A subset of a population selected for analysis when collecting data on every individual is impractical or impossible.
Sampling allows analysts to draw conclusions about a large population by studying a smaller, representative subset. The validity of conclusions depends on how the sample was selected. A random sample gives every member of the population an equal chance of being selected, reducing bias. A convenience sample (selecting whoever is easiest to reach) may not represent the population well.
Sample size matters: larger samples generally produce more reliable estimates, but there are diminishing returns. Statistical methods can calculate the sample size needed to achieve a desired level of confidence.
Example: Statistics Canada conducts the Labour Force Survey by sampling approximately 56,000 households each month rather than surveying all Canadian households. The sample is carefully designed to be representative of the Canadian population, allowing national employment statistics to be estimated with known confidence intervals.