The Law of Large Numbers states that as the number of trials of a random process increases, the relative frequency of an event gets closer and closer to its true (theoretical) probability, which is why simulations with more trials give better probability estimates on the AP Stats exam.
The Law of Large Numbers (LLN) is the rule that makes probability mean anything at all. It says that if you repeat a random process over and over, the proportion of times an event occurs (its relative frequency) will settle down toward the true probability of that event. Flip a fair coin 10 times and you might get 70% heads. Flip it 10,000 times and you'll be very close to 50%. Short-run results are wild; long-run results are stable.
In the CED, this lives inside Topic 4.2 (Estimating Probabilities Using Simulation). The essential knowledge for 4.2.A says the relative frequency of an outcome in simulated or empirical data can be used to estimate the probability of that event, and the Law of Large Numbers is the justification for that move. It's the bridge between what you observe (counts from trials) and what you want to know (the actual probability). One thing to keep straight is that LLN is about the long run only. It does not promise that a streak of tails will be "balanced out" by extra heads soon. The coin has no memory.
The Law of Large Numbers directly supports learning objective AP Stats 4.2.A, which asks you to estimate probabilities using simulation. Every simulation answer you write follows the same logic. You run many trials, count how often the event happens, and report the relative frequency as your estimate. LLN is the reason that estimate is trustworthy, and it's the reason more trials beat fewer trials.
It also quietly powers Topic 8.1 (Introducing Statistics: Are My Results Unexpected?). When you compare observed counts to expected counts in categorical data, those expected counts are long-run averages built on probabilities. LLN tells you what "expected" even means. More broadly, the entire frequentist interpretation of probability on the AP exam, including the meaning of a confidence level or a p-value as a long-run proportion, rests on this one idea.
Keep studying AP Statistics Unit 8
Estimating Probabilities Using Simulation (Unit 4)
LLN is the engine behind Topic 4.2. A simulation only works because relative frequency converges to true probability as trials pile up. When a question asks why 10,000 trials beat 100 trials, LLN is the answer they want by name.
Sample Size (Units 1, 4-9)
LLN is the probability version of a theme you see everywhere in AP Stats. Bigger samples give estimates that are closer to the truth, on average. That's why sample size shows up in margin of error, power, and simulation design.
Probability (Unit 4)
The AP exam defines probability as long-run relative frequency, and LLN is what makes that definition coherent. Without it, saying P(heads) = 0.5 would have no observable meaning.
Expected Counts and Chi-Square (Unit 8)
Topic 8.1 asks whether the gap between observed and expected counts is just random variation. Expected counts are what LLN predicts in the long run, so chi-square tests are really measuring how far short-run reality strayed from the long-run prediction.
On multiple choice, the Law of Large Numbers usually appears in a simulation context. A typical stem gives you trial counts and an event count (say, event A occurs 237 times in 1,000 simulations) and asks what justifies using 0.237 as an estimate of P(A), or shows estimates from 100, 500, 1,000, and 10,000 trials drifting toward a value and asks you to interpret the pattern. Another common setup compares two students, one with 100 trials and one with 10,000, and asks whose estimate is more reliable. The answer is always the larger trial count, because of LLN.
No released FRQ has asked you to define the term verbatim, but simulation FRQs in Unit 4 expect you to use its logic. When you describe a simulation, you should say you'll run many trials and use the relative frequency of the event as the probability estimate. Watch for trap answers built on the gambler's fallacy, like "after five heads, tails is due." Those are wrong precisely because LLN says nothing about short-run correction.
The Law of Large Numbers says relative frequency converges to true probability over many, many trials. The so-called law of averages is the false belief that random processes self-correct in the short run, like thinking tails is "due" after five heads in a row. LLN works by swamping early streaks with a huge number of new trials, not by reversing them. Each flip is still 50/50 no matter what came before. MCQ wrong-answer choices love to dress up the gambler's fallacy in LLN language, so check whether the answer claims short-run balancing. If it does, it's wrong.
The Law of Large Numbers states that as the number of trials increases, the relative frequency of an event converges to its true probability.
This is the justification for estimating probabilities with simulation in Topic 4.2, where your estimate is just the count of successes divided by the total number of trials.
More trials give a more accurate estimate, so an estimate from 10,000 trials is more trustworthy than one from 100 trials, even if you don't know the true probability.
LLN only describes the long run; it does not mean outcomes "even out" in the short run, and believing that is the gambler's fallacy.
Each trial of an independent random process is unaffected by previous results, so a streak of heads does not make tails more likely.
LLN underlies the meaning of expected counts in Unit 8 and the long-run interpretation of confidence levels and p-values in inference.
It's the rule that as the number of trials of a random process increases, the relative frequency of an event gets closer to the event's true probability. It's the reason simulation works as a probability-estimating tool in Topic 4.2.
No. That's the gambler's fallacy, and it's a classic wrong-answer trap on the exam. LLN says big streaks get diluted by thousands of future trials, not canceled by opposite results coming soon. A fair coin is still 50/50 after five heads in a row.
LLN says a sample proportion or mean converges to the true value as trials increase. The CLT describes the shape of the sampling distribution, saying it becomes approximately normal for large samples. LLN is about accuracy of the estimate; CLT is about the distribution of estimates.
There's no magic cutoff, but more is always better. Exam questions often show estimates stabilizing as trials go from 100 to 10,000, and the expected reasoning is that the larger trial count gives an estimate closer to the true probability.
Because of the Law of Large Numbers, the relative frequency from 10,000 trials will typically sit much closer to the true probability than one from 100 trials. A common MCQ gives two students with different trial counts and asks whose estimate is more reliable; pick the larger one.