Binomial Counts and Geometric Waiting Times as Dual Views

Two Questions About the Same Trials

Imagine repeating a process that can produce either a success or a failure: a component passes a test, a customer responds, or a trial produces a specified outcome. One question asks, how many successes occur in the first $n$ trials? Another asks, does the first success happen by trial $n$? These questions describe the same sequence from two different viewpoints.

A binomial random variable counts successes across a fixed number of trials. A geometric random variable records the trial number of the first success. The key connection is that “a first success by trial $n$” means there is at least one success among the first $n$ trials. Its complement is that all $n$ trials are failures.

Definition: Suppose trials are independent, each has the same success probability $p$, and $0<p\leq 1$. Let $T$ be the trial number on which the first success occurs. For a fixed positive integer $n$, let $X_n$ be the number of successes in the first $n$ trials. Then $T$ has a geometric model and $X_n\sim\operatorname{Bin}(n,p)$. The events $T\leq n$ and $X_n\geq 1$ are the same event.

The equality of events gives the central relationship:

P(T\leq n)=P(X_n\geq 1)=1-P(X_n=0)=1-(1-p)^n.

The notation matters. $T\leq n$ is a cumulative waiting-time event: the first success occurs on any of trials $1,2,\ldots,n$. It is not the event that the first success occurs exactly on trial $n$. Likewise, $X_n\geq1$ allows any positive number of successes, not just one.

Why the Relationship Holds

The binomial model applies because the number of trials is fixed at $n$, each trial has the same success probability $p$, and the outcomes of different trials are independent. In particular, the probability of failure on any one trial is $1-p$. Independence means the probability of getting failures on all $n$ trials is the product of those probabilities.

$$ P(X_n=0)=(1-p)^n. $$

If there are zero successes in the first $n$ trials, then the first success has not yet occurred, so $T>n$. Conversely, if $T>n$, every one of those first $n$ trials must have been a failure, so $X_n=0$. Thus $T>n$ and $X_n=0$ are equivalent events. Taking the complement of either event gives the result.

Result: Under the independent, constant-$p$ trial model, the probability of at least one success in the first $n$ trials equals the probability that the first success occurs by trial $n$: $$ P(T\leq n)=1-P(X_n=0)=1-(1-p)^n. $$

The same result can also be seen by dividing the event $T\leq n$ into the possible locations of the first success. These events are disjoint: the first success cannot occur on two different trials. In a geometric model, $P(T=k)=(1-p)^{k-1}p$, because the first $k-1$ trials must fail and trial $k$ must succeed. Therefore,

\begin{aligned} P(T\leq n) &=\sum_{k=1}^{n}P(T=k)\\ &=\sum_{k=1}^{n}(1-p)^{k-1}p\\ &=p\frac{1-(1-p)^n}{1-(1-p)}\\ &=1-(1-p)^n. \end{aligned}

The denominator in the finite geometric sum is $p$, so the cancellation is valid for $p>0$. This derivation and the binomial-complement derivation reach the same probability by organizing the event differently: one sums over the trial of the first success, while the other counts whether there is at least one success in a fixed block of trials.

Worked Example: At Least One Success in Five Trials

Suppose a hypothetical process has independent trials, each with success probability $p=0.20$. Let $T$ be the trial number of the first success, and let $X_5$ count successes in the first five trials. Find the probability that a success occurs by trial 5.

Check the model. For this hypothetical example, assume each trial has only two relevant outcomes, the success probability remains $0.20$, and trials are independent. A fixed block of five trials therefore has the binomial model $X_5\sim\operatorname{Bin}(5,0.20)$, and the waiting time has a geometric model.

A success by trial 5 means $T\leq5$, equivalently $X_5\geq1$. It is simpler to calculate the complement: zero successes in five trials. Each trial fails with probability $1-0.20=0.80$, so

\begin{aligned} P(T\leq5) &=1-P(X_5=0)\\ &=1-(1-0.20)^5\\ &=1-(0.80)^5\\ &=1-0.32768\\ &=0.67232. \end{aligned}

Thus, the probability of at least one success in the first five trials is about $0.6723$, or $67.23\%$. In context, under the stated hypothetical model, there is a $67.23\%$ chance that the first success occurs on one of trials 1 through 5. The calculation is not the probability of exactly one success; two or more successes also satisfy the event.

Reading the Two Views Correctly

The two random variables answer different kinds of questions, even though some events correspond exactly. $X_n$ is a count for a specified block of $n$ trials. $T$ is a waiting time, measured in trial number, and it can take values beyond $n$. The identity connects the cumulative event $T\leq n$ to the count event $X_n\geq1$; it does not make $T$ and $X_n$ interchangeable.

Question	Random-variable event	Meaning
Does at least one success occur within $n$ trials?	$X_n\geq1$, or $T\leq n$	At least one success occurs in trials 1 through $n$.
Are all of the first $n$ trials failures?	$X_n=0$, or $T>n$	The first success has not occurred by trial $n$.
Does the first success occur exactly on trial $n$?	$T=n$	Trials 1 through $n-1$ fail, and trial $n$ succeeds.
Are all $n$ trials successes?	$X_n=n$	Every trial in the fixed block is a success.

Worked Example: A First Success by Trial Four

Consider a hypothetical sequence of independent trials with success probability $p=0.30$. Find the probability that the first success occurs within the first four trials. State the corresponding binomial event.

Check the model. The calculation assumes that each trial has success probability $0.30$, that the probability is unchanged from trial to trial, and that outcomes are independent. For four fixed trials, $X_4\sim\operatorname{Bin}(4,0.30)$.

