Bias
Short definition
Bias is a systematic error in the design, conduct, analysis or publication of a study that pushes its results away from the true effect in one direction.
In plain language
Unlike random error, bias does not shrink with more participants. There are several types: selection bias (groups that differ from the outset), detection bias (assessors who know who received what), attrition bias (unequal drop-outs) or publication bias (negative studies that go unpublished). Systematic reviews assess each trial's risk of bias, for example with Cochrane's RoB 2 tool.