Sampling Bias

In the field of Statistics, sampling bias occurs when the collection of data is performed in a manner that favors certain outcomes or members of a population over others. This systematic error results in a non-random sample, which undermines the External-Validity of the findings and leads to incorrect Statistical-Inference.

One common form of this phenomenon is Selection-Bias, where the researcher’s method of choosing participants inherently excludes specific groups. For instance, Self-Selection-Bias happens when individuals volunteer for a study, potentially representing only those with strong opinions or high levels of motivation. Another notable subtype is Survivorship-Bias, which focuses only on the survivors of a process while ignoring those that failed or dropped out, a concept famously analyzed by Abraham-Wald during World War II.

A classic historical example of sampling bias is the Literary-Digest-Poll-of-1936, which incorrectly predicted that Alf Landon would defeat Franklin-D-Roosevelt. The bias arose because the poll relied on telephone directories and car registrations, which skewed the sample toward wealthier individuals during the Great-Depression. To mitigate these issues, researchers often employ Probability-Sampling techniques such as Stratified-Sampling or Simple-Random-Sampling.

For more information on the methodology of avoiding bias, see resources from the Pew Research Center and the Harvard Business Review.

• Selection-Bias
• Survivorship-Bias
• Statistical-Inference
• Probability-Sampling