An experiment on the extension of the political rights of foreigners in the Swiss city of Geneva used three different procedural ways to structure deliberation: participants take positions at the outset, do not take positions, and reflect first. Most opinion change occurred when participants did not have to take a position at the outset. However, no learning effects were recorded, the deliberative quality was poor and group influence had the greatest impact. When participants had to take a position at the outset, opinion change and group influence were least, but there was significant learning, and the deliberative quality was better. These results indicate a potential trade-off between opinion change – which many scholars equate with deliberative success – and good procedural deliberative quality.