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

Analysis is only possible under the untenable assumption that there is no clustering of individuals responses within communities.

More attention to the effects of clustering when determining sample size might help to eliminate designs which lack replication.