The idea for the first two graphs comes from .
In both graphs, the "count zeros" checkbox is to indicate the fact there is no way to count the occurrences when the photon does not pass through either filter, so no detector is triggered. If this happens, the Demonstration shows what would have happened if the zeros were counted. Also, increasing the number of repetitions will smooth the curve. However, the time goes up linearly with the number of repetitions (10,000 repetitions will take 100 times longer than 100 repetitions).
The first graph is the same as the graph in [1, Figure 3] except that it is simulating the polarization of entangled photons, which are always 90 degrees out of phase.
The second graph is produced by the Clauser–Horne–Shimony–Holt (CHSH) inequality [1, equations 20 and 21]. The CHSH inequality is a generalized Bell's inequality. The graph produced is a variant of the graph in [1, Figure 5], which plots the equation from 0 to 180 degrees. Again, the graph is phase shifted by 90 degrees.
The results of experiment 2 with the "count zeros" checkbox off shows two areas of below
that are violating the inequality. Thus, the photons are entangled, which directly contradicts the fact that the photons are not entangled in the Demonstration.
The third graph is produced by the CH–Eberhard (CH-E) inequality [2, equation 1]. According to [2, pages 2 and 4], this equation is used to close the fair-sampling loophole since the derivation does not make use of the fair-sampling assumption. The angle settings used were computed from line 2 of [3, Table 2].
The results of experiment 3 show areas in the red that are violating the inequalities. Thus, the photons are entangled, which directly contradicts the fact that the photons are not entangled in the demonstration.
 M. Giustina et al. "Significant-loophole-free test of Bell's theorem with entangled photons." arxiv.org/abs/1511.03190