Matched-Pair Design

Video Transcription

This tutorial examines matched-pair design, one type of experiment. In a matched-pair design you find experimental units that are as close of pairs as possible. Now, one person goes to the treatment group. The other person goes to control group. By having really close pairs, this helps to control for the other variables that we're not looking at.

So by choosing a pair of women who are the same age and the same gender who exercise the same amount and in the same area, this can help us to control for all the factors that are associated with any of those things. And let us narrow in and look only at the variable we're trying to study.

Now, doing this helps to give us some extra control. This is also pretty similar to a case control study, which is a type of observational study. But here, researchers are giving a treatment instead of just observing. Let's look at how this would get set up.

Here's an example of how it would set up. Here are some pairs that might rise. Two 30-year-old women who are highly active, a second pair, two 27-year-old men who are moderately active, and the third pair of two 50-year-old men who are highly active.

More likely than not, researchers would be controlling and making pairs based on more than just two variables, but this helps to simplify it down a little bit. Then you take your list of pairs, your first pair all the way down to your last pair. And you randomly assign each person in it to a treatment group or a control group.

Now, there's another way that matched-pair design can be done. Let's look at that now. Another type of matched-pair design compares the participant to him or herself. This is typically done in a before or after kind of study. If you've ever seen one of those late night commercials for weight loss drugs, they show you those before and after pictures.

So here's an example of that of someone before they lost weight and someone after they lost a lot of weight. So if our study we're trying to examine the effect of a drug or an activity program on weight loss, you would compare the person before and after the study. So matched-pair design helps us to control a bit for some confounding variables by pairing people with similar people. Another possibility is to do a before and after using the same person.