Written and maintained by CASRAI Editorial Board
Published · Last updated
On this page
A probability table lists every possible value of a variable next to its probability, and it is valid only if each probability is between 0 and 1 and all probabilities add up to exactly 1. To create one, define the variable, list every outcome once, assign a probability to each (from counts or from a model), then check both rules before you publish it.
Purpose of a probability table
A probability table is the tabular form of a discrete probability distribution. It lets a reader see the whole distribution at once, read off the probability of any single value, and add rows to get the probability of a range. In statistics texts the function behind the table is called a probability mass function; see the probability mass function formula for the notation and what a random variable is for the variable the table describes.
Two sources agree on the rules the table must obey. OpenStax Introductory Statistics 2e (section 4.1) states that “each probability is between zero and one, inclusive” and that “the sum of the probabilities is one”. The NIST/SEMATECH e-Handbook (section 1.3.6.1) says discrete probability functions are referred to as probability mass functions and that the sum of p(x) over all possible values of x is 1.
Steps to create a probability table
- Define the variable. Write X in words, including the unit and the time frame, for example “the number of revision rounds a manuscript needs”.
- List every possible value once. The values must not overlap and must cover everything that can happen, including 0 if 0 is possible.
- Get a probability for each value. From data, divide each count by the total count n. From a model (a fair die, a binomial model), compute each probability from the model.
- Check each probability. Every P(x) must be at least 0 and at most 1.
- Check the sum. Add the column. It must equal 1. If it does not, a category is missing, double counted, or a division is wrong.
- Add a cumulative column if readers need ranges. The running total P(X ≤ x) ends at 1 in the last row.
- Label the table. State the variable, the sample size or model, and where the probabilities came from.
Worked example from frequency counts
Illustrative example with invented counts, not real data. Suppose a lab tracks how many revision rounds each of 40 manuscripts needed. Let X be the number of revision rounds. The counts below are made up to show the arithmetic.
| x (rounds) | Count | P(x) = count / 40 | P(X ≤ x) |
|---|---|---|---|
| 0 | 6 | 0.15 | 0.15 |
| 1 | 14 | 0.35 | 0.50 |
| 2 | 12 | 0.30 | 0.80 |
| 3 | 6 | 0.15 | 0.95 |
| 4 | 2 | 0.05 | 1.00 |
| Total | 40 | 1.00 |
Check: every P(x) lies between 0 and 1, and 0.15 + 0.35 + 0.30 + 0.15 + 0.05 = 1.00, so the table is a valid probability distribution. The probability of needing at most two rounds is the cumulative value in the x = 2 row, 0.80. Because this table is built from observed counts, the probabilities are empirical; they describe the sample, not a model. If you need the mean of the distribution, see the expected value formula.
For the OpenStax worked table (Example 4.1, the number of times per week a newborn baby’s crying wakes its mother after midnight, with counts out of 50), see the OpenStax section listed under Sources.
Joint probability table with marginals
A table of one variable is a distribution. A table that cross-classifies two variables is a joint probability table: each cell is the count in that cell divided by the grand total, and the row and column totals are the marginal probabilities. Illustrative counts, 60 observations:
| Outcome yes | Outcome no | Row total (marginal) | |
|---|---|---|---|
| Group A | 18/60 = 0.30 | 12/60 = 0.20 | 0.50 |
| Group B | 6/60 = 0.10 | 24/60 = 0.40 | 0.50 |
| Column total (marginal) | 0.40 | 0.60 | 1.00 |
The same two checks apply: the four cell probabilities sum to 1, and so do the row totals and the column totals. See the joint probability formula for how a cell is used, and how to calculate marginal probability for the totals.
Common mistakes
- Probabilities that do not sum to 1. A missing category, a double-counted category or a wrong denominator. Recompute from the counts rather than adjusting a value to force the total.
- A probability above 1 or below 0. This usually means a count was divided by the wrong total, or a frequency was entered as a probability.
- Overlapping categories. If an observation can fall in two rows, the rows are not mutually exclusive and the column cannot be a distribution.
- Rounding drift. Three equally likely outcomes each have probability 1/3. Rounded to two decimals each becomes 0.33, and the column sums to 0.99. Keep fractions or more decimals in the working, or state in a table note that values are rounded and may not sum exactly to 1.
- Mixing a distribution with a cross-tab. A one-variable table sums to 1 down its single probability column. A two-way joint table sums to 1 over all cells, not down each column; dividing each column by its own total gives conditional probabilities, a different table.
Checklist
- The variable is defined in words, with its unit.
- Every possible value appears once, with no overlaps and no gaps.
- Each P(x) is between 0 and 1.
- The probabilities sum to 1 (state the rounding if they do not display as exactly 1).
- The source of the probabilities (observed counts with n, or a named model) is stated.
- If a cumulative column is shown, it ends at 1.
Presenting the table in a manuscript (writing guidance)
This is editorial guidance, not a sourced rule. Number the table and give it a caption that names the variable, for example “Distribution of revision rounds (n = 40)”. Use a column header such as P(x), report probabilities to a consistent number of decimals, say whether they are empirical or theoretical, and follow your target journal’s instructions for table formatting.
Building it in Excel
If your probabilities live in a spreadsheet, the PROB function can read them back; see how to calculate probability in Excel.
FAQ
What is a probability table?
A table that pairs each possible value of a variable with its probability. A valid one has every probability between 0 and 1 and a total of exactly 1.
How do I know my probability table is correct?
Run the two checks: each probability is between 0 and 1, and the column sums to 1. A table that fails either check is not a probability distribution.
Can I build a probability table from raw data?
Yes. Count how often each value occurs, divide each count by the total number of observations, and check that the results sum to 1.
What is the difference between a probability table and a frequency table?
A frequency table shows counts. A probability table shows each count divided by the total, so the column sums to 1 instead of to n.
Sources
- OpenStax, Introductory Statistics 2e, section 4.1, “Probability Distribution Function (PDF) for a Discrete Random Variable”: https://openstax.org/books/introductory-statistics-2e/pages/4-1-probability-distribution-function-pdf-for-a-discrete-random-variable (checked 2026-10-29).
- NIST/SEMATECH e-Handbook of Statistical Methods, section 1.3.6.1, “What is a Probability Distribution”: https://www.itl.nist.gov/div898/handbook/eda/section3/eda361.htm (checked 2026-10-29).







