Sorry for thinking on the fly.  Final comment.  This issue could be
answered once and for all in one season.  Let's say that we get beekeepers
all across the country (minimally U.S. and Canada, since we use similar
equipment) - but no reason beekeepers in other countries couldn't
participate.  Each beekeeper picks (say 30 hives at a site - just a guess,
I'd have to take a harder look at the stats)  10 controls, 10 bottom
entrance/ excluders, 10 top entrance/ excluders, intermixed - colonies
randomized by population size, etc.

If each beekeepers did one yard, and if we had 100 beekeepers all over the
country doing this, we'd have 100 different sites/regions.

That would be 3000 hives, 1000 for each treatment (control, bottom, top).

Everyone would weight colonies at beginning and end of trials, make similar
observations.  The dataset would speak for itself.  The stats folks would
be ecstatic.  1 production season and we should be able to get a real
handle on this.  Any takers?

Cheers