StatSolver Learning Center

How to choose a discrete probability distribution

The most important step in a probability calculation happens before you enter a number: identify what the random variable counts, what is fixed, and which assumptions describe the process.

Written and maintained by Mustafa Akilli · Updated September 9, 2026

Discrete probability models describe outcomes that can be counted or listed. A good model does not just produce a convenient answer; it gives the answer a clear meaning. Use the guide below to narrow the choice, then open the matching StatSolver calculator to see the calculation and interpretation together.

Quick decision rule: one yes/no trial suggests Bernoulli; a fixed number of repeated yes/no trials suggests Binomial; waiting for the first success suggests Geometric; counting events across an interval suggests Poisson; and observing two discrete variables together suggests a Joint distribution.

Compare the five models

Bernoulli: one binary outcome

Use this when one experiment has two possible outcomes. Code the outcome called success as 1 with probability p and the other as 0 with probability 1 − p. The mean is p and the variance is p(1 − p). Open the Bernoulli calculator.

Binomial: count successes in n trials

Use this when n is fixed, each trial has two outcomes, p stays constant, and trials are independent. The variable counts successes, not trials. Open the Binomial calculator.

Geometric: wait for the first success

Use this when trials continue until the first success. StatSolver uses the convention that X is the trial number of the first success, so X starts at 1 and E[X] = 1/p. Open the Geometric calculator.

Poisson: count events in an interval

Use this when the question counts events in time, space, area, or volume and the average rate is represented by λ. Check that the rate and interval use the same units. Open the Poisson calculator.

Joint distribution: two variables together

Use a joint table when outcomes for X and Y are observed at the same time. The table supports marginals, conditional probabilities, covariance, correlation, independence, and the distribution of X + Y. Open the Joint calculator.

A reliable modeling checklist

Define the random variable in words before choosing a formula.
Identify whether the number of trials or the exposure interval is fixed.
Check whether outcomes are binary, counts, or paired observations.
Write down the assumptions: constant probability or rate, independence, and any sampling limits.
Choose units carefully. A Poisson rate of four events per hour is λ = 2 for thirty minutes.
After calculating, ask whether the result answers “exactly,” “at least,” “at most,” or “by which trial.”

Worked comparison: fixed trials versus waiting time

Suppose an independent trial succeeds with probability 0.2. If the question asks for the number of successes in 10 trials, it is binomial: the random variable can take values from 0 through 10, with mean 10 × 0.2 = 2. If the question asks for the trial on which the first success occurs, it is geometric: P(X = 3) = (0.8)²(0.2) = 0.128, and the expected trial number is 1 / 0.2 = 5.

The same success probability appears in both examples, but the random variable and stopping rule are different. That distinction is more important than the arithmetic.

Common modeling mistakes

Entering 25 instead of 0.25 when a probability is given as 25%.
Calling observations independent simply because they are recorded separately.
Using a binomial model when the success probability changes from trial to trial.
Using a Poisson model without matching λ to the interval in the question.
Interpreting correlation as proof of causation or as a complete independence test.
Confusing “the first success occurs on trial 3” with “at least one success occurs in three trials.”

How StatSolver supports the calculation

The calculator pages state their assumptions, reject invalid numeric inputs, show intermediate values where appropriate, and explain how to read the output. The project is tested with regression cases and edge cases, but the results remain educational aids: verify important work against a textbook, instructor, or trusted reference.

For implementation scope, testing practices, and references used during maintenance, see Methodology & accuracy. For questions or corrections, contact the site author.