Sampling Bias

Intermediate

Sampling bias occurs when the method used to choose a sample makes some members of the population more likely to be included than others, so the sample does not accurately represent the population.

If a sample is biased, conclusions drawn from it can be wrong, no matter how large the sample is.

For example, suppose a company wants to know how satisfied its customers are and surveys only those who contact customer service. These customers are more likely to have had problems, so the results would probably understate overall satisfaction. Similarly, an online survey about internet use would overrepresent people who use the internet often.

Common causes of sampling bias include convenience sampling, where researchers choose whoever is easiest to reach; voluntary response, where people choose whether to take part and those with strong views are more likely to respond; and incomplete lists of the population, where some groups are left out entirely.

In finance, sampling bias can distort research and decisions. A study of investment returns that only looks at well-known funds, or a credit model built only on approved borrowers, may give misleading results. The best way to reduce sampling bias is to use random sampling from a complete list of the population and to think carefully about who might be missing from the data.

For worked examples and full lessons, visit notesbysara.org.