StatReason

Random Sampling Calculator

Draw a random sample of rows from a list you provide, such as IDs or names, with a seed that makes the exact selection fully reproducible, entirely in your own browser.

What this answers

This calculator answers "how do I pick a random subset of these specific rows, in a way I can reproduce or prove later?" a common need for randomly selecting audit samples, prize winners, or survey subsets from a known list.

How it is calculated

Rows are drawn using a seeded, fully reproducible pseudorandom generator: without replacement, every row can be selected at most once, matching a classic simple random sample; with replacement, the same row can appear more than once. The seed and population size are reported alongside the selection as a receipt, so the exact same draw can be reproduced or verified later by anyone using the same inputs and seed.

Worked example

For IDs 1 through 5, choosing 2 without replacement selects exactly 2 distinct IDs from that list, with the specific pair determined entirely by the seed you supplied; running the same seed and the same input list again always reproduces the identical pair.

Assumption audit

Calculated from your data: the selected rows and a receipt recording the population size and the exact seed used, so the draw is independently verifiable.
Evidence to review: whether your input list is genuinely the full population you intend to sample from; this calculator selects randomly from exactly the rows you provide, nothing more and nothing less.
You must verify: that random selection alone satisfies whatever requirement motivated the sample; this calculator provides randomness, not representativeness guarantees beyond what simple random sampling itself provides.

What this result does not mean

A random selection is not automatically a representative sample of some larger, different population; it is representative only of the specific list of rows you actually provided as input.

Common mistakes

Confusing this simple random sample with a stratified or systematic sample is a common mix-up; if your population has known subgroups you want proportionally represented, a plain random draw like this one does not guarantee that balance on its own, especially with a small selection count. A second mistake is forgetting to record the seed alongside the result; without it, the exact selection cannot be independently reproduced or verified later.

Limitations

Duplicate rows in your input list (such as duplicate header values) are treated as ordinary, individually selectable rows, not deduplicated automatically; deduplicate your list yourself first if that matters for your use case.