Probability and Statistics for Physicists: A Friendly Start

Probability assigns numbers between 0 and 1 to uncertain outcomes, and statistics uses data from repeated trials to estimate those numbers and judge how much to trust them. For physicists, this pair is not optional decoration: every measurement carries random scatter, and many phenomena, like radioactive decay, are random at their core.

This matters because a physics result is meaningless without its uncertainty. Saying "the length is 10.1 cm" claims perfection; saying "10.10 plus-or-minus 0.07 cm" honestly reports what the experiment actually established, and statistics is what turns raw repeated readings into that second statement.

Key Concepts

• Random variable: a quantity whose value varies from trial to trial, like the count of decays in one minute.
• Mean: mu = (sum of x(i)) / N, the average value over N measurements.
• Variance and standard deviation: sigma squared measures the average squared spread around the mean; sigma, its square root, has the same units as the data.
• Standard error of the mean: sigma / square root of N, the uncertainty of the average itself, which shrinks as more data is taken.
• Poisson distribution: P(k) = lambda to the power k times e to the power (minus lambda) / k factorial, for counting rare random events with mean lambda.

Formulas

• Sample mean: x-bar = (1/N) sum of x(i).
• Sample standard deviation: s = square root of [sum of (x(i) minus x-bar) squared / (N minus 1)].
• Standard error of the mean: s / square root of N.
• Error propagation for sums: if f = x + y, then sigma(f) = square root of (sigma(x) squared + sigma(y) squared).
• Poisson: mean = variance = lambda, so the standard deviation of a count N is square root of N.

Worked Example

Five length measurements give 10.1, 10.3, 9.9, 10.2, and 10.0 cm. Find the mean and its uncertainty.

1. Mean: (10.1 + 10.3 + 9.9 + 10.2 + 10.0) / 5 = 50.5 / 5 = 10.1 cm.
2. Deviations from the mean: 0.0, 0.2, minus 0.2, 0.1, minus 0.1 cm.
3. Squared deviations: 0, 0.04, 0.04, 0.01, 0.01; their sum is 0.10.
4. Sample variance = 0.10 / (5 minus 1) = 0.025, so s = square root of 0.025, about 0.158 cm.
5. Standard error of the mean = 0.158 / square root of 5, about 0.071 cm. Report: 10.10 plus-or-minus 0.07 cm.

Common Misconceptions

• "Standard deviation and standard error are the same." The standard deviation describes the scatter of single readings; the standard error of the mean describes the uncertainty of the average and is smaller by a factor of square root of N.
• "More data removes all error." Averaging shrinks random error as 1 / square root of N, but systematic error, like a miscalibrated ruler, never averages away.
• "A 95 percent confidence interval means the true value is inside with 95 percent probability." Strictly, it means the procedure captures the true value 95 percent of the time over many repeats; the distinction matters when interpreting results carefully.

Real-World Applications

• Particle physics: discoveries like the Higgs boson required counting statistics to rule out flukes at the five-sigma level.
• Least-squares fitting: extracting physical constants from data minimizes the weighted squared deviations.
• Monte Carlo simulations: random sampling estimates integrals and models complex systems from finance to neutron transport.

Probability models the randomness, and statistics converts scattered measurements into honest results with defensible uncertainties. For a physicist, these tools are as fundamental as the calculus itself.