For a class, we had to create a simple completely randomized experiment to investigate some topic. We chose to see if caffeine actually makes people feel more alert.

The experiment was on a class. We gave 16 subjects caff beverage, and 16 subjects decaf beverage. And the filled out a survey. The most relevant question was their self reported alertness level on a 1-7 scale before and 30 minutes after consuming sample.

Would a simulation be useful? Like premutations or bootstrapping? Which simulation would be useful? The resources I found via google are confusing to me.

Other than sampling...would I just carry about a two population means test? Using two-sample t-test?