systematic effects

Terms from Statistics for HCI: Making Sense of Quantitative Data

Aspects of data collection or analysis that are in some way connected to the thing that you are trying to measure, and hence interfere with analysis or lead to misleading effects or bias. There are many random or uncontrolled effects that influence the data you obtain, but so long as these are unrelated to the thing you are trying to measure (independent), they can be treated as random noise. Though potentially adding variability, they will eventually average out given a large enough sample. In contrast, systematic effects may cause bias.
For example, suppose you are using a laser to measure the heights of people coming and going from two different tourist attractions in London. As they will be wearing shoes and potentially hats, this will not be a true measure of height, but for the purposes of comparison between, say, Buckingham Palace and The Houses of Parliament, it is probably fine as there is no reason to suppose visitors at either wear substantially different attire. If, however, this method were used to compare the tourists at Buckingham Palace with the guards, since the guards would be likely to be wearing tall Bearskin hats, the size of hat would be a systematic effect, leading to bias and overestimating the height of guards compared with tourists.
Note that due to the nature of randomness apparently systematic effects sometimes happen by chance.

Used on pages 22, 29, 90