5.2: Empirical Probability and the Law of Large Numbers
- Page ID
- 237235
\( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)
\( \newcommand{\dsum}{\displaystyle\sum\limits} \)
\( \newcommand{\dint}{\displaystyle\int\limits} \)
\( \newcommand{\dlim}{\displaystyle\lim\limits} \)
\( \newcommand{\id}{\mathrm{id}}\) \( \newcommand{\Span}{\mathrm{span}}\)
( \newcommand{\kernel}{\mathrm{null}\,}\) \( \newcommand{\range}{\mathrm{range}\,}\)
\( \newcommand{\RealPart}{\mathrm{Re}}\) \( \newcommand{\ImaginaryPart}{\mathrm{Im}}\)
\( \newcommand{\Argument}{\mathrm{Arg}}\) \( \newcommand{\norm}[1]{\| #1 \|}\)
\( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\)
\( \newcommand{\Span}{\mathrm{span}}\)
\( \newcommand{\id}{\mathrm{id}}\)
\( \newcommand{\Span}{\mathrm{span}}\)
\( \newcommand{\kernel}{\mathrm{null}\,}\)
\( \newcommand{\range}{\mathrm{range}\,}\)
\( \newcommand{\RealPart}{\mathrm{Re}}\)
\( \newcommand{\ImaginaryPart}{\mathrm{Im}}\)
\( \newcommand{\Argument}{\mathrm{Arg}}\)
\( \newcommand{\norm}[1]{\| #1 \|}\)
\( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\)
\( \newcommand{\Span}{\mathrm{span}}\) \( \newcommand{\AA}{\unicode[.8,0]{x212B}}\)
\( \newcommand{\vectorA}[1]{\vec{#1}} % arrow\)
\( \newcommand{\vectorAt}[1]{\vec{\text{#1}}} % arrow\)
\( \newcommand{\vectorB}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\( \newcommand{\vectorC}[1]{\textbf{#1}} \)
\( \newcommand{\vectorD}[1]{\overrightarrow{#1}} \)
\( \newcommand{\vectorDt}[1]{\overrightarrow{\text{#1}}} \)
\( \newcommand{\vectE}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash{\mathbf {#1}}}} \)
\( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\(\newcommand{\longvect}{\overrightarrow}\)
\( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)
\(\newcommand{\ket}[1]{\left| #1 \right>}\)
\(\newcommand{\bra}[1]{\left< #1 \right|}\)
\(\newcommand{\braket}[2]{\left< #1 \vphantom{#2} \right| \left. #2 \vphantom{#1} \right>}\)
\(\newcommand{\braopket}[3]{\left< #1 \vphantom{#2}\vphantom{#3} \right| #2 \vphantom{#1}\vphantom{#3} \left| #3 \vphantom{#1}\vphantom{#2} \right>}\)
\(\newcommand{\qmvec}[1]{\mathbf{\vec{#1}}}\)
\(\newcommand{\op}[1]{\hat{\mathbf{#1}}}\)
\(\newcommand{\expect}[1]{\langle #1 \rangle}\)
\(\newcommand{\dfn}[1]{\emph{\textbf{#1}}}\)
- Differentiate theoretical and empirical probability.
- Calculate empirical probability.
- Explain the Law of Large Numbers.
Introduction
Probability began with the idea that it is based on analyzing the equally likely outcomes in a sample space. For example, the probability that a coin lands heads up is \(\frac{1}{2}=0.5=50\%\) because the sample space contains two equally likely outcomes, and only one of them -- heads -- belongs to the event.
However, data from real experiments do not always match these theoretical expectations. If you flip a coin 10 times, you might observe heads 40%, 60%, or even 70% of the time. Such variation is normal and expected whenever probability is measured through actual trials. This distinction leads us to two important concepts: theoretical probability and empirical probability.
Theoretical probability describes what should happen in the long run. It is calculated by analyzing the structure of the sample space and imagining the experiment repeated infinitely many times. It is the ratio of the number of outcomes in the event to the number of outcomes in the sample space.
Empirical probability describes what actually happens in practice. It is based on observed data collected from a finite number of trials. It is the relative frequency of an event after the experiment has been performed. Because it relies on real outcomes rather than idealized ones, empirical probability is also called experimental probability, and the two terms may be used interchangeably.
Together, these ideas highlight the difference between mathematical prediction and real‑world observation -- an essential theme in understanding probability.
