Abstract
The concept of parallel measurements. Measures that have identical true scores and linearly experimentally independent errors that have equal variances are named parallel measurements (Lord & Novick, 1968). Since the origin of the Classical Test Theory (CTT) the concept of parallel measurement has played a central role as a cornerstone of the concept and the estimation of reliability (Gulliksen, 1950; Lord & Novick, 1968). The difficulties in finding such similar comparable measurements were early known, and some other concepts, wording different levels of comparability, were proposed. The purpose of these relaxed concepts of comparability was to find a higher level of empirical availability yet capable of sustaining a theoretical base for the reliability coefficient. Although Wilks (1946) and Jöreskog (1971) provided soft methods to determine if two measures are parallel tests, the classical concept of parallel measurements remains untested and its empirical possibility is far from clear. The aim of this paper is to trace the development of the concept of parallel measurements as a base for reliability and discuss some of its limitations.
