I'm not sure you can compare the groups in that way without distorting the answer to the question being asked. In a previous study, a side-effect was discovere

I'm not sure you can compare the groups in that way without distorting the answer to the question being asked. In a previous study, a side-effect was discovered in which the gender of the researchers handling the rodents had some impact on their tumor rates. They have a more acute sense of smell and male hormones induced a higher level of stress which caused a higher rate of tumors. So, let's suppose that approximately 1 member of each group will develop a tumor for some reason unrelated to the effect being studied. If you have 5 groups, one control and 4 experiments, then potentially you'll see a 4x effect increase if you lump the experiment groups together. This would only hold of course if the effect isn't directly proportional to the population size -- if it's proportional to the number of groups instead. So, I dunno, maybe. But I'd be reluctant to draw any conclusions from numbers for "all experimental groups" vs "one control group". And while the percentages and ratios involved can look conclusive (varying from 0% to 2.2% -- that's an infinite percentage increase!), the effects in absolute numbers are still quite small.