To study a whole population, researchers almost never reach everyone — they study a sample and generalize. That only works if the sample is representative. This town of 100 lets you cheat: survey everyone (a census) to learn the true answer, then try to hit it by sampling. Pick a method and a size, then draw — and watch which methods land near the truth and which miss on purpose.
The town — 100 residents
Supports the measureOpposes itIn your sample
True support — the whole town (census)—
Your sample estimate —
Choose a method and size, then press Draw new sample.
Samplingpick a method and size, then draw
1 · How will you recruit residents?
2 · How many will you sample?
Bias vs. sampling error. A probability sample (random, stratified) may miss the truth by a little from chance alone — that gap is sampling error, and it shrinks as the sample grows. A non-probability sample (convenience, voluntary) misses in a consistent direction — that is bias, and more data only makes you confidently wrong. Representative samples are what let sociologists generalize from a few to the many — the empirical ambition Auguste Comte named positivism.