The group sizes should depend on how spread your data is. For example, say you have 30 pieces of data in group 10-19 and only 2 in 20-29 and 1 in 30-39 you could split the 10-19 group into something like 10-12, 13-15, 16-18... This would mean that you get a better idea of the distribution and you can write about it in your interpreataion of your results! Anyway, I don't think I explained that very well. Ask me in class if you're still stuck - or ask Scot, he's bound to know :P mweh eheheheh!