Abstract
This paper addresses the “inefficiency loophole” in the Bell theorem. We examine factorizable stochastic models for the Bell inequalities, where we allow the detection efficiency to depend both on the “hidden” state of the measured system and also its passage through an analyzer. We show that, nevertheless, if the efficiency functions are symmetric between the two wings of the experiment, one can dispense with supplementary assumptions and derive new inequalities that enable the models to be tested even for highly inefficient experiments