The geometric distribution models the number of independent trials needed to get the first success. It is useful when each trial has only two outcomes, such as success or failure, and the probability of success stays constant from trial to trial. This makes it a natural model for coin flips, quality checks, and repeated attempts at a task.
It helps answer questions about waiting time in simple random processes.
If the probability of success on each trial is , then the probability that the first success happens on trial is for . The factor represents failing on all previous trials, and the final represents succeeding on trial . The mean number of trials is , so rarer successes lead to longer expected waiting times.
A special feature of the geometric distribution is its memoryless property, which means the future waiting time does not depend on how many failures have already occurred.
Understanding Geometric Distribution
A geometric model depends on more than repeated attempts. Each attempt must be independent, meaning one result does not change the chance on the next attempt. The chance of success must stay fixed.
A fair coin meets these conditions when success means heads. Drawing cards from a deck without replacing them does not. After every card is removed, the deck changes, so the chance of drawing a chosen card changes too.
In real data, independence can fail when a machine warms up, a person learns from practice, or weather conditions change. Checking these assumptions matters before using any distribution formula.
There are two common ways to count geometric waiting time. One method counts the trial on which the first success occurs. This gives possible values starting at one.
Another method counts failures before the first success. That version can begin at zero. Both describe the same random process, but their averages differ by one.
Students often lose marks by using a formula from one convention with data defined by the other. Write down exactly what is being counted before calculating. If a basketball player makes a free throw with a fixed chance of one quarter, getting the first made shot on the third try means two misses followed by one make.
Probabilities for later trials get smaller because every later first success requires a longer run of failures. Suppose a device has a one in ten chance of passing a test on each independent attempt. The chance that its first pass is on attempt four comes from three failures followed by a pass.
The chance of waiting more than four attempts is found by considering four failures in a row. This type of result is useful for planning. A mean waiting time of ten attempts does not promise a pass within ten attempts.
Some runs end immediately, while others last much longer. The variance describes how spread out these waiting times can be. When success is rare, the spread is large.
The memoryless property has a precise meaning. If an independent process has already produced many failures, the next attempt still has the original success chance. A run of losses can feel meaningful, but it does not make a fair coin more likely to land heads next time.
This prevents the gambler's fallacy, which is the mistaken belief that a past streak must be corrected soon. Memorylessness does not mean past information is always useless. It applies only when the success chance truly remains constant and attempts are independent.
In a classroom experiment, record each trial carefully, define success before starting, and compare the observed waiting times with the model. Differences may reveal random variation, or they may show that the assumptions were not met.
Key Facts
- A geometric random variable X counts the trial number of the first success.
- Probability mass function: for
- The support is x = 1, 2, 3, ... because the first success cannot occur before trial 1.
- Mean: E(X) = 1/p
- Variance:
- Memoryless property:
Vocabulary
- Geometric distribution
- A probability distribution that gives the number of trials needed to get the first success in repeated independent trials.
- Independent trials
- Trials are independent when the outcome of one trial does not affect the outcome of any other trial.
- Success probability
- The value p is the probability that a single trial results in success.
- Probability mass function
- A formula that gives the probability that a discrete random variable takes a specific value.
- Memoryless property
- A property meaning that past failures do not change the probability distribution of the remaining waiting time.
Common Mistakes to Avoid
- Using x = 0 as a possible value, which is wrong because this version of the geometric distribution counts trials until the first success, so the smallest value is 1.
- Forgetting the final success factor , which is wrong because must include both failures and then one success.
- Applying the model when p changes from trial to trial, which is wrong because the geometric distribution requires a constant success probability on every trial.
- Confusing the geometric distribution with the binomial distribution, which is wrong because geometric counts how long until the first success, while binomial counts how many successes occur in a fixed number of trials.
Practice Questions
- 1 A biased coin has probability of heads p = 0.3 on each flip. What is the probability that the first head occurs on the 4th flip?
- 2 A machine produces a defective item with probability 0.08 on each inspection, independently. Let X be the trial number of the first defective item found. Find E(X) and Var(X).
- 3 Explain why repeated rolls of a fair die can be modeled by a geometric distribution if success is defined as rolling a 6, and state what assumption would fail if the die were changed after each roll.