It is widely accepted that for reversing double-step strain deformations, predictions base on the Doi-Edwards (DE) molecular theory without the independent alignment approximation (IAA) are superior to predictions obtained with the IAA, or equivalently, the Kaye-Bernstein-Kearsley-Zapas (K-BKZ) theory. This summation, however, is based on data obtained over limited ranges of strain and time : the time both between the step strains (t1) and following the second step strain (t - t1). In this study, a thorough evaluation of the DE theory ia carried out using a comprehensive double-step strain flow data set. The results of this study indicate that the DE theory is an improvement over the K-BKZ theory in flows with strain reversal but only for cases when the criteria t1, t - t1 much greater than tau(k) is satisfied. The constant tau(k) defined as the time beyond which the stress relaxation modulus is factorable : G (gamma t) = h (gamma) G (t), is believed to represent the end of the chain retraction process in the DE theory. It appears that the dynamics of chain retraction have an important influence on double-step strain behavior and, therefore, should be accounted for in molecular-based theories devised to have general validity in this important deformation history.