Calculate sampling error (standard error and margin of error) for proportions and means. Includes finite population correction, error decomposition, and sample size comparison.
Sampling error is the natural variability that occurs when you estimate a population parameter using a sample. Even with perfect random sampling, different samples yield slightly different results. The sampling error quantifies this unavoidable uncertainty and is the foundation of all inferential statistics.
This calculator computes the standard error and margin of error for both proportions (survey percentages) and means (continuous measurements). It breaks down the error components, applies finite population correction when appropriate, and shows how the error changes with different sample sizes.
Understanding sampling error is crucial for survey research (polling accuracy), quality control (measurement precision), scientific experiments (result reliability), and any context where sample data is used to make population-level conclusions. It is especially useful when you need to explain how much uncertainty comes from sampling alone rather than from measurement or model assumptions. That makes the output easier to defend in reports where precision is part of the decision.
Sampling error is often confused with margin of error or total survey error. This calculator clarifies the distinction, shows exactly how each component contributes, and demonstrates the relationship between sample size and precision. The visual comparison makes trade-offs immediately clear. It is useful when you need to justify whether a sample is precise enough for the decision you want to make.
Standard Error of Proportion: SE = √(p̂(1−p̂)/n) Standard Error of Mean: SE = s / √n Finite Population Correction: SE_adj = SE × √((N−n)/(N−1)) Margin of Error: MOE = z* × SE Sampling Error decreases as: SE ∝ 1/√n (inverse square root law)
Result: SE = 0.0158, MOE = ±3.10%
With a sample proportion of 0.52 from n = 1,000, the standard error is 0.0158. At 95% confidence (z* = 1.96), the margin of error is ±3.1%. The true proportion is likely between 48.9% and 55.1%. The FPC is not applied because no population size was specified.
Sampling error arises because a sample is only a subset of the population. Different random samples yield different statistics: different proportions, means, and variances. The standard error quantifies the typical size of this variation. Larger samples capture more of the population's diversity and produce more stable estimates.
The total error framework in survey methodology identifies: coverage error (who can be sampled), sampling error (who is sampled), non-response error (who responds), and measurement error (accuracy of responses). Sampling error is often the smallest contributor. Media reports of "margin of error" actually refer only to sampling error, understating true uncertainty.
When planning a study, work backward from the desired precision: choose your acceptable margin of error, set the confidence level, estimate the population variability, and solve for n. Add 10-20% for expected non-response. The diminishing returns of larger samples mean there's usually a practical maximum beyond which additional data isn't cost-effective.
Sampling error is the natural variability from using a sample instead of a census. Non-sampling errors include measurement error, non-response bias, coverage error, and processing mistakes. Sampling error decreases with larger samples; non-sampling errors may not.
Not exactly. Standard error measures the variability of the estimator. Margin of error = z* × SE includes the confidence level multiplier. At 95% confidence, MOE ≈ 1.96 × SE. The terms are related but not interchangeable.
Only by measuring the entire population (a census). As long as you use a sample, some sampling error exists. However, it can be made arbitrarily small by increasing the sample size.
This follows from the Central Limit Theorem. The variance of the sample mean is σ²/n, so the standard deviation (SE) is σ/√n. This inverse square root relationship is a fundamental property of averaging independent observations.
It depends on the application. Political polls typically target ±3%. Market research accepts ±5%. Medical studies may require ±1% or less. The acceptable level depends on the cost of being wrong and the cost of data collection.
Surprisingly little for large populations. A sample of 1,000 gives essentially the same precision whether the population is 100,000 or 100 million. Population size only matters through the FPC when you sample a substantial fraction (>5%).