ABSTRACT
The evolution of reliability has centered on its evaluation and framing of "measurement error," as its operational definition over the past century has remained focused on notions of consistency of scores across replications of a measurement procedure. A key element of the operational definition of reliability is the notion of expectation. Although generalizability-theory is often espoused as a conceptual centerpiece of modern psychometrics, it is important to separate the conceptual perspective G-theory offers on reliability from the estimation methods it proscribes. Historically, many treatments of reliability, whether explicitly or implicitly, have equated inconsistency in scores across replicates with "unpredictable" error. While classical test theory (CTT) certainly has its place in treatments of reliability, framing entire treatments around it serves to "trap" us within the CTT paradigm. The measurement models underlying CTT and G-theory actually share some important similarities.
