Hello Amy,

This is a correction for situations in which you are entering data from a cluster trial but the estimate of variance (standard error) has not taken account of the clustering. The adjustment uses the average cluster size and intracluster correlation coefficient (ICC) to calculate the 'design effect' - which essentially estimates the statistical power of the study taking account of the clustering. We then inflate the standard error that was originally calculated to incorporate this value.

You can find a bit more information (and some further references) here. https://en.wikipedia.org/wiki/Design_effect

Best regards,

Jeff