Random effects that are irrelevant for whatever you are trying to measure or study, but add noise, making the results harder to interpret. For example, if you are measuring someone's weight, they may sway slightly, meaning the scales register slightly more or less than their actual weight; or if you ask someone whether they like a new interface feature their response may depend on how good their day has been. Part of the job of statistics is to see back, past this noise, to the underlying facts about the real world.
Used on pages 19, 118