Title: A Robust Mathematical Model for Clauser-Horne Experiments, With Implications for Rigorous Statistical Analysis

Abstract: Recent experiments have reached detection efficiencies sufficient to close the detection loophole with photons, testing the Clauser-Horne (CH) version of Bell's inequality. However, the statistical analysis of the data has not been rigorous. For instance, the possibility of a local state varying over time, with possible dependence on earlier trials (the "memory loophole"), has not been accounted for, and such factors can allow a local state to have a greater chance of exhibiting a Bell violation due to chance fluctuation. In this paper, a mathematical model for a CH experiment is presented that allows for a new classification of CH inequalities that is inherently robust to the memory loophole. A correct statistical analysis of the experimental data is given, finding the results of both experiments to be statistically significant. A new method is introduced to calculate exact p-values for certain martingale-based statistics; previously, only un-sharp upper bounds had been available for these types of statistics. The effects of changing the setting probabilities are also explored.