Negative Binomial Distribution Calculator

JJ Ben-Joseph headshot Editorial review by: JJ Ben-Joseph

Understanding Negative Binomial Failures Before Successes

The negative binomial distribution counts failures accumulated before a chosen number of successes in independent, repeated trials. For example, when a coin is flipped until three heads occur, the number of tails seen before that third head follows this version of the negative binomial distribution. Since it records failures until a success target is reached, it is often right-skewed when success is unlikely. The geometric distribution is the special case that stops after one success.

For this failures-before-successes negative binomial model, the probability of observing exactly $k$ failures before the $r$ -th success is given by

P_{X} (k) = \frac{(k + r - 1)!}{k! (r - 1)!} {(1 - p)}^{k} p^{r}

Here, $p$ is the success probability on every trial. The binomial coefficient $\frac{(k + r - 1)!}{k! (r - 1)!}$ counts possible arrangements of the failures and successes. For this parameterization, the mean number of failures is

μ = r \frac{1 - p}{p}

and the variance is

σ^{2} = r \frac{1 - p}{p^{2}}

These negative binomial relationships show why lower success probabilities produce both more expected failures and substantially greater spread.

Introduction: Why the Negative Binomial Distribution Matters

Negative binomial counts arise whenever independent attempts continue until a fixed success goal is met. A reliability study may count defective components before a specified number of working components is found; a sales team may count rejections before closing a target number of sales. In each case, the model is appropriate only when the chance of success is treated as constant from trial to trial. Its long right tail represents the possibility that many failures occur before the final required success.

The shape of this negative binomial distribution is controlled by both inputs. A large success probability concentrates probability near zero failures, while a small probability creates a longer tail. For a fixed $r = 3$ , the following values illustrate how expected failures and variance increase as success becomes less likely.

Success Probability	Mean Failures	Variance
0.9	0.33	0.37
0.7	1.29	1.84
0.5	3	6
0.3	7	23.3

How to Use the Negative Binomial Calculator

Use this negative binomial calculator by entering the required number of successes $r$ , the per-trial success probability $p$ as a decimal from 0 to 1, and the observed number of failures $k$ . Select Compute to obtain the probability at exactly $k$ , the cumulative probability through $k$ , the survival probability above $k$ , and the mean and variance. An optional quantile probability asks for the smallest failure count whose cumulative probability reaches that probability. The cumulative and quantile calculations sum probabilities directly, so large counts can take longer to evaluate. If clipboard access is available, the displayed summary can be copied after calculation.

This calculator is intended for count processes that end at a success target, not processes with a preassigned trial total. A factory, for example, might record defects before the tenth acceptable item. A call center might record refusals before three positive responses. The ordinary binomial distribution instead counts successes in a fixed number of trials, so it answers a different question when the number of attempts is allowed to vary.

Negative Binomial Calculations Behind the Scenes

The negative binomial calculator evaluates the coefficient in the probability formula with an iterative combination routine. It then multiplies that coefficient by the failure term and success term to obtain the probability mass at the selected failure count. For the cumulative distribution, it adds the mass values from zero failures through the entered $k$ . The survival result is one minus that cumulative value. The optional quantile search adds mass values in order until the requested cumulative probability is reached.

Negative binomial notation can differ between sources. Some references count successes before a fixed number of failures instead. This calculator uses the convention shown in its form and formula: $r$ is a whole-number target of successes, and the random variable is the number of failures before the $r$ -th success. Confirm that convention before comparing results with a textbook or software package.

Practical Considerations for Negative Binomial Modeling

A negative binomial model assumes independent trials with the same success probability. In actual data, learning, fatigue, changing conditions, or clustering can make that probability vary, which can weaken the model fit. Even so, this distribution is often useful for over-dispersed counts because its variance can exceed its mean. That flexibility distinguishes it from a Poisson model, whose mean and variance are equal under the standard model.

When $p$ is high or the required success count $r$ is small, failures before completion are concentrated near zero. With $r = 1$ , this is the geometric distribution. Low success probabilities and larger values of $r$ yield a longer failure-count tail. Compare observed data with the model’s expected failure count and variance before treating it as a close description of a real process.

