Potential Problems with Sampling

Bias is a systematic error in a sampling method that makes the statistic consistently overshoot or undershoot the parameter. Undercoverage leaves part of the population out of the frame, as a landline survey leaves out the cell-only young. Voluntary response lets the sample assemble itself from volunteers with strong feelings. Nonresponse loses selected individuals who differ from those who answer. Response bias bends the answers themselves, through loaded wording or self-report. A complete bias claim names the mechanism, the direction of the error, and the reason for that direction.

The central trap is repair by enlargement: defending a tilted method by its count, as if 10,000 self-selected responses could outvote their own selection. Sample size reduces scatter around the method's target; it never moves the target. The other traps are name confusions: calling a convenience sample random because it felt unpredictable, filing voluntary response under nonresponse, or crediting a big response rate as proof of no bias while the answers themselves were bent by the question.