Interactive guide
One population. Eight ways to decide who gets asked.
Every sampling method answers the same question differently: does every member of the population have a known chance of being picked? Below, the exact same 32 university students get sampled eight different ways. Click any method to watch who actually gets selected, and how badly (or how well) that sample reflects reality.
The one question that splits everything in two
Probability sampling means every person's chance of selection is known ahead of time, not necessarily equal, just calculable. Non-probability sampling means it isn't: selection depends on convenience, judgment, referral chains, or fixed quotas, none of which have a formal probability attached at all.
That single distinction is why probability methods can support real statistical inference (margins of error, confidence intervals) and non-probability methods generally can't. It doesn't mean non-probability methods are useless; it means they're answering a different, more limited question.
32 students, three class years: 50% Freshman, 31% Sophomore, 19% Senior, split across 8 dorm buildings of 4 students each. Click any method below to sample them.
Reading each result
-
1
Who lit up
Highlighted dots were selected; dimmed dots weren't. Same 32 students, every time.
-
2
Known chance of selection?
The formal test for probability vs. non-probability, stated explicitly for every method.
-
3
Population vs. sample
The true class-year mix, next to what this method's sample actually contains.
Click any leaf to sample the population that way.
Probability Sampling
Every member has a known chance of selection
Non-Probability Sampling
Not every member has a known or equal chance
—
Freshman Sophomore Senior selected
- Sample size
- —
- Known chance of selection?
- —
Population vs. this sample
| Class year | Population | This sample |
|---|
The jargons
Every method below answers the same trade-off differently — how much control over who gets picked, versus how much it costs to actually reach them — from a fully randomized draw down to whoever happens to be easiest to find.
Every individual has an equal, known chance of selection: names drawn from a hat, with no structure at all.
Pick a random start, then take every k-th person from an ordered list. Equal chance, but not independent draws.
Split into groups first, then randomly sample within each: guarantees every group is represented in the right proportion.
Split into naturally-occurring groups, randomly pick a few whole groups, survey everyone inside them. Cheaper, but riskier if a chosen group isn't typical.
Survey whoever is easiest to reach. Fast and cheap, with no guarantee it represents anyone but itself.
The researcher deliberately selects individuals believed to be especially useful or informative: subjective by design.
Existing participants refer new ones. Great for reaching hidden or hard-to-find populations, but tends to cluster around the seed's own social circle.
Fix a target headcount per group, like stratified sampling, but fill each quota with whoever's easiest to find, not a random draw.