This document discusses statistics from simulation runs and analyzing simulation results. It provides an example of estimating the mean waiting time in a single server queueing system where arrivals and services are exponentially distributed. There are two main issues discussed: 1) observations may not be independent or the distribution may not be stationary, violating assumptions for confidence levels, and 2) the sample mean is biased for estimating the true mean waiting time, especially with early samples, as the distribution is not initially stationary. As the sample size increases, the bias diminishes but may not fully converge even with thousands of samples.