It might have to do with response rates. If you feel response rates to your experiment are going to be low for a certain group, you make the group bigger so yo

It might have to do with response rates. If you feel response rates to your experiment are going to be low for a certain group, you make the group bigger so you can have a more robust sample in which to characterize respondents of that group.