Suppose we are studying the effect of diet on height of children, and we have two diets to compare: diet A (a well balanced diet with lots of broccoli) and diet B (a diet rich in potato chips and candy bars).

We wish to find the diet that helps children grow (in height) fastest. We have decided to use 20 children in the experiment, and we are contemplating the following methods for matching children with diets:

1. Let them choose.

2. Take the first 10 for A, the second 10 for B.

3. Alternate A, B, A, B

4. Toss a coin for each child in the study: heads→ A, tails → B.

5. Get 20 children; choose 10 at random for A, the rest for B.

Which ones of these five methods are the most appropriate, and which ones are very wrong?