Let's say I have a new drug that I think helps diabetics. So I perform clinical trials in accordance with the FDA requirements. I give one randomly selected groups of diabetics the drug, and another sample gets a placebo pill. The diabetics who get the pill improve, the controls do not. I perform a t test on the data. What is the independent variable? The dependent variable? The null hypothesis?

The above is a cross sectional research design. Here is another strategy. I have only one group of diabetics. I evaluate them and give them the pill. I bring them in next week and evaluate them again, give them another pill. The following week I do the same, and on and on for several months. This is a longitudinal design, or a "repeated measures" design. I would have to analyze this with a "dependent" t test, or since the groups are measured over and over again, a repeated measures ANOVA. In these designs, the subjects are used as their own controls.

Which strategy or research design do you think is best?