Sampling bias is a type of selection bias. It occurs when the method used to sample the population means that some members of the intended population are more likely to be selected than others. A common example in emergency medicine occurs when patients are only enrolled in a trial during day time hours when a research assistant is available (a convenience sample), which results in a non-representational sample if patients who arrive in the middle of the night are significantly different from those who arrive during day time hours.
There are a large number of different ways that sampling bias can occur. Allowing study participants to self-select can result in response bias. Sampling bias can be built into the study design when inclusion or exclusion criteria result in a study population that is significantly different than the population of interest. It can also result from pre-screening of patients, or using a washout period to exclude patients with adverse reaction to a medication before the trial starts. Only studying patients in a hospital setting can be a type of sampling bias if the results need to be applied to patients in the community (referral bias). Survivorship bias is another type of sampling bias in which only “surviving” individuals are selected, ignoring those who fell out of view.
This post is part of a series of posts on bias in medical research. You can find the whole bias catalogue here.
Justin Morgenstern. Sampling bias, First10EM, 2018. Available at: