Random Draw Fairness: How to Check for Bias Yourself

Contents
When you're picking a place for a team dinner, splitting people into groups, or drawing a prize winner at an event, a random picker can start to feel like it keeps landing on the same person or number. Is the program favoring certain values, or is it just coincidence stacking up? You can't tell from a single draw. But if you run the draw many times and compare the results yourself, you can see it with your own eyes.
Summary ① One result tells you nothing about bias — you have to repeat the draw under the same conditions and compare the ratios ② If you don't handle the "modulo bias" that appears when you squeeze a random number into a fixed range, certain values come up slightly more often ③ The "no repeats" option only prevents duplicates within a single draw. It has nothing to do with the same person coming up again on the next draw
Why draw results seem to skew toward one side
When you split a group into teams of 5, or draw 1 prize winner out of 10 people, running it a few times will eventually produce the same person twice in a row. In that moment it's easy to suspect the program has a favorite value rather than to think about probability.
But if the conditions are identical every time you draw, each item has the same chance every time. With a list of 10 people, it's 1 in 10 each time. With a number range of 1–45, it's 1 in 45 each time, drawn independently. When the number of trials is small, some natural deviation from that probability is expected.
There are two things actually worth checking. First, does the deviation persist when you increase the number of trials enough? Second, does the underlying calculation for generating random numbers have a structural bias that makes certain values more likely?
How to repeat the draw and compare for yourself
The surest method is to run the same draw many times under the same conditions and write down the results. For example, draw 1 person at a time from a list of 10, 30 times, and count how many times each name came up. It won't come out to exactly 3 times each, but once you have around 30 draws, you can see by eye whether one name shows up 5 or more times.
The same goes for number ranges. Draw 1 number at a time from 1–45 several times and keep a frequency table on paper or in a spreadsheet. As the number of trials grows, you can check whether the counts for each number converge to a similar level.
Here's a spot that trips people up. The "no repeats" option keeps the same person or number from appearing twice within a single draw. Each new draw recalculates against the full list, so it's normal for the person picked in the last draw to come up again in the next one.
Check it with our tool. Every time you press the draw button, it recalculates with a cryptographic random number generator (crypto.getRandomValues), and the copy results button lets you move each draw's result straight into your notes — much faster than tallying by hand.
Free · No sign-up · Cryptographic randomness, drawn right in your browser

Common sticking points
| Situation | What to check |
|---|---|
| I only drew 5–10 times and the same person came up twice | That's normal deviation when the number of trials is small. You need 30 or more repetitions before the ratio means anything |
| I turned on "no repeats" but the same person came up on the next draw | No-repeat protection only applies within a single draw. Each new draw recalculates against the full list |
| I entered a draw count larger than the number of people on the list | It draws only as many as there are people, and shows a notice that the count was reduced to fit |
| I entered the min and max the wrong way around | The tool automatically sorts the smaller and larger values when it calculates the range, so the result isn't affected |
The bottom line
When a draw looks like it's leaning one way, don't judge it from a few impressions. Repeat the draw under the same conditions and count the frequencies. As the number of trials goes up, the share each item gets should move closer to its theoretical probability.
Remember two things and you won't misread your results: the "no repeats" option only prevents duplicates within a single draw, and the real thing to check is whether the random number calculation handles modulo bias.
Free · No sign-up · Cryptographic randomness, drawn right in your browser
Frequently asked questions
How is crypto.getRandomValues different from Math.random?
Math.random() is a pseudorandom generator built by the JavaScript engine for statistical distribution purposes, and it doesn't guarantee that the seed or the generation algorithm is unpredictable. crypto.getRandomValues() uses a cryptographically secure random number generator provided by the operating system, which makes it hard to predict any value coming up more often than another.
What exactly happens when I turn on the "no repeats" option?
It shuffles the entire pool at random, then takes the number of items you need from the front. With this method, a value already picked can't come up again within the same draw, and the whole pool is guaranteed to be shuffled evenly.
Are the names or number ranges I enter saved on a server?
No. Everything from entering the list to drawing numbers to showing the result runs inside your browser and is never sent to a server. If you refresh the page, your input and the draw history disappear with it.
This article was adapted from our Korean original. Read the Korean original