A Qualitative Study of Engineers’ Perception of Variability as 'Error'

Variability is an unavoidable reality: People have different heights, built parts have different dimensions, and manufactured components have different material properties. It is common in statistics to refer to certain variations as "error;" however, the term "error" has very different meanings across disciplines. This work was motivated by a concerning observation of some statisticians: a refusal to accept other, non-statistical perceptions of the term "error." As part of a larger study of practicing engineers (n=24), we used qualitative methods to investigate their interpretation of the term "error" and their ensuing approach to analyzing data. We find that the term "error' tends to erode trust in the data (11/24 participants) and can lead to a more dangerous interpretation of variability (2/24 participants). These results have important implications for communication on interdisciplinary teams and teaching statistics to engineering students.