Matched Pairs Design vs Random Assignment

Both terms below come up in the same part of the course, and students mix them up. Here is each one defined on its own, side by side, so you can see where they part company.

Matched pairs design

Collecting data and study design

A matched pairs design compares two treatments within pairs of similar units, or one unit measured twice, then analyzes the difference inside each pair.

A matched pairs design controls for variation between units by comparing treatments inside pairs that are alike, or on the same unit under both conditions. Because each pair serves as its own comparison, differences between individuals cancel out and the test focuses on the within-pair difference. For example, you measure each runner's time in old shoes and new shoes, then analyze the 20 time differences. The analysis is a one-sample tt procedure on the paired differences, using t=xˉdsd/nt = \frac{\bar{x}_d}{s_d / \sqrt{n}}, where xˉd\bar{x}_d is the mean difference, sds_d its standard deviation, and nn the number of pairs.

Full entry for matched pairs design

Random assignment

Collecting data and study design

Random assignment uses chance to sort experimental units into treatment groups, so the groups start out similar and confounding is balanced out.

Random assignment is how an experiment builds comparable groups. Because a coin flip or random number decides who gets which treatment, no systematic difference in age, health, or motivation piles up in one group, so a later difference in outcomes can be credited to the treatment. For example, flipping a coin for each of 40 patients to assign a drug or a placebo balances the other traits across the two groups on average. Random assignment builds the groups, while random selection picks the sample; assignment supports cause-and-effect claims, and selection supports generalizing to a population.

Full entry for random assignment

Where each one fits in the course