Resampling Calculator

Calculator Inputs

Resampling mode

Statistic

Iterations

Confidence level

Resample size

Use 0 to match Dataset A count.

Sample with replacement

Null value

Used for one dataset p-value. Leave blank if not needed.

Proportion threshold

Values at or above this number count as success.

Random seed

Decimal places

Dataset A

Dataset B

Required only for permutation mode.

Example Data Table

Example	Dataset A	Dataset B	Suggested mode	Statistic
Sales calls	12, 15, 14, 19, 13, 18	10, 11, 13, 14, 12, 15	Permutation	Mean
Delivery times	31, 35, 29, 40, 38, 33	None	One dataset	Median
Pass indicators	1, 0, 1, 1, 0, 1	0, 1, 0, 1, 0, 1	Permutation	Proportion

Formula Used

Observed statistic: θ = T(x), where T is the selected statistic and x is the original dataset.

Resampled statistic: θ* = T(x*), where x* is a repeated sample drawn from the original dataset.

Standard error: SE = sqrt Σ(θ* - mean(θ*))² / (B - 1), where B is the number of iterations.

Bias estimate: Bias = mean(θ*) - θ.

Percentile interval: Lower = Qα/2 and Upper = Q1-α/2 from the sorted resampled statistics.

Permutation p-value: p = (1 + count(|Δ*| ≥ |Δobs|)) / (B + 1).

How to Use This Calculator

Paste numeric values into Dataset A. Use commas, spaces, or line breaks.
Choose one dataset mode for interval estimation.
Choose permutation mode when comparing Dataset A with Dataset B.
Select the statistic that matches your question.
Set iterations, confidence level, seed, and decimal places.
Use the threshold only when calculating proportions.
Press the calculate button and review the result above the form.
Download the result table as CSV or PDF when needed.

Resampling for Practical Analysis

Resampling helps you study uncertainty without strict distribution rules. It reuses the values you already have. Each repeated sample creates a new estimate. Many estimates form an empirical distribution. That distribution can show spread, bias, and likely error.

Why Resampling Matters

Real data is often small, noisy, or uneven. A formula may assume normal shape. Your sample may not follow that shape. Resampling gives a practical check. It is useful for averages, medians, rates, and differences between groups. The method is also easy to explain. You are asking what could happen if similar samples were drawn again.

Bootstrap and Permutation Ideas

A bootstrap sample is drawn from one dataset with replacement. Some values may appear more than once. Some values may not appear. The statistic is calculated for each sample. Percentiles from those results form a confidence interval. The interval describes plausible values for the population statistic.

A permutation test answers a different question. It checks whether two groups look different after labels are mixed. If group labels do not matter, shuffled differences should resemble the observed difference. A small p-value means the observed difference is unusual under that label mixing rule.

Reading the Output

Start with the observed statistic. This is calculated from your original data. Then review the resampled mean. The difference between both values is the estimated bias. The standard error shows typical resampling spread. A wider interval means more uncertainty. A narrow interval suggests stable estimates, but it does not prove perfect accuracy.

Good Input Habits

Use clean numeric values. Remove symbols, units, and empty items. Keep methods aligned with the question. Use median for skewed data. Use proportion when values represent success or failure. Increase iterations for smoother results. A fixed seed helps you repeat the same calculation.

Limits and Care

Resampling does not repair poor sampling. It cannot remove measurement bias. It also depends on the data representing the process well. Treat results as decision support, not final proof. Combine them with subject knowledge, study design, and clear reporting.

Reporting Tips

Report iteration count, confidence level, statistic choice, and seed. Name the method used. Add the sample size. Share exported tables when another person must review results carefully later clearly.

FAQs

What is resampling?

Resampling repeats calculations on rearranged or redrawn data. It helps estimate uncertainty, standard error, confidence intervals, and p-values without relying only on strict theory formulas.

When should I use one dataset mode?

Use one dataset mode when you want an interval around one statistic. It works well for a mean, median, spread measure, sum, or proportion from one sample.

When should I use permutation mode?

Use permutation mode when comparing two groups. It checks whether the observed difference is unusual after group labels are shuffled many times.

How many iterations are enough?

Use at least 1,000 for quick checks. Use 5,000 or more for smoother intervals. More iterations improve stability but require more processing time.

What does the confidence interval mean?

The interval gives a plausible range for the selected statistic. It is based on the percentiles of the repeated resampled statistics.

Why does the seed matter?

The seed controls the random sequence. Using the same seed, data, and settings helps reproduce the same output again.

How is proportion calculated?

The calculator counts values greater than or equal to the threshold. It divides that success count by the total number of values.

Can resampling fix bad data?

No. Resampling measures uncertainty in the supplied data. It cannot correct biased sampling, missing context, wrong measurements, or poor study design.