Monte Carlo Flashcards
Explain the role of random sampling in a Monte Carlo simulation, and how it contributes to the accuracy of the results.
We need to draw different and unique random samples many times for a Monte Carlo simulation to perform well. In general, we want each replication in the study to have an independent random sample
from one replication to the next. To make sure that we have unique and reproducible samples, we should set a random number seed before the replication loop
Describe the concept of a Type I error and its relationship to the test size in a Monte Carlo simulation.
A type I error is the probability of rejecting a true null hypothesis, also known as false positive. We can run a Monte Carlo to calculate the Type I error for a procedure. If the procedure is working properly
the estimated probability of a Type I error should be equal to the level of a test, α.
Describe the relationship between the test size and the power of a Monte Carlo simulation, and explain why it is important to consider both of these factors when designing a simulation study.
Test size is the rate of false positives, test power is the rate of false negatives. We want to consider both in a simulation study as we want a procedure that both rejects a null which is false, and fails to
reject a null that is true.
