[chesser]

Since they only had two groups, one of them HAD to come out ahead.

I also noted it listed MEAN score instead of MEDIAN.

This would allow for a small number of people to spike the score. For example, a group of friends, or a study group. I have tutored students from a 60 or 70 to close to 100% in the space of a single math test.

If the divvying up happened to put a few more brainiacs in one group and a few more dullards in another, this could account for the entire swing.

The specifics of dividing up the groups is very important. For example, if one entire classroom got Essay A and another got Essay B, even if there were equal numbers of bodies, it would only take one good study group to account for the discrepancy.