Empirical Probability
Empirical probability is based on performing an actual experiment -- flipping coins, spinning spinners, drawing jellybeans from a jar, and similar activities. It represents the relative frequency of an event: the number of times the event occurs divided by the total number of trials. In other words, empirical probability reflects what actually happens when an identical experiment is repeated many times.
Lawrence has a standard 52‑card deck and wants to estimate the probability of drawing a single Queen.
The theoretical probability of drawing a Queen is \(\frac{4}{52}=\frac{1}{13} \approx 0.0769\) or about 7.69%, because there are 4 Queens in a deck of 52 equally likely cards.
Lawrence performs an experiment 25 times by drawing a card at random, recording it, and replacing it. If he draws a Queen 3 times in these 25 trials, then the empirical probability is
\(\dfrac{3}{25}=0.12 =12\%\).
This result differs from the theoretical probability. This is to be expected when working with a limited number of trials.
A survey was conducted to determine the programs that students were taking. The results are shown in the table.
\( \begin{array}{|l|c|c|c|c|c|} \hline \text{Program} & \text{English} & \text{Art} & \text{Nursing} & \text{Science} & \text{Education} \\ \hline \text{Students} & 3 & 8 & 2 & 12 & 10 \\ \hline \end{array} \nonumber \)
What is the probability that a student is enrolled in the Education program?
Solution
The total number of trials (students surveyed) is \(3+8+2+12+10=35\). Of these, 10 were in the Education program. The fraction of trials that resulted in the event "Education " gives the empirical probability:
\(P(\text{Education})=\dfrac{10}{35}=\dfrac{2}{7} \approx 0.2857 \text{ or } 28.57\%\).
In a given week, a veterinarian recorded how many times she treated each type of animal.
\( \begin{array}{c} \text{\textbf{Number of Animals Treated}} \\[6pt] \begin{array}{|l|c|} \hline \text{Animal} & \text{Number Treated} \\ \hline \text{Dog} & 40 \\ \hline \text{Cat} & 32 \\ \hline \text{Bird} & 9 \\ \hline \text{Rabbit} & 7 \\ \hline \text{Iguana} & 2 \\ \hline \end{array} \end{array} \nonumber \)
What is the probability that the next animal she treats is a cat?
- Answer
-
\(\frac{32}{90}=\frac{16}{45} \approx 0.3556 \text{ or } 35.56\%\).
Using the Law of Large Numbers to Estimate Theoretical Probability
Empirical probability is based on performing an experiment and observing how often an event occurs. By repeating the same process many times -- flipping coins, spinning spinners, drawing cards, or selecting jellybeans -- we can calculate the relative frequency of an event. This relative frequency serves as an estimate of the event’s theoretical probability, the probability we would expect if the experiment could be repeated infinitely many times.
The connection between empirical results and theoretical probability is explained by the Law of Large Numbers. This law states that as a probabilistic process is repeated a large number of times, the empirical probabilities (relative frequencies) of the outcomes will get closer and closer to their theoretical probabilities. In practice, this means that the more trials we perform, the more accurate our estimate becomes. Although we can never conduct infinitely many trials, a sufficiently large number of repetitions allows empirical probability to approximate theoretical probability quite well.
This principle is especially useful when theoretical probability is difficult to compute -- such as when the number of possible outcomes is extremely large -- or impossible to compute because the outcomes are not equally likely. In such cases, empirical probability provides a practical method for estimating the true probability of an event.
Consider the following example. Suppose we want to estimate the probability that a thumbtack lands point‑up when tossed. A thumbtack can land point‑up or on its side, but these outcomes are not equally likely, so theoretical probability cannot be determined by simple counting. Instead, we can toss a thumbtack many times (or toss many thumbtacks simultaneously) and record how often it lands point‑up. The resulting relative frequency gives an empirical estimate of the theoretical probability of the event.
A game uses a spinner like the one shown below. When spun, the arrow lands on either a white, yellow, blue, or red region. Let’s describe how we could use the Law of Large Numbers to estimate the probability of landing on the white region.