The waiting-time event $T\leq4$ is the same as the binomial event $X_4\geq1$. Its complement, $X_4=0$, requires four failures, each with probability $0.70$. Therefore,

\begin{aligned} P(T\leq4) &=P(X_4\geq1)\\ &=1-P(X_4=0)\\ &=1-(0.70)^4\\ &=1-0.2401\\ &=0.7599. \end{aligned}

The probability is $0.7599$, or $75.99\%$. This is a probability for the first four trials collectively. It includes a first success on trial 1, 2, 3, or 4.

Exact Events Are Not Cumulative Events

A frequent source of error is to confuse a cumulative waiting time with an exact waiting time. The event $T\leq n$ includes all possible first-success trial numbers from 1 to $n$. In contrast, $T=n$ requires failures on the first $n-1$ trials and a success on trial $n$. Similarly, $X_n\geq1$ includes one or more successes, while $X_n=1$ requires exactly one success.

For $T=n$, the relevant event is not “at least one success in $n$ trials”: that event also includes sequences where the first success happened earlier. Under the same trial assumptions, the exact waiting-time probability is $P(T=n)=(1-p)^{n-1}p$. For a binomial count, the probability of exactly one success is $P(X_n=1)=\binom{n}{1}p(1-p)^{n-1}$. These are different events: $T=n$ requires the success to be on the final trial, whereas $X_n=1$ allows its one success on any of the $n$ trials.

Worked Example: Separating “By Trial Three” From “On Trial Three”

Suppose a hypothetical sequence has independent trials with success probability $p=0.40$. Compare the probability that the first success occurs by trial 3 with the probability that it occurs exactly on trial 3. Also compare both with the probability that all three trials succeed.

Check the model. Assume the same $0.40$ success probability applies to each trial and the trials are independent. Let $X_3\sim\operatorname{Bin}(3,0.40)$, and let $T$ be the first-success trial.

For a first success by trial 3, use the complement of three failures:

\begin{aligned} P(T\leq3) &=1-P(X_3=0)\\ &=1-(0.60)^3\\ &=1-0.216\\ &=0.784. \end{aligned}

For a first success exactly on trial 3, trials 1 and 2 must fail, and trial 3 must succeed:

$$ P(T=3)=(0.60)^2(0.40)=0.144. $$

For all three trials to succeed, the probability is

$$ P(X_3=3)=(0.40)^3=0.064. $$

These probabilities answer distinct questions: $0.784$ is the chance of at least one success in the first three trials; $0.144$ is the chance the first success is specifically on trial 3; and $0.064$ is the chance that every one of the three trials succeeds. In particular, the last probability is not the probability of “a success by trial 3.”

When This Connection Is Useful

The binomial view is often convenient when a question sets a trial limit and asks whether any success has occurred by then. Instead of listing every possible first-success time, calculate the chance of zero successes and subtract from 1. The geometric view is useful when the quantity of interest is how long the process waits. Moving between the views helps select the simpler event without changing the question.

This connection also provides a consistency check. If the success probability $p$ is positive, then $(1-p)^n$ decreases as $n$ increases, so $1-(1-p)^n$ increases. That agrees with the event structure: allowing more trials cannot make “a success by then” less likely. The probability starts at $p$ for $n=1$, and approaches 1 as $n$ grows when $0<p<1$, because the chance of $n$ consecutive failures approaches zero.

The independence and constant-probability assumptions are essential. If the success probability changes by trial, or if one outcome affects the next, the probability of $n$ failures need not equal $(1-p)^n$. In that situation, the simple binomial model and this particular formula may not apply. The event equivalence itself remains logical—no success in the first $n$ trials means the first success is later than $n$—but its probability must be calculated using the actual trial model.

Common Mistakes

Using the probability of zero successes as the answer. $(1-p)^n$ is the probability that no success occurs by trial $n$. For “at least one success by trial $n$,” subtract it from 1.
Confusing “by trial $n$” with “on trial $n$.” $P(T\leq n)$ adds the possibilities of a first success on every trial from 1 through $n$; $P(T=n)$ covers only the final one.
Confusing at least one with exactly one. $P(X_n\geq1)$ includes two or more successes. It is not the same event as $P(X_n=1)$.
Applying the formula without checking the trial model. The expression $1-(1-p)^n$ requires independent trials with a common success probability $p$. A changing probability or dependence between trials requires a different calculation.

Key takeaway: For independent trials with common success probability $p$, a first success within $n$ trials is exactly the event of at least one success in those trials. The complement is zero successes, so $P(T\leq n)=1-P(X_n=0)=1-(1-p)^n$.

Check Your Understanding

For each question, identify the event carefully before choosing a probability calculation.

A process has independent trials with success probability $0.25$. What binomial event corresponds to the first success occurring by trial 6?
In the same setting, what is the probability of no successes in the first 6 trials? What is the probability of at least one success in those trials?
Explain why $T\leq5$ and $X_5\geq1$ describe the same event, but $T=5$ and $X_5=1$ do not.
For independent trials with success probability $p$, write the probability that the first success occurs after trial $n$ in terms of the binomial count $X_n$.
Which assumptions are needed to calculate $P(T\leq n)$ as $1-(1-p)^n$?

Question	Random-variable event	Meaning
Does at least one success occur within \(n\) trials?	\(X_n\geq1\), or \(T\leq n\)	At least one success occurs in trials 1 through \(n\).
Are all of the first \(n\) trials failures?	\(X_n=0\), or \(T>n\)	The first success has not occurred by trial \(n\).
Does the first success occur exactly on trial \(n\)?	\(T=n\)	Trials 1 through \(n-1\) fail, and trial \(n\) succeeds.
Are all \(n\) trials successes?	\(X_n=n\)	Every trial in the fixed block is a success.