In the classical theory of Doi and Edwards for entangled polymers the strain measure Q arises from the assumption of affine deformation of the tube segments. On the other hand, we have recently argued [G. Marrucci, F. Greco, and G. Ianniruberto, "Possible role of force balance on entanglements," Macromol. Symp. (in press)] that force balance on the nodes of the entangled network generates departures from affinity. An ad hoc "lattice" model of the network was then proposed to fulfill force balance automatically, at least for single-step deformations. The corresponding strain measure is linked to the square root of the Finger tensor, and is therefore easy to use. Predictions obtained with the new strain measure are here compared with a variety of step strain data.