Satterthwaite contributed to CIs for two-sample t tests, but he also contributed to CIs for a variance. So the first step is to define your problem. For example,
- What test are you running?
- What SAS procedure would you use to compute the statistics on one set of data? Figure out the syntax.
- The CI is for what statistic?
- The data are simulated from what distribution?
If your problem is a two-sample t test, you will also need to define the difference between the means of the two samples.
For an introduction to Monte Carlo simulation in SAS, see "Using simulation to estimate the power of a statistical test," which shows how to simulate data for a two-sample t test. The article contains links to other articles that discuss how to perform simulation studies efficiently. They are worth reading if you are learning about how to run Monte Carlo simulations in SAS.
