Observer Drift: A Drifting Definition

Abstract

Multiple meanings for a single descrip-tive term can contribute to conceptual confusion and a general inability to dis-tinguish reliably among different phe-nomena. A particular case in point is "observer drift." The current use ofthree definitions for this term may lead to var-ious problems ranging from simple, in-consistent interpretations of the phe-nomenon, and the evolution of new and perhaps unnecessary terms, to more complex problems such as the ineffective training of observers, improper applica-tion of countermeasures to attenuate observer drift, and failures to replicate studies concerning this and related phe-nomena. The purpose of this brief note is to describe these inconsistencies and suggest some possible alternative uses. First, the term "observer drift" has been used, by and large, to describe the implicit changes in code definitions made by observers over time (e.g., Kazdin, 1977). In some instances, the meaning of the term has quite reasonably been ex-tended to include the development of idiosyncratic versions of explicit defini-tions when observers work in pairs (e.g.,