The notions of three-particle entanglement and three-particle nonlocality are discussed in the light of Svetlichny's inequality Phys. Rev. D 35, 3066 (1987). It is shown that there exist sets of measurements, which can be used to prove three-particle entanglement, but which are nevertheless useless at proving three-particle nonlocality. In particular, it is shown that the quantum predictions giving a maximal violation of Mermin's three-particle Bell inequality Phys. Rev. Lett. 65, 1838 (1990) can be reproduced by a hybrid hidden variables model in which nonlocal correlations are present only between two of the particles. It should be possible, however, to test the existence of both three-particle entanglement and three-particle nonlocality for any given quantum state via Svetlichny's inequality.
José L. Cereceda (Mon,) studied this question.