I received a question from a customer about the "stability" of the simulated p-value for Anderson-Darling goodness of fit test. In other words, if you repeated the goodness of fit test multiple times, the simulated p-value will change. It is understood that this is expected given the random nature of the simulation. Here are the questions:
1. Would increasing the number of samples reduce the variability in the simulated p-value? In other words, would the simulated p-value be more stable over repeats of the goodness of fit test as N samples increases?
2. Is the N samples for Anderson-Darling 100,000?
3. Can you change the N samples for Anderson-Darling? Under Preferences > Platforms > Simulator you can change N Samples - would this apply to Anderson-Darling?
