6. Hypothesis Testing for Paired Data

Sometimes, data are "paired." For example, we might have a "before" and "after" measurement for each person on a weight loss plan. Ignoring the paired nature of the data would result in lost information. The best way to analyze paired data is to distill each pair into a single number. We do this by finding the difference for each pair. In the weight loss example, our single data point for each person is the amount of weight lost.