The regions do not appear to be the same size, so the four outcomes are not equally likely. We cannot apply our theoretical probability formula here unless we are told the areas of the regions or the measures of the central angles in the circular spinner. Instead, we can spin the spinner many times and record the results:
\( \begin{array}{c} \text{\textbf{Spinner Results}} \\[6pt] \begin{array}{|l|l|} \hline \text{Color} & \text{Tally} \\ \hline \text{White} & \cancel{||||}~\cancel{||||}~\cancel{||||}~||| \\ \hline \text{Red} & \cancel{||||}~\cancel{||||}~|| \\ \hline \text{Blue} & \cancel{||||}~| \\ \hline \text{Yellow} & |||| \\ \hline \end{array} \end{array} \nonumber \)
Using the data in the table, the spinner landed on "white" 18 times out of a total of \(18+12+6+4=40\) times. The empirical probability that the spinner lands on white is \(\frac{18}{40}=0.45\) or 45%.
This seems a reasonable estimate for the theoretical probability as the white region makes up slightly less than half the spinner. Randomness in the results will mean that we always get an approximation of the true probability, but the more trials we do, the more confidence we can have that the experimental probability will match the theoretical probability.ere.
Jennifer wants to compare the theoretical probability that a number cube lands on '3' with the empirical probability observed after performing 50 trials.
A fair six-sided die has 6 equally likely outcomes, and only one of them is the number '3.' Therefore, the theoretical probability is \(\frac{1}{6} \approx 0.1667\) or about 17%.
To compute an empirical probability, Jennifer rolls the number cube 50 times and records each outcome.
\( \begin{array}{|l|c|c|c|c|c|c|} \hline \text{Outcome} & 1 & 2 & 3 & 4 & 5 & 6 \\ \hline \text{Tally} & \cancel{||||}~||| & \cancel{||||}~| & \cancel{||||}~|||| & \cancel{||||}~\cancel{||||} & \cancel{||||}~|| & \cancel{||||}~\cancel{||||} \\ \hline \text{Frequency} & 8 & 6 & 9 & 10 & 7 & 10 \\ \hline \end{array} \nonumber \)
From her data, the number cube landed on '3' in 9 out of 50 rolls. The empirical probability is \(\frac{9}{50} = 0.18\) or 18%.
Although 18% does not match the theoretical probability exactly, it is reasonably close. This difference is expected when the number of trials is limited. According to the Law of Large Numbers, as the number of trials increases, the empirical probability tends to move closer to the theoretical probability. If Jennifer repeated the experiment 100, 1,000, or 10,000 times, the empirical probability would likely approach \(\frac{1}{6}\).
We also can use empirical probability to confirm theoretical probability to see for ourselves whether something happens as often as we’d expect it to. If it doesn’t, we may need to reconsider our thinking.
Nadia and Peter are playing a game with a six‑sided die, but Peter keeps winning, and Nadia begins to suspect that the die may be unfair. In particular, she notices how often Peter rolls a 6, so she decides to run an experiment. Nadia rolls the die 100 times, records each result, and organizes her data into a frequency table.
\( \begin{array}{|c|c|c|} \hline \text{Number on Die} & \text{Frequency} & P(\text{number}) \\ \hline 1 & 15 & \dfrac{15}{100} = 0.15 \\[4pt] \hline 2 & 14 & \dfrac{14}{100} = 0.14 \\[4pt] \hline 3 & 15 & \dfrac{15}{100} = 0.15 \\[4pt] \hline 4 & 13 & \dfrac{13}{100} = 0.13 \\[4pt] \hline 5 & 9 & \dfrac{9}{100} = 0.09 \\[4pt] \hline 6 & 34 & \dfrac{34}{100} = 0.34 \\[4pt] \hline \end{array} \nonumber \)
After reviewing the table, it becomes clear that something unusual may be happening: the outcome '6' appears roughly twice as often as the other numbers. This pattern suggests that the die could be weighted or otherwise biased. However, even with such a strong pattern, Nadia cannot conclude with absolute certainty that the die is unfair. Random variation can still produce surprising results, especially with a relatively small number of trials.
Because of this, Nadia must speak in terms of likelihood, not certainty. To strengthen her conclusion, she could repeat the experiment with more trials -- for example, 500 or 1,000 rolls. If the relative frequency of rolling a '6' continues to stand out as an outlier while the other outcomes remain roughly balanced, the evidence that the die is unfair becomes much stronger. This approach aligns with the Law of Large Numbers, which tells us that empirical probabilities tend to stabilize and reveal underlying patterns as the number of trials increases.
Stanley picks a card from a standard deck of cards and gets heart. He returns the card to the deck, picks a second time, and gets another heart. Stanley repeats this process a total of five times and gets a heart in four of those trials. Should Stanley conclude that the probability of selecting a heart is \(\frac{4}{5} = 0.80 = 80\%\)?
