the one thing almost everyone knows about studies like this is that a bigger sample size means the results are probably more significant. But if they’re not independent data points, then you have to treat it, in some respects, like a smaller sample, so the results become less significant. As statisticians would say, you must ‘correct for clustering’.
