Thomas Bayes (c. 1702–1761): English minister and mathematician whose posthumous "Essay Towards Solving a Problem in the Doctrine of Chances" (1763) introduced what became Bayes' theorem, revolutionizing probabilistic reasoning and modern statistical inference!
Population Proportion Estimation — Frequentist & Bayesian
Enter raw binary data (0/1) or summary (successes k in trials n). Compare frequentist intervals
(Wald, Wilson, Clopper–Pearson, Agresti–Coull) with a Beta-posterior credible interval.
Bayesian prior on p (Beta)
Default is Jeffreys Beta(0.5, 0.5). Uniform is Beta(1,1). Or set informative a,b.
Try examples:
Fill inputs and click Compute.
Frequentist Intervals
Wald is simple but often poor; prefer Wilson or Agresti–Coull for better coverage.
n = k = p̂ = level =
Method
Interval
Notes
Bayesian Credible Interval
Given prior + data, the interval contains p with the chosen posterior probability.
Formulas (teaching mode)
Wald: p̂ ± z √(p̂(1−p̂)/n) Wilson (score): [p̂ + z²/(2n) ± z √(p̂(1−p̂)/n + z²/(4n²))] / (1 + z²/n) Clopper–Pearson: [B⁻¹(α/2; k, n−k+1), B⁻¹(1−α/2; k+1, n−k)] Agresti–Coull: p̃ = (k + z²/2)/(n + z²), use Wald on (n + z², p̃) Bayesian: Posterior Beta(k+a, n−k+b), credible from Beta quantiles (equal-tail or HPD).
Plain-English summaries
Assumptions & when to worry
Independent Bernoulli trials; binomial model.
Wald unreliable for small n or p near 0/1—prefer Wilson or Bayesian.
Jeffreys prior (Beta(0.5,0.5)) is objective and boundary-friendly.