Tenth Grade Mathematics: Classical vs. Empirical Approaches to Probability
Uncertainty is an inherent feature of the natural world, governing everything from genetic inheritance and quantum mechanics to weather forecasting and financial risk. Probability is the mathematical science that quantifies uncertainty, assigning a numerical measure between 0 and 1 to the likelihood of an event occurring. In ninth grade, probability was introduced empirically through repeated trials and relative frequencies. In tenth grade, we formalize the transition from empirical (experimental) observation to classical (theoretical) deduction, establishing the Law of Large Numbers that bridges these two approaches.
The Empirical (Experimental) Approach
The empirical approach to probability is grounded in actual physical experimentation and historical data recording.
Formal Definition:
If an experiment is performed N times, and an event E occurs in m of those trials,
the empirical (experimental) probability P_emp(E) is defined as:
Number of trials in which event occurred
P_emp(E) = ------------------------------------------ = m / N
Total number of trials
Historical Coin-Tossing Experiments:
- Georges-Louis Leclerc, Comte de Buffon (18th Century): Tossed a coin 4,040 times;
obtained 2,048 heads. Empirical P(Heads) = 2048 / 4040 = 0.5069.
- J.E. Kerrich (during WWII internment): Tossed a coin 10,000 times;
obtained 5,067 heads. Empirical P(Heads) = 5067 / 10000 = 0.5067.
- Karl Pearson: Tossed a coin 24,000 times;
obtained 12,012 heads. Empirical P(Heads) = 12012 / 24000 = 0.5005.
Notice that while small trial samples can produce fluctuations (e.g., getting 7 heads in 10 tosses, giving P = 0.7), as the number of trials N grows into thousands, the experimental relative frequency stabilizes toward 0.5.
The Classical (Theoretical) Approach
Formulated systematically by Pierre-Simon Laplace in 1812, the classical approach calculates probabilities prior to any experimentation (a priori), relying entirely on mathematical symmetry and deductive reasoning.
The Fundamental Assumption:
The classical theory applies ONLY when all elementary outcomes of an experiment
are EQUALLY LIKELY (having the exact same symmetric chance of occurrence).
Formal Definition:
The theoretical (classical) probability of an event E, denoted P(E), is defined as:
Number of outcomes favorable to E
P(E) = ----------------------------------- = n(E) / n(S)
Total number of all possible outcomes
Comparison Table:
+-------------------+-----------------------------------+-----------------------------------+
| Feature | Empirical (Experimental) | Classical (Theoretical) |
+-------------------+-----------------------------------+-----------------------------------+
| Origin | Calculated AFTER trials (a post.) | Deduced BEFORE trials (a priori) |
+-------------------+-----------------------------------+-----------------------------------+
| Method | Physical testing, data tallying | Mathematical symmetry, counting |
+-------------------+-----------------------------------+-----------------------------------+
| Requirement | Large sample of recorded trials | Equally likely elementary outcomes|
+-------------------+-----------------------------------+-----------------------------------+
| Reliability | Fluctuates for small trial counts | Exact, constant, and invariant |
+-------------------+-----------------------------------+-----------------------------------+
The Law of Large Numbers (The Bridge)
The connection between experimental frequency and theoretical probability is governed by Jacob Bernoulli's celebrated Law of Large Numbers:
Theorem (The Weak Law of Large Numbers):
As the number of experimental trials N approaches infinity, the empirical
probability m / N converges mathematically to the theoretical probability P(E):
lim [ m / N ] = P_theoretical(E)
N ---> inf
Visualizing the Law of Large Numbers:
Relative Frequency
^
1.0 + / | / 0.5 +----+----+----\----------------------- Theoretical P = 0.5
| \___/\___/\___________ Stabilizes toward 0.5 as N grows!
0 +-------------------------------------> Number of Trials N
0 100 1,000 10,000
The Axiomatic Boundaries of Probability
Every valid probability P(E) satisfies three universal axioms:
1. Bounded Range:
For any event E:
0 <= P(E) <= 1
2. Impossible Event:
An event that cannot possibly occur under any circumstances has probability 0:
P(Impossible Event) = 0
Example: Rolling a 7 on a standard six-sided die.
3. Sure (Certain) Event:
An event that is guaranteed to occur on every trial has probability 1:
P(Sure Event) = 1
Example: Rolling a number less than 7 on a standard six-sided die.
Chapter Exercises
Exercise 1: An insurance company selected 2000 drivers at random in a city to find a relationship between age and accidents. The data showed that 440 drivers aged 18-29 had 1 accident in a year. What is the empirical probability that a randomly chosen driver aged 18-29 has 1 accident in a year?
Exercise 2: Explain in continuous prose why classical probability cannot be applied directly to predict whether a thumbtack will land with its point facing upwards or downwards, and state how its probability must be determined instead.
Exercise 3: A coin is tossed 100 times and head appears 58 times. (a) Calculate the empirical probability of getting a head and getting a tail. (b) Compare these values with their theoretical probabilities and explain the source of the difference.
Solutions and Explanations
Solution 1: The total number of drivers selected in the study is the total number of trials N = 2000. The number of drivers aged 18-29 who had exactly 1 accident in the year is the favorable count m = 440. The empirical probability is: P_emp = m / N = 440 / 2000. Dividing numerator and denominator by 40: P_emp = 11 / 50 = 0.22. Thus, the empirical probability is 0.22 (or 22 percent).
Solution 2: Classical probability requires as a strict prerequisite that all possible outcomes are equally likely due to geometric or physical symmetry (such as the symmetric faces of a coin or die). A thumbtack, however, possesses an asymmetric physical shape: it has a flat, heavy circular base and a sharp vertical pin. Because of this asymmetry, the center of gravity and aerodynamic profile do not give equal chances to landing point-up versus point-down. Because the two outcomes are not equally likely, the classical ratio 1/2 cannot be applied. Instead, the probability must be determined empirically by dropping the thumbtack thousands of times under controlled conditions and calculating the relative frequency of the pin landing upwards.
Solution 3: In the experiment of 100 tosses, total trials N = 100. Heads appeared 58 times, so tails appeared 100 - 58 = 42 times. For part (a), the empirical probability of getting a head is P_emp(Head) = 58 / 100 = 0.58. The empirical probability of getting a tail is P_emp(Tail) = 42 / 100 = 0.42. For part (b), a fair coin is physically symmetric with two equally likely outcomes (Head or Tail), giving a theoretical probability of P_theor(Head) = 1/2 = 0.50 and P_theor(Tail) = 1/2 = 0.50. The difference between the experimental values (0.58 and 0.42) and the theoretical value (0.50) is due to random sampling variance inherent in a finite sample size of only 100 tosses. By the Law of Large Numbers, if the coin were tossed 10,000 times, the relative frequencies would converge much closer to 0.500.