prev next front |1 |2 |3 |4 |5 |6 |7 |8 |9 |10 |11 |12 |13 |14 |15 |16 |17 |18 |19 |20 |21 |22 |23 |24 |25 |26 |27 |28 |29 |30 |31 |32 |33 |34 |35 |36 |37 |review
Trial randomizing between 30 and 50 individuals will tend to have almost the same statistical power as trials randomizing the same number of much larger units. But clusters of larger size are often recruited for very practical reasons (to reduce contamination, to avoid logistic or ethical problems, etc.).