Studies, including RCTs, are often affected by selection biases introduced (intentionally or unintentionally) by how participants are selected and/or self-selec

Studies, including RCTs, are often affected by selection biases introduced (intentionally or unintentionally) by how participants are selected and/or self-select. Even the gold standard RCTs are thus not guaranteed to produce valid (or rather, useful/non-misleading) results. They thoroughly researched the effectiveness of parachutes on a pool of candidates at various speeds and altitudes. While the initial pool of candidates covered a variety of situations, for some reason, the pool of participants willing to actually participate in the study was heavily biased towards people in an aircraft standing still on the airport, with about 60 cm from the door to the ground. The bias may have been further increased by the selection criteria where participants deemed mentally unfit by the investigator were excluded, and willingness to jump from an airplane at altitude with a 50% chance of an empty backpack increased the likelihood of exclusion. The bottom half of the right column on page 4 explains how this is relevant for real world studies.