A point estimate is a single number used as the best guess for an unknown value. For a proportion, it's successes ÷ trials: 42 yes answers out of 100 gives a point estimate of 0.42, or 42%. From a confidence interval, it's the midpoint.
From counts
p̂ = successes ÷ trials
42 ÷ 100 = 0.42, or 42%. Its standard error, √(p̂(1 − p̂) ÷ n), is √(0.42 × 0.58 ÷ 100) = 0.0494.
The standard error shows how much the estimate would vary from sample to sample. A bigger sample shrinks it: the same 42% from 400 trials would have half the standard error.
From a confidence interval
A symmetric confidence interval is built as estimate ± margin of error, so the estimate sits in the middle.
Point estimate = (lower + upper) ÷ 2 and margin of error = (upper − lower) ÷ 2
An interval of 0.32 to 0.52 has a point estimate of (0.32 + 0.52) ÷ 2 = 0.42 and a margin of error of (0.52 − 0.32) ÷ 2 = 0.1.
This only works for intervals built symmetrically around the estimate. Some intervals, like the Wilson interval for a proportion, aren't centered on p̂.
Adjusted estimates
With small samples or results near 0% or 100%, statisticians sometimes nudge the estimate toward the middle. The Laplace estimate adds one success and one failure: (x + 1) ÷ (n + 2). For 42 of 100, that's 43 ÷ 102 ≈ 0.4216. The calculator also shows the Wilson and Jeffreys versions.
Work it out
Enter your counts or interval in the point estimate calculator.