2.B Populations, Samples, and Sampling Bias

A population is everyone a researcher wants to draw conclusions about: every adult in the US, every student at a school. Since studying all of them is almost never possible, researchers study a sample, a subset of the population, and use what they find to make claims about the whole group. The trick is making sure the sample actually looks like the population in the ways that matter: age, income, opinions, whatever the study cares about. Random sampling does this by giving every member of the population an equal, known chance of being chosen, the same logic as drawing names from a hat. Because chance alone decides who gets in, the quirks of any one person tend to cancel out, and the sample ends up looking statistically like a smaller copy of the population.

Convenience sampling breaks that logic. Instead of giving everyone an equal shot, the researcher takes whoever is easiest to reach: students in one class, shoppers at one mall, callers who happen to pick up the phone. Each of those groups shares traits that the rest of the population does not, so the sample quietly tilts toward those traits. The bias is not something the researcher chose; it comes from who was available, which is exactly why it is easy to miss.

This connects directly to generalizability: a study's conclusions only extend to the population a sample actually represents. A huge convenience sample is still biased, just confidently so, while a smaller random sample can generalize further because every voice had an equal shot at being heard. This sets up later ideas about how bias shows up again in self-selection and nonresponse, where the people who drop out of a study are rarely a random slice of who started it.