions. If someone asks, “What’s the probability a household has 4 or fewer people?” you can read the answer directly from the cumulative column instead of adding individual probabilities: 0.898. To find the probability of a range in the middle, subtract one cumulative value from another. The probability of a household having between 3 and 5 people is the cumulative value at 5 (0.961) minus the cumulative value at 2 (0.603), which equals 0.358.
Two-Way Probability Tables
When you’re tracking two variables at once (for example, gender and voting choice), you need a two-way table, sometimes called a contingency table. Instead of a single column of probabilities, you get a grid where the rows represent one variable and the columns represent another.
To build one from raw data, start with counts. Say you have data on 1,000 drivers showing whether each was DUI and whether they had an accident:
- DUI + Accident: 70
- DUI + No Accident: 130
- Non-DUI + Accident: 30
- Non-DUI + No Accident: 770
To convert this into a joint probability table, divide every cell by the grand total (1,000). So the probability of being both DUI and in an accident is 70/1000 = 0.07. The row totals give you the overall probability of DUI vs. non-DUI (called marginal probabilities), and the column totals give you the overall probability of accident vs. no accident. All four inner cells plus nothing else should sum to 1.
If you’re starting from reported percentages rather than raw counts, pick a convenient total number (10,000 works well) and multiply it by your known probabilities to reconstruct counts. From there, the process is the same: fill in cells, add row and column totals, and verify everything sums correctly.
Calculating the Expected Value
Once your table is built, one of the most useful things you can do with it is calculate the expected value, which is the long-run average outcome. Multiply each outcome by its probability, then add up all those products.
For the two-dice example, the expected value is: (2 × 1/36) + (3 × 2/36) + (4 × 3/36) + … + (12 × 1/36) = 7. That means if you rolled two dice thousands of times, the average sum would settle at 7. For the household size table, the same calculation gives you the average household size in the U.S.
This works for any probability table. The expected value doesn’t have to be a value that can actually occur. Rolling two dice can never produce a 3.5, but that could theoretically be the expected value of a different distribution. It represents the center of gravity of all your outcomes.
Making a Probability Table in Excel
If you’re working with a dataset rather than a textbook problem, Excel can handle the counting for you. The key function is FREQUENCY. Say your raw data sits in cells A1 through A100 and you’ve defined your category boundaries (called bins) in D1 through D5. Highlight a column range one cell longer than your bin range, type =FREQUENCY(A1:A100, D1:D5), and press Ctrl+Shift+Enter (since it’s an array formula). Excel fills in the count for each bin.
From there, divide each frequency by the total count to get probabilities. If your frequencies are in column E, create a new column with a formula like =E1/SUM(E:E) and copy it down. You now have a probability table. To add a cumulative column, make the first cell equal to the first probability, then set each subsequent cell to the one above it plus the current row’s probability.
For quick summary statistics from a frequency-based table, Excel’s SUMPRODUCT function is helpful. If your outcomes are in one range and your frequencies in another, =SUMPRODUCT(outcomes, frequencies)/SUM(frequencies) gives you the mean directly from the table without needing to expand the data back into individual observations.