Solution
No. Stanley's reasoning is incorrect. He should not make a conclusion about the true probability of selecting a heart based only on five trials. If Stanley wants to estimate the chance of selecting a heart using empirical probability, he should perform the experiment a lot more than five times.
Juan suspects that his lucky coin is weighted so that he gets heads more often than tails in a coin toss. The reason he is suspicious is because he seems to get several heads in a row when he tosses it. He conducts an experiment and gets the following results.
\( \begin{array}{|c|c|c|c|c|c|c|c|} \hline \text{H} & \text{T} & \text{T} & \text{H} & \text{H} & \text{H} & \text{H} & \text{T} \\ \hline \text{H} & \text{H} & \text{T} & \text{T} & \text{T} & \text{H} & \text{H} & \text{H} \\ \hline \text{H} & \text{H} & \text{H} & \text{T} & \text{H} & \text{H} & \text{T} & \text{T} \\ \hline \text{H} & \text{T} & \text{H} & \text{H} & \text{T} & \text{H} & \text{T} & \text{T} \\ \hline \text{T} & \text{H} & \text{H} & \text{T} & \text{H} & \text{T} & \text{T} & \text{H} \\ \hline \end{array} \nonumber \)
Do you think he should conclude the coin is unfair? Explain.
- Answer
-
The 40 tosses resulted in 23 heads and 17 tails. So \(P(H)= \frac{23}{40} = 0.575 = 57.5\%\) and \(P(T) = \frac{17}{40} = 0.425= 42.5\%\). For the coin to be fair, we expect heads around 50% of the time. While 57.5% is higher than 50%, this difference may be due to random chance - especially since Juan performed the experiment only 40 times. Further experimentation would help to get a more certain answer to whether Juan's coin is weighted.
Section Summary
Empirical probability uses observed data from an experiment to estimate how likely an event is. It is calculated as the relative frequency of the event -- how often the event occurs divided by the total number of trials.
The Law of Large Numbers explains the relationship between empirical and theoretical probability. As an experiment is repeated many times, the empirical probability of an event tends to move closer to the event’s theoretical probability. In other words, more trials lead to more reliable estimates.
Together, these ideas show how actual experimental results can be used to approximate theoretical probabilities, especially in situations where theoretical calculations are difficult or impossible.
Section 5.2 Review Problems
- According to the results of a survey, 200 people said they prefer cats, 160 said they prefer dogs, while 40 said they don’t like either pet. Find the empirical probability that a person prefers dogs. Give the probability as a fraction in simplest form, as a decimal, and as a percent.
- Find the empirical probability that the next car passing you on the highway is red if the previous 25 car colors were as follows:
\( \begin{array}{|c|c|c|c|c|} \hline \text{red} & \text{blue} & \text{white} & \text{blue} & \text{silver} \\ \hline \text{red} & \text{black} & \text{black} & \text{white} & \text{red} \\ \hline \text{green} & \text{red} & \text{black} & \text{blue} & \text{white} \\ \hline \text{red} & \text{silver} & \text{white} & \text{red} & \text{black} \\ \hline \text{white} & \text{blue} & \text{silver} & \text{red} & \text{black} \\ \hline \end{array} \nonumber \)
- Moya is playing a board game with the spinner shown below. The spinner has 4 sections, each the same size. Moya recorded the results of her spins in a table as she played the game.

\( \begin{array}{|l|c|} \hline \text{Color} & \text{Frequency} \\ \hline \text{Blue} & 12 \\ \hline \text{Green} & 6 \\ \hline \text{Red} & 9 \\ \hline \text{White} & 13 \\ \hline \end{array} \nonumber \)
- If Moya's spinner is fair, find the theoretical probability that Moya spins 'green.'
- Find the empirical probability that Moya spins 'green.'
- Which color has an empirical probability closest to its theoretical probability?
- Decide whether you think this spinner is fair. Explain your choice.
- Determine whether you could calculate the theoretical probability of each event based on your knowledge of the possible outcomes, or whether you would have to find the empirical probability:
- Flipping a coin three times in a row and getting three heads.
- Pulling a nickel from your pocket when you know you have three nickels and five dimes in your pocket.
- Guessing the right answer on a four-option multiple choice question.
- Guessing the right answer on a free-response question.
- Sharing a birthday with one of your three best friends.
- Getting a flat tire while driving home.
- Having your left front tire be the one that goes flat, whenever you do get your next flat tire.
- Tossing a can of soda and it landing on its side.

