In the following video:
DrPhysicsA describes the results of an experiment involving three polarized mirrors randomly selected by the perennial Alice and Bob.
The first three columns represent possible pre-determined polarizations (hidden variables) of entangled photons sent to Alice's and Bob's randomly selected mirrors, A, B, and C. The second three columns represent the possible results of observations by Alice & Bob for three different combinations of mirrors. The results tabulated show that considering only the middle six rows there will be six occurrences when Bob and Alice measure the same result, either no photon is detected by both, or they both detect a photon. According to the presenter this means that they will make the same 33% or more of the time if there are hidden variables, while Quantum Mechanical analysis predicts, and experimental results show 25%. Thus the possibility of local hidden variables is eliminated.
My question is this: On what basis are the top and bottom row of the hypothetical data eliminated from the calculation? Considering all possible results there would be 12 "Same" observations out of 24 possible results, or 50% instead 33%. Also, when running the experiment, it would not be possible to exclude a top or bottom row from the results. Any "Same" looks like any other. So why are only the six middle rows counted?