Conclusion: Interpreting Negative Binomial Results

The negative binomial distribution gives a direct way to study failures encountered on the way to a specified number of successes. This calculator reports the probability of one exact failure count, the probability of reaching the success target with no more than that many failures, the complementary survival probability, and summary measures of the failure count. Varying $p$ or $r$ makes clear how the expected number and variability of failures change.

Alternative Negative Binomial Parameterizations

Some negative binomial references use a mean $μ$ and dispersion parameter $α$ rather than the calculator’s success target $r$ and probability $p$ . Under the convention used here, $μ = \frac{r (1 - p)}{p}$ and $α = r$ . In that notation, the variance is $μ + \frac{μ^{2}}{α}$ , which shows explicitly how it can be greater than the mean.

The Negative Binomial Poisson-Gamma Connection

The negative binomial distribution can also be understood as a Poisson count with a rate that varies according to a gamma distribution. After that varying rate is averaged out, the marginal count follows a negative binomial form. This connection helps explain why negative binomial models are commonly used when different observations have different underlying event rates, such as counts from locations with unequal exposure. It also appears naturally in Bayesian work that places a gamma prior on a Poisson rate.

Estimating Negative Binomial Parameters

Negative binomial parameters can be estimated from data with maximum likelihood or method-of-moments methods. For the latter, calculate the sample mean $\bar{x}$ and sample variance $s^{2}$ , then use $μ = \bar{x}$ and $s^{2} = \bar{x} + \frac{{\bar{x}}^{2}}{r}$ . This gives $r = \frac{{\bar{x}}^{2}}{s^{2} - \bar{x}}$ and $p = \frac{\bar{x}}{s^{2}}$ , provided the variance exceeds the mean.

Simulation and a Negative Binomial Practical Example

A negative binomial failure count can be simulated by generating the failures before each required success and adding them. For example, an email campaign seeking five positive replies could model each contact with a fixed probability of a positive reply; the simulated result is the number of rejections before the fifth reply. Repeated simulations reveal the skew and variability that a single mean value can hide, helping users set expectations for how many attempts may be needed.

Limitations and Assumptions: Negative Binomial Pitfalls

When applying a negative binomial distribution, check whether independent trials and a constant success probability are defensible. A learning effect may raise success chances over time, while unobserved conditions may cause failures to cluster. Parameter interpretation also matters: although some statistical formulations permit non-integer shape parameters, this calculator requires $r$ to be a whole number because it represents a literal number of required successes. Choose inputs that describe the process being modeled rather than mixing estimates from incompatible conditions.

Formula: how negative binomial probabilities are built

This calculator’s result is determined by the required successes $r$ , the decimal success probability $p$ , and the non-negative whole-number failure count $k$ . Enter $p$ as a fraction between 0 and 1, not as a percentage, and use counts for both $r$ and $k$ . Double-check which outcome is called a success before entering the probability, since reversing that definition changes every reported probability and summary value.

Arcade Mini-Game: Negative Binomial Distribution Calculator Calibration Run

Use this quick arcade run to practice separating useful scenario inputs from common planning mistakes before you rely on the calculator output.

Score: 0 Timer: 30s Best: 0

Start the game, then use your pointer or arrow keys to catch useful inputs and avoid bad assumptions.

Negative Binomial Distribution Calculator

Understanding Negative Binomial Failures Before Successes

Introduction: Why the Negative Binomial Distribution Matters

How to Use the Negative Binomial Calculator

Negative Binomial Calculations Behind the Scenes

Practical Considerations for Negative Binomial Modeling

Conclusion: Interpreting Negative Binomial Results

Alternative Negative Binomial Parameterizations

The Negative Binomial Poisson-Gamma Connection

Estimating Negative Binomial Parameters

Simulation and a Negative Binomial Practical Example

Limitations and Assumptions: Negative Binomial Pitfalls

Formula: how negative binomial probabilities are built

Embed this calculator

Related Calculators

Binomial Distribution Calculator for Exact, Cumulative, and Tail Probabilities

Geometric Distribution Calculator - First-Success Trial Probabilities

Pareto Distribution Calculator for PDF, CDF, Survival Probability, Quantiles, Mean, and Variance

Gamma Distribution Calculator | Density, Cumulative Probability & Quantiles

Student's t Distribution Calculator - PDF, CDF, Quantiles, and Critical Values

Quantum Annealing Time-to-Solution Calculator | Runs and Runtime Estimate